cmi dementia guidelines - kaiser...

© 2004 Kaiser Permanente Medical Care Program CMI Dementia GuidelinesFor use within Kaiser Permanente only. 01/04

CMI Dementia Guidelines

Published January, 2002Reviewed/Revised January, 2004

Care Management InstituteKaiser Permanente

Sponsored by theInterregional Guidelines Steering Group Quality Review

Subcommittee December, 2003

Revision Due: December, 2005

These guidelines are informational only. They are not intended or designed as asubstitute for the reasonable exercise of independent clinical judgment by practitioners,considering each patient’s needs on an individual basis.Guideline recommendations apply to populations of patients. Clinical judgment isnecessary to design treatment plans for individual patients.

Note: The guidelines focus on the diagnosis, treatment, and management of dementiain the outpatient, primary care setting. Inpatient, Emergency Department, and nursinghome settings are not addressed.

nadspaka

Text Box

Off-Cycle Update: May, 2004

© 2004 Kaiser Permanente Medical Care Program CMI Dementia GuidelinesFor use within Kaiser Permanente only. 01/04 i

Table of ContentsIntroduction................................................................................................................................... 1

I. What is the Care Management Institute? .................................................................. 1

II. What is the Interregional Guidelines Steering Group? ............................................ 1

III. What does it mean for a guideline to be IRGSG sponsored?................................... 1

IV. Purpose of CMI Dementia Guidelines........................................................................ 1

V. Overview........................................................................................................................ 2A. Guideline Development .......................................................................................... 2B. Methodology........................................................................................................... 2C. How is this manual organized? ............................................................................... 3

VI. How can we make this resource better? ..................................................................... 3

Acknowledgments ......................................................................................................................... 4

Guidelines Summary .................................................................................................................... 5

Supporting Documentation ........................................................................................................ 13

Prevention of Dementia .............................................................................................................. 14Problem Formulation ................................................................................................................ 14Evidence Search........................................................................................................................ 15Recommendations and Rationale Statement............................................................................. 19Evidence Tables ........................................................................................................................ 24

Screening/Casefinding ................................................................................................................ 25

Who to Screen:Problem Formulation ................................................................................................................ 25Evidence Search........................................................................................................................ 26Recommendations and Rationale Statements ........................................................................... 28How to Screen:Problem Formulation ................................................................................................................ 30Evidence Search........................................................................................................................ 31Recommendations and Rationale Statements ........................................................................... 33

Diagnosis – History ..................................................................................................................... 38Problem Formulation - History................................................................................................. 38Evidence Search - History......................................................................................................... 39Recommendations and Rationale Statement - History ............................................................. 41Problem Formulation - Functional Status Assessment ............................................................. 42Evidence Search - Functional Status Assessment..................................................................... 43Recommendations and Rationale Statement - Functional Status Assessment.......................... 46

CMI Dementia Guidelines © 2004 Kaiser Permanente Medical Care Programii For use within Kaiser Permanente only. 01/04

Diagnosis – Physical Exam......................................................................................................... 47Problem Formulation ................................................................................................................ 47Evidence Search........................................................................................................................ 48Recommendations and Rationale Statement............................................................................. 49

Diagnosis – Cognitive Status Assessment ................................................................................. 50Problem Formulation ................................................................................................................ 50Evidence Search........................................................................................................................ 51Recommendations and Rationale Statements ........................................................................... 54

Diagnosis – Diagnostic Tests ...................................................................................................... 57Problem Formulation ................................................................................................................ 57Evidence Search........................................................................................................................ 58Recommendations and Rationale Statement............................................................................. 60

Diagnosis – Neuroimaging.......................................................................................................... 63Problem Formulation ................................................................................................................ 63Evidence Search........................................................................................................................ 64Recommendations and Rationale Statement............................................................................. 67

Pharmacological Treatment of Dementia................................................................................. 70Problem Formulation ................................................................................................................ 70Evidence Search........................................................................................................................ 71Recommendations and Rationale Statement............................................................................. 75Evidence Tables ........................................................................................................................ 87

Pharmacologic Management – Memantine ............................................................................ 103Problem Formulation .............................................................................................................. 103Evidence Search...................................................................................................................... 104Recommendations and Rationale Statement........................................................................... 105Evidence Tables ...................................................................................................................... 108

Pharmacological Management of Behaviors .......................................................................... 112Problem Formulation .............................................................................................................. 112Evidence Search...................................................................................................................... 114Recommendations and Rationale Statement........................................................................... 118Evidence Tables ...................................................................................................................... 129

Non-Pharmacological Management of Behaviors.................................................................. 137Problem Formulation .............................................................................................................. 137Evidence Search...................................................................................................................... 138Recommendations and Rationale Statement........................................................................... 140Evidence Tables ...................................................................................................................... 142

© 2004 Kaiser Permanente Medical Care Program CMI Dementia GuidelinesFor use within Kaiser Permanente only. 01/04 iii

Specialty Referrals .................................................................................................................... 145Problem Formulation .............................................................................................................. 145Evidence Search...................................................................................................................... 146Recommendations and Rationale Statement........................................................................... 147

Special Considerations - Driving ............................................................................................. 148Problem Formulation .............................................................................................................. 148Evidence Search...................................................................................................................... 149Recommendations and Rationale Statement........................................................................... 150

Special Considerations - Caregivers........................................................................................ 153Problem Formulation .............................................................................................................. 153Evidence Search...................................................................................................................... 154Recommendations and Rationale Statement........................................................................... 155Evidence Tables ...................................................................................................................... 158

References.................................................................................................................................. 163

© 2004 Kaiser Permanente Medical Care Program CMI Dementia GuidelinesFor use within Kaiser Permanente only. 01/04 1

Introduction

I. What is the Care Management Institute?

The Kaiser Permanente Care Management Institute (CMI) is a national entity that synthesizesknowledge on the best clinical approaches both from within and outside Kaiser Permanente anddevelops integrated, evidence-based guidelines and care management programs that can betailored to local settings and for individual members. CMI facilitates implementation of theseguidelines and programs by working with health care professionals at the local level. CMI isinvolved in the creation of new care management programs for priority populations, as well as inthe identification and synthesis of existing successful clinical practices. In the development of itsprograms, CMI utilizes an evidence-based approach and relies on population managementprinciples and processes.

II. What is the Interregional Guidelines Steering Group?

The IRGSG is a group of practitioners and analytical managers from Kaiser Permanente regionswho facilitate the development, review, and exchange of evidence-based clinical practiceguidelines.

III. What does it mean for a guideline to be IRGSG sponsored?

A guideline that has been reviewed and approved by the IRGSG Quality Review Subcommittee(QRS) is said to be "IRGSG sponsored." This indicates that the IRGSG Common Methodologyhas been followed so that any KP Region could adapt or adopt that guideline knowing that it hadbeen based on the published evidence. In addition, the development process is fully documentedso that another Region could replicate the entire process, from problem formulation through toindividual recommendations of the guideline. IRGSG sponsorship is granted at the level of therecommendation, rather than the entire guideline. IRGSG sponsorship does not imply KaiserPermanente program-wide adoption of the guideline. KP Regions are free to use all, part, or noneof a sponsored guideline.

IV. Purpose of CMI Dementia Guidelines

The CMI Dementia Guidelines are an informational resource for primary care providers to helpthem better understand the evidence on diagnosing, treating, and managing people withdementia, and to better support the caregivers of these patients. This Dementia Guidelinesmanual is a technical review of the evidence and the methodology used to develop therecommendations, and is not intended for day-to-day use by clinicians due to its format anddetail. Additional tools that summarize and support the information in these guidelines will bedistributed to a much wider audience of Elder Care leaders and front-line clinicians.

CMI Dementia Guidelines © 2004 Kaiser Permanente Medical Care Program2 For use within Kaiser Permanente only. 01/04

V. Overview

A. Guideline Development

The medical community recognizes that clinical practice guidelines (CPGs) based on scientificevidence are an essential tool for improving and demonstrating the quality of care in the presentenvironment of burgeoning technology and resource limitations. Increasingly, clinicians mustconsider not only issues of quality but also the resource implications of their decisions. Thisinvolves addressing health problems in a way that maximizes the health of the populationgiven the available resources. As part of its work in synthesizing and disseminating knowledgerelated to priority populations, the Care Management Insitute (CMI) supports development ofexplicit, scientifically based recommendations for clinical practice to assist physicians,administrators, and other health care professionals from Kaiser Permanente in determining themost effective medical practices.

B. Methodology

To develop guidelines, CMI consultants work with a multidisciplinary team of physicians andother health care professionals. This Guideline Workgroup consists of a core, multidisciplinarygroup of physicians representing the medical specialties most affected by the guideline topic andother content experts from disciplines such as pharmacy, nursing, and social work as appropriate.During the guideline development process, the workgroup reviews evidence published in peer-reviewed scientific journals, existing evidence-based guidelines and consensus statements fromexternal professional societies and government health organizations, and clinical expert opinionof KP specialty groups. The workgroup also provides input into the content and format of theguideline. To keep current with changing medical practices, all guidelines are reviewed and, ifappropriate, revised at least every two years. Some guidelines are updated more frequently inresponse to the publication of important new evidence.

To review and update the CMI Dementia Guidelines released in January, 2002, amultidisciplinary, interregional CMI Dementia Guidelines Workgroup – including participantsfrom primary care, geriatrics/continuing care, neurology, pharmacy, nursing, and social work aswell as evidence-based medicine experts – was convened. This group met in April, 2003 todefine the scope of the revision. The Guidelines Project Management Team then performedsystematic reviews of the medical literature on each of the clinical questions identified by theworkgroup, sending draft versions for review by the Guidelines Workgroup. All of therecommendations and supporting evidence were reviewed by the Guidelines Workgroup in depththrough a series of conference calls in September, 2003, after which the guidelines weresubmitted to the KP Interregional Guidelines Steering Group Quality Review Subcommittee.That group reviewed and sponsored the guidelines on December 17, 2003.

Recommendations are classified as either “evidence-based” or “consensus.” Evidence-basedrecommendations are based on the workgroup’s review of relevant well-designed studies.Consensus recommendations were created where evidence was inadequate and the workgroupfelt that a recommendation was needed to guide clinicians.


C. How is this manual organized?

Section I Introduction – Provides a general introduction to the Dementia Guidelines.

Section II Guidelines Summary – Provides recommendations (evidence-based andconsensus-based) on the prevention, diagnosis, treatment, and management ofpatients with dementia.

Section III Supporting Documentation – Provides complete documentation, includingproblem formulations, literature search strategies, evidence tables, and rationalestatements for each of the recommendation areas presented in Section II.

Section IV References – Provides citations for all sources referred to in the guidelines

NOTE: This manual does not contain the old evidence tables from the resources first reviewedin the 2002 CMI Dementia Guidelines, but all tables are available on the Permanente KnowledgeConnection (PKC) at http://pkc.kp.org

VI. How can we make this resource better?

We are very interested in understanding the ways in which this resource is useful to you, andways in which you feel it could be improved. Your input is critical. First, let us knowimmediately of any suggestions or needed clarifications for this manual. In addition, after youutilize the existing approaches, please share your insights, experiences or materials that you feelshould be incorporated into the CMI Dementia Guidelines. The contents of this binder will beavailable on KP’s national clinical Web site, the Permanente Knowledge Connection:http://pkc.kp.org. Our collective effort will keep this resource useful and current for our healthcare clinicians, members, and their families.

Please direct your feedback to: Dementia GuidelinesCare Management InstituteOne Kaiser Plaza, 16th floorOakland, CA 94612

CMI Product Information Line: 510-271-6426Email address: [email protected]

http://pkc.kp.org/


AcknowledgmentsThe CMI Dementia Guidelines are the result of the extensive clinical expertise, collaborativeefforts, and outstanding personal contributions of the following participants:

CMI Dementia Guidelines Project Management Team

Richard Della Penna, MD Director, KP Aging Network Permanente FederationKate Heumann Meyers, MPP Elder Care Practice Leader Care Management InstituteJill Bowman Evidence-Based Medicine Methodologist Care Management Institute

CMI Dementia Guidelines Workgroup

Colorado Glenn Gade, MDRob Schabbing, MD

Georgia Carole Gardner, MDHawaii Steven Orimoto, MD

Warren Wong, MDMid-AtlanticStates

Charlotte Dean, MDMarlene Hayman, MDDavid Lipps, MD

Northwest Arthur Hayward, MDMary White, LCSW

NorthernCalifornia

Tim Lockyer, MDRik Smith, MDMarilyn Williams, MS, RNJeff Klingman, MD (ad hoc)

Ohio Mary Ann Dzurec, PharmDKathleen Grieser, MDWilliam Schwab, MD, PhD

SouthernCalifornia

Yuri Bronstein, MDDebbie Kubota, PharmDJames Mayo, MDRalph Yep, MD


Guidelines Summary

The Guidelines Summary is organized as follows:

I. PreventionII. Screening

A. Who to Screen B. How to Screen

III. DiagnosisA. HistoryB. Physical ExamC. Cognitive Status AssessmentD. Diagnostic TestsE. Neuroimaging

IV. Pharmacological Treatment of Dementia A. Acetylcholinesterase InhibitorsB. MemantineC. Re-evaluation of Drug TherapyD. Other Agents

V. Pharmacological Management of BehaviorsVI. Non-Pharmacological Management of BehaviorsVII. Specialty ReferralsVIII. Special Considerations

A. DrivingB. Caregivers

These guidelines are informational only. They are not intended or designed as a substitutefor the reasonable exercise of independent clinical judgment by practitioners, consideringeach patient’s needs on an individual basis.Guideline recommendations apply to populations of patients. Clinical judgment isnecessary to design treatment plans for individual patients.

Note: The guidelines focus on the diagnosis, treatment, and management of dementia in theoutpatient, primary care setting. Inpatient, Emergency Department, and nursing homesettings are not addressed.


I. Prevention

Estrogen plus progestin should not be used to prevent dementia, as their combined use hasdemonstrated increased risk of dementia.(Evidence-Based, IRGSG-sponsored)

There is insufficient evidence to recommend the use of any of the following agents to preventdementia: estrogen (alone), NSAIDs, Cox-2 inhibitors, ginkgo biloba, vitamin E, statins. (Evidence-Based, IRGSG-sponsored)

II. Screening

A. Who to Screen

There is insufficient evidence to recommend for or against routine screening for dementia inolder adults.(Evidence-Based, IRGSG-sponsored)

Targeted screening for dementia is recommended in:� Patients age 65 and over when clinical presentation suggests possibility of dementia, e.g.,

inability to provide a factually correct history, poor hygiene, poor compliance� Patients age 80 and over at some regular frequency(Consensus, IRGSG-sponsored)

B. How to Screen

Use of either the clock-drawing test (using a simple “normal” or abnormal” scoring method) orMini-Cog is recommended in initial targeted screening for dementia, with recognition that thesetools have greater reliability as severity of dementia increases.(Evidence-Based, IRGSG-sponsored)

Further evaluation with a more in-depth assessment process is recommended for:� Patients who fail a short cognitive screening test for dementia, or� Patients with a history or complaint (from self or informant) of forgetfulness (even if they

pass a short cognitive screen).(Consensus, IRGSG-sponsored)


III. Diagnosis

A. History

As part of an assessment for dementia, the recommended components of patient history are:� Medical history (systemic diseases, neurological disorders, head trauma, alcohol or substance

abuse, infectious or metabolic illnesses)� Functional history (ADLs and IADLs)� Medication history (non-prescription, prescription, and herbal/alternative)� Family history (early-onset dementia or other rare conditions leading to dementia)� Social history (education, literacy, preferred language)� Detailed description of the chief complaintIt is recommended that the history be obtained from the patient and a reliable informant whenpossible.(Consensus, IRGSG-sponsored)

It is necessary to confirm decline in functional status to make a diagnosis of dementia. It isrecommended to use the Functional Activities Questionnaire (FAQ) for this assessment, as longas an informant is present. (Evidence-Based, IRGSG-sponsored)

If an informant is not initially identified, efforts should be made to find one. (Consensus, IRGSG-sponsored)

B. Physical Exam

A basic physical exam is recommended as part of the assessment for dementia, with specialattention paid to: � Patient alertness, behavior, and appearance (i.e., hygiene and affect)� Neurological signs such as abnormal reflexes, gait, balance, strength, extrapyramidal signs

(including rigidity, bradykinesia, cogwheeling, and resting tremor)� Cardiac and carotid auscultation� Orthostasis(Consensus, IRGSG-sponsored)

C. Cognitive Status Assessment

Any of the widely studied mental status tests (MMSE, BIMC, BOMC, STMS) are recommendedas part of a dementia assessment.(Evidence-Based, IRGSG-sponsored)

The standard-form MMSE is recommended as the cognitive assessment tool of choice because itis widely studied and used. It is strongly recommended that scores be interpreted using norms


related to age, education, and primary language, that the test be considered just one piece ofinformation in the assessment process, not a test that can diagnose dementia.(Consensus, IRGSG-sponsored)

D. Diagnostic Tests

The following laboratory tests are recommended as part of the routine work-up for suspecteddementia: CBC, sodium, creatinine, calcium, glucose, thyroid function tests (TSH first; ifabnormal, then a T4 can be used), and Vitamin B12 levels. Additional tests are an option if the medical history or physical suggest that they are clinicallyindicated: MHATP or FTA for syphilis, methylmalonic acid, liver function, HIV, chest X-ray,urinalysis, toxicology screen, and EEG.

Assuming there are no contraindications, lumbar puncture is indicated only when the followingare present: cancer, suspicion of CNS infection, reactive serum antitreponemal syphilis serology,hydrocephalus, dementia in a person under age 55, rapidly progressive or unusual dementia,immunosuppression, or suspicion of CNS vasculitis, particularly in patients with connectivetissue diseases.

(Consensus, IRGSG-sponsored)

E. Neuroimaging

Non-contrast CT scanning is an option for all patients with suspected dementia, and isrecommended if:� Patient is under the age of 65OR� Patient is age 65 or over and has any of the following:

� Atypical presentation/unclear diagnosis� Rapid unexplained deterioration� Unexplained focal neurological signs or symptoms� History of head injury with a temporal relationship to symptoms� Urinary incontinence or gait ataxia early in the illness� Clinical suspicion of undiagnosed cerebrovascular disease

Contrast CT and MRI are generally not recommended for routine use in dementia assessment inprimary care because in most cases they offers no advantage over non-contrast CT for ruling outpotentially reversible causes.

PET and SPECT are not recommended in dementia assessment in primary care. It isrecommended that a specialist be consulted if questions arise about the usefulness of PET orSPECT in individual cases.

(Consensus, IRGSG-sponsored)


IV. Pharmacological Treatment of Dementia

A. Acetylcholinesterase Inhibitors

For mild to moderate Alzheimer’s disease (AD), a trial of an acetylcholinesterase inhibitor(donepezil, rivastigmine, and galantamine) is an option and should be considered. While there isno evidence of differences in efficacy, donepezil is preferred over the other two agents due to itsfavorable dosing schedule and moderate adverse effect profile.

For severe AD or vascular dementia, donepezil is an option. For “moderately severe” disease andAD with hypertension, rivastigmine is an option. (Note: These are not FDA-approvedindications.)

There is no published evidence to support or discourage a trial of a different acetylcholinesteraseinhibitor if the first agent fails.

Note: Formulary status of these agents varies from Region to Region – check with Pharmacy.

(Evidence-Based, IRGSG-sponsored)

B. Memantine

For patients with moderate to severe Alzheimer’s disease, a trial of memantine 20 mg/day is anoption.

There is no evidence to support the use of memantine in combination with anacetylcholinesterase inhibitor.

(Note: This drug is not on any KP formulary at this time. If added to formulary, it may berestricted to certain specialty groups.)


C. Re-evaluation of Drug Therapy

Evaluate patients being treated with acetylcholinesterase inhibitors and memantine 8-12 weeksafter initiation of therapy, then every 3-6 months. Recommended aspects of evaluation include: � Assessment of compliance� Assessment of significant side effects � Assessment of cognition� Assessment of function� A discussion with patient and/or caregiver to assess whether the benefit of therapy outweighs

the burden enough to justify continued use.(Consensus, IRGSG-sponsored)


D. Other Agents

There is insufficient evidence to recommend the use of vitamin E (2000 IU per day) or Ginkgobiloba for the treatment of cognition and function in Alzheimer’s disease and other dementias.(Evidence-Based, IRGSG-sponsored)

Estrogen, estrogen plus progestin, non-steroidal anti-inflammatory drugs (NSAIDs), and Cox-2inhibitors are not recommended for the treatment of cognition and function in Alzheimer’sdisease and other dementias.(Evidence-Based, IRGSG-sponsored)

V. Pharmacological Management of Behaviors

Certain antipsychotics or anticonvulsants are options for management of behavioral andpsychological symptoms related to dementia including aggression and agitation. (Note: Theseare not FDA-approved indications.)� Olanzapine is an option for agitation, hallucinations, and delusions. � Risperidone is an option for aggressive, verbal non-aggressive (e.g., grunting), and psychotic

symptoms, but there is no evidence to support its use for physical non-aggressive behaviorslike wandering, fidgeting, and hoarding.

� Haloperidol is an option for aggressive behavior, but there is no evidence to support its use inagitation (such as wandering, crying out).

� Carbamazepine is an option for management of agitation and aggression.With use of these agents, benefits with regard to behaviors must be weighed against risk of sideeffects.

There is no evidence to support the use of antipsychotics, anticonvulsants, or benzodiazepinesfor wandering, spontaneous vocalization, or apathy.

There is insufficient evidence to support the use of acetylcholinesterase inhibitors (donepezil,galantamine, rivastigmine), thioridazine, sodium valproate, or trazodone for the management ofbehavioral and psychological symptoms related to dementia, and risk of side effects should beconsidered.

There is no evidence to support the use of other antipsychotics (including quetiapine) andanticonvulsants, benzodiazepines, or buspirone for the management of behavioral andpsychological symptoms related to dementia.


VI. Non-Pharmacological Management of Behaviors

There is insufficient evidence to support the use of music therapy, validation therapy, realityorientation, and reminiscence therapy for managing behaviors associated with dementia.(Evidence-Based, IRGSG-sponsored)


VII. Specialty Referrals

It is recommended that primary care providers refer patients with suspected or diagnoseddementia to a specialist (geriatrician, neurologist, gero-psychiatrist as available) if any of thefollowing apply: � Cognitive loss is early-onset (i.e., before age 60)� Diagnosis is complex or remains unclear after basic work-up� PCP requires consultation to assist with management� Patient or family strongly desires the consultation of a specialist(Consensus, IRGSG-sponsored)

VIII. Special Considerations

A. Driving

For patients with dementia, it is recommended that patients and their families (or otherresponsible party) be told:� The patient is at increased risk for driving accidents and driving performance errors.

(Evidence-Based, IRGSG-sponsored)� The patient should not drive an automobile unless an on-road evaluation of driving ability

conducted by the state driver's licensing authority deems it appropriate. (Consensus, IRGSG-sponsored)

Physicians practicing in California are obliged by law to report all people with a diagnosis ofdementia severe enough to impair their ability to operate a motor vehicle to the Department ofHealth, who will then notify the Department of Motor Vehicles. (State law) Clinicians in otherstates should keep abreast of reporting requirements in their state. (Consensus, IRGSG-sponsored)

To support patients and families around the issue of driving, it is recommended that clinicians:� Advise caregivers of people with dementia of the early warning signs of driving problems.� Advise families of resources to help them manage this issue.� Encourage patients and families to plan for eventual cessation of driving privileges and to

develop alternate transportation plans. (Consensus, IRGSG-sponsored)


B. Caregivers

For caregivers of people with dementia, it is an option for clinicians to refer to programs andresources focusing on:� Counseling of caregivers and families� Education / training� Involvement of the patient� Support � Stress management(Evidence-Based, IRGSG-sponsored)

It is recommended that clinicians: � Consistently connect caregivers to resources for counseling, education, and training� Monitor caregivers for depression, fatigue, anger, anxiety, and other stress-related signs and

refer to KP and community resources as available.(Consensus, IRGSG-sponsored)


Supporting DocumentationFor each of the topic areas, the following supporting documentation is present:

� Problem Formulation – This explains the questions we seek to answer with these particularrecommendations.

� Evidence Search – This documents the approach taken to find a review literature on thistopic.

� Recommendations and Rationale Statement – This documents the recommendations andthe rationale (or basis) for the recommendation, including a summary of supporting evidenceand a description of how decisions were made in the face of conflicting or insufficientevidence.

� Evidence Tables – These outline the methods and major findings of relevant resourcesreviewed for each topic. NOTE: This manual does not contain the old evidence tables fromthe resources first reviewed in the 2002 CMI Dementia Guidelines, but all tables areavailable on the Permanente Knowledge Connection (PKC) at http://pkc.kp.org

http://pkc.kp.org/


Prevention of Dementia

Problem Formulation

Clinical Question: What agents should be used to prevent dementia?

Intended Use of theGuideline:

To assist primary care physicians and other health care professionals inthe outpatient primary care setting in preventing dementia among peoplewho do not have dementia

Population: Men and women without dementia (Re: estrogen and estrogen plus progestin, women without dementia)

Health Problem: Cognitive and functional decline related to development of dementia

Health Intervention:

� Estrogen (women only) � Estrogen plus Progestin

(women only)� NSAIDs� Cox-2 inhibitors

� Ginkgo biloba� Vitamin E� Statins� No treatment

Practitioners: KP physicians, physician assistants, nurse practitioners, nurses, andsocial workers in the Departments of Family Practice and InternalMedicine (primary care), health educators

Setting: Outpatient office visit Most Important

Health OutcomesAssociated with the

Intervention:

� Diagnosis of dementia

Side Effects of theIntervention:

� Drug adverse effects and interactions

Measures ofOutcomes:

� Score on cognitive scales� Score on functional scales


Evidence Search

Only RCTs or meta-analyses were included that studied the use of estrogen/estrogen plusprogestin (HRT), NSAIDs, Cox-2s, Ginkgo biloba, vitamin E, or statins (compared to placebo)for preventing or reducing the risk of dementia. Observational studies that examined theepidemiological relationship between these agents and the development of dementia were notreviewed because many such studies were included in the review for the 2002 CMI DementiaGuidelines and were deemed insufficient to draw conclusions. Because searches on this topic were conducted for the 2002 CMI Dementia Guidelines, thisguideline revision includes relevant RCTs or meta-analyses published since the original searcheswere performed. The most recent searches are listed first, followed by the original searches.Search strategy was changed from separate searches for each drug to one main search.

Database: Terms: Article typeand Limits:

TimeFrame:

#Found:

# inET:

New Searches

Cochrane dementia Systematic reviews 5/27/03 104 0

ClinicalEvidence dementia Systematic reviews

and RCTs 5/27/03 12 0

Meta-analysis,English, Human

4/1/01-8/5/03 6 0

PubMed

("Dementia"[MeSH] OR"Dementia/prevention andcontrol"[MeSH]) AND(("Estrogens"[MeSH] OR"Estrogen ReplacementTherapy"[MeSH] OR"Hormone ReplacementTherapy"[MeSH] OR"estrogen"[TEXT] OR"HRT"[TEXT]) OR ("Anti-Inflammatory Agents, Non-Steroidal"[MeSH] OR"NSAIDs"[TEXT]) OR("Ginkgo biloba"[MeSH] OR"ginkgo"[TEXT] OR"gingko"[TEXT]) OR("Vitamin E"[MeSH] OR"vitamin E"[TEXT]) OR("Hydroxymethylglutaryl-CoAReductase Inhibitors"[MeSH]OR "statin"[TEXT])

RandomizedControlled Trial,English, Human

4/1/01-8/5/03 19 1



TimeFrame:

#Found:

# inET:

PubMed

("Dementia"[MeSH] OR"Dementia/prevention andcontrol"[MeSH]) AND("rofecoxib"[TEXT] OR"celecoxib"[TEXT])


1965-10/1/03 1 0

Other Information Sources:A workgroup member identified one article on use of statins that did not come up in our literaturesearches and contained secondary outcomes relevant to dementia.(1) That information issummarized in the statins rationale section. Old Searches (from 2002 CMI Dementia Guidelines)

dementia[MESH] ANDEstrogens[MESH] AND (Meta-Analysis[PT] OR ClinicalTrial[PT])

Meta-analysis,Clinical Trial(embedded in search)

1965-5/23/01 14 1

(("Hormone ReplacementTherapy"[MESH] ORestrogens[MESH]) AND(dementia[MESH] OR"Alzheimer Disease"[MESH])AND ("case controlstudies"[MESH] OR "cohortstudies"[MESH]))

Case Control,Cohort (embedded insearch)

1/1/00-5/21/01 9 3

(("hormone replacementtherapy"[MESH] ORestrogens[MESH]) AND(cognition[MESH] ORdementia[MESH] OR"Alzheimer Disease"[MESH]))

RandomizedControlled Trial

1/1/00-5/21/01 11 0

PubMed

(("Hormone ReplacementTherapy"[MESH] OREstrogens[MESH]) AND(dementia[MESH] OR"Alzheimer Disease"[MESH]))

RandomizedControlled Trial

1965-5/21/01 12 3



TimeFrame:

#Found:

# inET:

(("Alzheimer Disease"[MESH]OR dementia[MESH]) AND("Anti-Inflammatory Agents,Non-Steroidal"[MESH] OR"anti-inflammatoryagents"[MESH] ORNSAID[TEXT] ORarthritis[MESH] ORarthritis[TEXT]) AND("epidemiologicstudies"[MESH] OR "Case-Control Studies"[MESH] OR"cohort studies"[MESH] OR"cross sectionalstudies"[MESH]))

Case Control,Cohort, Cross-sectional (embeddedin search)

1965-5/29/01 45 8

PubMed

(("Alzheimer Disease"[MESH]OR dementia[MESH]) AND("Anti-Inflammatory Agents,Non-Steroidal"[MESH] OR"anti-inflammatoryagents"[MESH] ORNSAID[TEXT]))

Clinical Trial 1965-5/29/01 48 2

PubMed

(("Alzheimer disease"[MESH]OR dementia[MESH] OR"Alzheimer disease"[TW] ORdementia[TW] ORcognition[TW]) AND ("ginkgobiloba"[MESH] ORginkgo[TW] OR gingko[TW]OR "EGb 761"[TW] ORTanakan[TW] OR Rokan[TW]OR Tebonin[TW] ORGinkoba[TW]))

English, Human 1965-6/1/01 70 5



TimeFrame:

#Found:

# inET:

(("Alzheimer Disease"[MESH]OR "Alzheimer'sDisease"[TEXT] ORdementia[MESH] ORdementia[TEXT]) AND("vitamin E"[MESH] OR"vitamin E"[TEXT]))


PubMed

(cognition[MESH] AND("vitamin E"[MESH] OR"alpha-tocopherol"[TW]))

English, Human 1965-6/29/01 17 4

PubMed

((lovastatin[MESH] ORstatins[TEXT]) AND(dementia[MESH] OR"Alzheimer Disease"[MESH]))

English, Human 1965-5/28/01 13 2

Cochrane dementia No Limits 4/12/01 76 0Other Information Sources (from 2002 CMI Dementia Guidelines)� American Academy of Neurology 2001 Practice Parameter: Management of Dementia (an

evidence-based review)(2)

� Methods: The literature review process included the following search terms: Alzheimer’sdisease, vascular or multi-infarct dementia, dementia with associated parkinsoniandisorder, progressive supranuclear palsy, frontotemporal dementia, and senile dementia,cholinesterase inhibitors, antioxidants, hormones, anti-inflammatory agents/drugs,nootropics, metabolic enhancers, neurotrophic agents/drugs, (complementary) alternativemedicines, treatment, and pharmacotherapy. For information on pharmacotherapies, thefollowing databases were searched: MEDLINE, Embase, Current Contents, PsychAbstracts, and Cochrane databases. Studies selected included the following: randomized,controlled studies in all languages and other types of studies limited to English; humansubjects with N greater than 20, regardless of outcome measured; and review articlespublished between January 1998 and November 1999. Articles were classified based onthe quality of the evidence. After review of the evidence, recommendations were drafted,reviewed by all committee members, and identified as a Practice Standard, Guideline, orOption.


Recommendations and Rationale Statement

Guideline: Prevention of dementia

Recommendations: Estrogen plus progestin should not be used to prevent dementia,as their combined use has demonstrated increased risk ofdementia.

There is insufficient evidence to recommend the use of any of thefollowing agents to prevent dementia: estrogen (alone), NSAIDs,Cox-2 inhibitors, ginkgo biloba, vitamin E, statins.

Methodology: � Evidence-Based � IRGSG-sponsored

Rationale:

Supporting Evidence for Estrogen/Estrogen + ProgestinEstrogen plus progestin versus placebo: � The Women’s Health Initiative Memory Study (WHIMS), an ancillary study to the Women’s

Health Initiative (WHI) hormone therapy trials, is a double-blind RCT examining the effectsof estrogen with and without progestin compared to placebo on all-cause dementia. Theestrogen plus progestin arm of WHI was discontinued early due to increased risks in theestrogen plus progestin group, providing the WHIMS data on that same arm. Results of thatstudy show a significant negative effect of estrogen plus progestin on risk for developingprobable dementia.(3) Over the five-year study, 40 women (out of 2229) in the estrogen plusprogestin group were diagnosed with probable dementia, compared with 21 out of 2303 inthe placebo group (Hazard Ratio: 2.05, 95% CI 1.21-3.48). This translates into a 97%increase in relative risk with estrogen plus progestin compared to placebo, and a 1% increasein absolute risk.

Estrogen versus placebo: � No RCTs were found examining the impact of estrogen alone on the development of

dementia. The estrogen only arm of WHIMS continues and will provide useful informationon this question when completed.

Estrogen plus progestin and estrogen alone:� Rationale from the 2002 CMI Dementia Guidelines:

� One systematic review (4) that examined the effect of estrogen plus progestin (HormoneReplacement Therapy, or HRT) on cognition and the association between HRT anddementia was determined to be more recent and more rigorous in design than two otherreviews that were found. (5, 6) The LeBlanc review (4) included 9 randomized controlledtrials (RCTs) and 8 cohort studies focused on the effect of HRT on cognition in womenwithout symptoms of dementia, as well as 2 cohort studies and 10 case-control studieslooking at the association between HRT and dementia. There were no randomizedcontrolled trials examining the impact of HRT on dementia. (Several large RCTsexamining this question are taking place at present; results will not be available forseveral years.)


� LeBlanc et al. found a lack of consistent improvement in cognition among women treatedwith varying formulations of estrogen. A decreased relative risk of dementia (-34%)associated with HRT was found, but lack of control for prior health status of users versusnon-users and lack of control for certain confounding variables limit interpretation.

� Following the search period covered in the LeBlanc period, there have been 2 cohortstudies (7, 8) on the effect of HRT on cognition, as well as 1 additional case-control study(9) on the association between HRT and Alzheimer’s disease. None of these producedresults that altered conclusions drawn in LeBlanc.

� The literature suggests that, while epidemiological studies have indicated an inverseassociation between HRT and dementia, no recommendations for the use ofestrogen/HRT for the prevention of dementia can be made. Prospective randomizedcontrolled trials are needed to assess the true effect of HRT on risk for dementia.

� Conclusion: The original 2002 CMI Dementia Guidelines cited epidemiologicalevidence supporting a relationship between use of estrogen alone or estrogen plusprogestin and reduced risk of dementia, but that was deemed insufficient to make arecommendation. Recent literature (from the WHIMS study) provided sufficientevidence to make a more definitive recommendation against the use of estrogen plusprogestin. Estrogen plus progestin should not be used to prevent dementia; use ofestrogen plus progestin has been demonstrated to increase the risk of dementia inpostmenopausal women over age 65. There is insufficient evidence to recommend theuse of estrogen alone for the prevention of dementia.

Supporting Evidence for Non-Steroidal Anti-Inflammatory Drugs (NSAIDs)� No RCTs were found examining the impact of NSAIDs on the development of dementia.� Rationale from the 2002 CMI Dementia Guidelines:

� Eight epidemiological studies (case-control, cohort, and cross-sectional designs)examining the association between historic or current use of NSAIDs and the presence ofAlzheimer’s disease (AD) were found and reviewed. (10-17)

� The results of these studies are mixed, as is the strength of the methodology and/orreporting for each study.

� Several of the studies reviewed do not support our recommendation, but there arelimitations to their methods and conclusions. The case-control analysis by Broe et al. (10)

found a significant inverse association between the usage of NSAIDs and the presence ofAD, but not between NSAIDs and vascular dementia or other dementia. The matchedcase-control analysis by Anthony et al. (11) found that current use of NSAIDs wassignificantly associated with reduced prevalence of AD. However, NSAID use wasassessed primarily through retrospective interviews, potentially biasing the study(patients with AD may be less likely than healthy patients to recall NSAID use correctly).Stewart et al. (14) conducted an unblinded cohort study, and found that reported use ofNSAIDs was significantly inversely related to risk of AD; however, this study did notreport the raw numbers of NSAID status by disease status (AD vs. non-AD), so theresults are difficult to interpret. The cross-sectional study by Andersen et al. (16) foundthat the relative risk of AD was significantly lower among NSAID users compared tonon-users. However, NSAID use was defined as use within 1 week before the interview,potentially biasing the results because patients who have AD may be less likely to receiveNSAIDs (since they may be less likely to articulate pain symptoms or their clinicians


may be less likely to treat their pain). The Canadian Study of Health and Aging (17) foundthat a history of use of NSAIDs was linked to a lower risk of AD. They did not definewhat qualified as a “history of use”.

� Several studies do support our recommendation. The nested case-control study by in‘tVeld et al. (12) found that NSAID use of at least 2 months or at least 6 months durationdid not significantly lower the risk of AD. The case-control analysis by Beard et al. (13)

also found that risk of AD was not significantly reduced with NSAID use. The cohortstudy by Fourrier et al. (15) did not show a significant difference between the incidencerate of dementia among a group of NSAID users compared to non-users.

� Using a conservative approach in the face of mixed results and flawed methodology forsome studies, our recommendation is that NSAIDs should not be used for the purpose ofpreventing AD. If NSAIDs are used to manage other conditions (i.e., arthritis), theliterature suggests that this may afford some protection against AD, but that is unclearand unconfirmed. Prospective randomized trials are needed to draw a conclusion aboutthe causal relationship between NSAID use and prevention of AD. Because of thedocumented risk of side effects (i.e., GI symptoms) associated with NSAID use, and thelack of clear benefit in terms of prevention of AD, NSAIDs can not be recommended forthis purpose.

� Conclusion: There is insufficient evidence to recommend use of NSAIDs to preventdementia.

Supporting Evidence for Cyclooxygenase-2 inhibitors (Cox-2 inhibitors)� No RCTs were found examining the impact of Cox-2 inhibitors on the development of

dementia.� Conclusion: There is insufficient evidence to recommend use of Cox-2 inhibitors to

prevent dementia.

Supporting Evidence for Ginkgo biloba� No RCTs were found examining the impact of Ginkgo biloba on the development of

dementia.� Rationale from the 2002 CMI Dementia Guidelines:

� Three randomized controlled trials on the effect of Ginkgo biloba on cognition, memory,or neuropsychologic functioning in older, healthy adults were found and reviewed. (18-20)

� Wesnes (18) studied 58 adults with mild impairment of everyday functioning, aged 62-85,randomized to receive either Tanakan (Ginkgo biloba) or placebo. Cognitive testing tookplace at baseline and after 12 weeks. Results were mixed. Data from several tests had tobe combined for the beneficial effects of Tanakan to emerge. Mix et al. (19) involved 48cognitively intact adults, aged 55-86, randomized to receive either 180 mg Ginkgo bilobaextract per day or placebo. The groups were compared after 6 weeks on severalneuropsychologic tests: Stroop Color and Word Test, Trail Making Test, WechslerMemory Scale-Revised, and Follow-up Self Report Questionnaire. Significantdifferences in favor of Ginkgo Biloba were found only on the Stroop Color and WordTest and the Follow-up Self Report Questionnaire. Rigney et al. (20) studied 36 healthyadults aged 30-59 in a 5-way crossover design randomized controlled trial. Each memberof the study took all 5 of the treatments (various doses of Ginkgo biloba, and placebo) fortwo days each, with a minimum 5 day washout period in between. Measures included the


Sternberg Short Term Memory scanning test, the Immediate and Delayed Word Recalltest, Critical Flicker Fusion, and Choice Reaction Time. All tests except the Immediateand Delayed Word Recall test showed statistically significant differences in memory infavor of Ginkgo biloba.

� These studies show some relationship between use of Ginkgo biloba and improvedmemory in healthy adults. However, the results were not consistent for all tests ofmemory, cognition, and neuropsychologic functioning used in these studies. Also, noneof these studies addressed the health outcome of Alzheimer’s disease or dementia, so noconclusion can be drawn as to their usefulness in preventing these diseases.

� Conclusion: There is insufficient evidence to recommend use of Ginkgo biloba toprevent dementia.

Supporting Evidence for Vitamin E� No RCTs were found examining the impact of Vitamin E on the development of dementia.� Rationale from the 2002 CMI Dementia Guidelines:

� Four epidemiological studies examining the relationship between Vitamin E andcognition were found and reviewed. (21-24) La Rue (21) used a cohort design, Peacock (22)

used a cross-sectional design, and Schmidt (24) used a case-control design – all to examinethe correlation between Vitamin E intake and cognitive function. Masaki (23) used acohort design to examine the relationship between Vitamin E use and various types ofdementia.

� La Rue found statistically significant associations between Vitamin E intake in healthypatients and performance on a variety of cognitive tests: Rey-Osterrieth Copy Test (testof visuospatial skill), Rey-Osterrieth Recall Test (measure of nonverbal learning andmemory), WMS Visual Reproduction Test (subtest from the Wechsler Memory Scale toassess nonverbal memory), and WMS Logical Memory Test (subtest from WechslerMemory Scale to assess verbal memory). However, Peacock found no statisticallysignificant association between Vitamin E intake in cognitively intact patients and othertests of cognitive performance: Word Recall Test, WAIS-R Digit Symbol, and WordFluency Test. Schmidt found a statistically significant difference in the mean Vitamin Eplasma levels in a group of 1,325 people with adequate cognitive function compared to agroup of 444 people with poor cognitive function – those with adequate cognitivefunction had higher mean plasma levels of Vitamin E. Further analysis of this data bylogistic regression, to control for possibly confounding variables, revealed that theassociation was weak but statistically significant. Masaki examined the relationshipbetween use of vitamins E and C and vascular dementia, Alzheimer’s disease, mixeddementia, and poor cognition (Cognitive Abilities Screening Instrument (CASI) scoreless than 74). That analysis found that Vitamin E use alone (without Vitamin C) wasassociated with slightly lower odds of poor cognition, vascular dementia, andAlzheimer’s disease (OR=.90, .86, and .84, respectively). These findings were notstatistically significant.

� Because these studies are epidemiological in design (not clinical trials), and because theevidence presented is mixed, there is insufficient evidence to recommend the use ofVitamin E for increasing cognitive performance or preventing Alzheimer’s disease andother dementias.


� Conclusion: There is insufficient evidence to recommend use of vitamin E to preventdementia.

Supporting Evidence for Statins� No RCTs were found examining the impact of statins on the development of dementia. A

workgroup member suggested the consideration of a study that did not come up in thisliterature search, but did have secondary outcomes relevant to this clinical question. Thatstudy, the Heart Protection Study, randomized 20,536 adults (aged 40-80) with coronarydisease, other occlusive arterial disease, or diabetes to simvastatin 40 mg daily or placebo.(1)

After a 5-year treatment period, the study found similar numbers of participants in eachgroup were reported to have developed dementia during follow-up (31 patients, or 0.3%, ineach group). Therefore, simvastatin was not found to have a preventive effect for dementia.

� Rationale from the 2002 CMI Dementia Guidelines:� There have been two recent publications examining the epidemiological relationship

between use of statins (HMG-CoA reductase inhibitors) and risk of dementia (includingAlzheimer’s disease, or AD). (25, 26) Jick et al. (25) used a case-control design to examinethe proportion of statin users (past or present) among a group of patients diagnosed withdementia or AD compared to a group of matched controls. Wolozin (26) used a cross-sectional design to examine the prevalence of AD in a group of patients receiving statinscompared to the entire population (using the databases from three hospitals) and alsocompared to patients receiving other medications for hypertension or cardiovasculardisease.

� Both studies found statistically significant inverse relationships between statin use andrisk of dementia (statin use was correlated with lower risk of dementia). Jick et al. foundthat the adjusted relative risk of dementia with current use of statins compared to noexposure to statins was statistically significant (RR=0.29; 95% CI: 0.13, 0.63; p=0.0002).Jick did not find any such relationship with the use of other lipid lowering agents or withhyperlipidemia alone. Wolozin found that the prevalence of AD was 60% to 73% lowerin one of the statin groups (lovastatin + pravastatin) compared to the entire patientpopulation or compared with patients receiving other medications for hypertension orcardiovascular disease (p� 0.001). Because of their design, both studies can only showassociations, not causal links.

� Although these studies present some interesting associations which merit further study,no conclusions on the causal relationship between statins and reduced risk of dementiacan be drawn. Therefore, at this time, statins can not be recommended for the purpose ofpreventing dementia.

� Conclusion: There is insufficient evidence to recommend use of statins to preventdementia.


Evidence Table: Estrogen plus progestin for the prevention of dementia

Study, Total n

TreatmentGroups Size

& Drug Study Population Results Comments

Shumaker et al., 2003 (RCT, double-blind) Follow up: up to 5 years Initial N: 4894 Final N: 4532 Intent-to-treat analyses used

Rx1 placebo(n = 2303) Rx2 estrogen +progestin(n = 2229)

� Healthy women age 65 orolder

� Intact uterus

Cases of probable dementia among all patients Rx1: 21 Rx2: 40 Hazard ratio: 2.05 (95% CI 1.21-3.48) Cases of probable dementia after excluding those with higherrisk at baseline Rx1: 10 Rx2: 24 Hazard ratio: 2.64 (95% CI 1.26-5.53) Relative Risk Reduction = - 0.97 97% increase in relative risk with estrogen + progestin comparedto placeboAbsolute Risk Reduction = -0.01 1% increase in absolute risk with estrogen + progestin comparedto placebo(These calculations were performed by article reviewer based ondata presented in the article.)

Conclusions� Estrogen plus

progestin therapyincreases the risk ofprobable dementia forwomen ages 65 andabove.

� The risks of estrogenplus progestin therapyoutweigh the benefits.

Biases� None noted.


Screening/Casefinding – Who to Screen

Problem Formulation

Clinical Question: Who should be screened for dementia?


To assist primary care physicians and other health care professionals inthe outpatient primary care setting in efficiently and effectivelyscreening for dementia.

Population: Men and women who have not been diagnosed with dementia

Health Problem: Identification of dementia


� Screening for dementia based on one of the following components:� Age 65 or older� Age 65 or older plus clinical symptoms� Age 80 or over

� No screening

Practitioners: KP physicians, physician assistants, nurse practitioners, nurses, andsocial workers in the Departments of Family Practice and InternalMedicine (primary care)



Intervention:

� Identifying appropriate population for full work-up


� False positives� False negatives

Intermediate(Biological)Outcomes:

� Identifying level of cognitive ability


Evidence Search

Only systematic reviews, clinical trials, or observational studies that identified the appropriatepopulation for dementia screening were included. Because searches on this topic were conducted for the 2002 CMI Dementia Guidelines, updateson those searches were performed (with limit of “clinical trial” removed). The most recentsearches are listed first, followed by the original searches.


TimeFrame:

#Found:

# inET:

New Searches



and RCTs 5/27/03 12 0


1965 –6/10/03 2 0

PubMed

("Mass Screening"[MeSH]OR "screening"[TEXT] OR"casefinding"[TEXT]) AND("Alzheimer Disease"[MeSH]OR "Dementia"[MeSH])

English, Human 2/1/01 –6/10/03 267 0


Other Information Sources� In May 2003, the Agency for Healthcare Research and Quality (AHRQ) released a systematic

evidence review conducted for the U.S. Preventive Services Task Force (USPSTF) onscreening for dementia. (27) This updated a 1996 review where the USPSTF found insufficientevidence to recommend for or against screening for dementia.(28) The new review wasconducted to consider more recent studies (literature published up to September 1, 2002)concerning screening tests as well as both pharmacologic and caregiver interventions. Thesystematic review of the literature sought to answer nine key questions:1. What is the direct effect of dementia screening on outcomes?2. How common is undiagnosed dementia?3. How accurate are the screening tests?4. How effective is primary treatment of potentially reversible or irreversible dementia?5. How effective is secondary treatment for dementia (pharmacologic and nonpharmacologic

interventions)? 6. How effective are interventions for caregivers of people with mild to moderate dementia?7. What are the adverse effects of screening and early treatment of dementia?8. What is the cost of dementia screening?9. What are the adverse effects of dementia treatment?

The methods of the systematic review were rigorous. Old Searches (from 2002 CMI Dementia Guidelines)

PubMed(("Mass Screening"[MESH]OR "casefinding"[TEXT])AND dementia[MESH])

Clinical Trial,English, Human

1965-4/1/01 13 0

Cochrane dementia No Limits 4/12/01 76 0Other Information Sources (from 2002 CMI Dementia Guidelines)A member of the CMI Dementia Guidelines Workgroup referred the group to a review article. (29)

This article was consulted and was used as a basis for discussion to develop a consensus-basedrecommendation.


Recommendations and Rationale Statements

Guideline: Screening for Dementia – Who to Screen

Recommendation 1: There is insufficient evidence to recommend for or againstroutine screening for dementia in older adults.

Methodology: � Evidence-Based� IRGSG-sponsored

Rationale:

� The USPSTF systematic evidence review on dementia screening concluded that, “theevidence is insufficient to recommend for or against routine screening for dementia in olderadults.” (30) This was based on their findings of:� Good evidence of good sensitivity but fair specificity of screening tests� Fair to good evidence that several drug therapies have beneficial effects on cognition and

function� Mixed evidence of effect on instrumental activities of daily living� Insufficient evidence to determine whether benefits observed in drug trials are

generalizable to those whose disease would be detected in primary care� Unknown accuracy of diagnosis, feasibility of screening and treatment in routine clinical

practice, and potential harms of screening (e.g., labeling effects)

Recommendation 2: Targeted screening for dementia is recommended in:� Patients age 65 and over when clinical presentation suggests

possibility of dementia, e.g., inability to provide a factuallycorrect history, poor hygiene, poor compliance

� Patients age 80 and over at some regular frequency

Methodology: � Consensus� IRGSG-sponsored

Rationale: (From 2002 CMI Dementia Guidelines with minor modifications)

Supporting Evidence for 65+ population:� The systematic evidence review on dementia screening conducted for the USPSTF did not

directly address the issue of targeted screening,(30) but the USPSTF’s “ClinicalConsiderations” summary concluded, “Although current evidence does not support routinescreening of patients in whom cognitive impairment is not otherwise suspected, cliniciansshould assess cognitive function whenever cognitive impairment or deterioration issuspected, based on direct observation, patient report, or concerns raised by family members,friends, or caretakers.” (30) This is consistent with the CMI Dementia Guidelines Workgrouptargeting strategy.


� A literature search on mass screening/casefinding for cognitive impairment yielded noclinical or observational studies on who should be targeted for screening in the 65 and overpopulation. The descriptive literature found (29) contained consensus-based recommendationsthat the CMI Dementia Guidelines Workgroup used as the basis for discussion on this issue.The Workgroup felt compelled to make a recommendation on this issue, even with the lackof solid evidence, because dementia is notoriously underdiagnosed and questions regardingthe need for global screening/casefinding are common in primary care.

� The prevalence literature shows that approximately 3-5% of those aged 65 and above havedementia. (31, 32) There is a problem in applying any screening test to a population in whichthe prevalence of the condition, or “base rate,” is low. Given the base rate in the general 65-and-above population, and because the review literature shows modest positive predictivevalues of cognitive screening for dementia in the general older adult population (29), the groupagreed that any screening criteria for the general population of patients age 65 and aboveneeds to be based on age plus clinical signs and symptoms (a group with a higher base rate).The group agreed that there are several readily apparent behaviors or characteristics that maysignal cognitive decline, including poor medication or appointment compliance, poorrecollection during the medical history, and poor hygiene.

Supporting Evidence for 80+ population:� A literature search on mass screening/casefinding for cognitive impairment yielded no

clinical or observational studies on who should be targeted for screening in the 80 and overpopulation. The descriptive literature found (29) contained consensus-based recommendationsthat the CMI Dementia Guidelines Workgroup used as the basis for discussion on this issue.The Workgroup felt compelled to make a recommendation on this issue, even with the lackof solid evidence, because dementia is notoriously underdiagnosed and frequency steadilyincreases with age, and questions regarding the need for routine screening/casefinding incertain populations are common in primary care.

� The prevalence literature estimates that approximately 15-40% of those aged 80 and abovehave dementia. (32) Given this high prevalence, the group agreed that a short cognitive screenamong the “older old” population at some regular frequency is appropriate.


Screening/Casefinding – How to Screen

Problem Formulation

Clinical Question: What tools/tests should be used to screen for dementia and how shouldthey be scored?


To assist primary care physicians and other health care professionals inthe outpatient primary care setting in efficiently and effectivelyscreening for dementia.

Population: Men and women who have not been diagnosed with dementia but whowarrant screening based on “Who to Screen” recommendations

Health Problem: Identification of dementia

Health Intervention: � Clock-drawing test (using formal or informal scoring system)� Mini-Cog (using published scoring system)




Intervention:

� Identifying appropriate population for full assessment




� Sensitivity� Specificity


Evidence Search

Only systematic reviews, clinical trials, or observational studies that examined the reliability andvalidity of the clock-drawing test and the Mini-Cog test for dementia screening were included. Because searches on this topic were conducted for the 2002 CMI Dementia Guidelines, updateson those searches were performed (combined into one search). The most recent searches arelisted first, followed by the original searches.


TimeFrame:

#Found:

# inET:

New Searches



and RCTs 5/27/03 12 0


1965 –6/10/03 0 0

PubMed

("Alzheimer Disease"[MeSH]OR "Dementia"[MeSH])AND ("clock-drawing"[TEXT] OR "Mini-Cog"[TEXT])

English, Human 01/01/01 –06/10/03 34 6*

* Six studies are described in detail in the rationale; no evidence tables were created because these descriptionscontained all the information that an evidence table would have.





The methods of the systematic review were rigorous. Old Searches (from 2002 CMI Dementia Guidelines)

dementia[MESH] ANDclock-drawing[TEXT] No Limits 1965-

3/7/01 50 3PubMed dementia[MESH] AND

“Mini-Cog”[TEXT] No Limits 1965-10/10/01 2 1

Cochrane dementia No Limits 4/12/01 76 0Other Information Sources (from 2002 CMI Dementia Guidelines)A member of the CMI Dementia Guidelines Workgroup referred the group to a review article.(29)

This article was consulted and was used as a basis for discussion to develop a consensus-basedrecommendation.



Guideline: Screening for dementia – how to screen

Recommendation 1: Use of either the clock-drawing test (using a simple “normal” orabnormal” scoring method) or Mini-Cog is recommended ininitial targeted screening for dementia, with recognition thatthese tools have greater reliability as severity of dementiaincreases.

Methodology: � Evidence-Based� IRGSG-sponsored

Rationale:

Supporting Evidence for Clock-Drawing Test: � The clock-drawing test (CDT) has been widely studied as a screening tool for dementia, with

assessments of validity using different test formats, scoring methods, and populations. As aresult, there is not clear agreement on how best to administer and score the test in primarycare. Several recent observational studies have examined the relative sensitivity andspecificity (or, in some studies, receiver operating characteristic [ROC] curve values) of thetest using several different published administration and scoring methods. � Scanlan et al. compared seven CDT scoring methods and the judgment of untrained raters

for 80 clock drawings drawn by community-dwelling people with dementia (using goldstandard of DSM-IV and NINCDS-ADRDA criteria for dementia). (33) Initially, clockswere scored to identify dementia status as normal, mild, moderate, and severe (n = 20 forall groups) according to CERAD (Consortium to Establish a Registry for Alzheimer’sdisease) clock scoring criteria. CERAD clock scoring was used as the basis forcomparison with 7 other published scoring methods, those by Shulman (1986), Morris(1989), Sunderland (1989),Wolf-Klein (1989), Mendez (1992), Manos and Wu (1994),and Lam (1998). Concordance of these methods with the CERAD findings was reported.All 8 clock-scoring system results were then compared for concordance with naïve raters.The study found that both naïve raters and the formal scoring systems showed 90-100%agreement in the normal, moderate, and severe categories, but only 39% agreement in themild dementia group. Naïve raters who assessed the clocks simply as “normal” or“abnormal” were as effective as 5 of the 7 CDT systems in identifying dementia.Assessment of accuracy of formal scoring systems compared to independent clinicaldementia diagnosis (yes/no) excluded the mild dementia group, and found that thoseshowing greatest percent correctly classified were Mendez (85.2%; sensitivity 90.7%;specificity 76%) using cut-point normal � 18 and CERAD (83.8%; sensitivity 95.3%;specificity 64%) using 0 as normal cut-point.

� Storey et al. examined the accuracy of five CDT scoring methods for detecting dementiain 127 English-speaking patients (34) and 93 non-English speaking patients (35) comparedto clinical diagnosis (demented or not demented) based on gold standard of DSM-IVcriteria. These prospective cohort studies took place in a general geriatric outpatientclinic in Australia.


� For English-speaking patients, clocks were scored using the methodspublished by Shulman (1986), Sunderland (1989),Wolf-Klein (1989), Mendez(1992), and Watson (1993) using receiver operating characteristic (ROC)curve analysis. (34) Authors stated that values of the area under the ROC curvebetween 0.50 and 0.70 represent low accuracy and usefulness, while highervalues may show usefulness. The area under the ROC curve was greatest forShulman (0.79, 95% CI 0.70 to 0.85) and Mendez (0.78, 95% CI 0.70 to 0.85)methods; both were significantly better than the Sunderland and Watsonmethods (p � 0.05 for all).

� For non-English speaking patients (patients born in non-English-speaking-background [NESB] countries), clocks were scored using the same methods asabove, plus the 4-point CERAD scale. (35) ROC analysis showed nosignificant difference between scoring methods – area under the ROC curveranged from 0.60 (Watson) to 0.72 (Sunderland). The most sensitive was theMendez method (98%) but specificity was only 16%. Authors identified theWolf-Klein method as the best in this population, with a sensitivity of 78%,specificity of 58%, and area under the curve of 0.66. Cut-points for thesescoring methods were not specified in the article, but it was noted that the cutpoints recommended by the original authors were used.

� Powlishta et al. examined six scoring methods in the CDTs of 75 participants classifiedas normal, very mild, mild, or moderate to severe dementia (according to ClinicalDementia Rating [CDR], not gold standards of DSM-IV or ADRDA-ADRDA) as part ofa longitudinal study. (36) The scoring methods used were those created by Manos (1994),Mendez (1992), Sunderland (1989), Rouleau (1992), Pfizer Inc. and Eisai Inc. (1997) andthe AD Cooperative Study (1999). Specificity ranged from 60% (Mendez, cut-pointnormal �18) to 93% (Pfizer Easai, cut-point normal �2; Rouleau, cut-point normal �5;and Sunderland, cut-point normal �5). Sensitivity was very low in the group with verymild dementia, ranging from 20% (Sunderland and Rouleau) to 60% (Mendez).Sensitivity in mild and moderate to severe dementia was good, ranging from 80% (PfizerEasai) to 97% (Mendez and Manos, cut-point normal �7).

� Kirby et al. assessed the CDT in a community-based sample of 41 elderly subjects withdementia (any type) (using Geriatric Mental State [GMS]-Automated GeriatricExamination for Computer Assisted Taxonomy as the standard, which has demonstratedgood concordance with the DSM-III definition of dementia), 84 elderly subjects withdepression, and 523 normal elderly subjects. (37) Using the Sunderland scoring methodand cut-point for normal score �5, they found the sensitivity of the test was 76%, and thespecificity was 81% against normal elderly and 77% against depressed elderly. Amongthe subgroup of patients with MMSE � 20, sensitivity dropped to 59%. The MMSE wasmore sensitive and specific than the CDT in detecting dementia.

� Lin et al. developed a simple 3-item scoring system for the CDT based on previouslypublished systems. (38) They assessed sensitivity and specificity of this method in asample of 84 Chinese subjects with dementia (according to gold standard of DSM-IV andNINCDS-ADRDA criteria) who were administered a CDT that included a drawing partand a copying part. The 3-item scoring system had similar sensitivity and specificity tomulti-item system – the 3-item system showed sensitivity of 72.9% and specificity of65.6% using a cut-point of 2.


� The USPSTF systematic evidence review on screening for dementia included as one ofits nine key questions, “Is there a reliable and valid screening test to detect dementia inprimary care populations?”(27) They concluded that the clock-drawing test and Mini-Cogare promising, but have not been adequately studied in primary care populations and thatthe MMSE has demonstrated validity in primary care. The CMI Dementia GuidelinesWorkgroup acknowledges the need for more studies on the clock-drawing test and Mini-Cog in primary care (although a few studies examined by the workgroup focused onambulatory outpatients). Because the CMI Dementia Guidelines Workgroup intends thescreening tool to be used as an initial screen with a targeted group (not the generalprimary care population as the USPSTF did), to be followed by a confirmatory diagnosticwork-up including the MMSE (which the USPSTF did not address), the workgroup feltthat the convenience and ease of use of the clock-drawing test and Mini-Cog make themdesirable for targeted initial screening.

� Rationale from the 2002 CMI Dementia Guidelines:� One systematic review of the clock-drawing test was found and reviewed.(39) The review

examined the mean sensitivity and specificity of the test, as well as inter-rater and test/re-test reliability, positive predictive value (PPV) and negative predictive value (NPV). Inaddition, one case-control study (40) and one cohort-based comparison of retrospectivedata (41) published after the search period of this study were found and reviewed. Thesestudies also looked at the sensitivity, specificity, PPV and NPV.

� Shulman (39) examined 13 studies published between 1983 and 1998 with original scoringsystems for the clock-drawing tests, as well as 7 replication studies published during thattime. Most of these studies used the DSM-III-R and NINCDS-ADRDA (NationalInstitute of Neurological and Communicative Disorders and Stroke and Alzheimer’sDisease and Related Disorders Associations) criteria as the gold standard, while a fewused clinical diagnosis as the gold standard. The CMI Dementia Guidelines Workgroupagreed that any of these are valid gold standards. In these studies, the sensitivity of theclock-drawing test ranged from 75% to 94%, with a mean sensitivity of 85%. Specificityranged from 72% to 96%, with a mean specificity of 85%. Among the four studies thatreported a PPV, the PPV ranged from 42% to 98%. Among the four studies that reportedNPV, the NPV ranged from 90% to 94%. Inter-rater reliability was high among all 14studies that reported it (range 0.75 to 0.98). Test/re-test reliability ranged from 0.63 to0.90 in the four studies that reported it.

� The studies by Manos (40) and Esteban-Santillan (41) supported the earlier findings ofShulman.

� Given the clock-drawing test’s ease of administration as a short, simple, well-toleratedtool with minimal language requirements, and its reliability in terms of sensitivity,specificity, and other measures, it is a useful tool for initial cognitive screening forsuspected dementia in a primary care setting.

� Conclusion: Although more studies of the clock-drawing test (CDT) in primary carepopulations are needed, the CMI Dementia Workgroup concluded that evidencesupports the validity of this tools as an initial screen, and its convenience and ease of useincreases the likelihood that screening will take place. Studies examining the relativeaccuracy of different CDT scoring methods produced inconsistent results. Most of thescoring methods are relatively complex, making them difficult to use in the busyprimary care setting. Recent studies show less sensitivity and specificity in very mild


dementia compared to mild or moderate to severe dementia. Overall, sensitivity andspecificity of the CDT is intermediate, no single scoring method emerges as optimal inthe primary care setting, and untrained raters using simple dichotomy of “normal” and“abnormal” can produce similar assessments of CDTs as expert raters using moreformal scoring methods.

Supporting Evidence for Mini-Cog: � Scanlan and Borson described the receiver operating characteristics (ROC) analysis for the

Mini-Cog and the accuracy of assessment using expert and naïve raters (42) in the populationwith dementia (according to gold standard of DSM-IV and NINCDS-ADRDA criteria) fromtheir earlier study described below. (43) A scoring algorithm for the Mini-Cog maximizingsensitivity and correct diagnosis was designed, producing an area under the ROC curvesimilar to the MMSE and CASI (Cognitive Abilities Screening Instrument). Assessment(normal or abnormal) was similar by naïve and expert raters for the normal subjects as wellas the moderate to severely demented subjects (98% agreement), but less so for those withmild dementia (60% agreement).

� The USPSTF systematic evidence review on screening for dementia included as one of itsnine key questions, “Is there a reliable and valid screening test to detect dementia in primarycare populations?”(27) They concluded that the clock-drawing test and Mini-Cog arepromising, but have not been adequately studied in primary care populations and that theMMSE has demonstrated validity in primary care. The CMI Dementia GuidelinesWorkgroup acknowledges the need for more studies on the clock-drawing test and Mini-Cogin primary care (although a few studies examined by the workgroup focused on ambulatoryoutpatients). Because the CMI Dementia Guidelines Workgroup intends the screening toolto be used as an initial screen with a targeted group (not the general primary care populationas the USPSTF did), to be followed by a confirmatory diagnostic work-up including theMMSE (which the USPSTF did not address), the workgroup felt that the convenience andease of use of the clock-drawing test and Mini-Cog make them desirable for targeted initialscreening.

� Rationale from the 2002 CMI Dementia Guidelines:� One study examining the sensitivity and specificity of the Mini-Cog was found and

reviewed. (43)

� This study by Borson et al. involved 249 subjects – 129 who met criteria for probabledementia and 120 who did not – and analyzed the sensitivity and specificity of the Mini-Cog in these populations, compared to a variety of cognitive tests and dementiaclassifications. Borson found that the Mini-Cog, composed of the clock-drawing test plusa 3-word recall, had higher sensitivity (96%) than either the MMSE (Mini-Mental StateExam) or the CASI (Cognitive Abilities Screening Instrument). They also found that theMini-Cog correctly classified the greatest percentage (96%) of patients, and that unlikethe two other tests, the Mini-Cog was not influenced by language or education.

� Because the Mini-Cog was shown to be sensitive, takes less than 5 minutes to complete,is unaffected by education or language, and can be scored by untrained raters withminimal loss of sensitivity and specificity, it is a useful option for initial cognitivescreening.

� Conclusion: Although more studies of the Mini-Cog in primary care populations areneeded, the CMI Dementia Workgroup concluded that evidence supports the validity of


this tools as an initial screen, and its convenience and ease of use increases thelikelihood that screening will take place. The scoring method identified for the Mini-Cog produces good sensitivity and specificity, and similar area under ROC curves asthe MMSE and CASI.

Recommendation 2: Further evaluation with a more in-depth assessment process isrecommended for:� Patients who fail a short cognitive screening test for dementia,

or� Patients with a history or complaint (from self or informant) of

forgetfulness (even if they pass a short cognitive screen).



� Although we did not find any clinical trials that tested the best approach to cognitiveassessment (i.e., a “step approach” versus an in-depth test up front), the CMI DementiaGuidelines Workgroup felt that it would be of value to make a recommendation in this realm.Primary care providers work under severe time constraints, and a step approach to cognitiveassessment may prevent unnecessary in-depth testing of individuals who do not havedementia. The group agreed that a more in-depth assessment would be appropriate once apatient has failed a short screen such as the clock-drawing test, or if the patient (or aninformant, such as a family member) raised a complaint of memory problem.


Diagnosis – History

Problem Formulation 1

Clinical Question: What information should be gathered as part of the history, and howshould it be obtained?


To assist primary care physicians and other health care professionals inthe outpatient primary care setting in taking a history that will elicitinformation helpful for diagnosing dementia

Population: Men and women who have not yet been diagnosed with dementia butwho exhibit some characteristic signs and are undergoing assessment

Health Problem: Determining whether signs are present that would support a diagnosis ofdementia, and if so, gathering information on possible cause� Administration of questions to patient and/or informant related to

one or more of the following aspects of a patient’s history:


Info on whether dementia is present:� Functional status/daily activities� Personality/behavior changes� Memory problems

Info that may help determine cause:� Medication use

� Prescription� Over the counter (OTC)

� Herbal and homeopathic remedies� Drug and alcohol use; smoking� Last medical evaluation� Psycho-social history� HIV risk factors� Travel history� Co-morbidities




Intervention:

� Diagnosis of disease


� False diagnosis� Unnecessary tests


Evidence Search 1

This literature search looked for cross-sectional or diagnostic studies that examined what shouldbe included in a history as part of a diagnostic work-up for dementia. Because searches on this topic were conducted for the 2002 CMI Dementia Guidelines, updateson those searches were performed. The most recent searches are listed first, followed by theoriginal searches.


TimeFrame:

#Found:

# inET:

New Searches



and RCTs 5/27/03 12 0

"Medical HistoryTaking"[MeSH] AND"Dementia"[MeSH]

English, Human 1/1/01-7/28/03 12 0

PubMed

"Mass Screening"[MESH]AND dementia[MESH] AND"Diagnosis,Differential"[MESH]

English, Human 1/1/01-7/28/03 5 0

Old Searches (from 2002 CMI Dementia Guidelines)“Medical HistoryTaking”[MESH] ANDdementia[MESH]

No Limits 1965-3/7/01 58 0

PubMed "Mass Screening"[MESH]AND dementia[MESH] AND"Diagnosis,Differential"[MESH]

No Limits 1965-3/7/01 24 0

BestEvidence dementia No Limits 4/12/01 59 0

Cochrane dementia No Limits 4/12/01 76 0


Other Information Sources (from 2002 CMI Dementia Guidelines)� Agency for Health Care Policy and Research (AHCPR) Clinical Practice Guideline Number

19, “Recognition and Assessment of Alzheimer’s Disease and Related Dementias” (44)

� This guideline was developed in 1996 by a panel of 18 members who identified the clinicalquestions to be addressed; developed systematic literature search strategies; reviewed,analyzed, and summarized the scientific evidence; and rated strength of evidenceaccording to a three-level system (A = Strong Evidence; B = Suggestive Evidence; C =Expert Opinion).

� The panel then wrote guideline recommendations based on data from the scientificliterature. When scientific evidence was missing or inconclusive, panel consensus wasused to support the guideline recommendation.

� American Medical Association Guideline, “Diagnosis, Management and Treatment ofDementia – A Practical Guide for Primary Care Physicians” (45)

� VA/University HealthSystem Consortium Guidelines (1997) (46)


Recommendations and Rationale Statement 1

Guideline: History

Recommendations: As part of an assessment for dementia, the recommendedcomponents of patient history are:� Medical history (systemic diseases, neurological disorders,

head trauma, alcohol or substance abuse, infectious ormetabolic illnesses)

� Functional history (ADLs and IADLs)� Medication history (non-prescription, prescription, and

herbal/alternative)� Family history (early-onset dementia or other rare conditions

leading to dementia)� Social history (education, literacy, preferred language)� Detailed description of the chief complaint

It is recommended that the history be obtained from the patientand a reliable informant when possible.

Methodology: � Consensus � IRGSG-sponsored


� This recommendation was developed based on the 1996 Agency for Health Care Policy andResearch (AHCPR) Clinical Practice Guideline Number 19, “Recognition and InitialAssessment of Alzheimer’s Disease and Related Dementias” (44) recommendations on takinga focused medical history, and was verified with the 1999 AMA guideline (45) and the 1997University Health System Consortium/U.S. Department of Veterans Affairs guideline. (46)

� Though much of the AHCPR guideline was evidence-based, this particular guideline wasbased on expert opinion due to the lack of hard clinical evidence in this realm. The CMIDementia Guidelines Workgroup reviewed the AHCPR guideline and recommendations inthe context of what the literature shows are the most common symptoms and signs ofdementia, as well as the most common causes of “reversible” dementia. Key components ofmedical history were also defined by CMI Dementia Guidelines Workgroup members asthose that gathered information on components of the DSM-IV definition of dementia (e.g.,history of cognitive/memory problems, gradual functional decline, etc.) The group thenagreed that the above recommendations, with additional detail as to the components of eachpart of the history, would enable primary care providers to gather information needed towork towards or eliminate a diagnosis of dementia.


Problem Formulation 2

Clinical Question: What tool(s)/test(s) should be used for functional status assessment aspart of a history?


To assist primary care physicians and other health care professionals inthe outpatient primary care setting in eliciting information helpful fordiagnosing dementia

Population: Men and women without a diagnosis of dementia

Health Problem: Questionable functional status

Health Intervention: � Functional ActivitiesQuestionnaire (FAQ)

� Katz ADL� No functional assessment




Intervention:

� Identifying level of functional ability


� Patient annoyance


Evidence Search 2

Only observational studies were included that examined the sensitivity and specificity of theFunctional Activities Questionnaire (FAQ) for diagnosing dementia. Because searches on this topic were conducted for the 2002 CMI Dementia Guidelines, updateson those searches were performed. The most recent searches are listed first, followed by theoriginal searches.


TimeFrame:

#Found:

# inET:

New Searches



and RCTs 5/27/03 12 0

PubMed

("sensitivity andspecificity"[MESH] OR"Predictive Value ofTests"[MESH]) AND("Functional ActivitiesQuestionnaire"[TEXT] OR"FAQ"[TEXT] OR "activitiesof daily living"[MESH] OR"functional status"[TEXT])ANDdementia/diagnosis[MESH]

English, Human 2/1/01-8/18/03 16* 0





The methods of the systematic review were rigorous.Old Searches (from 2002 CMI Dementia Guidelines)

PubMed

("sensitivity andspecificity"[MESH] OR"Predictive Value ofTests"[MESH]) AND("Functional ActivitiesQuestionnaire"[TEXT] OR"activities of dailyliving"[MESH] OR"functional status"[TEXT])ANDdementia/diagnosis[MESH]

English, Human 1965-4/13/01 22* 0

Cochrane dementia No Limits 4/12/01 76 0 * None of the studies obtained were more comprehensive reviews of functional assessment tools in general or theFunctional Activities Questionnaire (FAQ) specifically than the AHCPR guideline (see “Other Information Sources”below). No studies published after the search period of the AHCPR guideline examined the sensitivity andspecificity of the FAQ compared to any other tests of functional assessment.






� It included a detailed literature review and meta-analysis of assessments of functionalstatus tests.

� Searches on assessment tools (cognitive and functional) yielded a total of 8,198 records.An additional 961 abstracts were contributed by panel members and other sources, for atotal of 9,159 records.

� Panelists then rated the records to identify those sufficiently useful for the meta-analysis,based on: (a) relevance to assessment of early detection of mild dementia and (b) inclusionof quantitative information. 671 abstracts were retrieved and further evaluated.

� 178 of these 671 abstracts provided data relevant to the meta-analyses; 493 were rejected.The two most frequent reasons for rejection were lack of control or dementia group (40%)and no suitable screening instrument (20%). Review of new studies yielded an additional67 studies- of those, 32 were judged suitable for meta-analysis.

� A total of 210 articles (178 from the original search and 32 of the new studies) wereincluded in the meta-analysis.

� Because this meta-analysis was the most rigorous one found, it was used as the basis forthis recommendation.

� One evidence table was created.� American Medical Association Guideline, “Diagnosis, Management and Treatment of

Dementia – A Practical Guide for Primary Care Physicians” (45)



Recommendations and Rationale Statement 2

Guideline: History - Functional Status Assessment

Recommendations: It is necessary to confirm decline in functional status to make adiagnosis of dementia. It is recommended to use the FunctionalActivities Questionnaire (FAQ) for this assessment, as long as aninformant is present. (Evidence-Based)

If an informant is not initially identified, efforts should be made tofind one. (Consensus)

Methodology: � See specific recommendations � IRGSG-sponsored


� The need to confirm decline in functional status to make a diagnosis of dementia is based onthe DSM-IV criteria for dementia. (47)

� Use of the Functional Activities Questionnaire (FAQ) is based on the recommendation forfunctional status assessment made in the 1996 Agency for Health Care Policy and Research(AHCPR) Clinical Practice Guideline Number 19, “Recognition and Initial Assessment ofAlzheimer’s Disease and Related Dementias.” (44) AHCPR underwent an extensive, detailedliterature review of dementia assessment instruments, including functional status tests. Atotal of 210 articles were included in their meta-analysis.

� Two measures of functional status were included in the meta-analysis – the Katz ADL(Activities of Daily Living) and the FAQ (Functional Activities Questionnaire). The goldstandard used to identify patients with dementia in the studies included in the meta-analysiswas not indicated in the AHCPR Clinical Practice Guideline. The meta-analysis determinedthat the FAQ is a good discriminator of dementia, with an average effect size of 2.46 forpatients with mild-to-severe dementia and comorbid conditions. Effect size is the differencein outcomes between the intervention and control groups divided by the standard deviation.This effect size corresponds to sensitivity and specificity values in the range of about 85-90%. Neither the Katz ADL nor any other ADL or IADL (Instrumental Activities of DailyLiving) scale had data that met the most stringent criteria of the AHCPR meta-analysis.Because the FAQ is a survey given to an informant, not the patient him/herself, it is clearlyuseful only if an informant is present. If an informant is not present, other scales of ADLs orIADLs may be useful.

� Since the search period of the AHCPR guideline no data were found to contradict itsconclusions or point to any other clearly superior functional status test. The 2003 USPSTFsystematic evidence review on screening for dementia acknowledged the FAQ’s sensitivity(citing the AHCPR guideline that the CMI Dementia Guidelines Workgroup used), and thatfew high-quality studies have focused on functional assessment tests in primary care.(27)

� Therefore, the Functional Activities Questionnaire (FAQ) is recommended to assess forfunctional impairment as part of the dementia assessment.


Diagnosis – Physical Exam

Problem Formulation

Clinical Question: What information should be gathered as part of the physical examduring an assessment for dementia?




Health Problem: Determining whether signs are present that would support a diagnosis ofdementia, and if so, gathering information on possible cause

Health Intervention: � Physical examination focusing on identifying signs supporting adiagnosis and suggesting possible causes of dementia symptoms




Intervention:

� Identifying physical characteristics that may help determine cause ofdementia symptoms

� Finding signs that lead to diagnosing and treating reversibledementias/pseudo-dementias


� Incorrect diagnosis


Evidence Search

This literature search looked for observational studies that examined what should be included ina physical examination as part of a diagnostic work-up for dementia. Because searches on this topic were conducted for the 2002 CMI Dementia Guidelines, updateson those searches were performed. The most recent searches are listed first, followed by theoriginal searches.


TimeFrame:

#Found:

# inET:

New Searches



and RCTs 5/27/03 12 0

PubMed"PhysicalExamination"[MeSH] AND"Dementia/diagnosis"[MeSH]

English, Human 7/1/03-8/18/03 35 0

Old Searches (from 2002 CMI Dementia Guidelines)

PubMed"PhysicalExamination"[MESH] ANDdementia/diagnosis[MESH]

English, Human 1965-11/9/01 413 0

Cochrane dementia No Limits 4/12/01 76 0Other Information Sources (from 2002 CMI Dementia Guidelines)� Agency for Health Care Policy and Research (AHCPR) Clinical Practice Guideline Number



� The panel then wrote guideline recommendations based on data from the scientificliterature. When scientific evidence was missing or inconclusive, panel consensus wasused to support the guideline.





Guideline: Physical Exam

Recommendation: A basic physical exam is recommended as part of the assessmentfor dementia, with special attention paid to: � Patient alertness, behavior, and appearance (i.e., hygiene and

affect)� Neurological signs such as abnormal reflexes, gait, balance,

strength, extrapyramidal signs (including rigidity,bradykinesia, cogwheeling, and resting tremor)

� Cardiac and carotid auscultation� Orthostasis


Rationale: (From 2002 CMI Dementia Guidelines)

� A basic physical exam is part of any clinical work-up. Although the dementia assessmentliterature did not focus on the physical exam, and an evidence search on the physical examcomponent did not yield any clinical evidence on the impact of conducting a physical examon health outcomes, the CMI Dementia Guidelines Workgroup felt that a recommendationwas warranted. The recommendation encourages primary care providers to focus on severalkey elements during the physical exam that may help them make a diagnosis of dementia anddistinguish between different etiologies of dementia, based on the DSM-IV (47) definitions ofvarious types of dementia.


Diagnosis – Cognitive Status Assessment

Problem Formulation

Clinical Question: What tool(s)/test(s) should be used for cognitive status assessment aspart of a dementia work-up?




Health Problem: Questionable cognitive status


� Mini-Mental StateExamination (MMSE)

� Blessed Information-Memory-Concentration Test (BIMC)

� Blessed Orientation-Memory-Concentration Test (BOMC)

� Short Test of Mental Status(STMS)

� No cognitive assessment




Intervention:

� Identifying level of cognitive ability


� Misinterpreted results� Patient annoyance


Evidence Search

Only observational studies were included that compared the sensitivity and specificity of theMini-Mental State Examination (MMSE), Blessed Information-Memory-Concentration Test(BIMC), Blessed Orientation-Memory-Concentration Test (BOMC), or Short Test of MentalStatus (STMS) for diagnosing dementia. Because searches on this topic were conducted for the 2002 CMI Dementia Guidelines, updateson those searches were performed. The most recent searches are listed first, followed by theoriginal searches.


TimeFrame:

#Found:

# inET:

New Searches



and RCTs 5/27/03 12 0

PubMed

("sensitivity andspecificity"[MESH] OR"Predictive Value ofTests"[MESH]) AND((MMSE[TEXT] OR "Mini-Mental StateExamination"[TEXT]) OR(“Blessed Information-Memory-ConcentrationTest”[TEXT] OR“BIMC”[TEXT]) OR(“Blessed Orientation-Memory-ConcentrationTest”[TEXT] OR“BOMC”[TEXT]) OR (“ShortTest of MentalStatus”[TEXT] OR“STMS”[TEXT])) ANDdementia/diagnosis[MESH]

English, Human 1/1/01-8/18/03 56 0


Other Information Sources:� In May 2003, the Agency for Healthcare Research and Quality (AHRQ) released a systematic



The methods of the systematic review were rigorous.


TimeFrame:

#Found:

# inET:


PubMed

("sensitivity andspecificity"[MESH] OR"Predictive Value ofTests"[MESH]) AND(MMSE[TEXT] OR "Mini-Mental StateExamination"[TEXT]) ANDdementia/diagnosis[MESH]

English, Human 7/1/95-3/7/01* 68 1


* This search was conducted to find articles on the sensitivity and specificity of the MMSE published after thesearch period and peer review process of the AHCPR meta-analysis (see “Other Information Sources” below).






� It included a detailed literature review and meta-analysis of assessments of brief cognitivestatus tests.

� Searches on assessment tools (cognitive and functional) yielded a total of 8,198 records.An additional 961 abstracts were contributed by panel members and other sources, for atotal of 9,159 records.

� Panelists then rated the records to identify those sufficiently useful for the meta-analysis,based on: (a) relevance to assessment of early detection of mild dementia, and (b)inclusion of quantitative information. 671 abstracts were retrieved and further evaluated.

� 178 of these 671 abstracts provided data relevant to the meta-analyses; 493 were rejected.The two most frequent reasons for rejection were lack of control or dementia group (40%)and no suitable screening instrument (20%). Review of new studies yielded 67 additionalstudies; of those, 32 were judged suitable for meta-analysis.

� A total of 210 articles (178 from the original search and 32 of the new studies) wereincluded in the meta-analysis.

� Because this meta-analysis was the most rigorous one found, it was used as the basis fordeveloping the recommendation and for structuring additional literature searches.

� This meta-analysis is abstracted in an evidence table.� American Medical Association Guideline, “Diagnosis, Management and Treatment of

Dementia – A Practical Guide for Primary Care Physicians” (45)




Guideline: Cognitive Status Assessment

Recommendation 1: Any of the widely studied mental status tests (MMSE, BIMC,BOMC, STMS) are recommended as part of a dementiaassessment.


Rationale:

� The USPSTF systematic evidence review on screening for dementia in primary care (27) usedthe AHCPR Clinical Practice Guideline (detailed below) and additional literature on two ofthe tests of interest (the Mini-Mental State Exam [MMSE] and the Blessed Information-Memory-Concentration Test [BIMC]) to develop the following summary of test accuracy:

Instrument Sensitivity Specificity PPV* NPV*MMSE 71-92% 56-96% 15-72% 95-99%BIMC 90% 65-90% 22-50% 98-99%BOMC 69% 90% 43% 96%STMS 81% 90% 47% 98%

* Based on a dementia prevalence of 10%

� Rationale from the 2002 CMI Dementia Guidelines:� This guideline is based on the recommendation on mental status assessment made in the

1996 Agency for Health Care Policy and Research (AHCPR) Clinical Practice GuidelineNumber 19, “Recognition and Initial Assessment of Alzheimer’s Disease and RelatedDementias.” (44) AHCPR underwent an extensive, detailed literature review of dementiaassessment instruments, including brief mental status tests. A total of 210 articles wereincluded in their meta-analysis. The measure of interest in this meta-analysis was theeffect size for each instrument, which shows the differences in average test scoresbetween cases and controls, divided by a measure of variance. A higher score indicates abetter ability to distinguish case patients from control subjects.

� Taken as a whole, the results of the AHCPR meta-analyses involving the most relevantstudies -- those involving an expanded spectrum of case patients and control subjectswith interfering or comorbid conditions -- indicate that four tests are largely equivalent interms of their ability to differentiate between persons with and without dementia whohave comorbid conditions of a general medical, psychiatric, and/or neurologic nature.These four tests are the Mini-Mental State Exam (MMSE), the Blessed Information-Memory-Concentration Test (BIMC), the Blessed Orientation-Memory-ConcentrationTest (BOMC), and the Short Test of Mental Status (STMS). The effect sizes for theseinstruments are summarized in the table below, and are categorized by the types ofpatients included in the studies (Phase II = moderate to severe dementia in case patients,no comorbidity in control subjects, Phase III = mild to severe dementia in case patients,no comorbidity in control subjects, Phase IV = mild to severe dementia in case patients,


comorbidity in control subjects). In Phase IV studies, it is more difficult to discriminatebetween cases and controls, making these studies more relevant in determining a valuablecognitive assessment tool.

Effect Size by Study TypePhase II Phase III Phase IV

MMSE 4.08 2.40 1.78BIMC 4.72 3.75 2.49BOMC 6.51 3.79 1.63STMS 2.01

� A literature search of the period since the AHCPR guideline search period produced oneadditional article examining the sensitivity, specificity, and diagnostic value of theMMSE compared to any of the tests recommended in the AHCPR guideline. (48) Thisstudy by Stuss et al. found that the MMSE, BIMC, and STMS had comparable sensitivityand specificity and were virtually interchangeable in terms of their use in cognitiveassessment.

� Thus, any of these four mental status questionnaires -- the Mini-Mental State Exam(MMSE), the Blessed Information-Memory-Concentration Test (BIMC), the BlessedOrientation-Memory-Concentration Test (BOMC), or the Short Test of Mental Status(STMS) -- can be used to gather mental status information in the diagnostic work-up fordementia. For the MMSE, a score of 26 or below can be used as a cut-point for milddementia, though for older patients, those with less education, and those for whomEnglish is a second language this cut-point should be lower. Cut points for the other testswere not described in the AHCPR guideline, and given recommendation 2 below, werenot investigated.


Guideline: Cognitive Status Assessment

Recommendation 2: The standard-form MMSE is recommended as the cognitiveassessment tool of choice because it is widely studied and used.

It is strongly recommended that scores be interpreted usingnorms related to age, education, and primary language, that thetest be considered just one piece of information in the assessmentprocess, not a test that can diagnose dementia.


Rationale: (From 2002 CMI Dementia Guidelines, with minor modifications)

� Although the evidence shows that several mental status questionnaires are essentiallyequivalent in their ability to differentiate between persons with and without dementia, theCMI Dementia Guidelines Workgroup agreed that it makes sense to establish someconsistency in the approach used for cognitive assessment throughout Kaiser Permanente.The group agreed that the vast majority of primary care providers are already more familiarwith the Mini-Mental State Exam (MMSE) than any other cognitive test, and that this test ismore widely studied in a variety of settings and populations than any other cognitive test.Therefore, the group recommends that the standard-form MMSE be used as the cognitiveassessment tool of choice. This is supported by the conclusions of the USPSTF systematicreview on screening for dementia, which notes that the MMSE is the most widely studied andcommonly used test, with comparable accuracy to the other tests studied.(27)

� The group also felt it necessary to explicitly note the need for adjusted scoring methods thatconsider age, education, and primary language, as the test has been shown to be biased basedon these characteristics and usual scoring cut-points can be misleading in thesecircumstances. Also, because cognitive functioning is just one aspect of a diagnosis ofdementia, the group wanted to emphasize that the MMSE should be considered along withevery other piece of information gathered during the work-up, not in isolation.


Diagnosis – Diagnostic Tests

Problem Formulation

Clinical Question: What laboratory tests should be performed as part of the dementia work-up?


To assist primary care physicians and other health care professionals inthe outpatient primary care setting in administering lab tests that willelicit information helpful for diagnosing dementia


Health Problem: Determining possible cause of dementia symptoms and identifyingpossibly reversible conditionsAdministration of one of more of the following diagnostic tests as partof the dementia work-up


� Complete blood count (CBC)� Basic chemistries (calcium,glucose, electrolytes, creatinine,potassium)� Thyroid function test� Serum B12 � Methylmalonic acid� Homocysteine� Serology test (for syphilis)

� Liver function tests� Analysis of blood urea nitrogen

(BUN) (kidney function)� HIV test� Lumbar puncture� Copper level (ceruloplasmin

test)� No lab tests




Intervention:

� Finding and treating potentially reversible causes of dementia� Defining an etiology� Establishing baseline chemistries


� Pain and inconvenience oftests



Evidence Search

This literature search looked for observational studies that examined the health outcomes ofperforming particular laboratory tests as part of a diagnostic work-up for dementia. Because searches on this topic were conducted for the 2002 CMI Dementia Guidelines, updateson those searches were performed. The most recent searches are listed first, followed by theoriginal searches.


TimeFrame:

#Found:

# inET:

New Searches



and RCTs 5/27/03 12 0

PubMed

("Diagnostic Tests,Routine"[MESH] OR"laboratory tests"[TEXT])ANDdementia/diagnosis[MESH]

English, Human 2/1/01-7/29/03 11 0

Old Searches (from 2002 CMI Dementia Guidelines)"Diagnostic Tests,Routine"[MESH] ANDdementia/diagnosis[MESH]

No Limits 1965-5/4/01 12 0*

PubMedlaboratory tests ANDdementia No Limits 1965-

5/4/01 88 0*


* These searches resulted in review articles on recommended laboratory tests for the diagnostic work-up ofdementia. Because they were not observational studies that examined the health outcomes of performing certain labtests, they were not included in evidence tables. However, recommendations from several of the review articleswere included in a table summarizing those consensus recommendations as a basis for discussion by the CMIDementia Guidelines Workgroup.


Other Information Sources (from 2002 CMI Dementia Guidelines) � Agency for Health Care Policy and Research (AHCPR) Clinical Practice Guideline Number






CMI Dementia Guidelines DRAFT © 2004 Kaiser Permanente Medical Care Program60 For use within Kaiser Permanente only. 01/04


Guideline: Laboratory Tests

Recommendations: The following laboratory tests are recommended as part of theroutine work-up for suspected dementia: CBC, sodium,creatinine, calcium, glucose, thyroid function tests (TSH first; ifabnormal, then a T4 can be used), and Vitamin B12 levels.

Additional tests are an option if the medical history or physicalsuggest that they are clinically indicated: MHATP or FTA forsyphilis, methylmalonic acid, liver function, HIV, chest X-ray,urinalysis, toxicology screen, and EEG.

Assuming there are no contraindications, lumbar puncture isindicated only when the following are present: cancer, suspicion ofCNS infection, reactive serum antitreponemal syphilis serology,hydrocephalus, dementia in a person under age 55, rapidlyprogressive or unusual dementia, immunosuppression, orsuspicion of CNS vasculitis, particularly in patients withconnective tissue diseases.


Rationale: (From 2002 CMI Dementia Guidelines)

� Literature searches on diagnostic (lab) tests for dementia did not produce good prognosticdata (e.g., with reports of sensitivity and specificity for each lab test in identifying potentiallyreversible causes of dementia or in diagnosing dementia). The CMI Dementia GuidelinesWorkgroup turned to several rigorous review articles and published guidelines (45, 46, 49-55) anddiscussed the reasons for each potential test. Discussion centered on several key issues:What lab tests may reveal reversible causes of dementia symptoms? How frequently doeseach lab test reveal useful information in a dementia work-up? Is there a clinicalreason/justification for each test as part of a dementia work-up? Using these criteria, thegroup’s clinical experience, and summary information from several key articles andguidelines, the group agreed on the aforementioned recommendations.

© 2004 Kaiser Permanente Medical Care Program DRAFT CMI Dementia GuidelinesFor use within Kaiser Permanente only. 01/04 61

Diagnostic Tests Summary (49) (45, 46, 50-55)

Do = these tests should be administered as part of a routine dementia work-upOption = these tests may be helpful in certain circumstances, but are not recommended as routineNot routine = these tests should not be used in routine evaluation of dementia N/a = paper did not report on the use of these tests

AmericanAcademy of

Neurology (49)

NINCDS-ADRDA (50)

Corey-Bloomet al. (51)

CanadianConsensus

Conferences(52, 53)

Siu et al. (54) van Crevelet al. (55)

AMAGuideline (45)

VA/UHCGuideline (46)

Complete blood count do n/a do do do do do doSerum electrolytes do n/a do do do n/a do doCalcium do n/a do do do do do doGlucose do n/a n/a do do do do doBlood urea nitrogen do n/a do n/a do do do doCreatinine do n/a do n/a do n/a do doLiver function tests do n/a do n/a n/a do do doThyroid function tests do n/a do do do do do doVitamin B12 level do n/a do n/a do do do doSyphilis serology do n/a do n/a do do option doSedimentation rate option n/a do n/a do n/a option optionSerum folate level option n/a n/a n/a do do n/a n/aHIV testing option n/a do n/a n/a n/a option optionChest X-ray option n/a n/a n/a n/a n/a option n/aUrinalysis option n/a do n/a do n/a option optionToxicology screen option n/a n/a n/a n/a n/a option optionLumbar puncture not routine n/a * n/a not routine n/a ** *EEG n/a option option n/a do n/a n/a option*Assuming there are no contraindications, lumbar puncture is suggested only when the following are present: cancer, suspicion of CNS infection, reactive serumantitreponemal syphilis serology, hydrocephalus, dementia in a person under age 55, rapidly progressive or unusual dementia, immunosuppression, or suspicion ofCNS vasculitis, particularly in patients with connective tissue diseases.**Lumbar puncture is indicated for patients in whom there is a short duration of symptoms, or evidence of meningitis, fever, metastatic cancer, positive serumfluorescent treponemal antibody absorption.


Diagnosis – Neuroimaging

Problem Formulation

Clinical Question:Should neuroimaging technology be used to rule out potentiallyreversible causes of dementia including tumor, normal pressurehydrocephalus, or subdural hematoma?


To assist primary care physicians and other health care professionals inthe outpatient primary care setting identifying the cause of dementiasymptoms, making a diagnosis, and creating an appropriate care plan


Health Problem: Determining possible cause of dementia symptoms


Administration of one of more of the following neuroimaging tests aspart of the dementia work-up: � CT (computed tomography)� MRI (magnetic resonance imaging) � PET (positron emission tomography) scan� SPECT (single photon emission computed tomography) scan� No neuroimaging test




Intervention:

� Finding and treating tumor, normal pressure hydrocephalus, orsubdural hematoma, leading to improvement of dementia symptomsor lengthening of life


� Pain and inconvenience oftests



Evidence Search

This literature search looked for observational studies or clinical trials that examined the healthoutcomes of performing neuroimaging tests to detect potentially reversible causes of dementia. Because searches on this topic were conducted for the 2002 CMI Dementia Guidelines, updateson those searches were performed. The most recent searches are listed first, followed by theoriginal searches.


TimeFrame:

#Found:

# inET:

New Searches



and RCTs 5/27/03 12 0

Review, English,Human

2/1/01-7/29/03 75 0

PubMed

("Dementia/diagnosis"[MeSH]OR"Dementia/etiology"[MeSH])AND ("Magnetic ResonanceImaging"[MeSH] ORMRI[TEXT] OR "Tomography,X-Ray Computed"[MeSH] OR"Tomography, SpiralComputed"[MeSH] ORCT[TEXT] OR "Tomography,Emission-Computed"[MeSH]OR PET[TEXT] OR"Tomography, Emission-Computed, Single-Photon"[MeSH] ORSPECT[TEXT])


2/1/01-7/29/03 56 0



TimeFrame:

#Found:

# inET:


Review, English,Human

1965-4/20/01 246 1

PubMed

(dementia/diagnosis[MESH]ORdementia/etiology[MESH])AND ("Magnetic ResonanceImaging"[MESH] ORMRI[TEXT] OR"Tomography, X-RayComputed"[MESH] OR"Tomography Scanners, X-Ray Computed"[MESH] ORCT[TEXT])


1965-4/20/01 107 0



Other Information Sources (from 2002 CMI Dementia Guidelines)� Two articles on reversible dementia (56, 57) found during the literature review on differential

diagnosis of depression were used in this analysis.� American Academy of Neurology 2001 Practice Parameter: Management of Dementia (an

evidence-based review) (2)

� Methods: The literature review process included the following search terms: Alzheimer’sdisease, vascular or multi-infarct dementia, dementia with associated parkinsoniandisorder, progressive supranuclear palsy, frontotemporal dementia, and senile dementia,cholinesterase inhibitors, antioxidants, hormones, anti-inflammatory agents/drugs,nootropics, metabolic enhancers, neurotrophic agents/drugs, (complementary) alternativemedicines, treatment, and pharmacotherapy. For information on pharmacotherapies, thefollowing databases were searched: MEDLINE, Embase, Current Contents, PsychAbstracts, and Cochrane databases. Studies selected included the following: randomized,controlled studies in all languages and other types of studies limited to English; humansubjects with N greater than 20, regardless of outcome measured; and review articlespublished between January 1998 and November 1999. Articles were classified based onthe quality of the evidence. After review of the evidence, recommendations were drafted,reviewed by all committee members, and identified as a Practice Standard, Guideline, orOption.

� 1998 Canadian Consensus Conference on Dementia (58)

� Methods: Lead authors were chosen to write background papers that led to consensusstatements. The authors were responsible for literature search, critical review of articles,and preparation of a draft background document. Authors were instructed to use the rulesof evidence developed by the Canadian Task Force on the Periodic Health Examination forpurposes of grading the evidence. After an extensive review process, each conferenceparticipant voted on the recommendations.




Guideline: Neuroimaging

Recommendations: Non-contrast CT scanning is an option for all patients withsuspected dementia, and is recommended if:� Patient is under the age of 65OR� Patient is age 65 or over and has any of the following:

� Atypical presentation/unclear diagnosis� Rapid unexplained deterioration� Unexplained focal neurological signs or symptoms� History of head injury with a temporal relationship to

symptoms� Urinary incontinence or gait ataxia early in the illness� Clinical suspicion of undiagnosed cerebrovascular disease

Contrast CT and MRI are generally not recommended for routineuse in dementia assessment in primary care because in most casesthey offers no advantage over non-contrast CT for ruling outpotentially reversible causes.

PET and SPECT are not recommended in dementia assessment inprimary care. It is recommended that a specialist be consulted ifquestions arise about the usefulness of PET or SPECT inindividual cases.


Rationale: (From 2002 CMI Dementia Guidelines with minor modifications) Supporting Evidence for CT� The question of neuroimaging with a CT scan in the diagnosis of dementia was addressed by

examining the causes of potentially reversible dementia that can be detected by CT scan.These include brain tumor (neoplasm), normal pressure hydrocephalus (NPH), and subduralhematoma. The literature was then examined to seek answers to the following questions: � How often is each of these conditions the cause of dementia (prevalence of each

condition in the dementia population)?� How often can each of these conditions be detected without use of CT scan (i.e., accuracy

when using just symptom/clinical presentation)?� How often is CT scan able to detect each of these conditions if present (sensitivity,

specificity of CT)?� Does treatment of each of these conditions leads to improvement of dementia symptoms

(does the treatment improve outcomes)?


� Two systematic reviews (56, 57) that examined the literature on the frequency that brain tumor,NPH, and subdural hematoma are the cause of dementia were found and reviewed. � Clarfield (56) examined 32 studies (2889 subjects) from 1966 to 1987 that investigated the

prevalence of the various causes of dementia. This review found that, of all the causes ofdementia, neoplasm represented 1.5%, NPH represented 1.6%, and subdural hematomarepresented 0.4%. For the 11 studies in which follow-up data were provided, 8% ofpatients reversed partially and 3% completely. Of those 103 cases that reversed, 10.7%were caused by NPH, 5.8% were caused by subdural hematoma, and 4.0% were causedby neoplasm.

� Weytingh (57) carried out a quantitative review of 16 studies (1551 patients) reportedbetween 1972 and 1994 in which reversible dementia was diagnosed and outcome aftertreatment was assessed. An average of 15.2% of the dementia cases were potentiallyreversible. 9.3% partially reversed and 1.5% fully reversed. Of these 168 cases reportedwith fully and partially reversed dementia, 9.5% were caused by NPH, 6.5% were causedby neoplasm, and 6.0% were caused by subdural hematoma.

� One systematic review of the use of CT scanning in dementia was also found andreviewed.(59) In this study, Foster et al. used Clarfield’s prevalence data and assessed howoften treatment of a condition found by CT scan would result in improved health outcomes.This study concluded the following: � NPH: Occurs in less than 2% of dementia cases, and half of all cases are in patients

under age 60. Good evidence for effectiveness of shunting is lacking.� Tumors: Rare cause of dementia, occurring in 1-4% of cases. Most brain tumors are

malignant. Survival rates, even with treatment, are poor.� Subdural hematoma: Can respond well to surgery, especially if the dementia has been

of short duration. But, is the rarest of the potentially reversible conditions, occurring inonly 0.4% of demented patients. Patients usually have a suggestive history of headinjury, tend to be under 70, have significant focal neurological signs and symptoms, andduration of symptoms is less than 6 months. Therefore, evidence of subdural hematomacan be gathered through the medical history and clinical exam.

� Based on the prevalence data presented by Clarfield and Weytingh, and the health outcomesdata summarized by Foster, the CMI Dementia Guidelines Workgroup agreed that CTscanning is unlikely to benefit all patients who present with signs of dementia. Theconditions that could be detected by CT scan (NPH, neoplasm, and subdural hematoma)often have other characteristics (e.g., mainly occur in certain age groups; can present withspecific signs and symptoms). Therefore, more selective CT scanning can be undertakenusing the strategy outlined in the recommendation. None of the evidence presented discussedthe use of contrast CT, and the Workgroup agreed that non-contrast CT is preferred in mostcircumstances.

Supporting Evidence for MRI� Literature search on the use of MRI in the diagnosis of dementia did not produce any

compelling evidence of its superiority over CT scan in detecting normal pressurehydrocephalus (NPH), brain tumor, or subdural hematoma.


Supporting Evidence for PET and SPECT� PET and SPECT technologies are not readily available in Kaiser Permanente facilities and

use of these tools would not happen in the context of primary care. Therefore the CMIDementia Guidelines Workgroup consulted national guidelines rather than conducting an in-depth review of these technologies. The American Academy of Neurology’s (AAN) PracticeParameter: Diagnosis of Dementia (2001), an evidence-based review, includedrecommendations on the use of PET and SPECT. The AAN review found that the sensitivityof SPECT in the diagnosis of Alzheimer’s disease (AD) was lower than that of clinicaldiagnosis. They also found that PET scanning appears to have some promise for use as anadjunct to clinical diagnosis, but further prospective studies with PET are needed to establishthe value it brings to diagnosis over and above a competent clinical diagnosis. After thepublication of the AAN guidelines, a study by Silverman et al. demonstrated that regionalbrain metabolism as assessed by PET was a sensitive indicator of Alzheimer’s disease and ofneurodegenerative disease in general, using clinical diagnosis as the gold standard for some,and histopathological diagnosis on autopsy as the gold standard for others.(60) Given theemerging nature of this area, the lack of widespread availability of the technology in KP, andthe extent of information that can be gained without PET and SPECT, the CMI DementiaGuidelines Workgroup felt that there is insufficient evidence to support the use of thesetechnologies in the primary care setting for the diagnosis of dementia.


Pharmacological Treatment of Dementia

Problem Formulation

Clinical Question: How should pharmacological agents be used to treat cognitive andfunctional decline associated with dementia?


To assist primary care physicians and other health care professionals inthe outpatient primary care setting in providing effective and appropriatepharmacological treatment of dementia

Population: Men and women with diagnosed dementia

Health Problem: Cognitive and functional loss associated with dementia


� Donepezil� Rivastigmine� Galantamine� Estrogen� Estrogen plus progestin (HRT)

� NSAIDs� Cox-2 inhibitors� Ginkgo biloba� Vitamin E� Statins� No treatment




Intervention:

� Cognitive ability� Functional ability

� Patient/family satisfaction� Time to institutionalization� Caregiver time/burden


� Drug adverse effects

Measures ofOutcomes:

� Global caregiver impression� Score on cognitive scales

� Score on functional scales� Score on behavior scales


Evidence Search

Only RCTs, meta-analyses, or systematic reviews were included that studied the use ofacetylcholinesterase inhibitors (donepezil, galantamine, and rivastigmine), estrogen or estrogenplus progestin (HRT), NSAIDs, Cox-2 inhibitors, Ginkgo biloba, vitamin E, or statins (comparedto placebo or to each other) to improve cognition and/or function in people with dementia. Searches for meta-analyses had no time limits, while searches for RCTs went back to April 1,2001 to update the searches conducted for the 2002 CMI Dementia Guidelines on this topic.Some meta-analyses found had already been included in those earlier Guidelines so were notincluded again here. The most recent searches are listed first, followed by the original searches.


TimeFrame:

#Found:

# inET:

New Searches

Cochrane dementia Systematic reviews 5/27/03 104 3*


and RCTs 5/27/03 12 0**


1965 –6/18/03 22

1 (othersalreadyfound inCochranesearch)

PubMed

("Alzheimer Disease"[MESH]OR "Dementia"[MeSH])AND ("CholinesteraseInhibitors"[MeSH] OR"donepezil"[TEXT] OR"rivastigmine"[TEXT] OR"galantamine"[TEXT] OR"Hormone ReplacementTherapy"[MeSH] OR"Estrogens"[MeSH] OR"estrogen"[TEXT] OR "Anti-Inflammatory Agents, Non-Steroidal"[MeSH] OR"NSAID"[TEXT] OR"Ginkgo biloba"[MeSH] OR"ginkgo"[TEXT] OR"gingko"[TEXT] OR"Vitamin E"[MeSH] OR"Hydroxymethylglutaryl-CoAReductase Inhibitors"[MeSH]OR "statin"[TEXT])


4/1/01 –6/18/03 48 13



TimeFrame:

#Found:

# inET:

PubMed

(("AlzheimerDisease"[MeSH] OR"Dementia"[MeSH]) AND("rofecoxib"[TEXT] OR"celecoxib"[TEXT]))


1965-9/16/03 1

1(alreadyincludedin searchabove)

Other Information Sources� Two RCTs (61, 62) on use of donepezil in vascular dementia were published after our search

period, with full details of results already reported in an earlier article (63) included in ourreview. Brief descriptions of these two studies are included in our rationale statement.

� In May 2003, the Agency for Healthcare Research and Quality (AHRQ) released a systematicevidence review conducted for the U.S. Preventive Services Task Force (USPSTF) on screeningfor dementia. (27) This updated a 1996 review where the USPSTF found insufficient evidence torecommend for or against screening for dementia.(28) The new review was conducted toconsider more recent studies (literature published up to September 1, 2002) concerningscreening tests as well as both pharmacologic and caregiver interventions. The systematicreview of the literature sought to answer nine key questions:1. What is the direct effect of dementia screening on outcomes?2. How common is undiagnosed dementia?3. How accurate are the screening tests?4. How effective is primary treatment of potentially reversible or irreversible dementia?5. How effective is secondary treatment for dementia (pharmacologic and nonpharmacologic


The methods of the systematic review were rigorous. * 3 additional relevant Cochrane reviews were found, but all had already been included in 2002 CMI DementiaGuidelines. 1 additional review was found (on ibuprofen) but no evidence table was created, as the review found noRCTs or placebo controlled trials. ** Information from the Clinical Evidence review was not included in evidence tables but was reviewed to ensurethat our literature search captured all relevant articles published since the original search dates of the 2002 CMIDementia Guidelines.



TimeFrame:

#Found:

# inET:


Cochrane dementia No limits 4/12/01 76 0

PubMed

(("Hormone ReplacementTherapy"[MESH] OREstrogens[MESH]) AND(dementia[MESH] OR"AlzheimerDisease"[MESH]))

RandomizedControlled Trials

1965-5/29/01 12 3

PubMed

(("AlzheimerDisease"[MESH] ORdementia[MESH]) AND("Anti-Inflammatory Agents,Non-Steroidal"[MESH] OR"anti-inflammatoryagents"[MESH] ORNSAID[TEXT]))


PubMed

(("Alzheimerdisease"[MESH] ORdementia[MESH] OR"Alzheimer disease"[TW] ORdementia[TW] ORcognition[TW]) AND("ginkgo biloba"[MESH] ORginkgo[TW] OR gingko[TW]OR "EGb 761"[TW] ORTanakan[TW] ORRokan[TW] ORTebonin[TW] ORGinkoba[TW]))

English, Human 1965-6/1/01 70 3



TimeFrame:

#Found:

# inET:

PubMed

(("AlzheimerDisease"[MESH] OR"Alzheimer's Disease"[TEXT]OR dementia[MESH] ORdementia[TEXT]) AND("vitamin E"[MESH] OR"vitamin E"[TEXT]))


PubMed

((lovastatin[MESH] ORstatins[TEXT]) AND(dementia[MESH] OR"AlzheimerDisease"[MESH]))

English, Human 1965-5/28/01 13 0

Other Information Sources (from 2002 CMI Dementia Guidelines)� American Academy of Neurology 2001 Practice Parameter: Management of Dementia (an

evidence-based review) (2)

� Methods: The literature review process included the following search terms: Alzheimer’sdisease, vascular or multi-infarct dementia, dementia with associated parkinsonian disorder,progressive supranuclear palsy, frontotemporal dementia, and senile dementia,cholinesterase inhibitors, antioxidants, hormones, anti-inflammatory agents/drugs,nootropics, metabolic enhancers, neurotrophic agents/drugs, (complementary) alternativemedicines, treatment, and pharmacotherapy. For information on pharmacotherapies, thefollowing databases were searched: MEDLINE, Embase, Current Contents, PsychAbstracts, and Cochrane databases. Studies selected included the following: randomized,controlled studies in all languages and other types of studies limited to English; humansubjects with N greater than 20, regardless of outcome measured; and review articlespublished between January 1998 and November 1999. Articles were classified based on thequality of the evidence. After review of the evidence, recommendations were drafted,reviewed by all committee members, and identified as a Practice Standard, Guideline, orOption.



Guideline: Pharmacological treatment to impact cognition or function indementia

Recommendation 1: For mild to moderate Alzheimer’s disease (AD), a trial of anacetylcholinesterase inhibitor (donepezil, rivastigmine, andgalantamine) is an option and should be considered. While thereis no evidence of differences in efficacy, donepezil is preferredover the other two agents due to its favorable dosing schedule andmoderate adverse effect profile.

For severe AD or vascular dementia, donepezil is an option.(Note: This is not a FDA-approved indication.)

For “moderately severe” disease and AD with hypertension,rivastigmine is an option. (Note: This is not a FDA-approvedindication.)

There is no published evidence to support or discourage a trial ofa different acetylcholinesterase inhibitor if the first agent fails.

Note: Formulary status of these agents varies from Region toRegion – check with Pharmacy.


Recommendation 2: Evaluate patients being treated with acetylcholinesteraseinhibitors and memantine 8-12 weeks after initiation of therapy,then every 3-6 months. Recommended aspects of evaluationinclude: � Assessment of compliance� Assessment of significant side effects � Assessment of cognition� Assessment of function� A discussion with patient and/or caregiver to assess whether

the benefit of therapy outweighs the burden enough to justifycontinued use.



Recommendations3-4:

There is insufficient evidence to recommend the use of vitamin E(2000 IU per day) or Ginkgo biloba for the treatment of cognitionand function in Alzheimer’s disease and other dementias.

Estrogen, estrogen plus progestin, non-steroidal anti-inflammatory drugs (NSAIDs), and Cox-2 inhibitors are notrecommended for the treatment of cognition and function inAlzheimer’s disease and other dementias.


Rationale: Supporting Evidence for Acetylcholinesterase Inhibitors (AChEIs)

Donepezil versus placebo: � Two double-blind randomized controlled trials (64, 65) and one survival-to-endpoint

analysis (66) examined the effect on cognition, function, and global impression ofdonepezil versus placebo in people with Alzheimer’s disease (AD). All three of thesestudies were supported or funded Pfizer, Inc. and Easai, Inc., manufacturers of donepezil.� Winblad et al. randomized 286 patients with mild to moderate AD (MMSE score 10-

26, inclusive) to 52 weeks of donepezil 5 mg/day (increased to 10 mg/day after 4weeks) or placebo. (64) This study produced mixed results, as the more conservativeLOCF (last observation carried forward) analysis on the ITT (intention-to-treat)population showed no significant difference between donepezil and placebo on aglobal assessment (p = 0.054) or on 3 of the 4 components of that assessment, but didshow a significant difference on change in cognitive and functional scales.

� Feldman et al. randomized 290 patients with moderate to severe AD (MMSE score5-17 inclusive) to treatment with donepezil 5 mg/day (increased to 10 mg/day after 4weeks) or placebo for 24 weeks. (65) Primary outcome of clinician’s impression ofchange showed a statistically significant difference in favor of donepezil. Allsecondary outcomes (cognition, function, behaviors, and global function) also favoreddonepezil. These scores all had worsening trends by the endpoint of the study. Therate of discontinuation due to adverse events was higher in the donepezil group, butthe statistical significance was not reported.

� Mohs et al. randomized 431 patients with mild to moderate AD (MMSE score 12-20inclusive) to either donepezil 5 mg/day (increased to 10 mg/day after 4 weeks) orplacebo. (66) The endpoint of interest was clinically evident decline in functionalstatus (with parameters around activities of daily living [ADLs], instrumentalactivities of daily living [IADLs], and global Clinical Dementia Rating), at whichpoint patients were removed from the study. At the conclusion of the one-year study,only 111 patients of the original 431 remained. In comparing the two treatmentgroups, the study showed that the donepezil group reached the endpoint of clinicallyevident functional decline an average of 5 months later than the placebo group (p =0.0051). Secondary outcomes of function, Clinical Dementia Rating, and cognitionfavored donepezil at some of the earlier measurement points, but these advantages


were not sustained over time. The rate of discontinuation due to adverse events washigher in the donepezil group (10.7%) than in the placebo group (7.4%), butsignificance levels were not reported. The number of serious adverse events was alsohigher in the donepezil group (12.1%) versus placebo (8.8%); significance notreported.

� Four other papers examined the use of donepezil in AD with or without cerebrovasculardisease (CVD) (67) and in vascular dementia (VaD).(61-63)

� Tariot et al. studied the use of donepezil 5 mg/day (increased to 10 mg/day after 4weeks) or placebo in 208 nursing home patients with AD with or without CVD. (67)

These patients had mild to severe disease, with MMSE scores ranging from 5 to 26,inclusive. The primary outcome, change from baseline on the NeuropsychiatricInventory-Nursing Home version, showed more improvement in the placebo groupthan in the donepezil group, but this difference did not reach statistical significance(exact p value not reported). The donepezil group did show significant improvementcompared to placebo on the Clinical Dementia Rating at endpoint and on the MMSEat early measurement points only (not at endpoint). The rate of discontinuation due toadverse events was higher in the placebo group (18%) than in the donepezil group(11%) (p not reported), though the incidence of gastrointestinal adverse events wasmore common in the donepezil group. Because the nursing home population is olderand sicker then the population with AD in the community, these results may not begeneralizable to the community population.

� Pratt et al. combined the results of 2 double-blind RCTs with identical protocolsstudying use of donepezil in probable vascular dementia (VaD). (63) The first authoris the Senior Director of Clinical Research for Eisai, Inc., manufacturer of donepezil.The combined data on 893 patients randomized to donepezil 5 mg/day or 10 mg/day(5 mg/day for first 4 weeks; 10 mg/day thereafter) for 24 weeks showed statisticallysignificant improvement on the ADAS-Cog and the MMSE compared to placebo,though differences were small and all groups improved from baseline. Clinical andcaregiver global impression measures significantly favored donepezil 5 mg/day overplacebo, but not 10 mg/day. The rate of discontinuation due to adverse events washigher in the donepezil 10 mg/day group (19.9%) compared to both 5 mg/day (9.8%)and placebo (9.0%), but significance was not reported and adverse events were notdiscussed in the article.� The full results of the 2 RCTs described by Pratt et al. were also published

individually. (61, 62) These write-ups included the full study populations of bothpossible and probable VaD and the results described were consistent with thecombined results highlighted in Pratt et al.

� An integrated analysis of safety data from 4 donepezil trials was also reviewed. (68) Thefirst author is the Senior Director of Clinical Research for Eisai, Inc., manufacturer ofdonepezil. This analysis included several phase II and phase III trials of 12-24 weeks oftreatment with donepezil (various doses) versus placebo in a total of 1,920 patients withAlzheimer’s disease (MMSE score 10-26). Overall, the analysis found that donepezil 5mg/day was comparable to placebo in terms of serious adverse events and events that ledto withdrawal from the studies, though patients on this dose reported diarrhea, anorexia,and muscle cramps significantly more frequently than those on placebo. The 10 mg/daydose was related to higher likelihood of mild and serious adverse events compared to


placebo, and higher rates of withdrawal from the studies (statistical significance was notreported).

� Conclusion: There is evidence of efficacy of donepezil in people with mild to moderateAlzheimer’s disease in producing improvement or less decline during time-limitedstudies in measures of cognition, function, and global impression. There is someevidence of effectiveness in severe AD as well (with continued decline and narrowingdifferences compared to placebo over time). There is some recent favorable evidence ofdonepezil in vascular dementia, though differences were small and the author is anemployee of the company that manufactures donepezil. Overall, donepezil 5 mg/dayappears to be better tolerated than 10 mg/day, and side effects are comparable toplacebo except for higher rates of gastrointestinal side effects.

Rivastigmine versus placebo: � Doraiswamy et al. studied the use of rivastigmine for the treatment of “moderately

severe” AD through a post-hoc subgroup analysis (69) of an earlier RCT (70) and open-label extension. (71) The subgroup of interest consisted of 158 people with GlobalDeterioration Scale score of 5 or above at baseline, indicating “moderately severe”disease. The analysis found that the rivastigmine 1-4 mg/day and 6-12 mg/day groupsshowed significant improvement (p � 0.001) in cognition (as measured by the ADAS-Cog) versus placebo at 26 weeks. This was sustained only for the higher dose (p = 0.035)by week 52 (during the open-label extension period, during which the original placebogroup was started on rivastigmine). Adverse events were not discussed in detail, thoughit was noted that during the open-label phase the most common events weregastrointestinal (nausea, vomiting, anorexia).

� Erkinjuntti et al. examined the use of rivastigmine in people with AD and vascular riskfactors (hypertension) through a post-hoc subgroup analysis (72) of 178 patients from anearlier RCT. (73) The group randomized to rivastigmine 6-12 mg/day did significantlybetter on primary outcomes (ADAS-Cog [p = 0.047] and a clinician impression of changescale, CIBIC-plus [p = 0.004]) compared to placebo, but the lower dose group(rivastigmine 1-4 mg/day) did not. The higher-dose group reported gastrointestinaladverse events significantly more frequently than the lower-dose and placebo groups(significance not reported). Two of the authors work for Novartis, manufacturer ofrivastigmine, and the original study and this analysis were funded by Novartis.

� Conclusion: There is evidence of efficacy of rivastigmine in people with mild to“moderately severe” Alzheimer’s disease in producing improvement or less declineduring time-limited studies in measures of cognition, function, and global impressionversus placebo, with declining effect over time. There is also some evidence ofeffectiveness in people with AD and hypertension. Adverse events are significantlyhigher with higher doses of rivastigmine (6-12 mg/day) compared to lower doses andplacebo, but the higher doses are related to significantly greater improvement relativeto placebo or baseline while the lower doses are not.


Galantamine versus placebo: � The Cochrane Collaboration conducted a meta-analysis of 6 RCTs of galantamine for the

treatment of mild to moderate Alzheimer’s disease (AD). (74) The analysis foundoverall significant positive effects of galantamine versus placebo on global rating at 3months (all doses) and 6 months (all doses except 8 mg/day), and on ADAS-Cog at 3 and6 months (all doses). Gastrointestinal adverse events were most common in galantaminegroups, and were significantly higher than placebo for all doses of galantamine except 8mg/day. The small number of trials in this meta-analysis limited the power of subgroupanalyses.

� Erkinjuntti et al. conducted a 6-month double-blind RCT of galantamine 24 mg/dayversus placebo in 592 patients with probable vascular dementia (VaD) or possible ADwith evidence of cerebrovascular disease (CVD) of mild to moderate severity. (75)

This study was sponsored by Janssen Research Foundation, sister company to themanufacturer of galantamine. For the total study population, the galantamine groupshowed significant benefits in cognition (ADAS-Cog, p � 0.0001) and global impressionof change (CIBIC-plus, p = 0.001) compared to placebo. In subgroup analyses, thesedifferences favoring galantamine were maintained in the AD with CVD group, but not inthe VaD group. The study was not powered for these subgroup analyses, so it is possiblethat the VaD subgroup was too small for the change to be detected. The rate ofdiscontinuation due to adverse events was higher with galantamine (20%) compared toplacebo (8%) (significance not reported).

� Conclusion: There is evidence of efficacy of galantamine in people with mild tomoderate Alzheimer’s disease in producing improvement or less decline during time-limited studies in measures of cognition and global impression. GI side effects anddiscontinuation due to adverse events was higher with galantamine than with placebofor all but the lowest doses. There is insufficient evidence regarding efficacy ofgalantamine in vascular dementia.

One AChEI versus another (head-to-head trials):� An open-label randomized clinical trial by Wilkinson et al. comparing donepezil (up to

10 mg qd) and rivastigmine (up to 6 mg bid) in 111 patients over 12 weeks is the onlyhead-to-head study of any of the three AChEIs of interest. (76) The study found that thetwo drugs showed similar small favorable impact on cognitive measures (ADAS-Cog andMMSE), while donepezil was significantly favored in measures of clinician and caregiversatisfaction and ease of use. In addition, the donepezil group reported fewer adverseevents and had fewer discontinuations due to adverse events (p not reported). Two of thestudy authors work for Pfizer, Inc. and Easai, Inc., manufacturers of donepezil.

� Conclusion: The one head-to-head study available shows comparable efficacy betweendonepezil and rivastigmine, with greater satisfaction and ease of use and fewer adverseevents with donepezil. Two of the study authors work for the manufacturers ofdonepezil.


AChEIs examined collectively � The USPSTF systematic evidence review on screening for dementia included the key

question, “Do pharmacologic interventions of potentially reversible or irreversible dementiaimprove outcomes?”(27) Their conclusions for cholinesterase inhibitors (including donepezil,rivastigmine, galantamine, and tacrine) were that most studies showed significant positiveeffect of the drugs on cognition and global impression, but mixed results for functionalmeasures.

� Rationale from the 2002 CMI Dementia Guidelines:� For mild to moderate Alzheimer’s disease:

� The Health Technology Assessment (HTA) review of donepezil, rivastigmine, andgalantamine provided a thorough summary of the evidence published through January2000. (77) This systematic review examined well-designed systematic reviews andrandomized controlled trials (RCTs) on the effects of donepezil, rivastigmine, andgalantamine compared to placebo or another comparator. The HTA reviewsummarized information from three systematic reviews (78-80) and five publishedRCTs (81-85) of donepezil, three systematic reviews (79, 80, 86) and five published RCTs(70, 73, 87-89) of rivastigmine, and three RCTs (90-92) of galantamine. The primary healthoutcomes addressed included cognitive outcomes measured by ADAS-cog, and globaloutcomes measured by Clinician’s Global Impression of Change (CGIC) and CIBIC-plus. Secondary outcome measures included Mini-Mental State Exam (MMSE) scoreand quality of life indicators.

� Most of the studies showed some advantage of the acetylcholinesterase inhibitor overplacebo:� The three systematic reviews of donepezil reported significant improvements in

global outcome measures, and three of the RCTs of donepezil also demonstratedsignificant improvements in global outcomes. Four of the five RCTs of donepezilshowed significant changes in cognitive function using the ADAS-cog outcomemeasurement scale. Improvements on these scales are small and may not beclinically significant.

� All RCTs of rivastigmine that reported global outcome measurements showedsignificant improvements from baseline for higher doses of rivastigminecompared with placebo. Among the four published RCTs of rivastigmine thatincluded cognitive outcome measures, one reported statistically significantimprovement on the ADAS-cog scale for 9.7 mg/day rivastigmine compared withplacebo. Two others showed non-significant improvements on the ADAS-cogand the fourth showed no significant differences on the MMSE.

� Of the 3 RCTs published on galantamine, 2 were of better quality, and theyshowed significant improvements in cognitive and global function for higherdoses of galantamine (12 mg bid) compared to placebo.

� The three studies published after the search period of the HTA review (90, 93, 94)

showed similar results to those summarized in the HTA review and did not change theconclusions drawn from that review.

� The evidence to date has shown some advantage of three acetylcholinesteraseinhibitors (donepezil, rivastigmine, and galantamine) over placebo in patientsdiagnosed with Alzheimer’s disease in terms of improvement or slowed decline on


certain cognitive and global outcome measures. Changes on these scales may not beclinically significant.

� For Dementia with Lewy Bodies:� A literature search on the use of acetylcholinesterase inhibitors in patients

diagnosed with dementia with Lewy Bodies (a.k.a. Lewy Body disease)produced only one prospective clinical trial with more than 20 patients takingpart. (95)

� This study showed some improvement in neuropsychiatric symptomscommonly associated with dementia with Lewy Bodies (i.e., apathy,delusions, and hallucinations) in patients receiving rivastigmine. Theintention-to-treat analysis of a 4-item subset of the neuropsychiatric inventory(NPI-4) did not show significant differences between the treatment andcontrol groups, while the observed cases analysis did show significantdifferences. An assessment of clinical global change (Clinical GlobalChange-Plus) showed non-significant advantages of rivastigmine.

� Based on this study, there is currently insufficient evidence to support the useof acetylcholinesterase inhibitors to treat dementia with Lewy Bodies.Additional well-designed, prospective trials testing cognitive, functional, andbehavioral outcomes are needed.

� Overall conclusion: Each of the acetylcholinesterase inhibitors reviewed (donepezil,rivastigmine, and galantamine) have demonstrated some efficacy over placebo, howeverthe CMI Dementia Guidelines Workgroup felt that these agents should be an option(rather than recommended) due to several factors: small magnitude of benefitdemonstrated in most studies; unknown clinical significance of statistically significantchanges on measures of cognition and function in a study setting; side effects (especiallywith higher, more effective doses); and substantial cost of these agents with sometimesindeterminate benefit. While these agents are an option and should be considered, thestrength of evidence of benefit balanced with all of these factors led the CMI DementiaGuidelines Workgroup away from recommending these drugs for all patients withdementia.

Supporting Evidence for Estrogen and Estrogen plus Progestin (HRT)Estrogen/Estrogen plus Progestin (HRT) versus placebo: � The Cochrane Collaboration conducted a meta-analysis on the impact of estrogen therapy

and estrogen plus progestin (hormone replacement therapy, or HRT) on cognition andglobal change in women with mild to moderate Alzheimer’s disease (AD) and otherdementias in five randomized controlled trials (RCTs). (96) Three of these trials werereviewed as part of the 2002 CMI Dementia Guidelines (see below). The meta-analysisfound no positive effect beyond 2 months of estrogen or estrogen plus progestin onmeasures of global functioning, language tests, memory tests, informationprocessing/concentration tests, and clinical impression of change.

� A small study by Asthana et al. was not included in the Hogervorst analysis because itwas published after the search date. (97) That study, a double-blind RCT, examined theimpact of 8 weeks of treatment with an estradiol skin patch (0.10 mg/day) versus placeboin 20 women with mild to moderate Alzheimer’s disease. The study suggests that short-


term use of the estradiol patch is associated with statistically significant improvement inprimary outcomes of attention (p = 0.02), verbal memory (p = 0.049), and visual memory(p = 0.03) compared to placebo. However, the estradiol patch did not demonstratesuperiority over placebo on multiple measures of global impression of change, cognition,mood and psychiatric symptoms, and function. Due to the small sample size and thechoice of primary outcomes based on positive results from an earlier study, the results ofthis study are insufficient to recommend use of an estradiol skin patch for treatment ofAlzheimer’s disease.

� The USPSTF systematic evidence review on screening for dementia included the keyquestion, “Do pharmacologic interventions of potentially reversible or irreversibledementia improve outcomes?”(27) Their conclusion for estrogen was that there was noevidence of clinical benefit for estrogen therapy for women with mild to moderatedementia.

� Rationale from the 2002 CMI Dementia Guidelines: � Three double-blind, randomized controlled trials on the effect of estrogen treatment in

patients with Alzheimer’s disease (AD) were found. (98-100) Each study compared theuse of conjugated equine estrogens (CEE) (oral Premarin 0.625-1.25 mg/day) toplacebo in women 60 years of age and above with probable Alzheimer’s disease.Henderson (98) tested cognitive skill, overall global change, and functional status after16 weeks of therapy; Wang (99) examined cognitive skill, general clinical status after12 weeks; and Mulnard (100) tested global change, cognitive function, and othersecondary measures after 12 months of therapy.

� None of the 3 studies showed favorable statistically significant differences betweenthe treatment and control groups on any of the outcomes measured. Wang showedthat adverse events were similar between the two groups, except for vaginal bleeding,which occurred in 44% of women in the treatment group.

� The literature to date does not demonstrate any benefits of estrogen for treatment ofAD.

� Conclusion: Estrogen alone and estrogen plus progestin (HRT) are not recommendedfor the treatment of Alzheimer’s disease or other dementias.

Supporting Evidence for Non-Steroidal Anti-Inflammatory Drugs (NSAIDs) andCyclooxygenase-2 inhibitors (Cox-2 inhibitors)

NSAIDs or Cox-2 inhibitors versus placebo: � A Cochrane Systematic Review published in February 2003 found no completed

randomized controlled trials (RCTs) or placebo-controlled trials on whether ibuprofenprovides benefit for patients diagnosed with Alzheimer’s disease. (101)

� One new randomized controlled trial examining the use of a non-steroidal anti-inflammatory drug (NSAID) and a Cyclooxygenase-2 inhibitor (Cox-2 inhibitor) fortreatment of Alzheimer’s disease (AD) has been published since the search periods of the2002 CMI Dementia Guidelines and the Cochrane Review. (102) Aisen et al. compared theeffectiveness of naproxen sodium 220 mg bid (an NSAID) and rofecoxib 25 mg qd (aCox-2 inhibitor) with placebo for the treatment of mild to moderate AD in 351 patients.This study found no benefit of naproxen vs. placebo (p = 0.96) or rofecoxib vs. placebo(p = 0.09) on the primary outcome of interest, the Alzheimer’s Disease AssessmentScale-cognitive (ADAS-Cog). There was also no benefit with either drug on secondary


measures of dementia rating, activities of daily living, behaviors, or quality of life. Thenumber of adverse events (serious and mild) was higher with both active treatmentscompared to placebo, but this did not reach statistical significance.

� The USPSTF systematic evidence review on screening for dementia included the keyquestion, “Do pharmacologic interventions of potentially reversible or irreversibledementia improve outcomes?”(27) Their evidence review for NSAIDs included Aisen etal. and Scharf et al., which were also included in our review. Their conclusion was thatthese drugs had no effect on the cognitive, physical, global, or behavioral outcomes thatwere measured.

� Rationale from the 2002 CMI Dementia Guidelines:� Two small, double-blind, randomized controlled trials examining the effect of NSAID

therapy on various tests of cognition and function in patients with diagnosedAlzheimer’s disease (AD) were found and reviewed. (103, 104) Two non-steroidal anti-inflammatory agents were used in these trials. Scharf et al. (103) looked at use of acombination of diclofenac (an NSAID) with misoprostol (a prostaglandin) comparedto placebo for 25 weeks in patients age 50 and above with a diagnosis of mild tomoderate AD according to DSM-IV criteria. Rogers et al. (104) examined the effectsof treatment with indomethacin compared to placebo for 6 months in patients with adiagnosis of probable AD according to NINCDS-ADRDA criteria.

� Both studies compared the treatment and placebo groups in terms of mean changefrom baseline on tests of cognition and function. Neither study found statisticallysignificant differences between the treatment and placebo groups on any of theseindividual tests. Both studies found that withdrawal due to side effects (primarilyGI/abdominal symptoms) was greater in the NSAID treatment group.

� Based on these two trials, the only ones published to date, NSAIDs can not berecommended for the treatment of AD.

� Conclusion: Non-steroidal anti-inflammatory drugs (NSAIDs) and Cyclooxygenase-2inhibitors (Cox-2 inhibitors) are not recommended for the treatment of Alzheimer’sdisease and other dementias.

Supporting Evidence for Ginkgo bilobaGinkgo biloba versus placebo: � The Cochrane Collaboration conducted a meta-analysis on the use of Ginkgo biloba for

the treatment of cognitive impairment and dementia. (105) In this analysis, Birks et al.included 33 double-blind randomized controlled trials that examined the use of extracts ofGinkgo biloba (any strength) versus placebo for the treatment of cognitive impairmentincluding dementia (any severity). The duration of the studies ranged from 3 to 52weeks, with the majority being 12 weeks long. Using data from “completers” only (thosewith data at endpoint of the study), the meta-analysis found improvement in globalimpression, cognition, and function with use of Ginkgo biloba compared to placebo, andno significant differences in adverse events. However, the more recent studies in theanalysis, which were larger and more rigorously designed, showed mixed results.Because of this, and because the studies were pooled regardless of diagnosis (cognitiveimpairment or dementia, any severity), the authors did not draw firm conclusions aboutthe efficacy of Ginkgo biloba in treatment of dementia.


� Subsequent to the search period from the Cochrane meta-analysis, Le Bars et al.published a post-hoc subgroup analysis (106) of data from an earlier double-blindrandomized controlled trial. (107) This subgroup analysis stratified the earlier study’sintention-to-treat (ITT) Alzheimer’s disease population of 236 people based on baselineMMSE scores (stratum 1: MMSE score � 24; stratum 2: MMSE score � 23). Thispopulation was randomized to receive either Ginkgo biloba 120 mg/day or placebo. Theanalysis showed that subjects in severity stratum 1 (mild AD) showed insignificantdifferences between Ginkgo biloba and placebo on measures of cognition, function, andglobal impression. Subjects in stratum 2 (moderate to severe AD) did show statisticallysignificant differences favoring Ginkgo biloba on measures of cognition and function, butnot on clinical global impression. This mixed data leads to unclear conclusions about theusefulness of Ginkgo biloba.

� The USPSTF systematic evidence review on screening for dementia included the keyquestion, “Do pharmacologic interventions of potentially reversible or irreversibledementia improve outcomes?”(27) Their analysis of the literature for ginkgo bilobaincluded the same two meta-analyses that the CMI Dementia Guidelines Workgroup hadexamined for the 2002 CMI Dementia Guidelines (see below). Those two studiesincluded patients with mild to moderate dementia and found an approximate 3%difference in cognitive scales between groups taking ginkgo biloba and placebo. TheUSPSTF did not address the clinical significance of this difference.

� Rationale from the 2002 CMI Dementia Guidelines:� Two systematic reviews of the effect of treatment with Ginkgo biloba on patients

diagnosed with Alzheimer’s disease (AD) were found and reviewed. (108, 109) Oken(108) included 5 randomized controlled trials in the review; Ernst (109) reviewed 9randomized controlled trials, including 3 of the studies that were part of Oken’sreview. Oken limited the studies to those with randomized, placebo-controlled,double-blind study design, with clearly stated diagnosis of AD, and with inclusion ofat least one outcome measure as an objective assessment of cognitive function. Ernstincluded double-blind, randomized, placebo-controlled trials of Ginkgo biloba fordementia (excluding “cerebral insufficiency”). All studies included in the Ernstreview scored at least 3 out of 5 on the Jadad scale of methodological quality.

� Oken calculated the effect size associated with treatment in each of the four studies,and calculated a weighted mean effect size of 0.413, equivalent to a little less thanhalf of a standard deviation. Because no test for heterogeneity was reported, it isdifficult to assess the validity of combining the data into a mean effect size. Ernst etal. did not pool the results quantitatively, but did report the results of the 9 studiesincluded, and concluded that the studies collectively suggest that Ginkgo bilobaextract is more effective for dementia than placebo. However, there weremethodological limitations to the studies, including ill-defined inclusion/exclusioncriteria and unclear validity and reliability of clinical end points used in several of thetrials.

� Following the search period covered in the Oken and Ernst studies, there has been justone additional randomized controlled trial on the effect of Ginkgo biloba in ADpatients published. (110) This study by Van Dongen looked at the effect of 24 weeksof Ginkgo biloba treatment in patients with dementia and patients with age-associatedmemory impairment (AAMI) compared to placebo. A variety of neuropsychological


tests, tests of clinical assessment, and tests of functional assessment showed nosignificant differences in outcomes between the placebo and Gingko biloba groups.

� Based on the methodological limitations of the studies included in the two systematicreviews and of the reviews themselves, and based on the results of the Van Dongentrial, the conservative conclusion is that Ginkgo biloba can not be recommended forthe treatment of AD or other dementias at this time.

� Conclusion: There is insufficient evidence to recommend Ginkgo biloba for thetreatment of Alzheimer’s disease or other dementias.

Supporting Evidence for Vitamin EVitamin E versus placebo: � The Cochrane Collaboration conducted a systematic review on the use of Vitamin E for

Alzheimer’s disease in 2000 (see “Rationale for the 2002 CMI Dementia Guidelines”below). They sought new studies on this topic in June 2002, but found none. Ourliterature search in June 2003 also did not find any additional RCTs on this topic.Therefore, the rationale from our 2002 CMI Dementia Guidelines stands.

� In addition, the USPSTF systematic evidence review on screening for dementia includedthe key question, “Do pharmacologic interventions of potentially reversible or irreversibledementia improve outcomes?”(27) Their literature review for vitamin E produced the samearticle by Sano et al. as our review; the USPSTF concluded that the trial showed no effecton cognition but had limited evidence of delayed institutionalization.

� Rationale from the 2002 CMI Dementia Guidelines:� There has only been one well-designed randomized controlled trial published on the

effect of Vitamin E in patients with Alzheimer’s disease (AD). (111) This study bySano et al. was the only one identified from our search strategy, and was the onlystudy included in a review done by the Cochrane Collaboration. (112) The primaryoutcome measures addressed in this study were survival time to the first of fourendpoints: death, institutionalization, loss of ability to perform 2 out of 3 ADLs, andsevere dementia (defined as a Clinical Dementia Rating of 3). Secondary outcomemeasures included various tests of cognition, function, and behavior.

� The relative risk of reaching one of the primary outcome measures (death,institutionalization, etc.) with Vitamin E compared to placebo was not statisticallysignificant (RR=0.70, p=0.077). When the baseline MMSE (Mini-Mental StateExamination) was included as a covariate using a Cox proportional hazards model,the adjusted relative risk of reaching one of the primary outcome measures wasstatistically significant (RR=0.47, p=0.001). The details of the statistical methodsused to reach the adjusted relative risk were too sparse to assess fully. There were nostatistically significant differences for most of the secondary outcome measures,including the MMSE score and the ADAS-cog score, but there was a significantdifference (in favor of Vitamin E) on the Blessed Dementia Scale score. Rates offalls and syncope were higher in the Vitamin E group compared to placebo.

� Based on the fact that this one trial showed significant differences in primary outcomemeasures only after some statistical manipulation with unclear methods, we can notrecommend Vitamin E for the treatment of Alzheimer’s disease. Further evidence isneeded.


� Conclusion: There is insufficient evidence to recommend the use of vitamin (2000 IUper day) for Alzheimer’s disease and other dementias.


Evidence Table: Donepezil for treatment of mild to moderate Alzheimer’s disease (AD)

Study, Total n


& Drug Study Population

Results Least Squares (LS) Mean (� SE) � from baseline to 52 weeks

unless otherwise noted Comments

Winblad et al., 2001 (RCT, double-blind) Follow up: 52 weeks Initial N: 286 Final N: 192 (Most withdrawal from studycoincided with commercialavailability of the study drug) Study took place in 28 sites in 5northern European countries. Power calculations reported; targetsample size of 150 patients pertreatment group. Used observed cases (OC) analysisand intention-to-treat (ITT) analysiswith last observation carried forward(LOCF).

Rx1 placebo(n = 144) Rx2 donepezil 5mg/day for 28 days,then increased to 10mg/day(n = 142)

� Men and women of any race(though all randomized werewhite) between ages 40-90

� Diagnosis of possible orprobable Alzheimer’s diseaseconsistent with DSM-IV orNINCDS-ADRDA* criteria)

� MMSE* score �10 and �26� Otherwise healthy and

ambulatory or aided bywalker/cane

� Lab tests within normal limits� Reliable caregiver� No specific requirements

concerning vitamin E, ginkgobiloba, or memantine

Primary Outcomes GBS*LOCF analysis: Rx1: 11.5 Rx2: 8.0 (values estimated from graph) p = 0.054 OC analysis: Rx1: 13.5 (� 2.1) Rx2: 7.3 (� 2.1) p = 0.014 GBS Domains:� GBS-I (intellectual

impairment) favoreddonepezil in LOCF (p =0.012) and OC (p =0.004)analyses.

� For domains of function,emotional reaction, andbehavioral symptoms,placebo group improved morethan donepezil group, butdifference at endpoint (LOCF)was not significant.

Secondary Outcomes MMSE*LOCF analysis: Rx1: -2.2 Rx2: -0.5 (values estimated from graph) p � 0.001 PDS* Rx1: -15 Rx2: -11 p � 0.05 Telephone, memory, and self-careitems statistically significantly favoreddonepezil. All items worsened withboth placebo and donepezil, exceptmemory (which showed slightimprovement with donepezil). GDS* Favored donepezil at endpoint (LOCF)(p =0.047) NPI* Differences between groups were notsignificant

Adverse Events (AEs)# (%) withdrawn due to AEs Rx1: 9 (6.3%) Rx2: 10 (7.0%) p not reported Vertigo, asthenia, and syncope twiceas common in donepezil group.

Conclusions� The more conservative LOCF

analysis showed no significantdifference between donepeziland placebo on a clinician-andcaregiver-based globalassessment scale. OC analysisfavored donepezil.

� Only one domain of globalassessment scale (intellectualimpairment) favored donepezil.

� MMSE and ProgressiveDeterioration Scale favoreddonepezil; Global DeteriorationScale and NeuropsychiatricInventory did not.

� Overall, similar rates ofadverse events were noted.

Biases� Study funded by Pfizer, Inc.

* NINCDS-ADRDA: National Institute of Neurological Disorders and Stroke/Alzheimer’s Disease and RelatedDisorders Association

GBS: Gottfries-Brane-Steen (comprehensive global assessment based on interview between clinician and patientand caregiver; 27 items; score 0-162; lower is better)

MMSE: Mini-Mental State Examination (score 0-30; higher is better) PDS: Progressive Deterioration Scale (assesses ADLs and IADLs; 29 questions; higher is better)GDS: Global Deterioration Scale (7-point scale assessing cognition, behavior, and ADLs; mild AD=3, moderate AD=4,severe AD=6)NPI: Neuropsychiatric Inventory (12 items; score 0-144; lower is better)


Evidence Table: Donepezil for the treatment of moderate to severe Alzheimer’s disease (AD)

Study, Total n



Results Least Squares (LS) mean � SE unless otherwisenoted (LOCF); values estimated from graphs

Comments

Feldman et al., 2001 (RCT, double-blind) Follow up: 24 weeks Initial N: 290 Final N: 247 Computerized randomizationscheme cited. Power calculations described. Primary outcomes assessed usingintention-to-treat (ITT) analysis, withlast observation carried forward(LOCF) where there were missingvalues. Secondary outcomes assessedusing observed cases (OC) analysis.

Rx1 placebo(n = 146) Rx2 donepezil5mg/day for week 1-4,then 10mg/day as perclinician’s judgment(n = 144)

� Possible or probable AD byNINCDS-ADRDA* criteria

� sMMSE* score of 5-17inclusive

� FAST* score � 6 at baseline � Patients residing in

community or in assistedliving settings

� Most concomitantmedications allowed,including treatments forbehaviors

Primary OutcomeCIBIC+*Rx1: 4.60Rx2: 4.06p � 0.0001

% rated asimproved or nochange at week 24 Rx1: 42% Rx2: 63% p � 0.0001 Adverse Events Rate ofdiscontinuation dueto AEs Rx1: 6% Rx2: 8%

SecondaryOutcomesCognitivesMMSE* Rx1: -0.5 Rx2: 1.45 p � 0.0001 SIB* Rx1: -3.5 Rx2: 2.0 p � 0.0001 Functional DAD* Rx1: -8.98 Rx2: -0.74 p � 0.0001 IADL+* and PSMS+* Showed less meandecline in Rx2group versus Rx1group (bothsignificant)

BehaviorsNPI* Rx1: 1.0 Rx2: -4.7 p = 0.0005 (for subgroup onpsychoactive meds atbaseline, difference onNPI at week 24 notsignificant) Global Function FRS* Rx1: -1.66 Rx2: -0.38 p = 0.0002

Conclusions� Donepezil produced statistically

significant improvement in allprimary and secondary outcomesassessed. All but NPI showeddeclining trends by endpoint.

Biases� Study supported by Pfizer Inc. and

Eisai Inc.� Concomitant psychoactive

medications were taken by 33% ofdonepezil patient and 24% ofplacebo patients at baseline; thisdifference was sustained duringstudy.

* NINCDS-ADRDA: National Institute of Neurological Disorders and Stroke/Alzheimer’s Disease and RelatedDisorders Association

sMMSE: standardized Mini-Mental State Examination (score 0-30; higher is better) FAST: Functional Assessment Staging Test (16 stages [1-5, 6a-e, 7a-f]; lower stage is better) CIBIC-plus: Clinician’s Interview-Based Impression of Change with caregiver input (7 point Likert-like Scale

[LS] where 4=no change from baseline, 1=marked improvement, 7=marked worsening)

SIB: Severe Impairment Battery (51 items; score 0 to 100; higher is better)DAD: Disability Assessment in Dementia (10 domains, 40 items; higher is better)IADL+: modified Instrumental Activities of Daily Living Scale (scoring system not indicated)PSMS+: modified Physical Self-Maintenance Scale (scoring system not indicated)NPI: Neuropsychiatric Inventory (12 items; score 0-144; lower is better)FRS: Functional Rating Scale (8 domains; 40-point sum of the boxes’ maximum score; higher is better)


Evidence Table: Donepezil for the preservation of function in Alzheimer’s disease (AD)

Study, Total n



Mohs et al., 2001 (prospective, placebo-controlledstudy of survival to endpoint) Follow up: 1 year Initial N: 431 Final N: 111 Power calculations described. Efficacy analyses based on theintention-to-treat (ITT) population

Rx1 placebo (n = 217) (ITT=208) Rx2 donepezil5mg/day for weeks 1-4, 10mg day thereafter(n = 214) (ITT=207) Endpoint of interestwas clinically evidentdecline in functionalstatus, defined as:� Clinically evident

decline in ability toperform one ormore basic ADLspresent at baseline

� Clinically evidentdecline in ability toperform 20% ormore of IADLs

� Increase in globalCDR score of � 1point compared withbaseline

� Diagnosis of probable AD byDSM-IV and NINCDS-ADRDA* criteria

� MMSE score 12-20 inclusive� Clinical Dementia Rating

(CDR) of 1 or 2� Modified Hachinski ischemia

score � 4� Capable of performing 8 of 10

IADLs and 5 of 6 ADLs onADFACS* at screening andbaseline

� Reliable caregiver

Primary OutcomesMedian time in daysto reach endpoint*(95%CI) Rx1: 208 (165-252) Rx2: 357 (280-?) p = 0.0019 (log-rank) Hazard ratio forreaching endpoint:0.62 (patients treated withdonepezil 38% lesslikely to decline over1-year period) # (%) of ITTpopulation whoremained onassigned treatmentfor 54 weeks Rx1: 43 (21%) Rx2: 68 (33%) p not reported

Secondary OutcomesADFACS* Significant differencesin favor of Rx2 at weeks12, 24, and 36 (but notweeks 6, 18, 30, 42, 48,or 54) CDR-SB* Significant differencesin favor of Rx2 at weeks6, 18, 24, 36, and 42(but not weeks 12, 30,48, 54) MMSE*� Significant

differences in favor ofRx2 at weeks 6, 12,18, 24 (but none afterweek 24)

� At 6 of 9 evaluations,Rx1 group alsoshowed improvementfrom baseline

Adverse Events Rate ofdiscontinuation dueto AEs: Rx1: 7.4% Rx2: 10.7% p not reported AEs significantlyhigher in Rx2 groupwere headache, UTI,and GI-related. # (%) of serious AEs Rx1: 19 (8.8%) Rx2: 26 (12.1%) p not reported

Conclusions� Study found that donepezil

group reached endpoint ofclinically evident functionaldecline 5 months later thanplacebo group.

� Differences between placeboand donepezil groups onmeasures of function,cognition, and clinical stage ofdementia seemed to favordonepezil, but thesedifferences were notconsistently statisticallysignificant and were notsignificant at 54 weeks.

Biases� Study supported by Pfizer Inc.

and Eisai Inc.

* NINCDS-ADRDA: National Institute of Neurological Disorders and Stroke/Alzheimer’s Disease and Related Disorders Association ADFACS: Alzheimer’s Disease Functional Assessment and Change Scale (ADL and IADL scale; 16 items; score 0-54; lower is better) CDR: Clinical Dementia Rating (0=normal/no impairment, 3=severe impairment; 5 point scale [0, 0.5, 1, 2, 3]) CDR-SB: Clinical Dementia Rating Sum of Boxes (sum of ratings for each of the 6 CDR domains [“boxes”]; score 0-18; lower is better) MMSE: Mini-Mental State Examination (score 0-30; higher is better)


Evidence Table: Donepezil for the treatment of Alzheimer’s disease (AD) � cerebrovascular disease (CVD)

Study, Total n



Results Mean � from baseline to endpoint � SE unless

otherwise noted Comments

Tariot et al., 2001 (RCT, double-blind) Follow up: 24 weeks Initial N: 208 Final N: 162 Computerized randomizationscheme described. Power calculations described. Efficacy analyses used Intention-to-Treat (ITT) population.

Rx1 placebo(n = 105) Rx2 donepezil 5mg(week 1-4) and 10mg(week 5-24)(n = 103) (dose could bereduced to 5mg if10mg not welltolerated)

� Nursing home patients withdiagnosis of probable orpossible AD or AD withcerebrovascular disease (butnot vascular dementia)

� MMSE scores between 5 and26, inclusive, at screeningand baseline

� Item frequency of 3 or 4 for�1 symptom (domain) fromNPI-NH*

� Most concomitantmedications allowed(including psychotropic drugsfor behaviors)

Primary Outcome:NPI/NH* Rx1: -4.9 � 1.9 Rx2: -2.3 � 1.9p value not reported but described as notsignificant

No significant differences between groupsin mean � from baseline on individualitems of the NPI/NH

Secondary analysis of categorical �s inbehavior at endpoint found significantdifference only for agitation/aggression,which favored donepezil (p = .0442)Agitation/aggression:Improvement:Rx1: 28%Rx2: 45%No change:Rx1: 40%Rx2: 31%Worsening:Rx1: 32%Rx2: 24%

Secondary Outcomes:MMSE* Rx1: -0.8 Rx2: -0.1 (values estimated fromgraph)p = not significant

CDR (NH)* Rx1: 0.7 Rx2: -0.1 (values estimated fromgraph)p � .05

Adverse Events (AEs):Rate of discontinuationdue to AEs Rx1: 18% Rx2: 11% p = not reported Gastrointestinal AEsmore common in Rx2(donepezil)

Conclusions� Placebo group improved more than

donepezil group on NPI/NH score,but difference was not statisticallysignificant.

� Donepezil group showedstatistically significant improvementversus placebo on ClinicalDementia Rating.

� No significant difference on MMSEat 24 weeks; was significant atweeks 8, 16, and 20.

Biases� Study and authors supported by

Pfizer Inc. and Eisai Inc.� Nursing home population may not

be generalizable to populationseen in primary care.

� Authors say secondary analysis ofNPI/NH was planned, but articledoes not describe this plan inmethods section.

* NPI/NH: Neuropsychiatric Inventory/Nursing Home version (12 items; score 0-144; lower is better) MMSE: Mini-Mental State Examination (score 0 -30; higher score is better) CDR-NH: Clinical Dementia Rating-Nursing Home version (6 domains; sum of the item [box] scores ranges from 0-18; lower score is better)


Evidence Table: Donepezil for probable vascular dementia (VaD)

Study, Total n



Results Least Squares (LS) Mean � from baseline to 24 weeks

unless otherwise noted Comments

Pratt et al., 2002 (combined results of 2 double-blind RCTs with identical designsand protocols) Follow up: 24 weeks Initial N: 893 Final N: 707 Analyses performed on intention-to-treat (ITT) population.

Rx1 placebo(n = 290) Rx2 donepezil 5mg/day(n = 296) Rx2 donepezil 10mg/day (received 5mg/day for first 4weeks and 10 mg/daythereafter)(n = 307)

� Diagnosis of probablevascular dementia (VaD) byNINDS-AIREN* criteria

� Clinical or radiologicalevidence of cerebrovasculardisease (CVD)

� Generally healthy ambulatoryoutpatients


Primary OutcomesADAS-Cog* Rx1: -0.23 Rx2: -1.56 Rx3: -2.68 p = 0.001 (Rx2 v Rx1) p � 0.0001 (Rx3 v Rx1) (These p values were reporteddifferently in text and figures.Those reported here are fromtext.)

CIBIC-plus* - observed cases(OC) % w/ improvement Rx1: 32 Rx2: 46 Rx3: 36 % w/ no change Rx1: 44 Rx2: 37 Rx3: 40 % w/ worsening Rx1: 25 Rx2: 17 Rx3: 24 Overall: p = 0.002 Rx2 v Rx1: p = 0.0006 Rx3 v Rx1: p = 0.2096

Secondary Outcomes MMSE* Rx1: 0.7 Rx2: 0.84 Rx3: 1.06 p = 0.001 (Rx2 v Rx1) p � 0.0001 (Rx3 v Rx1) (Values for Rx2, Rx3, and pvalues were reported differentlyin text and figures. Thosereported here are from text.) Adverse Events# (%) who discontinued due toAEs Rx1: 26 (9.0%) Rx2: 29 (9.8%) Rx3: 61 (19.9%) p = not reported

Conclusions� Both donepezil groups (5 mg/day

and10 mg/day) showed statisticallysignificant improvement on theADAS-Cog at 24 weeks comparedto placebo.

� Clinical and caregiver globalimpression measures showed anadvantage of donepezil 5 mg/dayover placebo, but 10 mg/day was nomore effective than placebo.

� MMSE scores were slightlyimproved in all groups; donepezilshowed statistical significance overplacebo although differences werequite small.

� A much higher percentage of peopleon donepezil 10 mg/daydiscontinued due to adverse eventsthan either 5 mg or placebo;adverse events were not discussedin the article.

Biases� First author is the Senior Director of

Clinical Research for Eisai, Inc.� Combines results from 2 trials with

identical protocols without furtherexplanation.

� Numbers reported in text and inaccompanying figures wereinconsistent.

* NINDS-AIREN: National Institute of Neurological Disorders and Stroke and the Association Internationale pour la Recherche et l’Enseignement en Neurosciences ADAS-Cog: Alzheimer’s Disease Assessment Scale-cognitive: (11 items; score 0-70; lower is better) CIBIC-plus: Clinician’s Interview-Based Impression of Change plus caregiver input (7 point Likert-like Scale [LS] where 4=no change from baseline, 1=marked improvement, 7=marked worsening) MMSE: Mini-Mental State Examination (score 0-30; higher is better)


Evidence Table: Tolerability and safety of donepezil in treatment of Alzheimer’s disease (AD)

Study, Total n

TreatmentGroups Size &

Drug Study Population Results Comments

Pratt et al., 2002 (Integrated analysis of safetydata from one phase II andthree phase III donepezilRCTs) Follow up: 12-24 weeksdouble-blind treatment (followedby 2-6 weeks single-blindplacebo washout) Total Initial N: 1920 Final N: 1551

Rx1 placebo (n = 629) Rx2 donepezil 1 mg/ day(n = 42) Rx3 donepezil 3 mg/ day(n = 40) Rx4 donepezil 5 mg/ day(n = 621) Rx5 donepezil 10 mg/ day(n = 588) Total donepezil: n = 1291 Incidence of adverseevents (AEs) elicited ateach patient visit byquestioning patient andcaregiver and throughdirect observation byclinicians.

� Diagnostic evidence ofprobable Alzheimer’s disease(AD) by DSM-III andNINCDS-ADRDA* criteria

� Phase II trial: patients ages55-85; Phase III trials:patients ages 50 and above

� MMSE* scores 10-26� CDR* scores of 1-2� Ambulatory

# (%) withdrawn due to AEs Rx1:43 (7%) Rx2: 5 (12%) Rx3: 2 (5%) Rx4: 43 (7%) Rx5: 92 (16%) Total donepezil: 142 (11%) p not reported AEs that most frequently led towithdrawal in donepezil groupswere associated with digestivesystem (4% of patients) andnervous system (5%) ofpatients (compared to 1% and3%, respectively, for theplacebo group). � 1 AE reported: Rx1: 73% All donepezil: 78% Statistically significant for Rx5vs. Rx1, but not forRx4 vs.Rx1

AEs that occurred in more than5% of Rx5 group, and twice asmuch as in Rx1 were GI(nausea, diarrhea, vomiting,anorexia), asthenia/fatigue,insomnia, and muscle cramp. For Rx4, only diarrhea metthese criteria. AEs occurred more frequentlywith donepezil compared toplacebo in first several weeksof trials, then decreased tolevels similar to placebo byweek 9. % reporting serious AE Rx1: 8% Rx2: 5% Rx3: 8% Rx4: 7% Rx5: 10% p not reported # (%) clinically significanttreatment-emergent changesin vital signs: Rx1: 8 (1%) All donepezil: 24 (2%)

Conclusions� Overall, donepezil 5mg/day is

comparable to placebo in terms ofserious AEs and AEs that lead towithdrawal.

� Diarrhea, anorexia, and muscle crampswere reported significantly morefrequently with donepezil 5 mg/daycompared to placebo.

� Patients taking donepezil 10 mg/dayare more likely to experience mild andserious AEs than those on placebo,and are more likely to withdraw due toAEs. Statistical significance was notreported.

� The most common AEs with donepezilare gastrointestinal.

Biases� First author is the Senior Director of

Clinical Research for Eisai, Inc.

* NINCDS-ADRDA: National Institute of Neurological Disorders and Stroke/Alzheimer’s Disease and Related Disorders Association MMSE: Mini-Mental State Examination (score 0-30; higher is better) CDR: Clinical Dementia Rating (0=normal/no impairment, 3=severe impairment; 5 point scale [0, 0.5, 1, 2, 3])


Evidence Table: Rivastigmine for treatment of moderately severe Alzheimer’s disease (AD)

Study, Total n


Drug Study Population Results

Mean � from baseline (95% CI) unless otherwise noted Comments

Doraiswamy et al., 2002 (subgroup analysis of a double-blind RCT, Corey-Bloom et al.,1998, and an open-labelextension, Farlow et al., 2000) Follow up: 26 weeks (double-blind phase), followed by openlabel phase weeks 27-52 Final N (subgroup): 158 Intention-to-treat (ITT) lastobservation carried forward(LOCF) analysis and observedcases (OC) analysis conducted

Double-blind phase: Rx1 placebo (n = 51) Rx2 rivastigmine 1-4mg/day (n = 58) Rx3 rivastigmine 6-12mg/day (n = 49) Open-label phase: All groups: rivastigmine1 mg/bid for first week;dose increased 1 mg/bid,no faster than weeklyincreases (investigatorsand patient still blinded tooriginal group)

� Subgroup of originalstudy population with“moderately severe” ADbased on GDS* �5 atweek 0 and enteredopen-label treatmentphase

ADAS-Cog* Week 26, OC analysis Rx1: 5.8 (3.96, 7.61) Rx2: 1.2 (-0.31, 2.61) Rx3: -2.3 (-4.44, -0.09) p � 0.001 for Rx2 vs. Rx1 andRx3 vs. Rx1 Week 26, LOCF analysis All values not reported; p � 0.001 for Rx2 vs. Rx1 andRx3 vs. Rx1 Week 52, OC analysis Rx1: 5.5 (2.61, 8.36) Rx2: 3.4 (1.43, 5.43) Rx3: 1.2 (-1.4, 3.84) p = 0.310 for Rx2 vs. Rx1 p = 0.035 for Rx3 vs. Rx1 p � 0.001 for all drugs vs.“projected placebo” (modeledestimate of what Rx1 group’sresults would have been ifcontinued on placebo) Week 52, LOCF analysis p � 0.05 for Rx3 vs.Rx1 No results reported for Rx2 vs.Rx1 – assume this is because oflack of significant difference

Adverse Events (AEs) Incidence of AEs in open-labelphase was generally similar tothat in double-blind phase(numbers not reported). During open-label phase, mostcommon AEs weregastrointestinal (nausea,vomiting, anorexia)

Conclusions� Rivastigmine 1-4 mg/day and 6-12

mg/day showed significantimprovement in cognition compared toplacebo at 26 weeks (both OC andLOCF analyses).

� During open-label extension, theoriginal rivastigmine 6-12 mg/day groupshowed significantly less decline thanthe original placebo group. This wasnot the case for the originalrivastigmine 1-4 mg/day group, whichlooked similar to the original placebogroup at weeks 38, 44, and 52.

Biases� Analysis does not account for the 47

patients with GDS �5 at baseline whodid not complete the double-blindphase. Unknown how those patientswould have changed results.

� Conducted “projected placebo”modeling procedure to predictresponse at week 52 for Rx1 group ifthey had not been put on rivastigmineduring the open-label phase. Thismodel may overestimate the amount ofdecline that would occur.

* GDS: Global Deterioration Scale (7-point scale assessing cognition, behavior, and ADLs; mild AD=3, moderate AD=4, severe AD=6) ADAS-Cog: Alzheimer’s Disease Assessment Scale-cognitive (11 items; score 0-70; lower is better)


Evidence Table: Rivastigmine for treatment of Alzheimer’s disease (AD) with vascular risk factors

Study, Total n


Drug Study Population Results

Mean � from baseline (SD) at week 26 unless otherwise noted Comments

Erkinjuntti et al., 2002 (post hoc subgroup analysis ofRosler et al., 1999, a double-blind RCT) Follow up: 26 weeks Initial N: 178 (for this subgroup) Final N: 136 Observed cases (OC) analysisused.

Rx1 placebo (n = 52) Rx2 rivastigmine 1-4mg/day (n = 59) Rx3 rivastigmine 6-12mg/day (n = 67) For Rx2 and Rx3, a rapidforced titration schemewas used: dose ofrivastigmine wasincreased in incrementsof 3 mg/day every twoweeks to maximumtolerated dose, up to 12mg/day, over first 7weeks Weeks 8-26 were flexibledose phase

� Men and women ages � 50� MMSE* scores 10-26 � Satisfy DSM-IV criteria for

dementia� Probable Alzheimer’s disease

(AD) by NINCDS-ADRDA*criteria

� Presence of baseline arterialhypertension (“history orpresence of blood pressure�160/110 mmHg on threesuccessive readings, orrequiring diet modification ortreatment”)

Primary Outcomes ADAS-Cog* Rx1: 0.92 (6.28) Rx2: 1.30 (5.36) p = 0.771 vs. Rx1 Rx3: -1.71 (5.81) p = 0.047 vs. Rx1 CIBIC-plus* Mean rating (SD) Rx1: 4.61 (1.28) Rx2: 4.12 (1.19) p = 0.110 vs. Rx1 Rx3: 3.66 (1.39) p = 0.004 vs. Rx1

Secondary Outcomes PDS* Rx1: -3.99 (12.63) Rx2: -1.68 (12.81) p = 0.360 vs. Rx1 Rx3: 1.61 (13.44) p = 0.031 vs. Rx1 Adverse Events (AEs) # (%) reporting anorexia Rx1: 0 (0) Rx2: 1 (1.7) Rx3: 4 (6.0) p not reported # (%) reporting nausea Rx1: 1 (1.9) Rx2: 2 (3.4) Rx3: 15 (22.4) p not reported # (%) reporting vomiting Rx1: 0 (0) Rx2: 0 (0) Rx3: 11 (16.4) p not reported

Conclusions� Patients with AD and

hypertension takingrivastigmine 6-12mg/dayshowed statisticallysignificantly better outcomeson cognitive, globalimpression, and functionalscales than those onplacebo. This did not holdtrue for those taking lowerdoses of rivastigmine (1-4mg/day).

� Gastrointestinal adverseevents higher withrivastigmine 6-12mg/daycompared to lower dosesand placebo; significancenot reported.

Biases� Two of the authors work for

Novartis Pharmaceuticals.� Original study and this

analysis funded by Novartis.� Use of OC analysis (instead

of ITT) does not includethose patients who droppedout before week 26, whomay be different than thosewho completed the trial(e.g., less response todrug).

* MMSE: Mini-Mental State Examination (score 0-30; higher is better) NINCDS-ADRDA: National Institute of Neurological Disorders and Stroke/Alzheimer’s Disease and Related Disorders Association ADAS-Cog: Alzheimer’s Disease Assessment Scale-cognitive (11 items; score 0-70; lower is better) CIBIC-plus: Clinician’s Interview-Based Impression of Change plus caregiver input (7 point Likert-like Scale [LS] where 4=no change from baseline, 1=marked improvement, 7=marked worsening) PDS: Progressive Deterioration Scale (assesses ADLs and IADLs; 29 questions; higher is better)


Evidence Table: Galantamine for the treatment of Alzheimer’s disease

Study, Total n

Study Population Treatment Groups Size & Drug Results Comments

Olin et al., 2002(Meta-Analysis) # studies found: 7 met inclusioncriteria # studies included: 6 included inmeta-analysis (provided sufficientoutcome data) Intention-to-treat and observedcases analysis reported when datawere available. Note: 3 of the studies included inthis meta-analysis were alsoincluded in evidence tables from the2002 CMI Dementia Guidelines.

� Inclusion criteria for meta-analysis:� Clinical trials in Alzheimer’s disease (AD)� Double-blind, parallel-group, placebo-

controlled, randomized and unconfounded� Sample selection criteria specified� Outcome instruments specified� Duration specified

� Study duration:� 12 weeks (2 studies)� 13 weeks (1 study)� 5 months (1 study)� 29 weeks (1 study)� 6 months (2 studies)� Studies of 5 months to 29 weeks

aggregated as “6 months” for analysis� Participants:� # ranged from 95 to 978� Mild to moderate AD

� Intervention:� Rx1: galantamine 16-32 mg/day� Rx2: placebo

Global Rating* - 3 months Observed Cases (OC): 24-32mg/day: significant OR 2.3 (95%CI 1.3 to 3.9) 36 mg/day: significant OR 3.4 (95%CI 1.2 to 9.5) Intention-to-treat (ITT): 18mg/day: significant OR 2.44 (95%CI 1.2 to 5.0) 24mg/day: significant OR 2.11 (95%CI 1.0 to 4.6) 36mg/day: significant OR 2.7 (95%CI 1.2 to 6.2) Global Rating - 6 months Observed Cases (OC): 8 mg/day: not significant 16mg/day: significant OR 2.25 (95%CI 1.6 to 3.3) 24mg/day: significant OR 2.0 (95%CI 1.5 to 2.5) 32mg/day: significant OR 1.9 (95%CI 1.4 to 2.5) Intention-to-treat (ITT): 8 mg/day: not significant 16mg/day: significant OR 2.04 (95%CI 1.4 to 2.9) 24mg/day: significant OR 1.82 (95%CI 1.4 to 2.3) 32mg/day: significant OR 1.79 (95%CI 1.3 to 2.4)

ADAS-Cog – 6 months Observed Cases (OC): 8mg/day: WMD* -1.7 (95%CI -3.0 to 0.4) 16mg/day: WMD -3.3 (95%CI -4.5 to -2.2) 24mg/day: WMD -3.5 (95%CI -4.3 to -2.8) 32mg/day: WMD -4.0 (95%CI -5.0 to -3.0) Intention-to-treat (ITT): 8mg/day: WMD -1.3 (95%CI -2.6 to 0.03) 16mg/day: WMD -3.1 (95%CI -4.1 to -2.1) 24mg/day: WMD -3.3 (95%CI -3.9 to -2.7) 32mg/day: WMD -3.3 (95%CI -4.1 to -2.4) Adverse Events (AEs) � Trials did not report equivalent AE data,

limiting comparisons� GI events most common� Three 6-month studies reported GI

events appeared at least 5% morefrequently with galantamine vs. placebo

� AEs significantly more frequent with16mg, 24 mg, 32 mg vs. placebo

Conclusions� Meta-analysis shows overall

positive effects forgalantamine versus placeboin patients with mild tomoderate AD.

Biases� Six of the seven included

studies were Phase II or IIIstudies sponsored bypharmaceuticalmanufacturers.

� Small number of trials limitedpower of subgroup analyses.

� Patient population in thesetrials may not berepresentative of those thatwould receive galantamine(study subjects physicallyhealthier and within narrowerrange of cognitivefunctioning).

* Global Rating data were dichotomized into those who had improvement or no change versus those who worsened. WMD: Weighted Mean Difference


Evidence Table: Galantamine for the treatment of vascular dementia (VaD) and Alzheimer’s disease (AD) withcerebrovascular disease (CVD)

Study, Total n



Results Mean � from baseline to endpoint (SE) unless otherwise

noted, using ITT analysis (LOCF) Comments

Erkinjuntti et al., 2002 (RCT, double-blind) Follow up: 6 months Initial N: 592 Final N: 457 Power calculations (using datafrom a previous study) weredescribed.

Rx1 placebo (n = 196) Rx2 galantamine 24mg/day (n = 396) Used a one-weekdose escalationscheme (patientsstarted on 4mg/day inweek one, with weeklyincreases of 4mg/dayuntil reached 24mg/day in week 6.

� Probable vasculardementia (VaD) by NINDS-AIREN* criteria, OR

� Possible Alzheimer’sdisease (AD) by NINCDS-ADRDA* criteria

� Evidence ofcerebrovascular disease(CVD) on CT or MRI withinpast 12 months

� Mild to moderate dementia(score of 10-25 on MMSE,score �12 on ADAS-Cog*)

Primary Outcomes:ADAS-Cog/11* Rx1: 1.0 (0.5) Rx2: -1.7 (0.4) p � 0.0001 ADAS-Cog/11 subgroup analysis: Vascular dementia (n=188): Rx1: -0.4 (0.78) (not signif. baseline to end) Rx2: -2.4 (0.59) (p � 0.0001 baseline to end) p = 0.06 (between groups) AD w/ cerebrovascular disease (n=239): Rx1: 1.8 (0.6) (p � 0.05 baseline to end) Rx2: -1.0 (0.46) (p = 0.024 baseline to end) p =0.0006 (between groups) CIBIC-plus* # (%) patients improved or no change Rx1: 95 (59%) Rx2: 213 (74%) p = 0.001 CIBIC-plus* subgroup analysis: Vascular dementia (n=188): Rx1: 16 (23%) Rx2: 37 (31%) p = 0.238 AD w/ cerebrovascular disease (n=239): Rx1: 16 (19%) Rx2: 49 (32%) p =0.019

Secondary Outcomes:DAD* Rx1: -4.4 (1.3) Rx2: 0.2 (0.9) p = 0.0017 NPI* Rx1: 1.0 (0.9) Rx2: -1.2 (0.6)p = 0.0164

Adverse Events (AEs)Rate of discontinuationdue to AEs Rx1: 8% Rx2: 20% p = not reported

Conclusions� Galantamine showed statistically

significant benefits versus placeboin cognition and global impressionof change in people with VaD andAD with cerebrovascular disease.

� Subgroup analyses showedsignificant differences in measuresof cognition and impression ofchange between donepezil andplacebo for AD w/ cerebrovasculardisease, but not for VaD. Studywas not powered for subgroupanalysis, so differences may nothave been detected due to smallnumber of patients in subgroups.

Biases� Funded by Janssen Research

Foundation.� Lower study completion rate for

galantamine (74%) vs. placebo(83%); p not indicated.

� When study was designed,optimum dose escalation schemewas not known – now known to be4-weekly dose increments of8mg/day. The scheme used mayhave caused more nausea andvomiting.

* NINDS-AIREN: National Institute of Neurological Disorders and Stroke and the Association Internationale pour laRecherche et l’Enseignement en Neurosciences

NINCDS-ADRDA: National Institute of Neurological Disorders and Stroke/Alzheimer’s Disease and RelatedDisorders Association

ADAS-Cog/11: Alzheimer’s Disease Assessment Scale-cognitive (11 items; lower is better) CIBIC-plus: Clinician’s Interview-Based Impression of Change plus caregiver input (7 point Likert-like Scale [LS]where 4=no change from baseline, 1=marked improvement, 7=marked worsening)DAD: Disability Assessment in Dementia (10 domains; 40 items; higher is better)NPI: Neuropsychiatric Inventory (12 items; score 0-144; lower is better)


Evidence Table: Donepezil vs. Rivastigmine for the treatment of mild to moderate Alzheimer’s disease (AD)

Study, Total n


Drug

StudyPopulation

Results Least Squares (LS) mean � from baseline (�SE) at week 12 unless

otherwise noted; OC analysis Comments

Wilkinson et al., 2002 (Randomized clinical trial,open-label) Follow up: 12 weeks Initial N: 111 Final N: 88 Randomization schemereported. Observed cases (OC) andintention-to-treat (ITT) lastobservation carried forward(LOCF) analyses conducted;only OC analyses reportedbecause of “potential forbias resulting from largedifferences in studydiscontinuation ratesbetween the two treatmentgroups.”

Rx1 donepezil (up to 10mg qd) (n = 56) Rx2 rivastigmine (up to 6mg bid) (n = 55) Dose escalation: Rx1: Patients received 5mg qd for 28 days, then 10mg qd. Rx2: Patients received 1.5mg bid with food. At 14-day intervals, patientswere assessed and dosesincreased to 3 mg bid (day14), 4.5 mg bid (day 28),to a maximum of 6 mg bid(day 42) depending ontolerability. Either drug could bedecreased if not toleratedat higher doses.

� Patients ages � 50� Mild to moderate

possible or probableAlzheimer’s disease(AD) by DSM-IV andNINCDS-ADRDA*criteria

� MMSE* score 10-26inclusive

� Results of CT or MRIin past 12 monthsconsistent with AD


Primary OutcomesADAS-Cog*(Outcomes assessed byindependent ratersblinded to Rx group) Rx1: -0.90 (�0.56) Rx2: -1.05 (�0.67) Rx difference (95% CI): -0.15 (-1.85 to 1.55) Not significant

Secondary OutcomesMMSE*(Outcomes assessed byclinicians aware of Rxgroup) Rx1: 0.71 (�0.44) Rx2: 1.20 (�0.52) Rx difference (95% CI): 0.49 (-0.82 to 1.81) Not significant Satisfaction/ Ease of use* Mean (�SE) at week 12 Physicians: Rx1: 8.5 � 0.4 Rx2: 11.5 � 0.5 p � 0.001 Caregivers: Rx1: 10.9 � 0.5 Rx2: 12.4 � 0.6 p � 0.05 % reached max drug dose Rx1: 98.2% Rx2: 60.0% p not reported Stayed at max drug doseuntil study completion orfinal visit Rx1: 87.5% Rx2: 47.3% p not reported

Adverse Events (AEs)# (%) discontinued due toAEs Rx1: 6 (10.7%) Rx2: 12 (21.8%) p not reported # (%) w/ nausea Rx1: 6 (10.7%) Rx2: 23 (41.8%) p not reported # (%) w/ vomiting Rx1: 4 (7.1%) Rx2: 13 (23.6%) p not reported # (%) w/ headache Rx1: 4 (7.1%) Rx2: 10 (18.2%) p not reported

Conclusions� Donepezil and rivastigmine

showed similar small favorableimpact on cognitive measuresafter 12 weeks.

� Donepezil scored higher onsatisfaction/ease of usesurveys of physicians andcaregivers than rivastigmine.

� More patients on donepezilwere able to reach maximumdrug dose than those onrivastigmine.

� Donepezil produced feweradverse events (AEs) andfewer discontinuations due toAEs than rivastigmine.

Biases� Two authors work for Pfizer

and Eisai.� Satisfaction questionnaire

developed by Pfizer and Eisai.� Fairly small sample size.� Use of OC analysis (instead of

ITT) does not include thosepatients who dropped outbefore week 12, who may bedifferent than those whocompleted the trial.

* NINCDS-ADRDA: National Institute of Neurological Disorders and Stroke/Alzheimer’s Disease and Related Disorders Association MMSE: Mini-Mental State Examination (score 0-30; higher is better) ADAS-Cog: Alzheimer’s Disease Assessment Scale-cognitive (11 items; score 0-70; lower is better) Satisfaction/Ease of Use questionnaire: Assessing physicians’ and caregivers’ satisfaction with dosing frequency and titration of assigned treatment (developed by Pfizer and Easai w/ consultant from Columbia University)(physician survey

had 6 questions, total score 6-30; caregiver survey had 8 questions, total score 8-40; for both surveys lower is better)


Evidence Table: Estrogen and Estrogen plus Progestin (HRT) for treatment of dementia

Note: 3 of the studies included in this meta-analysis were included in the 2002 CMI Dementia Guidelines

Study, Total n

Study Population Treatment Groups Size & Drug

Results Mean difference (95% CI) from baseline to final assessment unless

otherwise noted Comments

Hogervorst et al., 2002(Meta-Analysis) # studies found: 7 met criteria # studies included: 5 had sufficientinformation

� Double-blind RCTs investigating effects ofestrogen replacement therapy (ERT) andhormone replacement therapy (HRT) (estrogenplus progestin) compared to placebo oncognition in post-menopausal women withAlzheimer’s disease (AD) and other dementiasyndromes.

� Number of subjects ranged from 14 to 120.� Subjects had mild to moderate dementia

(MMSE* score 10-28)� Duration of treatment: 8 weeks to 12 months

(average of 6 months)� Types of treatment:� 4 RCTs prescribed Premarin (conjugated

equine estrogen, CEE) 1.25 mg/day (3studies) or 0.625 mg/day (2 studies). Onestudy added a progestagen.

� 1 RCT used transdermal estradiol.

Global Cognitive Functioning MMSE* CEE both doses, 1-2 months WMD* = 1.00 (0.06 to 1.94) p � 0.05 2 months 0.625 mg/day dose showed significantdifference vs. placebo; 1.25 mg/daydose did not. 6 months, 12 months Neither dose significantly differentthan placebo When transdermal treatment includedin analysis, no significant effect ofestrogen on MMSE. Combined CEE (pills and transdermal)at 3, 6, and 12 months Not significant vs. placebo Language Tests No evidence of treatment effect ofestrogen on:� Boston Naming Test� Token test

Memory Tests No evidence of treatment effect ofestrogen on:� Paragraph Recall Tests (immediate and

delayed)� BSRT* immediate recall� CERAD word list� Paired Associate Learning Test� Digit Span Forward� Category Fluency� Visual Retention Tests Info Processing/Concentration Tests No evidence of treatment effect ofestrogen on:� Trail Making Test, part A or B (after 4

months)� Stroop Interference Test� Digit Symbol Substitution Test� Letter Cancellation test� Finger Tapping

Clinical Impression of Change No evidence of treatment effect ofestrogen on:� Clinician’s Global Impression of Change� Clinician Interview-Based Impression of

Change� Hamilton Depression ScaleFavored placebo:� Clinical Dementia Rating

Conclusions� This meta-analysis found no

positive effect of ERT orHRT for cognition or globalchange in women withdementia.

Biases� Some studies included were

small.

* MMSE: Mini-Mental State Examination (score 0-30; higher is better) WMD: Weighted mean difference BSRT: Buschke Selective Reminding Test (higher is better)


Evidence Table: Estrogen for treatment of Alzheimer’s disease (AD)

Study, Total n



Asthana et al., 2001 (RCT, double-blind) Follow up: 8 weeks oftreatment, followed by anotherobservation at week16 Initial N: 20 Final N: 20 Used intention-to-treat (ITT)analysis. Randomization schemereported.

Rx1 placebo (n = 10) Rx2 17�-estradiol0.10mg/day skin patch(n = 10)

� Diagnosis of probableAlzheimer’s disease (AD)by NINCDS-ADRDA*criteria.

� Mild to moderate severity

CognitionAttention: mean time to complete“interference” condition of StroopColor Word Interference Test Rx1: 110 seconds Rx2: 90 seconds (values estimated from graph) p = 0.02 Number of errors was similar. Verbal memory: mean # of wordsrecalled on BSRT* Rx1: 7.1 Rx2: 8.2 (values estimated from graph) p = 0.049 Visual memory: mean recall scoreon Figure Copy/Memory Test Rx1: 21 Rx2: 26 (values estimated from graph) p = 0.03

CIBIC*No effect of estrogen treatmentp = 0.32

Other measures No evidence of treatment effectof estrogen on:� MMSE*� BMICT*� BPRS*� PSMS*� IADL*

Adverse Events (AEs)AEs reported included breasttenderness (in two Rx2 patients)and skin irritation at the patchsite (3 patients in each group)

Conclusions� Study suggests that short-term

estradiol patch is associated withstatistically significant improvement inattention, verbal memory, and visualmemory compared to placebo.

� Estradiol patch did not demonstratesuperiority over placebo on severalmeasures of global impression,cognition, mood and psychiatricsymptoms, and function.

� Studies evaluating effect of estradioladministration with larger sample sizesand longer treatment duration areneeded.

Biases� Supported by Ciba-Geigy Corp.,

manufacturer of the 17�-estradiolpatch (Estraderm).

� Primary measures used were chosenbecause of positive results on thesemeasures for lower doses of estrogenin a previous trial.

� Very small n in each Rx group.* NINCDS-ADRDA: National Institute of Neurological Disorders and Stroke/Alzheimer’s Disease and Related Disorders Association BSRT: Buschke Selective Reminding Test (adapted to 8 items instead of 12; higher is better) CIBIC: Clinician Interview-Based Impression of Change (7 point Likert-like Scale [LS] where 4=no change from baseline, 1=marked improvement, 7=marked worsening) MMSE: Mini-Mental State Examination (score 0-30; higher is better) BMICT: Blessed Memory Information Concentration Test (score 0-26; lower is better) BPRS: Brief Psychiatric Rating Scale (24 symptoms rated 1-7 for severity; lower is better) PSMS: Physical Self-Maintenance Scale (score 0-6; higher is better) IADL: Instrumental Activities of Daily Living (score 0-8; higher is better)


Evidence Table: Naproxen (NSAID) or rofecoxib (Cox-2) for treatment of Alzheimer’s disease (AD)

Study, Total n


Drug Study Population

Results Mean � from baseline (SD) at week 52 endpoint unless otherwise

noted; ITT analysis Comments

Aisen et al., 2003 (RCT, double-blind) Follow up: 1 year of treatment,followed by 2 week washout Initial N: 351 Final N: 267 Randomization scheme reported. Power calculations reported. Intention-to-treat (ITT) analysisused. An alternate imputationscheme (rather than lastobservation carried forward, orLOCF) was used because ofexpectation that more dropoutswould occur in the active Rxgroups due to adverse events.Also did LOCF and observedcases (OC) analyses forcomparison.

Rx1 placebo (n = 111) Rx2 naproxen sodium220 mg bid (n = 118) Rx3 rofecoxib 25 mg qd(n = 122)

� Age � 50� Mild to moderate

Alzheimer’s disease(MMSE* 13-26)

� Ambulatory outpatients� Excluded if they had

comorbid conditions thatmight respond to NSAIDs

� Stable use ofcholinesterase inhibitorsallowed

Primary Outcomes ADAS-Cog* Rx1: 5.7 (8.2) Rx2: 5.8 (8.0) Rx3: 7.6 (7.7) p = 0.96 (Rx2 vs. Rx1) p = 0.09 (Rx3 vs. Rx1)

Secondary OutcomesCDR-SOB* Rx1: 2.2 (2.3) Rx2: 2.3 (2.3) Rx3: 2.2 (2.4) p = 0.89 (Rx2 vs. Rx1) p = 0.89 (Rx3 vs. Rx1) ADCS-ADL* Rx1: -11.5 (11.2) Rx2: -8.7 (10.5) Rx3: -8.3 (9.7) p = 0.14 (Rx2 vs. Rx1) p = 0.08 (Rx3 vs. Rx1) NPI* Rx1: 3.4 (11.9) Rx2: 3.7 (12.5) Rx3: 3.8 (10.9) p = 0.76 (Rx2 vs. Rx1) p = 0.76 (Rx3 vs. Rx1) QOL-AD* Rx1: -2.4 (5.1) Rx2: -2.6 (5.5) Rx3: -2.9 (5.6) p = 0.93 (Rx2 vs. Rx1) p = 0.85 (Rx3 vs. Rx1)

Adverse Events (AEs) # of serious AEs Rx1: 15 Rx2: 25 Rx3: 26 p not significant # (%) w/ fatigue Rx1: 7 (6%) Rx2: 17 (14%) Rx3: 18 (15%) p = 0.06 (Rx2 vs. Rx1) p = 0.05 (Rx3 vs. Rx1) # (%) w/ dizziness Rx1: 5 (5%) Rx2: 14 (12%) Rx3: 14 (11%) p = 0.06 (Rx2 vs. Rx1) p = 0.06 (Rx3 vs. Rx1) # (%) w/ dry mouth Rx1: 1 (1%) Rx2: 6 (5%) Rx3: 1 (1%) p = 0.06 (Rx2 vs. Rx1) p � 0.99 (Rx3 vs. Rx1) # (%) w/ hypertension Rx1: 0 Rx2: 6 (5%) Rx3: 9 (7%) p = 0.03 (Rx2 vs. Rx1) p = 0.004 (Rx3 vs. Rx1)

Conclusions� No benefit of either

naproxen (an NSAID) orrofecoxib (a COX-2inhibitor) on cognitivedecline compared toplacebo.

� Both drugs produced moreserious and mild adverseevents than placebo,though this only reachedstatistical significance forhypertension (both drugs)and fatigue (for rofecoxib).

Biases� None noted.

* MMSE: Mini-Mental State Examination (score 0-30; higher is better) ADAS-Cog: Alzheimer’s Disease Assessment Scale-cognitive (11 items; score 0-70; lower is better) CDR: Clinical Dementia Rating scale, sum-of-boxes score (score 0-18; lower is better)

ADCS-ADL: Alzheimer Disease Cooperative Study – Activities of Daily Living (score 0-78; higher is better)NPI: Neuropsychiatric Inventory (12 items; score 0-144; lower is better)QOL-AD: Quality of Life, Alzheimer’s disease (score 13-52; higher is better)


Evidence Table: Ginkgo biloba for the treatment of cognitive impairment and dementia

Study, Total n


Results Using data from completers only (observed cases) Comments

Birks et al., 2002(Meta-Analysis) # studies included: 33

� Included double-blind RCTs examining extractsof Ginkgo biloba at any strength and over anyperiod with placebo for effects on people withacquired cognitive impairment, includingdementia, of any degree of severity.

� Rx1: placebo� Rx2: Ginkgo biloba (dose ranged from 80-600

mg/day, usually � 200 mg/day)� Duration of studies range from 3-52 weeks, with

majority 12 weeks long

Global Impression - CGI* Ginkgo � 200 mg/day, �12 weeks, oddsof improvement OR 15.32, 95% CI 5.90 to 39.80 p � 0.001 in favor of Ginkgo Ginkgo � 200 mg/day, 24 weeks, odds ofimprovement OR 2.16, 95% CI 1.11 to 4.20 p = 0.02 in favor of Ginkgo Cognition Ginkgo � 200 mg/day, �12 weeks SMD* –0.57, 95% CI –1.09 to –0.05 p = 0.03 in favor of Ginkgo Ginkgo � 200 mg/day, 12 weeks SMD –0.56, 95% CI –1.12 to 0 p = 0.05 in favor of Ginkgo Ginkgo any dose, 24 weeks SMD –0.17, 95% CI –0.32 to –0.02 p = 0.03 in favor of Ginkgo Ginkgo � 200 mg/day, 52 weeks SMD –0.41, 95% CI –0.71 to –0.11 p � 0.01 in favor of Ginkgo

Activities of Daily Living Ginkgo � 200 mg/day, 12 weeks SMD –1.10, 95% CI –1.79 to –0.41 p � 0.01 in favor of Ginkgo Ginkgo � 200 mg/day, 24 weeks SMD –0.25, 95% CI –0.49 to 0 p = 0.05 in favor of Ginkgo Ginkgo � 200 mg/day, 52 weeks SMD –0.41, 95% CI –0.71 to –0.11 p � 0.01 in favor of Ginkgo Mood & Emotional Function Ginkgo � 200 mg/day, �12 weeks SMD –0.51, 95% CI –0.99 to –0.03 p = 0.04 in favor of Ginkgo Ginkgo � 200 mg/day, 12 weeks SMD –1.94, 95% CI –2.73 to –1.15 p � 0.0001 in favor of Ginkgo Adverse Events (AEs) No significant differences

Conclusions� Meta-analysis showed promising

evidence of improvement in globalimpression, cognition, and function withuse of Ginkgo biloba; however, resultsfrom the larger, better designed, morerecent trials are mixed.

� A large, well-designed trial usingintention-to-treat (ITT) methodology isneeded.

Biases� Meta-analysis pooled results from all

studies regardless of diagnosis(cognitive decline or dementia, anydegree of severity).

� Methodology of many of the early trialswas unsatisfactory and many weresmall.

� Publication bias.� Because most studies reported data

only for those subjects who completedthe trial, this meta-analysis reportscompleter analyses only, not intention-to-treat (ITT). This may bias the resultsdue differences between those whodropped out and those who completedthe trial.

* CGI: Clinical Global Impression (7-point Likert-type scale; 1=improved much, 4=no change, 7=extreme worsening) SMD: Standardized mean difference (used because of different assessment scales in trials)


Evidence Table: Influence of severity of cognitive impairment on effect of Ginkgo biloba in Alzheimer’s disease

Study, Total n



Results Mean � from baseline (95% CI) at week 52 endpoint unless

otherwise noted; ITT analysis Comments

Le Bars et al., 2002 (post hoc analysis from LeBars et al., 1997, a double-blind RCT) Follow up: 52 weeks Intention-to-treat (ITT) N forthis subgroup analysis: 236

Rx1 placebo(n = 116) Rx2 Ginkgo biloba(Egb) 120 mg/day(n = 120)

� Stratified intention-to-treat(ITT) population withAlzheimer’s disease (AD)from original study usingcutoff points of 23 and 14 onbaseline MMSE* to indicateseverity subgroups:� Severity stratum 1:

MMSE � 23 (n = 122)� Severity stratum 2:

MMSE � 24 (n = 114)� Note: This is how severity

strata were described inthe article. It is equivalentto:� Stratum 1: MMSE �24� Stratum 2: MMSE �23

ADAS-Cog* MMSE� 23 Rx1: -0.7 (-1.8 to 0.4) Rx2: -1.7 (-2.7 to -0.7) Rx difference: 1.0 (-0.4 to 2.4) Not significant MMSE � 24 Rx1: 4.1 (2.1 to 6.1); p = 0.0001

from baseline Rx2: 1.6 (-0.2 to 3.3); p = 0.07

from baseline Rx difference: 2.5 (0.0 to 5.1) p = 0.05 GERRI* MMSE� 23 Rx1: 0.01 (-0.09 to 0.11) Rx2: -0.09 (-0.18 to 0.00) Rx difference: 0.10 (-0.03 to 0.23) Not significant MMSE � 24 Rx1: 0.18 (0.07 to 0.29); p = 0.002

from baselineRx2: -0.08 (-0.19 to 0.03); NS*from baseline Rx difference: 0.25 (0.10 to 0.41) p = 0.001

CGIC*Rating Mean (95% CI) MMSE� 23 Rx1: 3.9 (3.7 to 4.0) Rx2: 3.9 (3.7 to 4.0) Rx difference: 0.0 (-0.2 to 0.2) Not significant MMSE � 24 Rx1: 4.6 (4.5 to 4.8) Rx2: 4.7 (4.5 to 4.9) Rx difference: 0.0 (-0.3 to 0.2) Not significant Adverse Events (AEs) Not reported

Conclusions� Subjects with mild AD showed

insignificant differences betweenplacebo and Ginkgo biloba onmeasures of cognition and function.

� Subjects with moderate AD did showsignificant differences favoring Ginkgobiloba compared to placebo.

� Neither severity strata showeddifferences in clinician globalimpression scores between Ginkgobiloba and placebo.

Biases� Unclear whether this subgroup analysis

was planned at outset of original study� Cochrane meta-analysis on Ginkgo

biloba pointed out a methodologicalflaw in Le Bars et al., 1997, wherebypatients could be removed from thetrial at any time if they appeared to bedeclining more than 1 point on theCGIC. This should not be the case inan RCT and biases the results ofcompleters to those who did notdecline.

* MMSE: Mini-Mental State Examination (score 0-30; higher is better) ADAS-Cog: Alzheimer’s Disease Assessment Scale-cognitive (11 items; score 0-70; lower is better) GERRI: Geriatric Evaluation by Relatives Rating Instrument (caregiver assessment of ADLs and social function; score 1-5; lower is better) CGIC: Clinical Global Impression of Change (7-point Likert-type scale; 1=improved much, 4=no change, 7=extreme worsening) NS: Non-significant

© 2004 Kaiser Permanente Medical Care Program CMI Dementia Guidelines For use within Kaiser Permanente only. 01/04 103

Pharmacologic Management – Memantine

Problem Formulation

Clinical Question: Should memantine be used for the treatment of dementia? Intended Use of the

Guideline: To assist primary care physicians and other health care professionals in the outpatient primary care setting in managing dementia

Population: Men and women with dementia

Health Problem: Cognitive and functional decline

Health Intervention: • Memantine • Memantine plus AChEI • No treatment

Practitioners: KP physicians, physician assistants, nurse practitioners, nurses, and social workers in the Departments of Family Practice and Internal Medicine (primary care)


Health Outcomes Associated with the

Intervention:

• Cognitive ability • Functional ability • Behavior

• Patient/family satisfaction • Time to institutionalization • Caregiver time/burden

Side Effects of the Intervention:

• Drug adverse effects (i.e., restlessness)

Measures of Outcomes:

• Global impression • Score on cognitive scales • Score on functional scales

• Score on behavior scales

CMI Dementia Guidelines © 2004 Kaiser Permanente Medical Care Program 104 For use within Kaiser Permanente only. 01/04

Evidence Search

Only RCTs, meta-analyses, or systematic reviews were included that studied the use of memantine compared to placebo for cognition and/or function in people with dementia. Because this topic was not addressed in the 2002 version of the CMI Dementia Guidelines, literature searches extend back to 1965.

Database: Terms: Article type and Limits:

Time Frame:

# Found:

# in ET:


Clinical Evidence dementia Systematic reviews

and RCTs 5/27/03 12 0

Meta-analysis, English, Human

1965-7/29/03

2 (1 is

Cochrane) 0

PubMed

("Dementia"[MeSH] OR "Dementia/therapy"[MeSH]) AND ("Memantine"[MeSH] OR memantine[TEXT]) Randomized

Controlled Trial, English, Human

1965-7/29/03 10 3

Off-cycle revision (May 2004): An article published after the search date addressed the use of memantine in combination with an acetylcholinesterase inhibitor, donepezil.∗ Because our earlier search had yielded no articles on combination treatment, we had indicated in our original 2004 guidelines that, “There is no evidence to support the use of memantine in combination with an acetylcholinesterase inhibitor.” We are revising this guideline to reflect the more recently published article.

∗ Tariot PN, Farlow MR, Grossberg GT, Graham SM, McDonald S, Gergel I at al. Memantine treatment in patients with moderate to severe Alzheimer disease already receiving donepezil: a randomized controlled trial. JAMA 2004; 291(3):317-324.



Guideline: Use of memantine for treatment of dementia

Recommendations: For patients with moderate to severe Alzheimer’s disease, a trial of memantine 20 mg/day is an option.

There is also limited evidence to support the option of memantine plus donepezil in moderate to severe AD, but use of memantine alone is recommended over this combination due to the weight of the evidence. (Off-cycle revision, May 2004)

(Note: This drug is not on all KP formularies at this time. When on formulary, it may be restricted to certain specialty groups.)

Methodology: g Evidence-Based g IRGSG-sponsored

Rationale: Supporting Evidence for Memantine vs. Placebo

BENEFITS: A systematic review of the clinical efficacy and safety of memantine for people with

Alzheimer’s disease, vascular, or mixed dementia was performed by the Cochrane Collaboration. (113) This review included 7 double-blind RCTs, 5 of which included adequate data for analysis.(114-118) Pooling of all data was impossible due to major differences in the trials. Overall review of results demonstrated in those trials showed statistically significant positive effects of memantine on cognition, mood and behavior, and global clinical impression compared to placebo. Effect on ADLs was not significant for memantine in some studies and difficult to interpret in others due to validity issues. Several limitations make interpretation difficult: The review does not seek to parse out differences in results based on type or severity

of dementia, probably because the studies themselves often did not define their populations very well. Only one of the five studies included in the data analysis focused exclusively on patients with vascular dementia.(117) That study by Orgogozo et al. found significant differences in favor of memantine in terms of change from baseline on the ADAS-Cog, but found no difference on global change scales – making it difficult to assess the clinical impact of the drug in this patient population.

Three additional double-blind RCTs have been published on this topic since the search period of the Cochrane review. Wilcock et al. examined the effect of memantine 20 mg/day versus placebo in 579

outpatients with mild to moderate vascular dementia for 28 weeks. (119) Although both groups declined (worsened) on the ADAS-Cog, the memantine group declined statistically significantly less than the placebo group (p = 0.000505). More benefit was seen in those memantine patients with more serious disease, as a predefined subgroup analysis showed the greatest treatment effect on the ADAS-Cog for those with baseline MMSE < 15. There was no significant difference between memantine


and placebo groups on the Clinical Global Impression of Change scale or any of the secondary measures assessing cognition (MMSE), function, and global assessment based on patient/caregiver feedback.

Reisberg et al. studied the effects of memantine 20mg/day versus placebo in 252 patients with moderate-to-severe Alzheimer’s disease for 28 weeks. (120) Assessment of global clinical impression showed statistically significant improvement with memantine compared to placebo for an analysis of observed cases (OC) at week 28 (p = 0.03) but did not achieve significance with an analysis of last observation carried forward (LOCF) at end point (p = 0.06). Assessment of function showed significantly less deterioration among patients taking memantine versus placebo (Activities of Daily Living inventory: p = 0.003 for OC analysis and p = 0.02 for LOCF analysis. Severe Impairment Battery: p = 0.002 for OC analysis and p < 0.001 for LOCF analysis). No significant differences were observed between groups for cognitive ability (MMSE), Global Deterioration Scale stage, or Neuropsychiatric Inventory score. The study used the “Resource Utilization in Dementia score” and found that caregivers of people in the memantine group spent 45.8 fewer hours per month caring for the patients than those in the placebo group (95% CI 10.37 to 81.27, p = 0.01). More information on resource utilization was contained in next article.

Wimo et al. examined resource utilization (using the Resource Utilization in Dementia [RUD] scale) and costs for the treated-per-protocol population (n=166) from the Reisberg study described above. (121) Their analysis showed that caregivers of people in the memantine group spent a mean of 42 fewer hours per month on caring tasks and supervision than their counterparts in the placebo group (p not reported). Costs related to caregivers and to society as a whole favored memantine. Patient direct medical costs were higher in the memantine group, but direct non-medical costs were lower.

HARMS: The Cochrane meta-analysis found no statistically significant difference in overall

adverse events between placebo and memantine (for the three studies that reported adverse events), though one included study did find significantly higher incidence of restlessness at 6 weeks in the memantine 30 mg/day group compared to placebo (Odds Ratio 13.50, 95% CI 2.71 to 67.19, p=0.001). (113)

Wilcock et al. found that dizziness, constipation, and other general, gastrointestinal, and respiratory disorders were more common in the memantine group, but there were no differences in serious adverse events or withdrawal due to adverse events. (119)

Reisberg et al. found no significant differences in adverse events between the memantine and placebo groups. (120)

Wimo et al. captured costs of adverse events in their overall resource and cost analysis. (121)

Other considerations: On October 16, 2003, the FDA approved memantine for the treatment of moderate to severe

dementia of the Alzheimer’s type.


• Conclusion: Effect on cognition (as measured by ADAS-Cog) tends to show positive benefit of memantine compared to placebo in moderate to severe Alzheimer’s disease, vascular dementia, and mixed dementia. Effect on global clinical impression was demonstrated in patients with Alzheimer’s disease, but not in vascular dementia. Impact of memantine on MMSE and Neuropsychiatric Inventory (behaviors) has not been consistently demonstrated. Memantine has demonstrated positive impact on caregiver time/burden, though these data are self-report, not observed. Overall, given the positive impact on cognition and global change in patients with moderate to severe Alzheimer’s disease, memantine is an option for those patients. Since evidence is weaker in other types of dementia, it is not recommended for other types at this time.

Supporting Evidence for Combination Therapy with Memantine Plus an Acetylcholinesterase Inhibitor (Off-cycle revision, May 2004) One RCT has been published examining the efficacy and safety of memantine versus placebo

in patients with moderate to severe Alzheimer’s disease (AD) already receiving stable treatment with donepezil, an acetylcholinesterase inhibitor.∗ In this study, Tariot et al. randomized 404 patients with mild to moderate AD who were on stable doses of donepezil to either memantine 5mg/day (increased to 20mg/day) (n=203) or placebo (n=201) for 24 weeks. Changes in measures of cognition, function, caregiver impression, and behaviors were assessed.

The study demonstrated statistically significant differences favoring memantine over placebo on the Severe Impairment Battery (test of cognitive ability) and the AD Cooperative Study – Activities of Daily Living Inventory (test of functional ability) and all secondary outcomes measures using both last observation carried forward and observed cases analyses. Comparable rates of adverse events were seen, except for significantly more reports of confusion in the memantine group (p = 0.01), more reports of headache in the memantine group (not significant), and higher incidences of diarrhea and fecal incontinence in the placebo group (p not reported).

This is the first RCT to examine use of memantine and donepezil in combination, and it demonstrates that memantine plus donepezil was more effective than donepezil alone. However, it is impossible to tell from this study whether memantine alone would have been just as effective as the combination in this population of moderate to severe AD.

Given its demonstrated efficacy in one fair quality study, the combination of memantine and donepezil has limited evidence to support its use in moderate to severe AD. But due to the unknown relative efficacy of memantine alone, the stronger body of evidence on memantine alone in this population, and the high cost of combination therapy, the CMI Dementia Guidelines Workgroup concludes that memantine alone is recommended over the combination of memantine and donepezil.

∗ Tariot PN, Farlow MR, Grossberg GT, Graham SM, McDonald S, Gergel I at al. Memantine treatment in patients with moderate to severe Alzheimer disease already receiving donepezil: a randomized controlled trial. JAMA 2004; 291(3):317-324.


Evidence Table: Memantine for the treatment of dementia

Study, Total n


Results (results are for one study [not pooled] unless noted otherwise) Comments

Areosa Sastre et al., 2003 (Meta-Analysis) # studies found: 7 met inclusion criteria # studies included: 5 had sufficient data for analysis Not possible to pool all the data due to differences in length of trials, doses of treatment, and populations studied

Inclusion criteria for meta-analysis: Clinical trials in Alzheimer’s disease (AD), vascular dementia (VaD) or mixed dementia (MD)

Double-blind, parallel-group, placebo controlled, with randomized and unconfounded treatment assignment to placebo or memantine

Sample selection criteria specified Outcome instruments specified Duration specified

Study populations Number of subjects ranged from 60 to 579 2 studies: vascular dementia (VaD) only 1 study: Alzheimer’s disease (AD) only 3 studies: both VaD and AD 1 trial did not distinguish between different types of dementia

4 studies included mild to moderate disease 1 included mild, moderate, and severe 2 included severe disease only

Study duration: 3 studies lasted 4 or 6 weeks (Phase II trials) 4 studies lasted 28 or 12 weeks (Phase III trials)

Drug: Rx1: memantine 10 to 30mg/day; most common was 20mg/day

Rx2: placebo

Cognition ADAS-Cog* OC* analysis ∆ baseline to week 28: Mean difference –2.83 95% CI –4.37 to –1.29 p=0.0003 favoring memantine 20mg SKT* ∆ baseline to week 6: Mean difference –3.04 95% CI –5.68 to –0.40 p=0.02 favoring memantine 30mg Function - Activities of Daily Living Pooled data for two studies showed advantage of memantine 30 mg/day for 6 weeks (p =0.0003), but the validity of pooling these data is questionable. BGP* subscore showed no advantage of memantine 10 mg/day at 12 weeks and memantine 20 mg/day at 28 weeks. Mood and behavior NOSIE* OC analysis ∆ baseline to week 6: Mean difference 23.30 95% CI 17.83 to 28.77 p<0.00001 favoring memantine 30mg

Global scales CGIC* # improved since baseline: Memantine 10mg, at 12 weeks Rx1: 60/82 Rx2: 38/84 OR 3.30 95% CI 1.72 to 6.33, p=0.0003 Memantine 20mg, at 6 weeks Rx1: 24/41 Rx2: 11/41 OR 3.85 95% CI 1.52 to 9.75, p=0.004 Memantine 20mg, at 28 weeks Rx1: 52/93 Rx2: 38/95 OR 1.90 95% CI 1.07 to 3.40, p=0.03 Memantine 30mg, at 6 weeks Rx1: 20/30 Rx2: 8/29 OR 5.25 95% CI 1.72 to 15.98, p=0.004 Adverse Effects (AEs) Overall: no sig. differences One study: higher rates of restlessness in Rx1(30mg) (OR 13.50; 95% CI 2.71 to 67.19, p=0.001)

Conclusions Results suggest a small beneficial effect of 20 or 30mg/day of memantine on measures of cognitive ability and possible benefits for behaviors.

Effect on ADLs is difficult to interpret due to validity issues for some studies, and showed no significant effect of memantine in others.

A positive effect on global clinical impression has also been shown.

Adverse effects are generally comparable to placebo except for restlessness.

Biases All of the included studies were sponsored by Merz Pharma, the German pharmaceutical company that first patented memantine for CNS and neuorological purposes.

Populations studied varied in type and stage of dementia, and it was difficult to assess efficacy in these different populations.

Trials were short.

* ADAS-Cog: Alzheimer’s Disease Assessment Scale cognitive (11 items; score 0-70; lower is better) OC: Observed cases SKT: Syndrom-Kurtz Test (scoring system not indicated)

BGP: Behavioral Rating Scale for Geriatric Patients (scoring system not indicated) NOSIE: Nurses Observation Scale for Inpatient Evaluation (30 items; higher is better) CGIC: Clinical Global Impression of Change (7-point Likert-type scale; 1=improved much, 4=no change, 7=extreme worsening)


Evidence Table: Memantine for the treatment of dementia

Study, Total n

Treatment Groups Size

& Drug Study Population Results

(∆ from baseline ± Standard Deviation) Comments

Reisberg et al., 2003 (RCT, double-blind) Follow up: 28 weeks Initial N: 252 Final N: 181

Rx1 placebo (n = 126) Rx2 memantine 20mg (n = 126)

Men and women ≥ age 50 Living in community Probable Alzheimer’s disease (AD) according to DSM-IV and NINCDS-ADRDA criteria

Moderate to severe AD Baseline MMSE* 3-14 Stage 5 or 6 on GDS* Stage ≥6a FAS*

Primary Outcomes CIBIC-Plus* LOCF* analysis Rx1: 4.8 ± 1.09 Rx2: 4.5 ± 1.12 p = 0.06 OC* analysis Rx1: 4.7 ± 1.13 Rx2: 4.4 ± 1.12 p = 0.03 ADCDS-ADLsev* LOCF* analysis Rx1: -5.2 ± 6.33 Rx2: -3.1 ± 6.79 p = 0.02 OC* analysis Rx1: -5.9 ± 6.78 Rx2: -2.5 ± 6.27 p = 0.003 Adverse Events (AEs) # (%) discontinued due to AEs Rx1: 22 (17%) Rx2: 13 (10%) p not reported

Secondary Outcomes MMSE* LOCF* analysis Rx1: -1.2 ± 3.02 Rx2: -0.5 ± 2.40 p = 0.18 OC* analysis Rx1: -0.9 ± 3.09 Rx2: -0.6 ± 2.61 p = 0.68 NPI* LOCF* analysis Rx1: 3.8 ± 16.06 Rx2: 0.5 ± 15.76 p = 0.33 OC* analysis Rx1: 2.9 ± 16.13 Rx2: 0.1 ± 15.92 p = 0.60 Resource Utilization in Dementia score Caregivers spent mean of 45.8 hours less per month with patients receiving memantine compared to placebo (p=0.01, 95% CI 10.37 to 81.27). Methods for this score are not specified.

Conclusions Results suggest that memantine is more effective than placebo in patients with moderate to severe dementia in terms of global impression of change and function.

There was no difference between memantine and placebo on impact on cognitive and neuropsychiatric measures.

Adverse effects with memantine were similar to placebo.

Biases Dropout rate was 28% (33% of placebo group and 23% of memantine group); data was not able to be collected for most of these at week 28.

Study supported by Merz Pharma.

* MMSE: Mini-Mental State Examination (score 0-30; higher is better) GDS: Global Deterioration Scale (7-point scale assessing cognition, behavior, and ADLs; mild AD=3, moderate

AD=4, severe AD=6) FAS: Functional Assessment Staging (16 stages [1-5, 6a-e, 7a-f]; lower stage is better) CIBIC-plus: Clinician’s Interview-Based Impression of Change Plus Caregiver Input (7 point Likert-like Scale [LS]

where 4=no change from baseline, 1=marked improvement, 7=marked worsening)

LOCF: Last observation carried forward (performed with intention-to-treat population) OC: Observed cases ADCDS-ADLsev: Alzheimer’s Disease Cooperative Study Activities of Daily Living Inventory (modified for more

severe dementia) (higher is better) NPI: Neuropsychiatric Inventory (12 items; score 0-144; lower is better)


Evidence Table: Memantine for the treatment of mild to moderate vascular dementia (VaD)

Study, Total n

Treatment Groups Size &

Drug

Study Population

Results Mean ∆ from baseline (±SD) unless otherwise noted

Comments

Wilcock et al., 2002 (RCT, double-blind) Follow up: 28 weeks Initial N: 579 Final N: 464 Used intention-to-treat population (ITT) using the last observation carried forward (LOCF) for all those with at least one post-baseline efficacy assessment (n = 548)

Rx1 placebo (n = 284) Rx2 memantine 20 mg/day (10 mg bid) (starting at 5 mg/day and titrating up in weekly increments of 5 mg, thus reaching 20 mg/day in week 4 (n = 295)

Outpatients from geriatric centers or neurology

Probable vascular dementia by DSM-III-R and NINDS-AIREN* criteria, confirmed by a Hachinski ischemic score ≥4

Mild to moderate severity (MMSE* from 10-22 inclusive)

Primary Outcomes (ITT population, LOCF analysis) ADAS-Cog* Rx1: 2.28 (±7.77) Rx2: 0.53 (±7.02) Rx difference (95% CI): 1.75 (0.49 to 3.023) p = 0.000505 Predefined subgroup analysis showed that group with baseline MMSE < 15 showed largest treatment effect on ADAS-Cog (mean Rx difference: 3.17, p = 0.04) CGI-C* Values not reported p = 0.1103 Not significant

Secondary Outcomes (Treated-per-protocol, n = 368) NOSGER* Rx1: 3.45 (±11.08) Rx2: 2.32 (±11.12) Not significant (p value not reported) GBS* Rx1: 2.48 (±15.95) Rx2: 1.65 (±12.00) Not significant (p value not reported) MMSE* Rx1: 0.51 (±3.9) Rx2: 0.24 (±3.8) Not significant (p value not reported) Adverse Events (AEs) # (%) w/ dizziness: Rx1: 23 (8%) Rx2: 33 (11%) p not reported # (%) w/ constipation: Rx1: 12 (4%) Rx2: 30 (10%) p not reported Slightly more general, GI, and respiratory disorders in Rx2 group. No significant differences in serious AEs between groups.

Conclusions Significant differences on ADAS-Cog scores were demonstrated, favoring the memantine group. Although both groups declined on this cognitive measure, the memantine group showed statistically significantly less decline.

No significant differences were found on a clinical global change scale or any of the secondary measures.

Dizziness, constipation, and other general, gastrointestinal, and respiratory disorders appeared in the memantine group compared to placebo, but there were no significant differences in serious AEs or withdrawal due to AEs.

Biases None noted.

* NINDS-AIREN: National Institute of Neurological Disorders and Stroke and the Association Internationale pour la Recherche et l’Enseignement en Neurosciences ADAS-Cog: Alzheimer’s Disease Assessment Scale cognitive (11 items; score 0-70; lower is better) CGI-C: Clinical Global Impression of Change (7-point Likert-type scale; 1=improved much, 4=no change, 7=extreme worsening) NOSGER: Nurse’s Observation Scale for Geriatric Patients (includes rating of memory, IADLs, self-care, mood, social behavior, and disturbing behavior (30 items in 6 domains, each on 5-point scale; direction not indicated) GBS: Gottfries-Brane-Steen Scale (comprehensive global assessment based on interview between clinician and patient and caregiver; 27 items; score 0-162; lower is better) MMSE: Mini-Mental State Examination (score 0-30; higher is better)


Evidence Table: Resource utilization analysis of memantine for moderate to severe Alzheimer’s disease (AD)

Study, Total n

Treatment Groups Size & Drug Study Population Results

Mean ± SD per month Comments

Wimo et al., 2003 (RCT, double-blind) This study was prospectively included as part of a phase III pharmacoeconomic trial Follow up: 28 weeks Initial N: 252 Final N: 166 formed the treated-per-protocol (TPP) subset for the health economic analyses

Rx1 placebo (n = 76) Rx2 memantine 20 mg/day (10 mg bid) (n = 90) Costing process consisted of 2 phases: Phase 1: use of resources was quantified

Phase 2: costs calculated by multiplication of resource use by a per item/per diem cost

Patient and caregiver resource utilization data were captured by Resource Utilization in Dementia (RUD*) questionnaire at baseline, week 12, and week 28.

Outpatients age ≥ 50 with probable Alzheimer’s disease (AD) by DSM-IV and NINCDS-ADRDA* criteria

Moderate to severe disease (MMSE* 3-14, stage 5 or 6 on the GDS*, and stage 6a or greater on the FAST*)

Primary Outcomes Resource Utilization (RUD* scale – TPP population) # hours caregivers spent on caring tasks and supervision Rx1: 455.64 ± 232.70 Rx2: 413.46 ± 234.27 p not reported LOCF analysis on ITT population showed similar results favoring memantine – values not reported. (difference = 45.82 hours, p = 0.01) # hours work loss of caregivers Rx1: 14.05 ± 38.72 Rx2: 9.81 ± 29.60 p not reported Resource utilization of caregivers (e.g., hospital days, primary care visits) No significant differences # patients moved from community to institutional setting during study Rx1: 5 Rx2: 1 p = 0.04 Time to institutionalization Values not reported favoring memantine; p = 0.052

Costs (1999 US dollars) (Cost sources were identified, including literature, Medicare fee schedule, surveys of small numbers of providers, and Bureau of Labor Statistics) Data were positively skewed (mean costs > median costs)

Data highly variable (large SDs) Total caregiver costs Rx1: 7231.18 ± 4025.14 Rx2: 6934.99 ± 4348.25 Difference: 295.19 favoring Rx2 p not reported Total societal costs Rx1: 7834.51 ± 4278.54 Rx2: 7247.22 ± 4346.65 Difference: 587.29 favoring Rx2 p not reported Patient direct medical costs Rx1: 74.32 ± 129.68 Rx2: 234.00 ± 95.01 Difference: 159.68 favoring Rx1 p not reported Patient non-direct medical costs Results favored memantine (lower costs). Text cites lower rates of institutionalization, but utilization table does not outline those rates.

Conclusions Resource utilization analysis showed an advantage of memantine in terms of caregiver time.

Costs related to caregivers and to society as a whole favored memantine, though statistical significance of these differences was not reported.

Patient direct medical costs were higher in the memantine group (mainly due to the cost of the medication).

Patient direct non-medical costs were lower in the memantine group.

Biases The TPP population does not include those who dropped out of the study, who may have been different from the population that completed the study (e.g., less effect on resource utilization).

First two authors have acted as consultants to several pharmaceutical companies that make anti-dementia drugs, but have no employment, stocks, or research grants from these companies.

Other authors are employees of Merz Pharmaceuticals, manufacturer of memantine.

* NINCDS-ADRDA: National Institute of Neurological Disorders and Stroke/Alzheimer’s Disease and Related Disorders Association MMSE: Mini-Mental State Examination (score 0-30; higher is better) GDS: Global Deterioration Scale (7-point scale assessing cognition, behavior, and ADLs; mild AD=3, moderate AD=4, severe AD=6)

FAST: Functional Assessment Staging (16 stages [1-5, 6a-e, 7a-f]; lower stage is better) RUD: Resource Utilization in Dementia (structured interview of patient’s caregiver consisting of baseline and follow-up questionnaires on caregiver time, caregiver productivity losses, caregiver and patient direct medical resource utilization, and patient direct non-medical resource utilization)


Evidence Table: Addition of Memantine to Donepezil Therapy (Off-cycle revision, May 2004)

Study, Total n

Treatment Groups Size


(∆ from baseline least squares mean [SE]) Comments

Tariot et al., 2004 (RCT, double-blind) Follow up: 24 weeks Initial N: 404 Final N: 322 Randomization scheme reported. Power calculations reported. Efficacy analyses based on modified intent-to-treat (ITT) population (patients who received ≥1 dose of study medication and completed ≥1 post-baseline assessment of primary outcomes [n=395]) using both observed cases (OC) and last observation carried forward (LOCF)

Rx1 placebo (n = 201) Rx2 memantine 5 mg/day (increased to 20 mg/day in 5 mg weekly increments) (n = 203)

Men and women ≥ age 50 Diagnosis of probable Alzheimer’s disease (AD) by NINCDS-ADRDA* criteria

MMSE* score of 5-14 at both screening and baseline

Ongoing therapy with donepezil for >6 months before entrance into trial; stable dose (5-10 mg/day) for >3 months

Reliable caregiver Residence in community

Primary Outcomes SIB* LOCF* analysis Rx1: -2.5 (0.69) Rx2: 0.9 (0.67) p < 0.001 OC* analysis Rx1: -2.4 (0.74) Rx2: 1.0 (0.70) p < 0.001 ADCS-ADL19* LOCF* analysis Rx1: -3.4 (0.51) Rx2: -2.0 (0.50) p = 0.03 OC* analysis Rx1: -3.3 (0.55) Rx2: -1.7 (0.51) p = 0.02 Adverse Events (AEs) # (%) discontinued due to AEs Rx1: 25 (12.4%) Rx2: 15 (7.4%) p not reported Confusion: Headache: Rx1: 2.0% Rx1: 2.5% Rx2: 7.9% Rx2: 6.4% p = 0.01 p = 0.09

Secondary Outcomes CIBIC-Plus* LOCF* analysis Rx1: 4.66 (0.075) Rx2: 4.41 (0.074) p = 0.03 OC* analysis Rx1: 4.64 (0.087) Rx2: 4.38 (0.081) p = 0.03 NPI* LOCF* analysis Rx1: 3.7 (0.99) Rx2: -0.1 (0.98) p = 0.002 OC* analysis Rx1: 2.9 (1.06) Rx2: -0.5 (0.99) p = 0.01 BGP* LOCF* analysis Rx1: 2.3 (0.38) Rx2: 0.8 (0.37) p = 0.001 OC* analysis Rx1: 2.2 (0.40) Rx2: 0.6 (0.37) p = 0.001

Conclusions Memantine demonstrated statistically significant benefits in primary and secondary outcomes compared to placebo in patients already taking donepezil.

Biases/Limitations Donepezil approved for mild to moderate AD and memantine approved for moderate to severe AD, so would expect memantine to have benefit in this study population. Unable to determine if combination is more effective than memantine only would have been.

Study does not adequately compare use of combination to draw firm conclusions. An RCT that examines use of memantine only, donepezil only, and combination in groups segregated by dementia severity would yield more definitive results.

Significant differences between groups in completion of study (Rx1: 74.6% vs. Rx2: 85.1%, p = 0.01)

* NINCDS-ADRDA: National Institute of Neurological Disorders and Stroke/Alzheimer’s Disease and Related Disorders Association

MMSE: Mini-Mental State Examination (score 0-30; higher is better) SIB: Severe Impairment Battery (test of cognitive ability) (score 0-100; higher is better) ADCS-ADL19: AD Cooperative Study – Activities of Daily Living Inventory (19-item subset of original 42-item

inventory) (score 0-54; higher is better)

CIBIC-plus: Clinician’s Interview-Based Impression of Change Plus Caregiver Input (7 point Likert-like Scale [LS] where 4=no change from baseline, 1=marked improvement, 7=marked worsening)

NPI: Neuropsychiatric Inventory (12 items; score 0-144; lower is better) BGP: Behavioral Rating Scale for Geriatric Patients (35 items; score 0-70; lower is better)


Pharmacological Management of Behaviors

Problem Formulation

Clinical Question:What is the role of antipsychotics, anticonvulsants, benzodiazepines,acetylcholinesterase inhibitors, trazadone, or buspirone in themanagement of behavioral and psychological symptoms in dementia?


To assist primary care physicians and other health care professionals inthe outpatient primary care setting to effectively manage behaviorsassociated with dementia.

Population: Men and women with dementia who exhibit behaviors that aredisturbing to themselves or others

Health Problem: Aggression, agitation, hallucinations, delusions, restlessness, orcombativeness associated with dementia


� Antipsychotics:� Haloperidol� Olanzapine� Risperidone� Thioridazine� Quetiapine� Ziprasidone� Aripiprazole

� Benzodiazepines:� Alprazolam� Oxazepam� Lorazepam� Diazepam� Chlordiazepoxide� Clonazepam� Triazolam

� Anticonvulsants:� Valproic acid� Gabapentin� Topiramate� Tiagabine� Carbamazepine� Oxcarbazepine� Phenytoin� Levetiracetam� Zonisamide� Lamotrigine

� Acetylcholinesterase inhibitors:� Donepezil� Rivastigmine� Galantamine

� Buspirone� Trazadone� No treatment


Setting: Outpatient office visit

Most ImportantHealth Outcomes

� Hallucinations� Delusions� Aggression� Agitation

� Restlessness� Combativeness� Patient/family satisfaction� Cerebrovascular events



� Drug adverse effects (i.e., extrapyramidal symptoms [EPS]/tardivedyskinesia, somnolence, falls, syncope, hypotension, stroke,hepatotoxicity, bone marrow toxicity)

� Impact on cognition� Impact on function


� Global caregiver impression


Evidence Search

Only RCTs, meta-analyses, or systematic reviews were included that studied the use ofantipsychotics, anticonvulsants, benzodiazepines, acetylcholinesterase inhibitors, buspirone, ortrazodone (compared to placebo or each other) in people with behaviors related to dementia. Because this topic was not addressed in the 2002 version of the CMI Dementia Guidelines,literature searches extended back to 1965.


TimeFrame:

#Found:

# inET:



and RCTs 5/27/03 12 *


1965 –6/6/03 1

1(Coch-rane)

PubMed

("AntipsychoticAgents"[MeSH] OR"Haloperidol"[MeSH] OR"Risperidone"[MeSH] OR"Thioridazine"[MeSH] OR"Olanzapine"[TEXT] OR"Quetiapine"[TEXT] OR"Ziprasidone"[TEXT] OR"Aripiprazole"[TEXT]) AND("Dementia"[MeSH] OR"Alzheimer Disease"[MeSH])AND ("Aggression"[MeSH]OR "Hallucinations"[MeSH]OR "Delusions"[MeSH] OR"Agitation"[TEXT])


1965 –6/6/03 50 5


1965 –6/6/03 0 0

PubMed

("Anticonvulsants"[MeSH]OR "Valproic Acid"[MeSH]OR "Carbamazepine"[MeSH]OR "Phenytoin"[MeSH] OR"Gabapentin"[TEXT] OR"Topiramate"[TEXT] OR"Tiagabine"[TEXT] OR"Oxcarbazepine"[TEXT] OR"Levetiracetam"[TEXT] OR"Zonisamide"[TEXT] OR"Lamotrigine"[TEXT]) AND("Alzheimer Disease"[MeSH]OR "Dementia"[MeSH])


1965 –6/6/03 47 1

* Information from the Clinical Evidence review was not included in evidence tables but was included verbatim inthe rationale statement. When articles found in the evidence search were represented in the Clinical Evidencesummary, evidence tables were not created.



TimeFrame:

#Found:

# inET:


1965 –6/6/03 0 0

PubMed

"Benzodiazepines"[MeSH]AND ("AlzheimerDisease"[MeSH] OR"Dementia"[MeSH])


1965 –6/6/03 35 0


1965 –6/18/03 0 0

PubMed

("Alzheimer Disease"[MeSH]OR "Dementia"[MeSH] OR"mixed dementia"[TEXT] OR"Mild CognitiveImpairment"[TEXT] OR"MCI"[TEXT]) AND("donepezil"[TEXT] OR"rivastigmine"[TEXT] OR"galantamine"[TEXT]) AND("Aggression"[MeSH] OR"Psychotic Disorders"[MeSH]OR "psychosis"[TEXT] OR"Hallucinations"[MeSH] OR"hallucinations"[TEXT] OR"agitation"[TEXT] OR"restlessness"[TEXT] OR"combativeness"[TEXT])


1965 –6/18/03 8 0


1965 –6/12/03 0 0

PubMed

("Buspirone"[MeSH] OR"buspar"[TEXT] OR"buspirone"[TEXT]) AND("Alzheimer Disease"[MeSH]OR "Dementia"[MeSH] OR"dementia"[TEXT]) Clinical Trial,

English, Human1965 –6/12/03 1 0



TimeFrame:

#Found:

# inET:


1965 –8/19/03 0 0

PubMed

("Dementia"[MeSH] OR"Dementia/drugtherapy"[MeSH]) AND("Trazodone"[MeSH] OR"trazodone"[TEXT]) AND("Aggression"[MeSH] OR"Hallucinations"[MeSH] OR"Delusions"[MeSH] OR"Agitation"[TEXT])


1965 –8/19/03 6 0


1965 –9/19/03 0 0

PubMed

(("Dementia"[MeSH] OR"Dementia/drug therapy"[MeSH]OR "Alzheimer Disease"[MeSH])AND (("AntipsychoticAgents"[MeSH] OR"Haloperidol"[MeSH] OR"Risperidone"[MeSH] OR"Thioridazine"[MeSH] OR"Olanzapine"[TEXT] OR"Quetiapine"[TEXT] OR"Ziprasidone"[TEXT] OR"Aripiprazole"[TEXT]) OR("Anticonvulsants"[MeSH] OR"Valproic Acid"[MeSH] OR"Carbamazepine"[MeSH] OR"Phenytoin"[MeSH] OR"Gabapentin"[TEXT] OR"Topiramate"[TEXT] OR"Tiagabine"[TEXT] OR"Oxcarbazepine"[TEXT] OR"Levetiracetam"[TEXT] OR"Zonisamide"[TEXT] OR"Lamotrigine"[TEXT]) OR"Benzodiazepines"[MeSH] OR("donepezil"[TEXT] OR"rivastigmine"[TEXT] OR"galantamine"[TEXT]) OR("Buspirone"[MeSH] OR"buspar"[TEXT] OR"buspirone"[TEXT]) OR("Trazodone"[MeSH] OR"trazodone"[TEXT]) AND("wandering"[TEXT] OR"wander"[TEXT] OR"yelling"[TEXT] OR"apathy"[TEXT] OR"vocalization"[TEXT]))


1965 –9/19/03 5 0


Other Information Sources� In May 2003, the Agency for Healthcare Research and Quality (AHRQ) released a

systematic evidence review conducted for the U.S. Preventive Services Task Force(USPSTF) on screening for dementia. (27) This updated a 1996 review where the USPSTFfound insufficient evidence to recommend for or against screening for dementia.(28) The newreview was conducted to consider more recent studies (literature published up to September1, 2002) concerning screening tests as well as both pharmacologic and caregiverinterventions. The systematic review of the literature sought to answer nine key questions:1. What is the direct effect of dementia screening on outcomes?2. How common is undiagnosed dementia?3. How accurate are the screening tests?4. How effective is primary treatment of potentially reversible or irreversible dementia?5. How effective is secondary treatment for dementia (pharmacologic and

nonpharmacologic interventions)? 6. How effective are interventions for caregivers of people with mild to moderate dementia?7. What are the adverse effects of screening and early treatment of dementia?8. What is the cost of dementia screening?9. What are the adverse effects of dementia treatment?




Guideline: Pharmacological management of behaviors

Recommendations: Certain antipsychotics or anticonvulsants are options formanagement of behavioral and psychological symptoms related todementia including aggression and agitation. (Note: These are notFDA-approved indications.)� Olanzapine is an option for agitation, hallucinations, and

delusions. � Risperidone is an option for aggressive, verbal non-aggressive

(e.g., grunting), and psychotic symptoms, but there is noevidence to support its use for physical non-aggressivebehaviors like wandering, fidgeting, and hoarding.

� Haloperidol is an option for aggressive behavior, but there isno evidence to support its use in agitation (such as wandering,crying out).

� Carbamazepine is an option for management of agitation andaggression.

With use of these agents, benefits with regard to behaviors mustbe weighed against risk of side effects. There is no evidence to support the use of antipsychotics,anticonvulsants, or benzodiazepines for wandering, spontaneousvocalization, or apathy.

There is insufficient evidence to support the use ofacetylcholinesterase inhibitors (donepezil, galantamine,rivastigmine), thioridazine, sodium valproate, or trazodone forthe management of behavioral and psychological symptomsrelated to dementia, and risk of side effects should be considered.

There is no evidence to support the use of other antipsychotics(including quetiapine) and anticonvulsants, benzodiazepines, orbuspirone for the management of behavioral and psychologicalsymptoms related to dementia.



Rationale: Supporting Evidence for Pharmacological Management of Behaviors� Clinical Evidence contains a summary of the evidence related to the effects of treatments

(including antipsychotics) on behavioral and psychological symptoms of dementia (searchdate: October 2002). (122) Excerpts from that review are included here verbatim. In instanceswhere additional relevant articles were found or where drugs of interest were not included inthe Clinical Evidence review, they were reviewed and included here.

� Note: many of the studies examining management of behaviors took place in the nursinghome setting. It is unknown how these results translate to the general population of peoplewith dementia-related behaviors living in the community.

Supporting Evidence for Antipsychotics versus Placebo Haloperidol versus placebo: BENEFITS: � Clinical Evidence said: “We found one systematic review of haloperidol for agitation in

various types of dementia, including Alzheimer's disease and vascular dementia (searchdate 2000, 5 RCTs, none of which were included in the older review). (123) It found nosignificant difference in agitation at 6–16 weeks between haloperidol and placebo(change in symptoms from baseline measured by the Cohen-Mansfield AgitationInventory or the psychomotor score of the Behavioural Symptoms Scale for Dementia;WMD –0.48, 95% CI –1.43 to +0.53). However, it found that haloperidol significantlyreduced aggression from baseline at 3–6 weeks compared with placebo (2 RCTs, 240people: WMD –1.11, 95% CI –2.02 to –0.11).” (122)

� Conclusion: Haloperidol is more effective than placebo at improving aggression, butnot agitation. CMI Dementia Guidelines Workgroup members cited risk of side effectswith haloperidol in rationale for making this an option for aggression, rather thanrecommending it.

Risperidone versus placebo: BENEFITS: � Brodaty et al. conducted a double-blind RCT comparing risperidone to placebo in 345

nursing home patients with dementia and aggression. (124) This study found that oralrisperidone (flexible dose – average dose 0.95 mg/day) was significantly more effective atreducing aggressive, verbal non-aggressive, and psychotic symptoms in these patients.Least-squares mean change from baseline to endpoint on the Cohen-Mansfield AgitationInventory (CMAI) total aggression score was –3.1 points for placebo and –7.5 points forrisperidone (p� .001); LS mean change on the CMAI total non-aggression score was –2.8for placebo and –7.3 for risperidone (p=.002). The LS mean change on the psychoticsymptom subscale of the BEHAVE-AD tool (sum of the paranoid and delusional ideationand hallucination subscales) was –0.7 for placebo and –2.0 for risperidone (p=.004).Clinical Global Impression of investigator and caregiver also indicated greaterimprovement in the risperidone group (p�.001).

� Clinical Evidence said: “We found two RCTs. (125, 126) The first RCT (double blind, 625people with moderate to severe dementia plus behavioural and psychological symptoms,73% with Alzheimer's disease, mean age 83 years, 68% women) compared risperidone


versus placebo over 12 weeks. (125) A response was defined as a reduction of at least 50%in the Behave-AD scale. It found that risperidone (1 and 2 mg daily) significantlyimproved the chance of responding over 12 weeks compared with placebo (45% withrisperidone 1 mg v 33% with placebo v 50% with risperidone 2 mg; for risperidone 1 mgv placebo NNT 9, 95% CI 5 to 100; for risperidone 2 mg v placebo NNT 6, 95% CI 4 to17). Gender and the type of dementia did not significantly affect the results. The secondRCT (344 people with agitation and severe dementia, 67% with Alzheimer's, 26% withvascular dementia, mean age 81 years, 56% women) compared adjusted doses ofrisperidone (mean dose 1.1 mg) versus placebo or haloperidol (mean dose 1.2 mg) over13 weeks for the treatment of behavioural symptoms. (126) A response was defined as areduction of at least 30% in the Behave-AD scale. It found no significant difference in theproportion of people who responded over 13 weeks between risperidone and placebo(37/68 [54%] with risperidone v 35/74 [47%] with placebo; ARI +7%, 95% CI, –9% to+23%; see below for risperidone v haloperidol).” (122)

HARMS: � On April 25, 2003, Janssen Pharmaceutica and the FDA revised the warnings section of

the prescribing information for Risperidal (risperidone) to indicate that cerebrovascularevents including fatalities were reported in trials of risperidone in elderly patients withdementia-related psychosis. They stated that Risperidal has not been shown to be safe oreffective in the treatment of patients with dementia-related psychosis.

� The RCT by Brodaty et al. found that the incidence of serious adverse events (includinginjury, cerebrovascular disorder, pneumonia, and accidental “overdose”) occurred in16.8% of the risperidone group compared to 8.8% of the placebo group. (124) Statisticalsignificance of that difference was not reported. In the risperidone group, 5 patients had astroke and 1 had a transient ischemic attack (TIA). The authors cited predisposingmedical risk factors in these patients. The incidence of extrapyramidal signs was higherin the risperidone group compared to baseline (p =.043, significant) and placebo (p =.407, not significant). Somnolence and urinary tract infection were among the mostcommon side effects, and occurred more frequently in the risperidone group (p notreported).

� Clinical Evidence said: “The first RCT found that discontinuation because of adverseevents was more common with high dose risperidone than with low dose risperidone orplacebo (16% with 1 mg risperidone v 24% with 2 mg v 12% with placebo v 8% with 0.5mg). (125)” (122)

� The DrugDex Drug Evaluation for risperidone lists many precautions, including QTprolongation, stroke and transient ischemic attack in elderly treated for dementia-relatedpsychosis, and tardive dyskinesia.

� Conclusion: Risperidone is more effective than placebo at improving aggressive, verbalnon-aggressive, and psychotic symptoms in people with dementia. However, the risk ofserious adverse events, including stroke and ischemic attack, precludes routine use ofthis drug in older adults with dementia. It is an option for the behaviors outlinedabove, but the risk of these serious adverse events must be carefully balanced with thepotential benefits.


Olanzapine versus placebo: BENEFITS: � Clinical Evidence: “We found one RCT (double blind, 6 wks' duration, 206 elderly US

nursing home residents with Alzheimer's disease [177 people] or Lewy body dementia[29 people] plus psychotic or behavioural symptoms). (127) The RCT compared olanzapine(given as a fixed dose of 5, 10, or 15 mg daily) versus placebo. Agitation, hallucinations,and delusions were improved by the two lower doses but not by the highest dose ofolanzapine compared with placebo (subscale of the Neuropsychiatric Inventory [nursinghome version]: –7.6 with olanzapine 5 mg v –6.1 with olanzapine 10 mg v –4.9 witholanzapine 15 mg v –3.7 with placebo).” (122)

� Two post-hoc subgroup analyses of the RCT by Street et al. were also reviewed. (128, 129)

Cummings et al. examined the impact of olanzapine (5mg, 10mg, 15mg) compared toplacebo in the 29 people from the original study who met criteria for Dementia withLewy Bodies (DLB). (128) The study found a significant decrease in delusions (olanzapine5mg and 10mg) and hallucinations (olanzapine 5mg) compared to placebo. There was nosignificant effect of any dose of olanzapine on agitation. Olanzapine 15mg was similar toplacebo. Clark et al. examined the effect of olanzapine (5mg, 10mg, 15mg) on those 165patients from the original RCT who had no or low levels of psychotic symptoms(hallucinations and delusions). (129) The study found that subjects with no or fewpsychotic symptoms at baseline who received olanzapine 10mg and 15mg hadsignificantly less change from baseline on total psychotic symptom scores compared toplacebo. The 5mg dose did not show a statistically significant difference with regard tohallucinations, delusions, or both. No dose of olanzapine was significantly different thanplacebo in delusions score among patients without delusions at baseline. The short-termfollow up, small subgroups (in some cases), and retrospective post-hoc structure of theseanalyses lead to the determination that further studies are needed to understand the impactof olanzapine on behaviors among these subgroups.

HARMS: � Clinical Evidence said: “The RCT found that olanzapine increased sedation (25% with 5

mg olanzapine v 26% with 10 mg v 36% with 15 mg v 6% with placebo) and gaitdisturbance compared with placebo (20% with 5 mg olanzapine v 14% with 10 mg v 17%with 15 mg v 2% with placebo). (127)” (122)

� The DrugDex Drug Evaluation for olanzapine lists several precautions, including signsand symptoms of tardive dyskinesia.

� Conclusion: Olanzapine has been shown to be more effective than placebo at reducingagitation, hallucinations, and delusions in people with Alzheimer’s disease and reducinghallucinations and delusions in people with Dementia with Lewy Bodies, thoughstatistical significance of these differences was not reported by Clinical Evidence. Thesestudies showed that increased sedation and gait disturbance is found in 20-35% ofpatients taking olanzapine. This risk precludes recommending use of this drug for allpatients with behavioral disturbances related to dementia. Instead, it is an option forbehaviors cited above; risk of side effects should be carefully balanced with potentialbenefits.


Thioridazine versus placebo: BENEFITS: � A meta-analysis conducted for the Cochrane Collaboration examined the literature on

efficacy of thioridazine for treatment of behaviors associated with dementia, including 8studies of 3-8 weeks duration. (130) Two of the trials included in the meta-analysiscompared thioridazine to placebo, and the meta-analysis revealed that thioridazine wasmore effective than placebo at reducing anxiety as measured by the Hamilton AnxietyScale. However, there was no significant effect on Clinical Global Change scales. Noother placebo-controlled data were analyzable.

HARMS:� The meta-analysis found a non-significant trend for adverse effects with thioridazine

compared to placebo. (130) Because the studies included were of short duration, it ispossible that adverse events had not yet emerged.

� Conclusion: Thioridazine is more effective than placebo at reducing anxiety scores inpatients with dementia, but clinical significance of this is unknown. In addition, CMIDementia Workgroup members cited side effects as a deterrent against use of thisagent.

Other Considerations: � Clinical Evidence said: “High response rates with placebo indicate that many behavioural

problems resolve spontaneously in the short term. Most people with dementia aresensitive to adverse effects from antipsychotics, especially sedation and extrapyramidalsymptoms. People with Lewy body dementia are particularly sensitive to these adverseeffects, (131) suggesting that antipsychotics have a poor balance of benefits and harms inpeople with Lewy body dementia. More studies are needed to determine whether neweratypical antipsychotics have a better ratio of benefits to harms than older antipsychotics.”(122)

� A small (36 subjects) double-blind randomized baseline-treatment controlled trial wasconducted by Bridges-Parlet et al. to examine the impact of withdrawal of antipsychotictherapy among demented nursing home patients with physically aggressive behavior(PAB). (132) This study found no significant difference in observed instances of PABbetween those patients who had been withdrawn from antipsychotic therapy and thosewho had not. Wide variations in individual levels of PAB and baseline levels of PABmade it difficult to draw firm conclusions.

� No recent (2000 or later) national guidelines were found that specifically and thoroughlyaddress the issue of managing dementia-related behaviors with antipsychotics. However,the USPSTF systematic evidence review on screening for dementia included the keyquestion, “Do pharmacologic interventions of potentially reversible or irreversibledementia improve outcomes?”(27) Regarding the management of behaviors, they onlyconsidered neuroleptics (antipsychotics). Most of the articles included in their reviewwere also included in the Clinical Evidence reviews used by the CMI DementiaGuidelines Workgroup. Their overall conclusion was that there is some evidence thatantipsychotics improve the behavioral problems related to dementia among patients withmoderate to severe stages and who are living in institutional settings.


� Overall conclusion: If the decision is made to manage dementia-related behaviors withantipsychotics, these agents should be used conservatively. Some behaviors will resolvewith time, and some patients can be successfully withdrawn from antipsychoticswithout increasing behaviors. In addition, adverse effects are common withantipsychotics (more so in Dementia with Lewy Bodies) and should be weighed againstpotential benefit of these agents.

Supporting Evidence for Anticonvulsants versus PlaceboCarbamazepine versus placebo: BENEFITS: � Clinical Evidence said: “We found one RCT (single blind, 51 nursing home patients with

agitation and Alzheimer's disease, vascular dementia or mixed Alzheimer's disease andvascular dementia, 6 wks' duration) comparing carbamazepine (individualised doses;modal dose 300 mg; mean serum level 5.3 µg/mL) versus placebo. (133) It found thatcarbamazepine significantly improved a measure of agitation and aggression (mean totalBrief Psychiatric Rating Scale score: 7.7 with carbamazepine v 0.9 with placebo) and ameasure of global status compared with placebo (Clinical Global Impressions rating:77% with carbamazepine v 21% with placebo).” (122)

� An extension of this study, including a 3-week washout period followed by 12 weeks ofopen carbamazepine, supported the earlier findings. (134) Patients who had previouslybeen taking carbamazepine demonstrated significant worsening of behaviors compared toplacebo during the washout period. Efficacy in reducing behavioral symptoms continuedover 12 weeks. However, the study did not use an intent-to-treat analysis and dropoutover the 12-week open study was large.

HARMS: � Clinical Evidence said: “The RCT found that adverse effects were significantly more

common with carbamazepine than with placebo (16/27 [59%] with carbamazepine v 7/24[29%] with placebo; P = 0.003). (133) These were considered clinically significant in twocases: 1 person with tics, 1 with ataxia. Carbamazepine in the elderly may cause cardiactoxicity.” (122)

� The extension of that study found that 4 cases of serious adverse events (AEs) (includingataxia, UTI and deep venous thrombosis) and 2 deaths. (134) The authors concluded thatonly the ataxia (which was reversed upon discontinuation) was likely to have beenattributable to carbamazepine therapy. In addition, there were 17 falls (11 with injury)during the open use of carbamazepine. The lack of a control group during this phasemakes it difficult to determine the relationship between the AEs and the drug

� The DrugDex Drug Evaluation for carbamazepine lists a precaution for use of this drug inelderly patients.

� Conclusion: Carbamazepine has been found in one RCT with 51 people (and itsextension studies) to be more effective than placebo at improving agitation, aggression,and global status, but serious adverse events including ataxia and falls were shown to bemore common with carbamazepine than placebo. Because there is just one small study(plus extensions) supporting the use of carbamazepine for behaviors related todementia, it is premature to recommend the drug for this population. Instead, it is anoption; risk of side effects should be carefully balanced with potential benefits.


Sodium valproate versus placebo: BENEFITS: � Clinical Evidence said: “We found two RCTs. (135, 136) The first RCT (single blind, 56

people with Alzheimer's disease or vascular dementia in nursing homes, 6 wks' duration)compared sodium valproate versus placebo. (135) It found that when several covariateswere taken into account, sodium valproate significantly improved agitation andaggression compared with placebo (measured by Brief Psychiatric Rating Scale score; P= 0.05 only after adjustment) and a measure of global status (Clinical Global Impressionsrating: 68% with sodium valproate v 52% with placebo; P = 0.06). The second RCT (43people with various types of dementia plus behavioural problems, crossover design)comparing sodium valproate (480 mg daily) versus placebo for 3 weeks found nosignificant difference in aggressive behaviour over 8 weeks after crossover (mean changein Social Dysfunction and Aggression Scale-9 score –0.72 with valproate v –0.72 withplacebo; P = 0.99). (136)” (122)

HARMS: � Clinical Evidence said: “The first RCT found that adverse effects, generally rated as

mild, were significantly more common with sodium valproate than with placebo (68%with sodium valproate v 33% with placebo; P = 0.003). (135)” (122)

� Conclusion: Evidence regarding efficacy of sodium valproate versus placebo foragitation and aggression is conflicting.

Supporting Evidence for Benzodiazepines versus Placebo� There were no randomized controlled trials examining the effect of benzodiazepines versus

placebo in managing behaviors related to dementia. � Conclusion: There is no rigorous evidence to support the use of benzodiazepines for

behaviors related to dementia.

Supporting Evidence for Cholinesterase Inhibitors versus PlaceboDonepezil versus placebo: BENEFITS: � Clinical Evidence said: “We found no systematic review but found two RCTs. (65, 67) The

first RCT (290 people with moderate to severe Alzheimer's disease aged 48–92 years,Mini Mental State Examination [MMSE] score 5–17) compared donepezil (5–10 mgdaily) versus placebo. (65) It found that donepezil significantly improved functional andbehavioural symptoms at 24 weeks compared with placebo (Disability Assessment forDementia score; mean difference 8.23, no 95% CI provided; P < 0.001; NeuropsychiatricInventory score; mean difference 5.64, no 95% CI provided; P < 0.0001). The secondRCT (208 people with mild to moderate Alzheimer's disease, at least 1 symptom on theNeuropsychiatric Inventory Nursing Home version and living in a nursing home) foundno significant difference in psychiatric symptoms after 24 weeks of treatment betweendonepezil and placebo (change in mean Neuropsychiatric Inventory Nursing Homeversion scores –4.9 with donepezil v –2.3 with placebo; reported as non-significant; nofurther data provided). (67)” (122)

� A subanalysis of Feldman et al. was subsequently published by Gauthier et al., focusingon the effect of donepezil on those patients who had behavioral symptoms. (137) It foundsignificant improvement in patients with anxiety, apathy/indifference, and irritability who


were taking donepezil (p � .05), based on the change from baseline on theNeuropsychiatric Inventory (NPI) in the intent-to-treat population. It is not clear that thissubanalysis was planned at the outset of the original study.

HARMS: � Clinical Evidence said: “Adverse effects common to all cholinesterase inhibitors include

anorexia, nausea, vomiting, and diarrhoea. Versus placebo: The RCTs identified by thereview found that donepezil was associated with nausea, vomiting, and diarrhoea, whichtended to be mild and transient. (78) The review found no difference between donepeziland placebo in the proportion of people who withdrew for any cause (27% with 10 mgdonepezil v 20% with 5 mg v 21% with placebo). (78) Long term follow up of peopletaking donepezil (� 10 mg; open label extension) found that 86% experienced at least oneadverse effect, often occurring later in the study. Common adverse events includedagitation (24%), pain (20%), insomnia (11%), and diarrhoea (9%). (138) The firstsubsequent RCT found no significant difference between donepezil and placebo in theproportion of people who experienced any adverse event over 24 weeks (120/144 [83%]v 117/146 [80%]; RR 1.04, 95% CI 0.93 to 1.16). (65)” (122)

� Conclusion: There is insufficient evidence to support the use of donepezil for behaviorsrelated to Alzheimer’s disease.

Galantamine versus placebo: BENEFITS: � Clinical Evidence said: “We found one systematic review (search date 2002), which

identified two RCTs that assessed the effects of galantamine on behavioural andpsychological symptoms. (74) A meta-analysis was not performed because of differencesin length of follow up between the trials. Both trials used the Neuropsychiatric Inventory(NPI); scores range from 0–120, reduction indicates improvement. The first RCT (386people with mild to moderate Alzheimer's disease; MMSE score 10–22) found nosignificant difference in psychiatric symptoms at 3 months between galantamine (12–16mg twice daily) and placebo (mean reduction in NPI score: –0.30 with galantamine v+0.50 with placebo; WMD –0.80, 95% CI –2.67 to +1.07). The second RCT (978 peoplewith mild to moderate Alzheimer's disease; MMSE score 12–24) found that galantamine(16 mg daily) significantly reduced psychiatric symptoms at 6 months compared withplacebo (mean reduction in NPI score –0.10 with galantamine v +2.00 with placebo;WMD –2.10, 95% CI –4.04 to –0.16), but found no significant difference withgalantamine (8 mg daily or 24 mg daily). (74)” (122)

HARMS: � Clinical Evidence said: “Adverse effects common to all cholinesterase inhibitors include

anorexia, nausea, vomiting, and diarrhoea. The review in people with Alzheimer's diseasefound that adverse effects over 6 months were more frequent with larger doses ofgalantamine, including nausea (42% with galantamine 16 mg twice daily v 25% withplacebo) and vomiting (21% with galantamine 16 mg twice daily v 7% with placebo). Italso found that higher doses of galantamine increased the proportion of people whodiscontinued treatment because of adverse effects over 6 months (27% with galantamine16 mg twice daily v 15% with galantamine 12 mg twice daily v 8% with placebo). (74)

The additional RCT comparing galantamine versus placebo in people with vascular


dementia found that more people taking galantamine withdrew because of adverse effects(20% with galantamine v 8% with placebo). (75)” (122)

� Conclusion: There is insufficient evidence to support the use of galantamine forbehaviors related to Alzheimer’s disease.

Rivastigmine versus placebo: BENEFITS:� McKeith et al. conducted a double-blind RCT comparing rivastigmine and placebo in 120

patients with Dementia with Lewy Bodies for 20 weeks plus 3 weeks of washout. (95)

This study found that there was improvement on most items of the NeuropsychiatricInventory with rivastigmine, but only one of these (apathy/indifference) was statisticallysignificant compared to placebo. There was no significant difference betweenrivastigmine and placebo in global impressions of change.

HARMS:� This RCT found that the frequency of nausea, vomiting, anorexia, and somnolence was

significantly higher with rivastigmine compared to placebo. (95)

� Conclusion: There is insufficient evidence to support the use of rivastigmine forbehaviors related to Dementia with Lewy Bodies.

Supporting Evidence for Buspirone versus Placebo� No studies were found that examined the use of buspirone versus placebo for the treatment of

behavioral and psychological symptoms of dementia.� Conclusion: There is no evidence to support the use of buspirone for behaviors related

to dementia.

Supporting Evidence for Trazodone (versus placebo and other drug classes)BENEFITS:� Clinical Evidence said: “We found no systematic review but found two RCTs. (139, 140)

The first RCT (double blind, 28 elderly people with agitated behaviour associated withAlzheimer's disease, vascular dementia or mixed Alzheimer's disease and vasculardementia, 9 wks' duration) compared trazodone (50-250 mg daily) versus haloperidol (1-5 mg daily). (139) It found no significant difference in agitation between the groups, butthe trial was too small to exclude a clinically important difference. The second RCT(double blind, 149 people with Alzheimer's disease and agitated behaviours, 16 wks'duration) compared four treatments: haloperidol (mean dose 1.1 mg daily); trazodone(mean dose 200 mg daily); behaviour management techniques; or placebo. (140) It foundno significant difference in outcome (Alzheimer's Disease Co-operative Study ClinicalGlobal Impression of Change) between the four interventions, but it may have been toosmall to exclude a clinically important difference.” (122)

HARMS:� Clinical Evidence said: “In the first RCT, adverse effects were more common in the

group treated with haloperidol than trazodone. (139) In the second RCT no significantdifferences in adverse events were seen between the trazodone group and the placebogroup. (140) Priapism has been reported with trazodone, occurring in about 1/10 000people.” (122)

Other considerations:


� Clinical Evidence said: “The RCTs were too small to exclude clinically importantdifferences between the interventions. (139)” (122, 140)

� Conclusion: There is insufficient evidence to support the use of trazodone for behaviorsrelated to dementia.

Supporting Evidence for Comparisons within or between Drug Class (Antipsychotics,Anticonvulsants, Benzodiazepines, Cholinesterase Inhibitors, Buspirone)

Comparisons between antipsychotics: BENEFITS: � Chan et al. conducted a double-blind RCT comparing risperidone to haloperidol in 58

Chinese patients with dementia and behavioral symptoms. (141) This study found thatboth drugs significantly reduced the severity of behaviors from baseline to endpoint (asmeasured by the Cohen-Mansfield Agitation Inventory). Risperidone producedstatistically significant improvements from baseline to endpoint on more BEHAVE-ADsub-scales than haloperidol did, but overall there was no significant difference betweenthe efficacy of the two drugs.

� A Cochrane Collaboration meta-analysis on the efficacy of thioridazine included 3studies that compared thioridazine to other antipsychotics (loxapine and zuclopenthixol).(130) The analysis found that thioridazine was not superior on any measures except forfewer adverse events than loxapine.

� Clinical Evidence said: “The review (search date 1995) identified eleven RCTscomparing different antipsychotics. (142) [CMI note: some of these RCTs also includedother agents such as benzodiazepines.] It found no significant difference in efficacyamong haloperidol, diazepam, thioridazine, loxapine, or oxazepam. One subsequent RCTcomparing adjusted doses of risperidone versus haloperidol or placebo found nosignificant difference in the proportion of people who responded over 13 weeks betweenrisperidone and haloperidol. (126) A response was defined as a reduction of at least 30%in the Behave-AD scale.” (122)

HARMS: � The RCT by Chan et al. found that the haloperidol group experienced significant

worsening of extrapyramidal symptoms from baseline to endpoint (p � 0.001) andcompared to risperidone at endpoint (p = 0.001). (141)

� Clinical Evidence said: “In the RCT comparing risperidone versus placebo orhaloperidol, about 18% of people withdrew because of adverse effects from each of thethree arms. (126) Extrapyramidal adverse effects were more common in people receivinghaloperidol than placebo (22% with haloperidol v 15% with risperidone v 11% withplacebo).” (122)

� Conclusion: There is insufficient evidence to recommend one antipsychotic agent overanother for the management of behaviors associated with dementia. Extrapyramidalsymptoms have been shown to be worse with haloperidol than with risperidone.

Comparisons between antipsychotics and benzodiazepines:� A Cochrane Collaboration meta-analysis on the efficacy of thioridazine included 3

studies that compared thioridazine to diazepam, a benzodiazepine. (130) The analysisfound that thioridazine was superior in terms of some anxiety symptoms with similaradverse event profiles. Global clinical evaluations did not favor either treatment.


� Conclusion: There is insufficient evidence to recommend antipsychotics overbenzodiazepines for the management of behaviors associated with dementia. Thoughthioridazine has been shown to improve anxiety compared to diazepam, the clinicalsignificance of that effect is unknown.


Evidence Table: Risperidone for managing behaviors associated with dementia

Study, Total n



Efficacy measured as � from baseline to endpoint Comments

Brodaty et al., 2003 (RCT, double-blind) Follow up: 12 weeks Initial N: 345 randomized (337received at least one dose of studydrug) Final N: 236 Power calculation was done –assumed that a difference of 4.15points on the CMAI total aggressionscale is clinically relevant; calculatedthat 218 patients (109 per treatmentgroup) were required to detectdifference at 5% significance levelwith 80% power. Analysis used intent-to-treatpopulation at included sites.

Rx1: placebo(n = 170) Rx2: risperidonesolution up to 2mg/day(flexible dose)(n = 167) Mean dose was0.95mg/day Randomizationprocess explained;double-blind treatmentperiod was precededby a maximum 7-daysingle-blind washoutperiod

Nursing home patients (age�55) with DSM-IV diagnosis ofdementia of the Alzheimer’stype, vascular dementia, or acombination of the two (mixeddementia) and aggressivebehaviors as exhibited byaggression score on CMAI*:� score of at �4 on at least 1

aggression item, OR� score of 3 on at least 2

aggression items, OR� score of 2 on at least 3

aggression items, OR� 2 aggressive items occurring

at a frequency of 2 and 1 at afrequency of 3

Patients resided in nursinghome for at least one monthprior to enrollment

Primary outcome:CMAI* totalaggression scoreMean change: Rx1: -2.8 Rx2: -7.7 (approximate basedon figure) p � .01 LS mean* Rx1: -3.1 Rx2: -7.5 p � .001

Secondaryoutcomes:CMAI* total non-aggression score LS mean: Rx1: -2.8 Rx2: -7.3 p = .002 BEHAVE-AD totalscore LS mean: Rx1: -2.3 Rx2: -6.8 p � .001 BEHAVE-ADpsychotic score LS mean: Rx1: -0.7 Rx2: -2.0 p = .004 MMSE, FAST No changes wereobserved

CGI* score Significantly greaterimprovement with risperidonecompared to placebo asassessed by investigators andcaregivers p � .001 Adverse Events (AEs)� Injury, somnolence, falls,

and UTI most common AEs;somnolence and UTI higherwith risperidone (p notreported)

� Serious AEs:� Rx1: 15 (8.8%)� Rx2: 28 (16.8%)� p not reported� 5 in Rx2 had stroke, 1

had TIA� ESRS* mean � (SE):� Rx1: 0.5 (0.48)� Rx2: 0.7 (0.35)p�.05 vs. baseline

Conclusions� Oral risperidone (flexible

dose) was significantlymore effective atreducing aggressive andnon-aggressivesymptoms in dementedpatients with aggression.

� Cerebrovascular AEsmore common in patientsreceiving risperidone(authors citedpredisposing medical riskfactors).

� Incidence of anyextrapyramidal signs wasmore common inrisperidone group, but notsignificant vs. placebo.

Biases� Main author is consultant

for Janssen(manufacturer ofrisperidone). Otherauthors have receivedsupport from Janssen.

* CMAI: Cohen-Mansfield Agitation Inventory (29 items; 7 point frequency scale; lower is better) LS mean: Least-squares mean, adjusted for baseline score and investigator MMSE: Mini-Mental State Examination (score 0-30; higher is better)

FAST: Functional Assessment Staging Test (16 stages [1-5, 6a-e, 7a-f]; lower stage is better) CGI: Clinical Global Impression scale (7 point Likert-type scale; 1=improved much, 4=no change, 7=extreme worsening)ESRS: Extrapyramidal Symptom Rating Scale (lower is better)


Evidence Table: Olanzapine for managing behaviors associated with Dementia with Lewy Bodies

Study, Total n



� from baseline on NPI/NH* subscales (p value) Comments

Cummings et al., 2002 (post-hoc analysis of RCT) Follow up: Retrospective post-hocanalysis of double-blind placebocontrolled trial (6 weeks) (127)

Initial N: 29 (for post hoc analysis) Final N: 29

Rx1: placebo (n = 10) Rx2: olanzapine 5mg(n = 5) Rx3: olanzapine 10mg(n = 7) Rx4: olanzapine 15mg(n = 7)

From the complete set of 206nursing home patients in theoriginal study, a subset meetingcriteria for Dementia with LewyBodies were extracted. (Scoresof �1 on hallucination subscaleof NPI/NH* and anextrapyramidal syndrome asexemplified by a non-zero scoreon the Simpson-Angus Scale.)

Delusions Rx1: -1.8 Rx2: -5.6 (0.009) Rx3: -2.9 (0.018) Rx4: -0.71 (0.706) Hallucinations Rx1: -0.1 Rx2: -6.0 (0.009) Rx3: -1.3 (0.432) Rx4: -1.5 (0.606) Agitation Rx1: -0.8 Rx2: -3.4 (0.243) Rx3: -4.0 (0.392) Rx4: -3.1 (0.595)

Delusions/ disruptiveness Rx1: -0.5 Rx2: -2.2 (0.004) Rx3: -1.0 (0.008) Rx4: -0.6 (0.190) Hallucinations/ disruptiveness Rx1: -0.1 Rx2: -2.6 (0.002) Rx3: -0.4 (0.110) Rx4: -1.3 (0.142) Adverse Events (AEs)� No significant difference in

exacerbation of extrapyramidalsymptoms

� No significant difference in otherAEs assessed (cardiovasculardisorders, confusion, delirium,hypotension, somnolence)

Conclusions� Significant decrease in

delusions and hallucinationswith 5 mg olanzapine comparedto placebo; significant decreasein delusions with 10mgolanzapine compared toplacebo.

� No significant effect on agitation.� 15mg olanzapine similar to

placebo. Biases� Small N for each group. � Some variation in baseline

NPI/NH scores.� Unclear whether this post-hoc

analysis was planned fromoutset of original study.

* NPI/NH: Neuropsychiatric Inventory/Nursing Home version (12 items; score 0-144; lower is better)


Evidence Table: Olanzapine for managing behaviors associated with dementia

Study, Total n

Treatment GroupsSize & Drug Study Population Results Comments

Clark et al., 2001 (post-hoc analysis of RCT) Follow up: Retrospectivepost-hoc analysis of double-blind placebo controlled trial(6 weeks) (127)

Initial N: 165 (for post hocanalysis) Final N: 165

“No hallucinations” (n=153) Rx1: placebo (n = 32) Rx2: olanzapine 5mg (n = 42) Rx3: olanzapine 10mg (n = 41) Rx4: olanzapine 15mg (n = 38) “No delusions” (n=87) Rx1: placebo (n = 18) Rx2: olanzapine 5mg (n = 23) Rx3: olanzapine 10mg (n = 21) Rx4: olanzapine 15mg (n = 25) “No psychotic symptoms”(n= 75) (n for each treatment arm notreported.) Rx1: placebo Rx2: olanzapine 5mg Rx3: olanzapine 10mg Rx4: olanzapine 15mg

From the complete set of 206nursing home patients in theoriginal study, a subgroup withno or low levels of psychoticsymptoms at baseline wereidentified:� “No hallucinations”: (n = 153)

Those with no or minimalhallucinations at baseline (�2on hallucinations item ofNPI/NH*)

� “No delusions”: (n = 87)Those with no or minimaldelusions at baseline (�2 ondelusions item of NPI/NH*)

� “No psychotic symptoms”:(n = 75) Those with no orminimal hallucinations ordelusions at baseline (�2 onboth items of NPI/NH*)

Mean change in NPI/NH*psychosis total score among“No psychotic symptoms”subset (p value) Rx1: 2.73 Olanzapine (all dosagegroups): 0.27 (.006) (individual dosage groupsreported in a bar chart only;exact mean change is notlisted) Mean change in NPI/NH*hallucinations total scoreamong “No hallucinationssymptoms” subset (p value) Rx1: 1.25 Rx2: 0.50 (.113) Rx3: 0.29 (.050) Rx4: 0.18 (.033) All olanzapine: 0.33 (.026)

Mean change in NPI/NH*delusions total score among“No delusions symptoms”subset (p value) Rx1: 0.94 Rx2: 0.13 (.169) Rx3: 0.38 (.435) Rx4: 0.08 (.174) All olanzapine: 0.19 (.153)

Conclusions� Significantly less worsening on

psychosis total score andhallucinations total score comparedto placebo.

� Higher doses of olanzapine (10mg,15mg) produced bigger differencein those groups.

� No significant differences ondelusions score.

Biases� Unclear whether this post-hoc

analysis was planned from outsetof original study.

� Absence of historical dataregarding patients’ pre-existingpsychosis.

� Short duration of study; longer termprospective studies are needed.

* NPI/NH: Neuropsychiatric Inventory/Nursing Home version (12 items; score 0-144; lower is better)


Evidence Table: Thioridazine for managing behaviors associated with dementia

Study, Total n

Study Population Treatment Groups Size & Drug Results Comments

Kirchner, 2002 For Cochrane Collaboration(Meta-Analysis) # studies found: 52 # studies included: 8 (double-blind RCTs of 3-8 weeksduration; none describedrandomization process)

� Sample sizes ranged from 30-610 subjects� Mean age was 72.5-80 years� 6 trials did not specify the type of dementia, 1 specified

Alzheimer’s, 1 specified vascular dementia� All trials compared thioridazine (dose range 10-250mg;

mean doses 32.9-95mg) with a control:� 3 trials: diazepam (mean dose 7.2-12mg); 1 of these

had a placebo group� 2 trials: loxapine (mean dose 10.5-17.5mg); 1 of

these had a placebo group� 1 trial: zuclopenthixol (mean dose 6.8mg)� 1 trial: etoperidone (mean dose 100mg)� 1 trial: chlormethiazole (mean dose 723.2mg)

Note: Only thioridazine, diazepam, and loxapine are FDA-approved drugs available in the United States.

vs. placebo Hamilton Anxiety Scale Anxious mood/tension/fear/insomnia items: OR 4.91 95% CI: 3.21-7.50 (significant) Intellect/agitation/ depressedmood items: OR 3.64 95% CI: 2.39-5.54 (significant) Clinical Global Change OR 1.58 95% CI: 0.42-5.96 (not significant) Adverse Events (AEs) OR 0.41 95% CI: 0.09-1.86 (non-significant trend towardADs with thioridazine) vs. etoperidone, loxapine,zuclopenthixol Thioridazine not superior onany measure (except fewerAEs than loxapine)

vs. diazepam Anxiety OR 1.80 95% CI: 1.04-3.10 (significant) – only in one studythat combined items Clinical Global Change OR 2.87 95% CI: 1.30-6.32 (significant) – data pooled onlyfor 2 of the 3 studies found Adverse Events (AEs)OR 0.8495% CI: 0.25-2.82(not significant)

vs. chlormethiazole Crichton Geriatric BehaviouralScaleConfusion/alertness itemsOR 0.3195% CI: 0.11-0.89Continence itemOR 0.2395% CI: 0.07-0.76(Thioridazine significantly worse)

Conclusions� Extremely limited data to support

the use of thioridazine in thetreatment of dementia.

� Only positive effect versus placebowas to reduce anxiety.

� Had same or more AEs thanplacebo, other neuroleptics andsedatives.

� Minimal or no effect on globalratings.

Biases� Studies were generally of short

duration and had a small N.� Inadequacy of published results

made synthesis across trialsdifficult.

� Wide variety of outcome measuresused constrained pooling of data.


Evidence Table: Effects of antipsychotic (neuroleptic) withdrawal in people with dementia

Study, Total n



Observed PAB, (mean � SD)

Baseline Week 4Rx1all

3.86 � 6.97 1.27 � 3.95

Rx1comp-leters

2.60 � 3.62 1.45 � 4.16

Rx2 2.14 � 3.71 4.50 � 8.93

Bridges-Parlet et al., 1997(Randomized double-blindbaseline-treatment-controlledtrial) Follow up: 4 weeks Initial N: 36 Final N: 34 No power calculations werereported. Sample size was based onresources available to conduct thestudy. Completer analysis showed nodifference in results betweencompleters and total.

Rx1 Withdrawal fromneuroleptic (n = 22) Rx2 No withdrawalfrom neuroleptic(n = 14)

Patients from 5 nursing homeswho:� Were receiving a “traditional

neuroleptic” with a stabledose for 3 months prior

� Had a diagnosis of dementia� Had a history of physically

aggressive behavior (PAB)according to referring nursinghome supervisor

� Who resided in a nursinghome

Completion of 4-week study* Rx1: 20 (91%) Rx2: 14 (100%) p � .05 (not significant)

No group differences were statisticallysignificant (p �.05)

Amount of PAB varied widely betweenparticipants, resulting in large standarddeviations for pooled data.

Additional analyses were conducted tominimize inter-subject variability:Change in PAB for individuals (week 5compared to baseline):Rx1: 5 improved, 5 worsened (including the

2 who were withdrawn), 12unchanged

Rx2: 5 worsened, 9 unchanged(no significant difference between groups)

Conclusions� No significant difference in

observed instances of physicallyaggressive behavior betweensubjects who were withdrawn fromantipsychotics and those not.

Biases� These institutionalized patients

may be different (e.g., more severebehaviors) than those being seenin primary care, so applicability ofresults is unknown.

� All eligible participants may nothave been included, so studypatients may have different fromnon-participants.

* Completion of the 4-week study was an outcome of interest because nursing personnel were encouraged to halt a patient’s participation of there were unacceptable levels of physically aggressive behavior.


Evidence Table: Impact of withdrawal of carbamazepine vs. placebo and efficacy of re-starting therapy

Study, Total n



Tariot et al., 1999 (Extension of earlier 6-week,randomized, parallel-group studyof placebo vs. carbamazepine in51 patients) (133)

Follow up: 12 weeks For this extension study: Initial N: 47 Final N:� after 3 weeks of withdrawal: 45� after 6 weeks of open

carbamazepine: 32� after 12 weeks of open

carbamazepine: 25

Original study: Rx1: placebo Rx2: carbamazepine All 47 remainingpatients from originalstudy underwentwithdrawal/ washoutand subsequenttreatment withcarbamazepine Modal dose ofcarbamazepine after 6and 12 weeks of openuse was 300 mg/day.

Criteria from original study :� Probable or possible

Alzheimer’s disease(NINCDS-ADRDA criteria),vascular dementia (DSM-IVcriteria), or mixed dementias

� Exhibited disturbedbehaviors for �2 weeks withBRPS* score �3 on itemsrating tension, hostility,uncooperativeness orexcitement

� Medically stable at time ofenrollment

� Free of other psychotropicsfor �2 weeks beforerandomization (except asneeded chloral hydrate250mg p.o. [max 2 g in 24hours])

Mean � (SD) inTotal BPRS* duringwashout Rx1: 1.9 (6.2) Rx2: 9.8 (5.8) p = .001 (Rx1 vs.Rx2) Mean � (SD) duringopen treatment At 6 weeks vs.washout: Total BPRS*: -10.2 (9.4) p = .0001 BPRS agitation: -3.7 (3.2) p = .0001 BPRS hostility: -2.2 (2.1) p = .0001 BPRS anxiety/depression: -0.9 (2.7) p = .0585 BPRS psychosis: -1.2 (1.7) p =.0006

Mean � (SD) duringopen treatment At 12 weeks vs.washout: Total BPRS*: -18.3 (8.7) p = .0001 BPRS agitation: -5.6 (2.8) p = .0001 BPRS hostility: -3.6 (2.0) p = .0001 BPRS anxiety/depression: -2.7 (2.7) p = .0001 BPRS psychosis: -2.3 (1.7) p = .0001

Adverse Events(AEs) Dropouts duringopen treatment:� 2 died � 4 serious AEs

(including UTIand deep venousthrombosis)

Authors concludethat 1 seriousadverse event waslikely to have beenattributable tocarbamazepinetherapy (ataxia,which reversedwhen therapydiscontinued) Non-dropouts:� 17 falls (11 with

injury)� 10 cases

posturalinstability

Conclusions� During the washout period, there

was a significant worsening ofbehaviors among those who hadbeen taking carbamazepine versusplacebo.

� During open use ofcarbamazepine, efficacy continuedto be observed over 12 weeks.

Biases� Study did not use an intent-to-treat

analysis; drop-out over 12 weekopen study was large.

� Without placebo control during 12week extension of carbamazepineuse, unable to discern relation ofAEs to drug.

* BRPS: Brief Psychiatric Rating Scale (24 symptoms rated 1-7 for severity; lower is better)


Evidence Table: Rivastigmine for managing behaviors in Dementia with Lewy Bodies

Study, Total n



Mean � from baseline to week 20 (SD) Comments

McKeith et al., 2000 (RCT, double-blind) Follow up: 23 weeks (20 weeks oftreatment, 3 weeks of washout) Initial N: 120 Final N: 92 (after 20 weeks oftreatment); 89 (after 3 weekwashout) Power calculations not discussed.Authors stated that original desiredsample size was 172 (based on“feasibility considerations”). Safety analysis included those whoreceived �1 dose of studymedication and �1 safetyevaluation. Efficacy analysis included classicintent-to-treat (ITT), traditional lastobservation carried forward (LOCF)(randomized patients with �1assessment while treated) andobserved cases (OC) (randomizedpatients with review done whiletaking rivastigmine at designatedassessment times).

Rx1 placebo(n = 61) Rx2 rivastigmine1.5mg bid (escalatedby 1.5mg twice dailyfor maximum of 2weeks at each doseuntil 6 mg bid or amaximum well-tolerated maintenancedose was reached.Titration lasted 8weeks)(n = 59)

� Clinical diagnosis of probableLewy-body dementia

� MMSE* � 9� Contact with responsible

caregiver �5 days per week

� People with severeextrapyramidal signs wereexcluded

Primary :NPI-4* ITT analysis Rx1: 0.8 (7.3) Rx2: 2.5 (8.4) p = 0.088 LOCF analysis Rx1: 0.8 (7.4) Rx2: 3.1 (9.1) p =0.045 Secondary:NPI-10* ITT analysis Not reported LOCF analysis Rx1: 1.2 (10.7) Rx2: 5.0 (16.2) p = 0.048 CGC-plus* No significantdifference betweenRx1 and Rx2 inmean score

Adverse Events (AEs)Count (%) Rx1: 46 (75%) Rx2: 54 (92%) p = not reported Frequency of nausea (22; 37%), vomiting(15, 25%), anorexia (11, 19%), andsomnolence (5, 9%) was significantly higherin Rx2 than Rx1. Severe AEs: Rx1: 8 Rx2: 10 Distribution of severe AEs by body systemdid not differ, except for 3 cases of severeagitation in Rx2. UPDRS* motor subscale: No change in parkinsonian symptoms onRx2 compared to baseline and to Rx1.

Conclusions� Improvement on individual items of

NPI seen with rivastigminecompared to placebo, but 9 of 10were not statistically significantdifferences.

� Only significant difference was onapathy/indifference item.

� ITT analysis of NPI-4 subscale didnot show significant difference.

� Anticholinergic AEs significantlyhigher in rivastigmine group.

Biases� Two of the authors are employees

of Novartis (manufacturer ofrivastigmine).

� ITT analysis not reported for NPI-10.

� Efficacy on the full 12-item NPIscale not reported (individualefficacy shown).

� Study may be under-powered.

* MMSE: Mini-Mental State Examination (score 0-30; higher is better) NPI-4: Neuropsychiatric Inventory, 4-item sub-score (sum of scores for delusions, hallucinations, apathy, and depression; lower is better) NPI-10: Neuropsychiatric Inventory, items 1-10 (lower is better) CGC-plus: Clinical Global Change-plus Scale (scoring not indicated) UPDRS: Unified Parkinson’s Disease Rating Scale (measures rigidity, tremor, bradykinesia; score 0-199; lower is better)


Evidence Table: Comparison of antipsychotic treatment (risperidone and haloperidol) for behaviors

Study, Total n



CMAI* score (mean � SD)baseline endpoint p value (pre/post)

Rx1 46.4 � 10.5 36.3 � 10.4 0.000Rx2 48.9 � 14.5 40.8 � 16.9 0.002

Chan et al., 2001 (RCT; double-blind) Follow up: 12 weeks Initial N: 58 Final N: 55 Randomization process notdescribed. Power calculations not reported. Used per treatment analysis, notintent-to-treat analysis.

Rx1 0.5-2mg/day ofhaloperidol (flexibledose) (n = 29) Mean dose: 0.90 mg/day Rx2 0.5-2mg/day ofrisperidone (flexibledose) (n = 29) Mean dose: 0.85 mg/day

� Recruited through inpatient oroutpatient units ofpsychogeriatric departmentsof three hospitals in HongKong

� Chinese patients age � 55with DSM-IV diagnosis ofdementia of Alzheimer’s typeor vascular dementia (orboth)

� Active behavioral symptoms(frequency score of �4 onone and �3 on another itemof the CMAI*)

� Total score of �8 on theBEHAVE-AD* scale

� No statistically significantdifferences between groupsat baseline

BEHAVE-AD* score (mean � SD)Rx1: Significant improvement from baseline to endpoint on scores

for aggressiveness (p = 0.011) but no other subscalesRx2: Significant improvement from baseline to endpoint on scores

for psychosis (p =0.004), activity disturbances (p =0.011),aggressiveness (p =0.019), and diurnal rhythm disturbances(p =0.025).

No statistically significant differences in CMAI total score andBEHAVE-AD sub-scales between Rx1 and Rx2 groups (tested byrepeated measures ANOVA).

Adverse Events (AEs)Rx1: Significant increase in Simpson-Angus Scale score from

baseline to endpoint (p � 0.001) and compared to Rx2 atendpoint (p = 0.001), indicating worsening of extrapyramidalsymptoms (EPS)

Rx2: No significant increase in AEs assessed

Conclusions� Both agents significantly reduced

the severity of behavioral andpsychological symptoms.

� No significant differences betweengroups in terms of efficacy.

� Worsening of extrapyramidalsymptoms with haloperidol.

Biases� Unknown how results in this group

of Chinese dementia patients canbe applied to other ethnic groups.

* CMAI: Cohen-Mansfield Agitation Inventory (29 items; 7 point frequency scale; lower is better) BEHAVE-AD: Behavioral Pathology in Alzheimer’s Disease Rating Scale (lower is better)

FAST: Functional Assessment Staging Rating Scale (16 stages [1-5, 6a-e, 7a-f]; lower stage is better) CMMSE: Cantonese version of Mini-Mental State Exam (score 0-30; lower is better)


Non-Pharmacological Management of Behaviors

Problem Formulation

Clinical Question:What is the role of music therapy, validation therapy, reminiscencetherapy, and reality orientation in the management of behavioral andpsychological symptoms in dementia?


To assist primary care physicians and other health care professionals inthe outpatient primary care setting to effectively manage behaviorsassociated with dementia.

Population: Men and women with dementia who exhibit behaviors that aredisturbing to themselves or others

Health Problem: Aggression, agitation, anxiety, hallucinations, delusions, restlessness, orcombativeness associated with dementia

Health Intervention:� Music therapy� Validation therapy

� Reminiscence therapy� Reality orientation� No treatment




Severity of:� Hallucinations� Delusions� Aggression

� Agitation� Restlessness� Combativeness� Patient/family satisfaction


� Global caregiver impression� Caregiver burden

Side Effects: � Wasted resources


Evidence Search

Only RCTs, meta-analyses, or systematic reviews were included that studied the use of musictherapy, validation therapy, reminiscence therapy, and reality orientation (compared to placeboor compared to each other) in people with behaviors related to dementia. Because this topic was not addressed in the 2002 version of the CMI Dementia Guidelines,literature searches extended back to 1965. A systematic evidence review was found thataddressed this topic (see “Other Information Sources” below). The literature search for thatreview went up to September 2002. We updated the search on this topic to see if any additionalrigorous evidence had been published since that search period.


TimeFrame:

#Found:

# inET:



and RCTs 5/27/03 12 0**


7/1/02 –8/4/03 0 0

PubMed

"Dementia"[MeSH] AND("Music Therapy"[MeSH] OR"validation therapy"[TEXT]OR "reminiscencetherapy"[TEXT] OR "realityorientation"[TEXT]) AND("Anxiety"[MeSH] OR"anxiety"[TEXT] OR"PsychomotorAgitation"[MeSH] OR"Aggression"[MeSH] OR"Hallucinations"[MeSH] OR"Delusions"[MeSH] OR"Agitation"[TEXT])


7/1/02 –8/4/03 0 0

* The Cochrane Database of Systematic Reviews did contain relevant reviews on all four topics of interest, but sincethose were included as part of the USPSTF systematic evidence review, separate evidence tables were not created.

** Clinical Evidence did contain a section on reality orientation that summarized the Cochrane Review describedabove. Again, a separate evidence table was not created.



Guideline: Non-pharmacological management of behaviors

Recommendation: There is insufficient evidence to support the use of music therapy,validation therapy, reality orientation, and reminiscence therapyfor managing behaviors associated with dementia.


Rationale:

Note: All of these studies summarized below included nursing home populations, not peopleliving in the community (our real population of interest for this primary care-based guideline).Because it was the only available evidence on the topic, it was reviewed with this discrepancyacknowledged. Supporting Evidence for Music Therapy

BENEFITS: � The USPSTF systematic evidence review (27) found one systematic review on music

therapy for managing behaviors in dementia. (143) In that review, Koger and Brotonsfound no RCTs but reviewed 126 articles that overall supported the use of music therapyin dementia.

� No RCTs on this topic were found after the search period of the systematic review.HARMS: � None cited.

� Conclusion: Since no RCTs have been conducted, there is insufficient evidence torecommend use of music therapy to manage behaviors associated with dementia.

Supporting Evidence for Validation Therapy

BENEFITS: � The USPSTF systematic evidence review (27) found one meta-analysis on validation

therapy for managing behaviors in dementia. (144) In that review, Neal and Briggsidentified 3 relevant RCTs, but were only able to obtain and review 2. (145, 146) From the2 studies, data on 87 subjects were included in the Cochrane meta-analysis. There wereno significant findings for validation therapy versus usual care or validation therapyversus social contact in any of the subgroup analyses (99% CI). Given that, the Cochranereview and the AHRQ reviewers concluded that there was insufficient evidence tosupport any conclusion about the efficacy of validation therapy. The Cochrane reviewdid say that observational studies suggest there may be some positive effects.

� No RCTs on this topic were found after the search period of the systematic review.HARMS: � None cited

� Conclusion: There is insufficient evidence to recommend use of validation therapy tomanage behaviors associated with dementia.


Supporting Evidence for Reminiscence Therapy BENEFITS: � The USPSTF systematic evidence review (27) found one meta-analysis on reminiscence

therapy for managing behaviors in dementia. (147) Spector et al. found two trials that mettheir inclusion criteria, (148, 149) but only one of those studies (148) with 15 patients had datathat could be extracted into the Cochrane meta-analysis. Analysis of effect on cognitionand behavior both showed insignificant results. For the Information/Orientation subscaleof the CAPE (Clifton Assessment Procedure for the Elderly), the weighted meandifference between the group who received reminiscence therapy compared to the groupwho received no treatment was 0.049 (95% CI –4.371 to 4.771). For the behaviorsubscale of the CAPE, the weighted mean difference between the groups was –3.3 (95%CI –14.190 to 7.590).

HARMS: � None cited.

� Conclusion: There is insufficient evidence to recommend use of reminiscence therapy tomanage behaviors associated with dementia.

Supporting Evidence for Reality Orientation BENEFITS: � The USPSTF systematic evidence review (27) found one meta-analysis on reality

orientation for managing behaviors in dementia. (150) Spector et al. found 8 studies thatmet their inclusion criteria but were only able to include data from 6 RCTs (125 patients)into their analysis, examining cognitive and behavioral outcomes. Only 3 studies (48subjects) included behavioral outcomes; the total result for these showed a significantdifference in favor of reality orientation, with a standard mean difference of –0.659 (95%CI –1.268 to –0.050). The three individual studies had insignificant results with trends infavor of treatment. The behavioral measures used in each of those three studies weredifferent. Results were also positive for impact on cognition, with a standard meandifference of –0.586 (95% CI –0.952 to –0.220).

HARMS: � No reported side effects.Other considerations:� The interventions in each trial differed in length and frequency of treatment as well as

content of sessions, making it difficult to determine which aspects are most beneficial.� Conclusion: There is insufficient evidence to support the use of reality orientation for

cognition and behavior in dementia. The USPSTF systematic evidence review agreedwith the conclusions of the meta-analysis by Spector et al. that no firm conclusions canbe drawn about the effectiveness of reality orientation for dementia, and the CMIDementia Guidelines Workgroup also came to this conclusion.


Evidence Table: USPSTF Systematic Evidence Review on Screening for Dementia (Topic 5: Efficacy ofNonpharmacologic Interventions)

This table is taken directly from: Agency for Healthcare Research and Quality Systematic Evidence Review on screening for dementia(Technical Support for the U.S. Preventive Services Task Force). (27) Reprinted with permission of the author.

AuthorN

RCTPatients Intervention

Outcomes Results Quality andComments

Koger andBrotons, 2000

0 RCTs Music therapy ---- No RCT was found Good

MSQ WMD:-1.8 (99% CI, -9.7 to +6.1) NS

PGCMS WMD: 1.1 (95% CI, -7.5 to +5.3) NS

Self-care MOSES WMD: -1.1 (99% CI, -4.9 to +2.7) NS

Verbal agitation CMAI WMD: 3.9 (99% CI, -4.1 to +11.9) NS

Withdrawal MOSES WMD: 1.6 (99% CI, -6.0 to +2.8) NS

Confusion MOSES WMD: 3.0 (99% CI, -2.8 to +8.8) NS

Neal andBriggs, 2000

2 RCTs124 (incorrectn of 102 waslisted inoriginal AHRQsystematicreview)

Validation therapy 2-4times wk for 36-52 wks

Cognitive (MSQ,PGCMS)Functional: MOSESBehavioral: CMAI,MOSES, MSBS

Social behavior MSBS WMD: 1.1 (99% CI, -10.3 to +8.1) NS

Good

Both studies conductedin long term carefacility; dementia wasmoderate to severe in 1RCT and at leastmoderate in the second

MSQ = Mental Status QuestionnairePGCMS = Philadelphia Geriatric Center Morale ScaleMOSES = Multi Observational Scale for Elderly SubjectsCMAI = Cohen Mansfield Agitation InventoryMSBS = Minimal Social Behavior ScaleWMD = Weighted Mean DifferenceCI = Confidence IntervalNS = Non-significant


(continued)

AuthorN

RCTPatients Intervention

Outcomes Results Quality andComments

Information/orientation CAPE WMD: 0.05 (95% CI, -4.37 to +4.77) NSSpector et al.,2000

2 RCTs15*

Reminiscence therapy30 min 2-5 times weeklyfor 4-5 weeks

Cognitive CAPE, MMSE, Behavior CAPE BDI

Behavioral CAPE WMD: -3.3 (95% CI, -14.2 to +7.60) NS

Good

Clinical setting anddementia severity notspecified; in 1 RCT,patients had moderateto severe dementia

Cognitive SMD: -0.59 (95% CI, -0.95 to -0.22) SignificantSpector et al.,2000

6 RCTs125**

Reality orientation 30-60 minutes 2-5 timesweekly for 4-21 weeks

Cognitive multiple scalesBehavior multiple scales

Behavior SMD: -0.64 (95% CI, -1.20 to -0.08) Significant

Good

Clinical setting anddementia severity notspecified; patients in 1RCT had severecognitive impairment,other trial had milddementia. Patients in 1RCT wereinstitutionalized

CAPE = Clifton Assessment Procedure for the ElderlyWMD = Weighted Mean DifferenceCI = Confidence IntervalNS = Non-significantSMD = Standard Mean Difference

* Data for 15 patients (from Baines, 1987) were analyzed to produce the results highlighted here. Data needed for MetaView were not available in the other relevant paper on this topic (Goldwasser, 1987).** Out of the 8 relevant studies found for this review, data from 6 studies (125 patients) were analyzed to produce the results highlighted here. Data needed for MetaView were not available for the other 2 studies.


Specialty Referrals

Problem Formulation

Clinical Question: When should primary care physicians refer patients with suspecteddementia to specialists?


To assist primary care physicians and other health care professionals inthe outpatient primary care setting to appropriately refer patientsundergoing assessment for dementia


Health Problem: Determining cause of dementia symptoms and making diagnosis


Referral to one or more of the following specialists:� Neurologist� Psychiatrist� Geriatrician




Intervention:

� Finding and treating reversible dementias/pseudo-dementias� Patient and family satisfaction


� Inconvenience of seeingmultiple physicians

� Potential unclear accountabilityfor long-term management

� Possible miscommunicationre: treatment plan


Evidence Search

Only observational studies, clinical trials, or meta-analyses were included that studied the impactof specialty referral on dementia diagnosis. Because searches on this topic were conducted for the 2002 CMI Dementia Guidelines, updateson those searches were performed. Additional search terms were added for the PubMed search toensure that no relevant articles were missed. The most recent searches are listed first, followedby the original searches.


TimeFrame:

#Found:

# inET:

New Searches



and RCTs 5/27/03 12 0

PubMed

("Referral andConsultation"[MESH] OR"referral"[TEXT] OR"specialist"[TEXT]) AND"dementia/diagnosis"[MESH]

English, Human 2/1/01-8/7/03 44 0


PubMed"Referral andConsultation"[MESH] ANDdementia/diagnosis[MESH]

English, Human 1965-4/14/01 99 0




Guideline: Specialty referrals

Recommendation: It is recommended that primary care providers refer patients withsuspected or diagnosed dementia to a specialist (geriatrician,neurologist, gero-psychiatrist as available) if any of the followingapply: � Cognitive loss is early-onset (i.e., before age 60)� Diagnosis is complex or remains unclear after basic work-up� PCP requires consultation to assist with management� Patient or family strongly desires the consultation of a

specialist



� Although there is no direct evidence about the ideal time for a primary care clinician to refera patient with suspected dementia for further evaluation, the CMI Dementia GuidelinesWorkgroup felt that primary care providers need some guidance on this issue. The groupagreed that there are several key triggers that should lead to a referral, based on thecomplexity of the case (i.e., age of the patient or murky diagnosis), the needs of the primarycare clinician, and the desires of the patient and/or family for specialist consultation.Specifications about when to refer to which type of specialist are not included due tovariation in availability in each KP Region.


Special Considerations - Driving

Problem Formulation

Clinical Question: How should primary care clinicians manage the issue of driving inpeople with dementia?


To assist primary care physicians and other health care professionals incounseling people with dementia and their families regarding drivingissues and reporting potentially hazardous drivers as required

Population: Men and women with dementia

Health Problem: Safety hazards related people with dementia driving


� Determine driving status� Discuss issue of driving with

patient and family� Identify alternative

transportation

� Report unsafe drivers to stateauthorities

� Refer for driving evaluation� No intervention




Intervention:

� Traffic accidents/injury/deathof patient and others

� Independence/function� Patient/family satisfaction


� Anger/sadness of patient and family� Disruption of clinician-patient relationship� Social isolation


Evidence Search

Clinical practice guidelines, systematic reviews, and observational or retrospective studies thatexamined the risk of vehicle accidents in people with dementia and the use of tools andassessments to identify unsafe drivers among people with dementia were included. Reviewarticles and other information sources (see below) were also consulted. Because this topic was not addressed in the 2002 version of the CMI Dementia Guidelines,literature searches extended back to 1965.


TimeFrame:

#Found:

# inET:



and RCTs 5/27/03 12 0

PubMed

("AutomobileDriving"[MeSH] OR"AutomobileDriving/standards"[MeSH]OR "driving"[TEXT]) AND"Dementia"[MeSH]

English, Human 1965-7/31/03 153 0

Other Information Sources� “At the Crossroads: A Guide to Alzheimer’s Disease, Dementia & Driving” developed by The

Hartford Financial Services Group, Inc., the MIT Age Lab and Connecticut Community Care,Inc. http://www.thehartford.com/alzheimers

� Alzheimer’s Association Position Statement on Driving, October 2001.http://www.alz.org/AboutUs/PositionStatements/overview.htm#driving

� Alzheimer’s Association Fact Sheets on driving and dementia:http://www.alz.org/ResourceCenter/FactSheets/FSDriving.pdfhttp://www.alz.org/ResourceCenter/FactSheets/drivingEI.pdf

� Family Caregiver Alliance Fact Sheets on driving and dementia:http://www.caregiver.org/caregiver/jsp/content_node.jsp?nodeid=432http://www.caregiver.org/caregiver/jsp/content_node.jsp?nodeid=433

http://www.thehartford.com/alzheimers

http://www.alz.org/AboutUs/PositionStatements/overview.htm#driving

http://www.alz.org/ResourceCenter/FactSheets/FSDriving.pdf

http://www.alz.org/ResourceCenter/FactSheets/drivingEI.pdf

http://www.caregiver.org/caregiver/jsp/content_node.jsp?nodeid=432




Guideline: Managing the issue of driving with patients, families, and stateauthorities

Recommendations: 1. For patients with dementia, it is recommended that patientsand their families (or other responsible party) be told:� The patient is at increased risk for driving accidents and

driving performance errors. (Evidence-Based)� The patient should not drive an automobile unless an on-road

evaluation of driving ability conducted by the state driver'slicensing authority deems it appropriate. (Consensus)

2. Physicians practicing in California are obliged by law toreport all people with a diagnosis of dementia severe enough toimpair their ability to operate a motor vehicle to the Departmentof Health, who will then notify the Department of Motor Vehicles.(State law) Clinicians in other states should keep abreast ofreporting requirements in their state. (Consensus)

3. To support patients and families around the issue of driving, itis recommended that clinicians:� Advise caregivers of people with dementia of the early

warning signs of driving problems.� Advise families of resources to help them manage this issue.� Encourage patients and families to plan for eventual cessation

of driving privileges and to develop alternate transportationplans.

(Consensus)

Methodology: � Varied – see specific recommendations � IRGSG-sponsored


Rationale: Supporting Evidence for Recommendation 1� The Quality Standards Subcommittee of the American Academy of Neurology (AAN) issued

an evidence-based Practice Parameter on the risk of driving and Alzheimer’s disease (AD) in2000. (151) They used an evidence-based review of the medical literature on the topic, finding14 articles that met their criteria for Class I, II, or III evidence.* The articles focused on crashstatistics, evaluation of driving performance, and analysis of driving task components. � Almost all articles on crash rate demonstrated an increased crash rate for drivers with AD

compared to non-demented older adults. � The studies on driving performance testing methods (on-the-road driving evaluations and

driving simulators) demonstrated significant impairment in driving abilities of peoplewith AD.

� The studies focused on testing driving-related tasks (visual perception, visual processing,and recognition of common traffic signs) demonstrated significant problems with visualprocessing in drivers with AD.

� Based on these findings, the Practice Parameter stated (verbatim):1. Patients and their families should be told that patients with AD with a severity of CDR

(Clinical Dementia Rating) of 1 or greater have a substantially increased accident rateand driving performance errors, and therefore should not drive an automobile. (Standard)

2. Patients and their families should be told that patients with possible AD with a severity ofCDR 0.5 pose a significant traffic safety problem when compared to other elder drivers.Referral of the patient for a driving performance evaluation by a qualified examinershould be considered. (Guideline) Because of the high likelihood of progression to aseverity of CDR 1 within a few years, clinicians should reassess dementia severity andappropriateness of continued driving every 6 months. (Standard)� Standard: A principle for patient management that reflects a high degree of clinical

certainty (usually requires class I evidence that directly addresses clinical question, oroverwhelming class II evidence when RCTs are not possible)

� Guideline: A recommendation for patient management that reflects moderate clinicalcertainty (usually requires class II evidence or a strong consensus of class IIIevidence.

� The CMI Dementia Guidelines Workgroup did not think the CDR level specifications fromthe Practice Parameter would be useful to a primary care clinician, so a more generalrecommendation was made based on the Practice Parameter.

� One additional study on this topic published after the search period of the AAN PracticeParameter does not change the conclusions drawn by the AAN.� Carr et al. examined the differences in crash rates in people with Alzheimer’s disease

(n=63) versus controls (n=58) in a 5-year retrospective analysis of state-recorded crash * Class I: Must have all of a to d: a) a prospective study of well-defined cohorts that includes a description of the nature of thepopulation, the inclusion or exclusion criteria, demographic characteristics such as age, sex, and commonly used staging ofdementia severity; b) the sample size must be adequate with enough statistical power to justify the conclusions or foridentification of subgroups for whom testing does or does not yield significant information; c) the interpretation of evaluationsperformed must be done blinded to subject status; d) the methodology used for evaluations must be adequate.Class II: a) a retrospective study of a well-defined cohort which otherwise meets criteria for class 1a, 1b, and 1d; b) a prospectiveor retrospective study which lacks any of the following: adequate sample size, a well defined control group, adequatemethodology, blinding of evaluators, a description of the inclusion or exclusion criteria, and information such as age, sex, andcommonly used staging of dementia severity.Class III: May have either a, b, or c: a) criteria of a class II article published in a nonpeer reviewed format; b) a small cohort orcase report; c) evidence from expert opinion or consensus


data for people enrolled in a longitudinal study of aging and Alzheimer’s disease. (152)

Unlike almost all the studies reviewed for the AAN Practice Parameter, this studyshowed no statistically significant difference in crash rates for those with very mild AD(CDR = 0.5), mild AD (CDR = 1), and controls, even when adjusting for roadwayexposure. Crashes in all groups were infrequent, so that may have made differencesdifficult to detect in this small sample. In addition, roadway exposure (miles driven peryear) was calculated from daily driving diaries for one week, which were thenextrapolated to calculate driving exposure per year and related to the 5 years of crashdata. These diaries may have been less accurate in the very mild and mild AD groupscompared to controls. Also, it is difficult to determine if the extrapolation method fordriving exposure per year was sound. Finally, subjects willing to volunteer for a drivingstudy may not be representative of drivers with AD in the community.

Supporting Evidence for Recommendation 2� The following information is taken verbatim from the Family Caregiver Alliance Fact Sheet,

“Dementia, Driving, and California State Law”(http://www.caregiver.org/caregiver/jsp/content_node.jsp?nodeid=433) � “California’s Health & Safety Code [Section 103900] requires physicians to submit a

confidential report to the county health department when an individual is diagnosed ashaving Alzheimer's disease or related disorders, including dementia, severe enough toimpair a person's ability to operate a motor vehicle. This information is forwarded to theDepartment of Motor Vehicles (DMV), which is authorized to take action against thedriving privileges of any individual who is unable to safely operate a motor vehicle. If thephysician’s report indicates that an individual has moderate or severe dementia, thatindividual will no longer be permitted to operate a motor vehicle. DMV has determinedthat only drivers with dementia in the mild stages may still have the cognitive functionsnecessary to continue driving safely. DMV requires re-examination for all individualsreported to have mild dementia.”

� Given that this is state law and physicians can be held liable if an accident occurs when adiagnosis of dementia had been made but the driver had not been reported, the CMIDementia Guidelines Workgroup felt it important to raise awareness among physicians ofthis law.

� The text of the California Senate Bill 335 on this topic is located at:http://www.leginfo.ca.gov/pub/99-00/bill/sen/sb_0301-0350/sb_335_bill_20000930_chaptered.html

Supporting Evidence for Recommendation 3� This recommendation was created because of the importance of educating patients and

families early on about the issue of driving with dementia, and giving them tools andinformation to help them detect problems and plan for future needs.


http://www.leginfo.ca.gov/pub/99-00/bill/sen/sb_0301-0350/sb_335_bill_20000930_chaptered.html

http://www.leginfo.ca.gov/pub/99-00/bill/sen/sb_0301-0350/sb_335_bill_20000930_chaptered.html


Special Considerations - Caregivers

Problem Formulation

Clinical Question: How should primary care clinicians and staff support and counselcaregivers of people with dementia?


To assist primary care physicians and other health care professionals inproviding support, information, and assistance to the caregivers ofpeople with dementia

Population: Caregivers of men and women with dementia

Health Problem: Adequate knowledge, skills, and support to care for someone withdementia

Health Intervention:� Refer to KP social services� Refer to Alzheimer’s

Association

� Identify needs and concerns

Practitioners: KP physicians, physician assistants, nurse practitioners, nurses, andsocial workers in the Departments of Family Practice and InternalMedicine (primary care); health educators



Associated with theIntervention:

� Patient safety� Patient function� Time to patient

institutionalization� Patient/family satisfaction

� Caregiver depression� Caregiver fatigue� Caregiver anxiety� Caregiver anger� Caregiver self-efficacy


Evidence Search

Clinical practice guidelines, systematic reviews, and meta-analyses that examined the efficacy ofinterventions or approaches for assisting caregivers of people with dementia were included.Other information sources (see below) were also consulted. Because this topic was not addressed in the 2002 version of the CMI Dementia Guidelines,literature searches extend back to 1965.


TimeFrame:

#Found:

# inET:



and RCTs 5/27/03 12 0

Practice Guideline,English, Human

1965-9/16/03 3 0

PubMed"Caregivers"[MeSH] AND("Dementia"[MeSH] OR

"Alzheimer Disease"[MeSH]) Meta-analysis,English, Human

1965-9/16/03 5 1





� Alzheimer’s Association Caregiver Resources: http://www.alz.org/Caregivers/overview.asp � Family Caregiver Alliance: http://www.caregiver.org � National Family Caregivers Association: http://www.nfcacares.org * A Cochrane Systematic Review on the topic was included in the USPSTF systematic evidence review, so aseparate evidence table was not created for the Cochrane review.

http://www.alz.org/Caregivers/overview.asp

http://www.caregiver.org/

http://www.nfcacares.org/



Guideline: Providing support and counseling to caregivers of people withdementia

Recommendations: For caregivers of people with dementia, it is an option forclinicians to refer to programs and resources focusing on:� Counseling of caregivers and families� Education / training� Involvement of the patient� Support � Stress management(Evidence-Based)

It is recommended that clinicians: � Consistently connect caregivers to resources for counseling,

education, and training� Monitor caregivers for depression, fatigue, anger, anxiety,

and other stress-related signs and refer to KP and communityresources as available.

(Consensus)

Methodology: � Varied – see specific recommendations � IRGSG-sponsored

Rationale:

Supporting Evidence for Effect of Caregiver Interventions on Caregiver Outcomes � As part of the systematic evidence review conducted for the US Preventive Services Task

Force (USPSTF) on screening for dementia, evidence on the efficacy of caregiverinterventions was reviewed. (27) Out of 17 RCTs and 9 systematic reviews that met thegroup’s initial screening criteria, only 1 review was rated good and 5 RCTs were rated fair togood. The review categorized all study outcomes as outcomes that target caregivers’ stressand coping or outcomes that target the patients’ functioning, behavioral problems, andinstitutionalization. The following summary is taken from the USPSTF review (notverbatim).� The one good quality systematic review found in the USPSTF review, by Thompson and

Briggs, evaluated four types of caregiver assessments: individualized services assessmentand planning, technology-based interventions, caregiver education and training, and amulti-component program versus conventional care or support. (153) They found nosignificant differences between any of the intervention and control groups, and concludedthat little or no evidence of quantifiable benefit of these interventions exists.

� The 5 fair-to-good RCTs reviewed included 3 studies evaluating the effect of caregiverintervention on caregiver burden, 1 that examined the effect of intervention on


caregivers’ sleep, and 2 that studied the effect of intervention on caregiver reaction topatients’ behavioral problems.� Of the studies focusing on caregiver burden and depressive symptoms, 2 (with small

sample sizes of 36 to 45) found no effect (154-156) and 1 found a positive effect of a 14-session multi-component intervention. (157)

� In the study that examined impact on caregiver sleep, a multi-component interventionin 36 caregivers with significant sleep problems was found to improve quality ofsleep, with 50% of those in the intervention arm considered to have no significantsleep problems at the end of the intervention. (154)

� Of those studies that examined impact on caregiver reaction to patients’ behavioralproblems, the authors found no significant effect. (154-156)

� A meta-analysis by Brodaty et al. examined 30 studies of caregiver interventions, includingcounseling of carers, education, family counseling or extended family involvement, patientinvolvement, support groups or programs, stress management, and training. (158) Impact ofinterventions was calculated as effect size (ES) on primary outcomes of caregiverpsychological morbidity and caregiver burden, and secondary outcomes including caregiverknowledge and patient mood/behavior. Moderate effect sizes were seen for caregiverknowledge (ES=0.51, 95% CI 0.05 to 0.98, n=8 studies) and mood/behavior of patient(ES=0.68, 95% CI 0.30 to 1.06, n=5 studies), while effect on psychological morbidity wasweaker (ES = 0.31, 95% CI 0.13 to 0.50, n=26 studies) and effect on caregiver burden wasquite weak (ES=0.09, 95% CI –0.09 to 0.26, n=20 studies). Time to nursing home placementwas longer with caregiver interventions in a few studies, but effect sizes and significancelevels were not consistently reported.

� Conclusion: Results of studies of caregiver interventions are quite mixed. While ameta-analysis showed moderate effect on caregiver knowledge and mood/behavior ofthe dementia patient, it showed weaker effect on psychological morbidity and caregiverburden. A different systematic review found that most trials are not able todemonstrate statistically significant benefits of caregiver interventions, though onestudy of a multi-component intervention demonstrated decreased burden anddepression, and another showed positive effect on sleep problems. Overall, it is difficultto determine which types of interventions produce the most positive results.

Supporting Evidence for Effect of Caregiver Interventions on Patient Outcomes� The USPSTF review also looked for evidence of impact of caregiver interventions on patient

outcomes. No systematic reviews on this topic were found, but 3 of the fair-to-good RCTsexamined the effect of caregiver interventions on patients’ cognition, function, andbehavioral problems. The following summary is taken from the USPSTF review (notverbatim).� Marriott et al. found statistically significant positive effect on patients’ behavioral

problems and functional status, but did not report caregivers’ perceptions of thesechanges. (157)

� Studies by Mittelman et al. and Brodaty et al. found that a comprehensive caregiverintervention significantly increased the amount of time before institutionalizationcompared to controls. (159, 160)


� Herbert et al. found that a multi-component intervention had some effect on probabilityof nursing home placement within 2 years, though this did not reach statisticalsignificance in this small population (n=45). (155, 156)

� Conclusion: Caregiver interventions have demonstrated some positive impact onpatient outcomes, including behavior, function, and time to institutionalization, butthese results are not consistently statistically significant.

Supporting Evidence for Focusing on Caregiver Capabilities and Health� It is well documented that caregivers for people with dementia experience adverse

psychological, physical, social, and financial outcomes including increased rates ofdepression, poorer health than age-matched counterparts, social isolation and loneliness, anddirect and indirect financial burden and loss. (158) Because informal caregivers are crucial formaintaining people with dementia in the community, it is important for clinicians to helpthese people get the support and resources they need to care for the dementia patient and tomaintain their own good physical, mental, and emotional health.

� Conclusion: Caregivers need to gain skills related to caring for the dementia patientand are at risk for psychological and physical stress, so clinicians should assess needsand refer to support services and resources as available.


Evidence Table: Efficacy of Caregiver Interventions – Systematic Review

This table is taken directly from: Agency for Healthcare Research and Quality Systematic Evidence review on screening for dementia(Technical Support for the U.S. Preventive Services Task Force, Topic 6: Efficacy of Caregiver Interventions. (27) Reprinted withpermission of the author.

Author Number of Studies Interventions Outcomes Results Comments

Thompson and Briggs, 1998 6 RCTsN (33-102)

(1) Individualized serviceassessment and planning vs.conventional care or support(2) Technology-basedinterventions vs. conventionalcare or support(3) Career education/ trainingvs. conventional care/support(4) Multi-faceted/ dimensionalstrategies vs. conventionalcare/support

(1) Caregiver burden, strain,support, quality of life(2) Caregiver mental health:depression, anxiety(3) Service utilization and cost(4) Others: knowledge ofAlzheimer's disease, asking forhelp, decision-makingconfidence

No significant differencesbetween experimental andcontrol groups for any of theseoutcomes

Good

Limited to 1998 and tocaregiver outcomes only


Evidence Table: Efficacy of Caregiver Interventions – Studies

This table is taken directly from: Agency for Healthcare Research and Quality Systematic Evidence review on screening for dementia(Technical Support for the U.S. Preventive Services Task Force, Topic 6: Efficacy of Caregiver Interventions. (27) Reprinted withpermission of the author.

Intervention ResultsAuthor NSupportGroup

SkillsTraining

Counsel-ing

Education-al

OutcomesMeasured Scale Treatment

ArmControlArm

P valueQuality and Comments

Hebert etal., 1994and 1995

45 + + + + Caregiver:-Burden: BI-Depression: BSI-Reaction to pt’s BPRDRMBPC-ADKT-Health care utilizationHCUQ

Patient:-Nursing homeplacement-Cognition: 3MS-Functional: SMAF-BPRD: RMBPC (F)

BIBSIRMBPC(R)ADKTHCUQ

P (nursing homeplacement)3MSSMAFRMBPC (F)

34.9033.571.399.52----

0.33

40.6335.671.58

36.0630.201.736.53----

0.45

36.5336.731.63

NSNSNSP = 0.004NS

P = 0.31

NSNSNS

Good

Difficult to implement

Mittelmanet al., 1996

206 + + + + Patient:Median time to nursinghome placement

Days to nursinghome placement

1,203 874 P = 0.02 Fair (not all outcomesconsidered)

Difficult to implementBrodaty etal., 1997

96 + + + + Patient:Time to nursing homeplacementTime to death

Time to nursinghome placement(in months)Time to death (inmonths)

47.5

65

27.6

53

P < 0.05

P = 0.08

Fair

Difficult to implement in-patientsetting

McCurry etal., 1998

36 -- + + + Caregiver:-Burden: SCB-Depression: CES-D-Sleep problems: PSQI-Reaction to patients'BPRD: RMBPC

SCBCES-DPSQIRMBPC

--------7.8----

--------10.6----

NSNSP < 0.05NS

FairDifficult to implement; attritionrate varied between the 2 groupsat follow-up, but not at immediatepost treatment; (included only theresults of post treatment);caregivers had sleep problemsbefore entry


(continued)

Intervention ResultsAuthor NSupportGroup

SkillsTraining

Counsel-ing

Education-al

OutcomesMeasured Scale Treatment

ArmControlArm

P valueQuality and Comments

Marriott etal., 2000

42 -- + + + Caregiver:-Burden: GHQ-Depression: BDI

Patient:-Cognition: MMSE-Depression: CSDD-BPRD: MOUSEPAD-Functional status: CDR

GHQ @ 9 msGHQ @ 12 msBDI @ 9 msBDI @12ms

MMSECSDDMOUSEPAD @9 msMOUSEPAD@12msCDR(ADL)@9msCDR(ADL) @12ms

6.03.96.96.1

--------4.95.35.45.5

12.710.811.811.8

--------5.65.25.16.4

P = 0.001P = 0.001P < 0.01P = 0.001

NSNSP = 0.01NSNSP = 0.043

Good

Difficult to implement; caregiverhad significant psychologicalmorbidity at entry to the trial.

Key:BI: Burden InterviewBSI: Brief Symptoms InventoryRMBPC: Revised Memory Behavior Problem ChecklistADKT: Alzheimer's Disease Knowledge TestHCUQ: Healthcare Cost and Utilization Project3MS: Modified Mini-Mental Status ExamSMAF: Functional Autonomy MeasuresRMBPC(F): Revised Memory Behavior Problem Checklist – FrequencySCB: Screen of Caregiver BurdenCES-D: Center of Epidemiology Studies of DepressionPSQI: Pittsburg Sleep Quality IndexGHQ: General Health QuestionnaireBDI: Beck Depression InventoryMMSE: Mini-Mental State ExaminationCSDD: Cornell Scale for Depression in DementiaMOUSEPAD: Manchester and Oxford Universities Scales for the Psychopathological Assessment of DementiaCDR: Clinical Dementia Rating


Evidence Table: Psychosocial Interventions for Caregivers of People with Dementia

Study, Total n


Results Weighted average effect size* (95% CI, # of studies) Comments

Brodaty et al., 2003(Meta-Analysis) # studies found: 52 # studies included: 30 Results were compared for fixed-effects and random-effects models,and in most cases there was nosubstantial difference between thetwo models. Results for random-effects models are displayedbecause tests for homogeneity andheterogeneity of studies underexamination support the use of arandom-effects model for most ofthe pooled estimates.

� Randomized or quasi-experimental trials in whichinformal caregivers (persons providing unpaid care, athome of a non-institutional environment) of peoplewith diagnosed Alzheimer’s disease (AD) wereallocated to intervention or control groups

� Outcomes of interest related to caregiver-targetedinterventions, not patient-targeted interventions

� Sample size ranged from 16-206, median = 53� Intervention types studied were:� Counseling of carers� Education� Family counseling/extended family involvement� Patient involvement� Support group/program� Stress management� Training

Primary Outcomes (Caregiver) Psychological morbidity 0.31 (0.13 to 0.50, n=26) 77% of studies showed positiveeffect size for this outcome, butonly 5 of the 20 positiveinterventions were statisticallysignificant. Burden 0.09 (-0.09 to 0.26, n=20) 1 out of 20 interventions showeda statistically significant effect oncaregiver burden

Secondary Outcomes Caregiver knowledge 0.51 (0.05 to 0.98, n=8) Mood/behavior of dementiapatient 0.68 (0.30 to 1.06, n=5) Overall effect on any mainoutcome measure 0.32 (0.15 to 0.48, n=30) Time until nursing homeplacement (n=7 studies)� 2 studies showed

significant effect sizes (ESvalue not reported)

� 2 studies reported alonger median time of homecare until institutionalizationin the intervention group vs.control group (p value onlyreported for one of these at0.02)

Intervention characteristicsassociated with effect� Involvement of caregiver

and patient� Higher “dosage” (These are based on smallnumbers in heterogeneousstudies – validity unknown)

Conclusions� Caregiver interventions show

mixed results, including little impacton caregiver burden, weak tomoderate impact on caregiverpsychological morbidity, moderateimpact on caregiver knowledge,and moderately strong impact onmood or behavior of the dementiapatient.

� Involvement of the patient inaddition to the caregiver in astructured program may have someimpact.

Biases� Interventions, patients,

caregivers, and methods ofrecruitment were heterogeneous,so validity of combining data isunknown.

� Intention-to-treat analyseslargely not performed.

* Effect size of 0.2 is considered weak, 0.5 is considered moderate, and 0.8 or above is considered strong.


References(1) MRC/BHF Heart Protection Study of cholesterol lowering with simvastatin in 20,536

high-risk individuals: a randomised placebo-controlled trial. Lancet 2002; 360(9326):7-22.

(2) Knopman DS, DeKosky ST, Cummings JL, Chui H, Corey-Bloom J, Relkin N et al.Practice parameter: diagnosis of dementia (an evidence-based review). Report of theQuality Standards Subcommittee of the American Academy of Neurology. Neurology2001; 56(9):1143-1153.

(3) Shumaker SA, Legault C, Thal L, Wallace RB, Ockene JK, Hendrix SL et al. Estrogenplus progestin and the incidence of dementia and mild cognitive impairment inpostmenopausal women: the Women's Health Initiative Memory Study: a randomizedcontrolled trial. JAMA 2003; 289(20):2651-2662.

(4) LeBlanc ES, Janowsky J, Chan BK, Nelson HD. Hormone replacement therapy andcognition: systematic review and meta-analysis. JAMA 2001; 285(11):1489-1499.

(5) Yaffe K, Sawaya G, Lieberburg I, Grady D. Estrogen therapy in postmenopausal women:effects on cognitive function and dementia. JAMA 1998; 279(9):688-695.

(6) Hogervorst E, Williams J, Budge M, Riedel W, Jolles J. The nature of the effect offemale gonadal hormone replacement therapy on cognitive function in post-menopausalwomen: a meta-analysis. Neuroscience 2000; 101(3):485-512.

(7) Fillenbaum GG, Hanlon JT, Landerman LR, Schmader KE. Impact of estrogen use ondecline in cognitive function in a representative sample of older community-residentwomen. Am J Epidemiol 2001; 153(2):137-144.

(8) Grodstein F, Chen J, Pollen DA, Albert MS, Wilson RS, Folstein MF et al.Postmenopausal hormone therapy and cognitive function in healthy older women. J AmGeriatr Soc 2000; 48(7):746-752.

(9) Seshadri S, Zornberg GL, Derby LE, Myers MW, Jick H, Drachman DA.Postmenopausal estrogen replacement therapy and the risk of Alzheimer disease. ArchNeurol 2001; 58(3):435-440.

(10) Broe GA, Grayson DA, Creasey HM, Waite LM, Casey BJ, Bennett HP et al. Anti-inflammatory drugs protect against Alzheimer disease at low doses. Arch Neurol 2000;57(11):1586-1591.

(11) Anthony JC, Breitner JC, Zandi PP, Meyer MR, Jurasova I, Norton MC et al. Reducedprevalence of AD in users of NSAIDs and H2 receptor antagonists: the Cache Countystudy. Neurology 2000; 54(11):2066-2071.


(12) in't Veld BA, Launer LJ, Hoes AW, Ott A, Hofman A, Breteler MM et al. NSAIDs andincident Alzheimer's disease. The Rotterdam Study. Neurobiol Aging 1998; 19(6):607-611.

(13) Beard CM, Waring SC, O'Brien PC, Kurland LT, Kokmen E. Nonsteroidal anti-inflammatory drug use and Alzheimer's disease: a case-control study in Rochester,Minnesota, 1980 through 1984. Mayo Clin Proc 1998; 73(10):951-955.

(14) Stewart WF, Kawas C, Corrada M, Metter EJ. Risk of Alzheimer's disease and durationof NSAID use. Neurology 1997; 48(3):626-632.

(15) Fourrier A, Letenneur L, Begaud B, Dartigues JF. Nonsteroidal antiinflammatory druguse and cognitive function in the elderly: inconclusive results from a population-basedcohort study. J Clin Epidemiol 1996; 49(10):1201.

(16) Andersen K, Launer LJ, Ott A, Hoes AW, Breteler MM, Hofman A. Do nonsteroidalanti-inflammatory drugs decrease the risk for Alzheimer's disease? The Rotterdam Study.Neurology 1995; 45(8):1441-1445.

(17) The Canadian Study of Health and Aging: risk factors for Alzheimer's disease in Canada.Neurology 1994; 44(11):2073-2080.

(18) Wesnes K, Simmons D, Rook M, Simpson P. A Double-blind Placebo-controlled Trial ofTanakan in the Treatment of Idiopathic Cognitive Impairment in the Elderly. HumanPsychopharmacology 1987; 2:159-169.

(19) Mix JA, Crews WD, Jr. An examination of the efficacy of Ginkgo biloba extract EGb761on the neuropsychologic functioning of cognitively intact older adults. J AlternComplement Med 2000; 6(3):219-229.

(20) Rigney U, Kimber S, Hindmarch I. The effects of acute doses of standardized Ginkgobiloba extract on memory and psychomotor performance in volunteers. Phytother Res1999; 13(5):408-415.

(21) La Rue A, Koehler KM, Wayne SJ, Chiulli SJ, Haaland KY, Garry PJ. Nutritional statusand cognitive functioning in a normally aging sample: a 6-y reassessment. Am J ClinNutr 1997; 65(1):20-29.

(22) Peacock JM, Folsom AR, Knopman DS, Mosley TH, Goff DC, Jr., Szklo M. Dietaryantioxidant intake and cognitive performance in middle-aged adults. The AtherosclerosisRisk in Communities (ARIC) Study investigators. Public Health Nutr 2000; 3(3):337-343.

(23) Masaki KH, Losonczy KG, Izmirlian G, Foley DJ, Ross GW, Petrovitch H et al.Association of vitamin E and C supplement use with cognitive function and dementia inelderly men. Neurology 2000; 54(6):1265-1272.


(24) Schmidt R, Hayn M, Reinhart B, Roob G, Schmidt H, Schumacher M et al. Plasmaantioxidants and cognitive performance in middle-aged and older adults: results of theAustrian Stroke Prevention Study. J Am Geriatr Soc 1998; 46(11):1407-1410.

(25) Jick H, Zornberg GL, Jick SS, Seshadri S, Drachman DA. Statins and the risk ofdementia. Lancet 2000; 356(9242):1627-1631.

(26) Wolozin B, Kellman W, Ruosseau P, Celesia GG, Siegel G. Decreased prevalence ofAlzheimer disease associated with 3-hydroxy-3-methyglutaryl coenzyme A reductaseinhibitors. Arch Neurol 2000; 57(10):1439-1443.

(27) Boustani M, Peterson B, Hanson L, Harris R, Krasnov C. Screening for Dementia. ASystematic Evidence Review. 2002. Rockville, MD, Agency for Healthcare Researchand Quality.

(28) United States Preventive Services Task Force. Guide to Clinical Preventive Services.Second Edition ed. Washington, DC: Office of Disease Prevention and Health Promotion,1996.

(29) Brodaty H, Clarke J, Ganguli M, Grek A, Jorm AF, Khachaturian Z et al. Screening forcognitive impairment in general practice: toward a consensus. Alzheimer Dis AssocDisord 1998; 12(1):1-13.

(30) Screening for dementia: recommendation and rationale. Ann Intern Med 2003;138(11):925-926.

(31) Brookmeyer R, Gray S, Kawas C. Projections of Alzheimer's disease in the United Statesand the public health impact of delaying disease onset. Am J Public Health 1998;88(9):1337-1342.

(32) Hy LX, Keller DM. Prevalence of AD among whites: a summary by levels of severity.Neurology 2000; 55(2):198-204.

(33) Scanlan JM, Brush M, Quijano C, Borson S. Comparing clock tests for dementiascreening: naive judgments vs formal systems--what is optimal? Int J Geriatr Psychiatry2002; 17(1):14-21.

(34) Storey JE, Rowland JT, Basic D, Conforti DA. A comparison of five clock scoringmethods using ROC (receiver operating characteristic) curve analysis. Int J GeriatrPsychiatry 2001; 16(4):394-399.

(35) Storey JE, Rowland JT, Basic D, Conforti DA. Accuracy of the clock drawing test fordetecting dementia in a multicultural sample of elderly Australian patients. IntPsychogeriatr 2002; 14(3):259-271.

(36) Powlishta KK, Von Dras DD, Stanford A, Carr DB, Tsering C, Miller JP et al. The clockdrawing test is a poor screen for very mild dementia. Neurology 2002; 59(6):898-903.


(37) Kirby M, Denihan A, Bruce I, Coakley D, Lawlor BA. The clock drawing test in primarycare: sensitivity in dementia detection and specificity against normal and depressedelderly. Int J Geriatr Psychiatry 2001; 16(10):935-940.

(38) Lin KN, Wang PN, Chen C, Chiu YH, Kuo CC, Chuang YY et al. The three-item clock-drawing test: a simplified screening test for Alzheimer's disease. Eur Neurol 2003;49(1):53-58.

(39) Shulman KI. Clock-drawing: is it the ideal cognitive screening test? Int J GeriatrPsychiatry 2000; 15(6):548-561.

(40) Manos PJ. Ten-point clock test sensitivity for Alzheimer's disease in patients with MMSEscores greater than 23. Int J Geriatr Psychiatry 1999; 14(6):454-458.

(41) Esteban-Santillan C, Praditsuwan R, Ueda H, Geldmacher DS. Clock drawing test in verymild Alzheimer's disease. J Am Geriatr Soc 1998; 46(10):1266-1269.

(42) Scanlan J, Borson S. The Mini-Cog: receiver operating characteristics with expert andnaive raters. Int J Geriatr Psychiatry 2001; 16(2):216-222.

(43) Borson S, Scanlan J, Brush M, Vitaliano P, Dokmak A. The mini-cog: a cognitive 'vitalsigns' measure for dementia screening in multi-lingual elderly. Int J Geriatr Psychiatry2000; 15(11):1021-1027.

(44) Recognition and Initial Assessment of Alzheimer's Disease and Related Dementias.AHCPR Publication [Clinical Practice Guideline #19], 1-143. 96. Rockville, Maryland,Agency for Health Care Policy and Research.

(45) American Medical Association. Diagnosis, Management and Treatment of Dementia: APractical Guide for Primary Care Physicians. Rosalie Guttman, Miriam Seleski, editors.1999. Chicago, IL, American Medical Association, Program on Aging and CommunityHealth.

(46) University HealthSystem Consortium, U.S.Department of Veterans Affairs. DementiaIdentification and Assessment: Guidelines for Primary Care Practitioners. 1997.Washington DC and Oakbrook IL.

(47) American Psychiatric Association. Diagnostic and Statistical Manual of MentalDisorders. Fourth Edition. 1994. Washington, D.C., American Psychiatric Association.

(48) Stuss DT, Meiran N, Guzman DA, Lafleche G, Willmer J. Do long tests yield a moreaccurate diagnosis of dementia than short tests? A comparison of 5 neuropsychologicaltests. Arch Neurol 1996; 53(10):1033-1039.

(49) Practice parameter for diagnosis and evaluation of dementia. (summary statement) Reportof the Quality Standards Subcommittee of the American Academy of Neurology.Neurology 1994; 44(11):2203-2206.


(50) McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. Clinicaldiagnosis of Alzheimer's disease: report of the NINCDS-ADRDA Work Group under theauspices of Department of Health and Human Services Task Force on Alzheimer'sDisease. Neurology 1984; 34(7):939-944.

(51) Corey-Bloom J, Thal LJ, Galasko D, Folstein M, Drachman D, Raskind M et al.Diagnosis and evaluation of dementia. Neurology 1995; 45(2):211-218.

(52) Organizing Committee CCCotAoD. Assessing dementia: the Canadian consensus. CanMed Assoc J 1991; 144(7):851-853.

(53) Chertkow H, Bergman H, Schipper HM, Gauthier S, Bouchard R, Fontaine S et al.Assessment of suspected dementia. Can J Neurol Sci 2001; 28 Suppl 1:S28-S41.

(54) Siu AL. Screening for dementia and investigating its causes. Ann Intern Med 1991;115(2):122-132.

(55) van Crevel H, van Gool WA, Walstra GJ. Early diagnosis of dementia: which tests areindicated? What are their costs? J Neurol 1999; 246(2):73-78.

(56) Clarfield AM. The reversible dementias: do they reverse? Ann Intern Med 1988;109(6):476-486.

(57) Weytingh MD, Bossuyt PM, van Crevel H. Reversible dementia: more than 10% or lessthan 1%? A quantitative review. J Neurol 1995; 242(7):466-471.

(58) Patterson C, Gauthier S, Bergman H, Cohen C, Feightner JW, Feldman H et al. Therecognition, assessment and management of dementing disorders: conclusions from theCanadian Consensus Conference on Dementia. Can J Neurol Sci 2001; 28 Suppl 1:S3-16.

(59) Foster GR, Scott DA, Payne S. The use of CT scanning in dementia. A systematicreview. Int J Technol Assess Health Care 1999; 15(2):406-423.

(60) Silverman DH, Small GW, Chang CY, Lu CS, Kung De Aburto MA, Chen W et al.Positron emission tomography in evaluation of dementia: Regional brain metabolism andlong-term outcome. JAMA 2001; 286(17):2120-2127.

(61) Wilkinson D, Doody R, Helme R, Taubman K, Mintzer J, Kertesz A et al. Donepezil invascular dementia: A randomized, placebo-controlled study. Neurology 2003; 61(4):479-486.

(62) Black S, Roman GC, Geldmacher DS, Salloway S, Hecker J, Burns A et al. Efficacy andtolerability of donepezil in vascular dementia: positive results of a 24-week, multicenter,international, randomized, placebo-controlled clinical trial. Stroke 2003; 34(10):2323-2330.

(63) Pratt RD, Perdomo CA. Donepezil-treated patients with probable vascular dementiademonstrate cognitive benefits. Ann N Y Acad Sci 2002; 977:513-522.


(64) Winblad B, Engedal K, Soininen H, Verhey F, Waldemar G, Wimo A et al. A 1-year,randomized, placebo-controlled study of donepezil in patients with mild to moderate AD.Neurology 2001; 57(3):489-495.

(65) Feldman H, Gauthier S, Hecker J, Vellas B, Subbiah P, Whalen E. A 24-week,randomized, double-blind study of donepezil in moderate to severe Alzheimer's disease.Neurology 2001; 57(4):613-620.

(66) Mohs RC, Doody RS, Morris JC, Ieni JR, Rogers SL, Perdomo CA et al. A 1-year,placebo-controlled preservation of function survival study of donepezil in AD patients.Neurology 2001; 57(3):481-488.

(67) Tariot PN, Cummings JL, Katz IR, Mintzer J, Perdomo CA, Schwam EM et al. Arandomized, double-blind, placebo-controlled study of the efficacy and safety ofdonepezil in patients with Alzheimer's disease in the nursing home setting. J Am GeriatrSoc 2001; 49(12):1590-1599.

(68) Pratt RD, Perdomo CA, Surick IW, Ieni JR. Donepezil: tolerability and safety inAlzheimer's disease. Int J Clin Pract 2002; 56(9):710-717.

(69) Doraiswamy PM, Krishnan KR, Anand R, Sohn H, Danyluk J, Hartman RD et al. Long-term effects of rivastigmine in moderately severe Alzheimer's disease: does earlyinitiation of therapy offer sustained benefits? Prog Neuropsychopharmacol BiolPsychiatry 2002; 26(4):705-712.

(70) Corey-Bloom J, Anand R, Veach J. A randomized trial evaluating the efficacy and safetyof ENA 713 (rivastigmine tartrate), a new acetylcholinesterase inhibitor, in patients withmild to moderately severe Alzheimer's disease. Int J Geriatr Psychopharmacol 1998;6(4):55-65.

(71) Farlow M, Anand R, Messina J, Jr., Hartman R, Veach J. A 52-week study of the efficacyof rivastigmine in patients with mild to moderately severe Alzheimer's disease. EurNeurol 2000; 44(4):236-241.

(72) Erkinjuntti T, Skoog I, Lane R, Andrews C. Rivastigmine in patients with Alzheimer'sdisease and concurrent hypertension. Int J Clin Pract 2002; 56(10):791-796.

(73) Rosler M, Anand R, Cicin-Sain A, Gauthier S, Agid Y, Dal Bianco P et al. Efficacy andsafety of rivastigmine in patients with Alzheimer's disease: international randomisedcontrolled trial. BMJ 1999; 318(7184):633-638.

(74) Olin J, Schneider L. Galantamine for Alzheimer's disease. Cochrane Database Syst Rev2002;(3):CD001747.

(75) Erkinjuntti T, Kurz A, Gauthier S, Bullock R, Lilienfeld S, Damaraju CV. Efficacy ofgalantamine in probable vascular dementia and Alzheimer's disease combined withcerebrovascular disease: a randomised trial. Lancet 2002; 359(9314):1283-1290.


(76) Wilkinson DG, Passmore AP, Bullock R, Hopker SW, Smith R, Potocnik FC et al. Amultinational, randomised, 12-week, comparative study of donepezil and rivastigmine inpatients with mild to moderate Alzheimer's disease. Int J Clin Pract 2002; 56(6):441-446.

(77) Bryant J, Clegg A, Nicholson T, McIntyre L, De Broe S, Gerard K et al. Clinical andcost-effectiveness of donepezil, rivastigmine and galantamine for Alzheimer's disease: arapid and systematic review. Health Technol Assess 2001; 5(1):1-137.

(78) Birks JS, Melzer D, Beppu H. Donepezil for mild and moderate Alzheimer's disease.Cochrane Database Syst Rev 2000;(4):CD001190.

(79) Wolfson C, Moride Y, Perrault A, Momoli F, Demers L, Oremus M. Drug treatments forAlzheimer's disease. A comparative analysis of clinical trials. Ottawa: CanadianCoordinating Office Of Health Technology Assessment (CCOHTA), 2000.

(80) Livingston G, Katona C. How useful are cholinesterase inhibitors in the treatment ofAlzheimer's disease? A number needed to treat analysis. Int J Geriatr Psychiatry 2000;15(3):203-207.

(81) Rogers SL, Friedhoff LT. The efficacy and safety of donepezil in patients withAlzheimer's disease: results of a US Multicentre, Randomized, Double-Blind, Placebo-Controlled Trial. The Donepezil Study Group. Dementia 1996; 7(6):293-303.

(82) Rogers SL, Doody RS, Mohs RC, Friedhoff LT. Donepezil improves cognition andglobal function in Alzheimer disease: a 15-week, double-blind, placebo-controlled study.Donepezil Study Group. Arch Intern Med 1998; 158(9):1021-1031.

(83) Rogers SL, Farlow MR, Doody RS, Mohs R, Friedhoff LT, Donepezil Study Group. A24-week, double-blind, placebo-controlled trial of donepezil in patients with Alzheimer'sdisease. Neurology 1998; 50(1):136-145.

(84) Burns A, Rossor M, Hecker J, Gauthier S, Petit H, Moller JH et al. The effects ofdonepezil in Alzheimer's disease - results from a multinational trial. Dementia GeriatrCogn Disord 1999; 10(3):237-244.

(85) Greenberg SM, Tennis MK, Brown LB, Gomez-Isla T, Hayden DL, Schoenfeld DA et al.Donepezil therapy in clinical practice: a randomized crossover study. Arch Neurol 2000;57(1):94-99.

(86) Birks J, Grimley Evans J, Iakovidou V, Tsolaki M. Rivastigmine for Alzheimer's disease[Systematic Review]. The Cochrane Library 2000;(4).

(87) Sramek JJ, Anand R, Wardle TS, Irwin P, Hartman R, Cutler NR. Safety/tolerability trialof SDZ ENA 713 in patients with probable Alzheimer's disease. Life Sci 1996;58(15):1201-1207.


(88) Agid Y, Dubois B, Anand R, Gharabawi G. Efficacy and tolerability of rivastigmine inpatients with dementia of the Alzheimer type. Curr Ther Res Clin Exp 1998; 59(12):837-845.

(89) Forette F, Anand R, Gharabawi G. A phase II study in patients with Alzheimer's diseaseto assess the preliminary efficacy and maximum tolerated dose of rivastigmine (Exelon).Eur J Neurol 1999; 6(4):423-429.

(90) Wilcock GK, Lilienfeld S, Gaens E. Efficacy and safety of galantamine in patients withmild to moderate Alzheimer's disease: multicentre randomised controlled trial.Galantamine International-1 Study Group. BMJ 2000; 321(7274):1445-1449.

(91) Tariot PN, Solomon PR, Morris JC, Kershaw P, Lilienfeld S, Ding C et al. A 5-month,randomized, placebo-controlled trial of galantamine in AD. Neurology 2000;54(12):2269-2276.

(92) Raskind MA, Peskind ER, Wessel T, Yuan W, the Galantamine USA- Study Group.Galantamine in AD: A 6-month randomized, placebo-controlled trial with a 6-monthextension. Neurology 2000; 54(12):2261-2268.

(93) Doody RS, Geldmacher DS, Gordon B, Perdomo CA, Pratt RD. Open-label, multicenter,phase 3 extension study of the safety and efficacy of donepezil in patients with Alzheimerdisease. Arch Neurol 2001; 58(3):427-433.

(94) Homma A, Takeda M, Imai Y, Udaka F, Hasegawa K, Kameyama M et al. Clinicalefficacy and safety of donepezil on cognitive and global function in patients withAlzheimer's disease. A 24-week, multicenter, double-blind, placebo-controlled study inJapan. E2020 Study Group. Dement Geriatr Cogn Disord 2000; 11(6):299-313.

(95) McKeith I, Del Ser T, Spano P, Emre M, Wesnes K, Anand R et al. Efficacy ofrivastigmine in dementia with Lewy bodies: a randomised, double-blind, placebo-controlled international study. Lancet 2000; 356(9247):2031-2036.

(96) Hogervorst E, Yaffe K, Richards M, Huppert F. Hormone replacement therapy tomaintain cognitive function in women with dementia. Cochrane Database Syst Rev2002;(3):CD003799.

(97) Asthana S, Baker LD, Craft S, Stanczyk FZ, Veith RC, Raskind MA et al. High-doseestradiol improves cognition for women with AD: results of a randomized study.Neurology 2001; 57(4):605-612.

(98) Henderson VW, Paganini-Hill A, Miller BL, Elble RJ, Reyes PF, Shoupe D et al.Estrogen for Alzheimer's disease in women: randomized, double-blind, placebo-controlled trial. Neurology 2000; 54(2):295-301.


(99) Wang PN, Liao SQ, Liu RS, Liu CY, Chao HT, Lu SR et al. Effects of estrogen oncognition, mood, and cerebral blood flow in AD: a controlled study. Neurology 2000;54(11):2061-2066.

(100) Mulnard RA, Cotman CW, Kawas C, van Dyck CH, Sano M, Doody R et al. Estrogenreplacement therapy for treatment of mild to moderate Alzheimer disease: a randomizedcontrolled trial. Alzheimer's Disease Cooperative Study. JAMA 2000; 283(8):1007-1015.

(101) Tabet N, Feldmand H. Ibuprofen for Alzheimer's disease. Cochrane Database Syst Rev2003;(2):CD004031.

(102) Aisen PS, Schafer KA, Grundman M, Pfeiffer E, Sano M, Davis KL et al. Effects ofrofecoxib or naproxen vs placebo on Alzheimer disease progression: a randomizedcontrolled trial. JAMA 2003; 289(21):2819-2826.

(103) Scharf S, Mander A, Ugoni A, Vajda F, Christophidis N. A double-blind, placebo-controlled trial of diclofenac/misoprostol in Alzheimer's disease. Neurology 1999;53(1):197-201.

(104) Rogers J, Kirby LC, Hempelman SR, Berry DL, McGeer PL, Kaszniak AW et al. Clinicaltrial of indomethacin in Alzheimer's disease. Neurology 1993; 43(8):1609-1611.

(105) Birks J, Grimley EV, van Dongen M. Ginkgo biloba for cognitive impairment anddementia. Cochrane Database Syst Rev 2002;(4):CD003120.

(106) Le Bars PL, Velasco FM, Ferguson JM, Dessain EC, Kieser M, Hoerr R. Influence of theseverity of cognitive impairment on the effect of the Gnkgo biloba extract EGb 761 inAlzheimer's disease. Neuropsychobiology 2002; 45(1):19-26.

(107) Le Bars PL, Katz MM, Berman N, Itil TM, Freedman AM, Schatzberg AF. A placebo-controlled, double-blind, randomized trial of an extract of Ginkgo biloba for dementia.North American EGb Study Group. JAMA 1997; 278(16):1327-1332.

(108) Oken BS, Storzbach DM, Kaye JA. The efficacy of Ginkgo biloba on cognitive functionin Alzheimer disease. Arch Neurol 1998; 55(11):1409-1415.

(109) Ernst E, Pittler MH. Ginkgo biloba for Dementia: A Systematic Review of Double-Blind,Placebo-Controlled Trials. Clin Drug Invest 1999; 17(4):301-308.

(110) van Dongen MC, van Rossum E, Kessels AG, Sielhorst HJ, Knipschild PG. The efficacyof ginkgo for elderly people with dementia and age-associated memory impairment: newresults of a randomized clinical trial. J Am Geriatr Soc 2000; 48(10):1183-1194.

(111) Sano M, Ernesto C, Thomas RG, Klauber MR, Schafer K, Grundman M et al. Acontrolled trial of selegiline, alpha-tocopherol, or both as treatment for Alzheimer'sdisease. The Alzheimer's Disease Cooperative Study. N Engl J Med 1997; 336(17):1216-1222.


(112) Tabet N, Birks J, Grimley Evans J, Orrel M, Spector A. Vitamin E for Alzheimer'sdisease. Cochrane Dementia and Cognitive Improvement, editor. 2001. The CochraneCollaboration. The Cochrane Database of Systematic Reviews.

(113) Areosa SA, Sherriff F. Memantine for dementia. Cochrane Database Syst Rev2003;(1):CD003154.

(114) Pantev M, Ritter R, Gortelmeyer R. Clinical and behavioural evaluation in long-term carepatients with mild to moderate dementia under memantine treatment. Zeitschrift fuerGerontopsychologie und Psychiatrie 1993; 6(2):103-117.

(115) Ditzler K. Efficacy and tolerability of memantine in patients with dementia syndrome. Adouble-blind, placebo controlled trial. Arzneimittelforschung 1991; 41(8):773-780.

(116) Gortelmeyer R, Erbler H. Memantine in the treatment of mild to moderate dementiasyndrome. A double-blind placebo-controlled study. Arzneimittelforschung 1992;42(7):904-913.

(117) Orgogozo JM, Rigaud AS, Stoffler A, Mobius HJ, Forette F. Efficacy and safety ofmemantine in patients with mild to moderate vascular dementia: a randomized, placebo-controlled trial (MMM 300). Stroke 2002; 33(7):1834-1839.

(118) Winblad B, Poritis N. Memantine in severe dementia: results of the 9M-Best Study(Benefit and efficacy in severely demented patients during treatment with memantine).Int J Geriatr Psychiatry 1999; 14(2):135-146.

(119) Wilcock G, Mobius HJ, Stoffler A. A double-blind, placebo-controlled multicentre studyof memantine in mild to moderate vascular dementia (MMM500). Int ClinPsychopharmacol 2002; 17(6):297-305.

(120) Reisberg B, Doody R, Stoffler A, Schmitt F, Ferris S, Mobius HJ. Memantine inmoderate-to-severe Alzheimer's disease. N Engl J Med 2003; 348(14):1333-1341.

(121) Wimo A, Winblad B, Stoffler A, Wirth Y, Mobius HJ. Resource utilisation and costanalysis of memantine in patients with moderate to severe Alzheimer's disease.Pharmacoeconomics 2003; 21(5):327-340.

(122) Clinical Evidence, Issue 10, December 2003. Mental Health section, search date:October, 2002 (on-line version).

(123) Lonergan E, Luxenberg J, Colford J. Haloperidol for agitation in dementia. CochraneDatabase Syst Rev 2002;(2):CD002852.

(124) Brodaty H, Ames D, Snowdon J, Woodward M, Kirwan J, Clarnette R et al. Arandomized placebo-controlled trial of risperidone for the treatment of aggression,agitation, and psychosis of dementia. J Clin Psychiatry 2003; 64(2):134-143.


(125) Katz IR, Jeste DV, Mintzer JE, Clyde C, Napolitano J, Brecher M. Comparison ofrisperidone and placebo for psychosis and behavioral disturbances associated withdementia: a randomized, double-blind trial. Risperidone Study Group. J Clin Psychiatry1999; 60(2):107-115.

(126) De Deyn PP, Rabheru K, Rasmussen A, Bocksberger JP, Dautzenberg PL, Eriksson S etal. A randomized trial of risperidone, placebo, and haloperidol for behavioral symptomsof dementia. Neurology 1999; 53(5):946-955.

(127) Street JS, Clark WS, Gannon KS, Cummings JL, Bymaster FP, Tamura RN et al.Olanzapine treatment of psychotic and behavioral symptoms in patients with Alzheimerdisease in nursing care facilities: a double-blind, randomized, placebo-controlled trial.The HGEU Study Group. Arch Gen Psychiatry 2000; 57(10):968-976.

(128) Cummings JL, Street J, Masterman D, Clark WS. Efficacy of olanzapine in the treatmentof psychosis in dementia with lewy bodies. Dement Geriatr Cogn Disord 2002; 13(2):67-73.

(129) Clark WS, Street JS, Feldman PD, Breier A. The effects of olanzapine in reducing theemergence of psychosis among nursing home patients with Alzheimer's disease. J ClinPsychiatry 2001; 62(1):34-40.

(130) Kirchner V, Kelly CA, Harvey RJ. Thioridazine for dementia. Cochrane Database SystRev 2001;(3):CD000464.

(131) McKeith IG. Dementia with Lewy bodies. Br J Psychiatry 2002; 180:144-147.

(132) Bridges-Parlet S, Knopman D, Steffes S. Withdrawal of neuroleptic medications frominstitutionalized dementia patients: results of a double-blind, baseline-treatment-controlled pilot study. J Geriatr Psychiatry Neurol 1997; 10(3):119-126.

(133) Tariot PN, Erb R, Podgorski CA, Cox C, Patel S, Jakimovich L et al. Efficacy andtolerability of carbamazepine for agitation and aggression in dementia. Am J Psychiatry1998; 155(1):54-61.

(134) Tariot PN, Jakimovich LJ, Erb R, Cox C, Lanning B, Irvine C et al. Withdrawal fromcontrolled carbamazepine therapy followed by further carbamazepine treatment inpatients with dementia. J Clin Psychiatry 1999; 60(10):684-689.

(135) Porsteinsson AP, Tariot PN, Erb R, Cox C, Smith E, Jakimovich L et al. Placebo-controlled study of divalproex sodium for agitation in dementia. Am J Geriatr Psychiatry2001; 9(1):58-66.

(136) Sival RC, Haffmans PM, Jansen PA, Duursma SA, Eikelenboom P. Sodium valproate inthe treatment of aggressive behavior in patients with dementia--a randomized placebocontrolled clinical trial. Int J Geriatr Psychiatry 2002; 17(6):579-585.


(137) Gauthier S, Feldman H, Hecker J, Vellas B, Ames D, Subbiah P et al. Efficacy ofdonepezil on behavioral symptoms in patients with moderate to severe Alzheimer'sdisease. Int Psychogeriatr 2002; 14(4):389-404.

(138) Rogers SL, Doody RS, Pratt RD, Ieni JR. Long-term efficacy and safety of donepezil inthe treatment of Alzheimer's disease: final analysis of a US multicentre open-label study.Eur Neuropsychopharmacol 2000; 10(3):195-203.

(139) Sultzer DL, Gray KF, Gunay I, Berisford MA, Mahler ME. A double-blind comparisonof trazodone and haloperidol for treatment of agitation in patients with dementia. Am JGeriatr Psychiatry 1997; 5(1):60-69.

(140) Teri L, Logsdon RG, Peskind E, Raskind M, Weiner MF, Tractenberg RE et al.Treatment of agitation in AD: a randomized, placebo-controlled clinical trial. Neurology2000; 55(9):1271-1278.

(141) Chan WC, Lam LC, Choy CN, Leung VP, Li SW, Chiu HF. A double-blind randomisedcomparison of risperidone and haloperidol in the treatment of behavioural andpsychological symptoms in Chinese dementia patients. Int J Geriatr Psychiatry 2001;16(12):1156-1162.

(142) Lanctot KL, Best TS, Mittmann N, Liu BA, Oh PI, Einarson TR et al. Efficacy and safetyof neuroleptics in behavioral disorders associated with dementia. J Clin Psychiatry 1998;59(10):550-561.

(143) Koger SM, Brotons M. Music therapy for dementia symptoms. Cochrane Database SystRev 2000;(3):CD001121.

(144) Neal M, Briggs M. Validation therapy for dementia. Cochrane Database Syst Rev2000;(2):CD001394.

(145) Robb SS, Stegman CE, Wolanin MO. No research versus research with compromisedresults: a study of validation therapy. Nurs Res 1986; 35(2):113-118.

(146) Toseland RW, Diehl M, Freeman K, Manzanares T, Naleppa M, McCallion P. Theimpact of validation group therapy on nursing home residents with dementia. Journal ofApplied Gerontology 1997; 16(1):31-50.

(147) Spector A, Orrell M, Davies S, Woods RT. Reminiscence therapy for dementia.Cochrane Database Syst Rev 2000;(4):CD001120.

(148) Baines S, Saxby P, Ehlert K. Reality orientation and reminiscence therapy. A controlledcross-over study of elderly confused people. Br J Psychiatry 1987; 151:222-231.

(149) Goldwasser AN, Auerbach SM, Harkins SW. Cognitive, affective, and behavioral effectsof reminiscence group therapy on demented elderly. Int J Aging Hum Dev 1987;25(3):209-222.


(150) Spector A, Orrell M, Davies S, Woods B. Reality orientation for dementia. CochraneDatabase Syst Rev 2000;(4):CD001119.

(151) Dubinsky RM, Stein AC, Lyons K. Practice parameter: risk of driving and Alzheimer'sdisease (an evidence-based review): report of the quality standards subcommittee of theAmerican Academy of Neurology. Neurology 2000; 54(12):2205-2211.

(152) Carr DB, Duchek J, Morris JC. Characteristics of motor vehicle crashes of drivers withdementia of the Alzheimer type. J Am Geriatr Soc 2000; 48(1):18-22.

(153) Thompson C, Briggs M. Support for carers of people with Alzheimer's type dementia.Cochrane Database Syst Rev 2000;(2):CD000454.

(154) McCurry SM, Logsdon RG, Vitiello MV, Teri L. Successful behavioral treatment forreported sleep problems in elderly caregivers of dementia patients: a controlled study. JGerontol B Psychol Sci Soc Sci 1998; 53(2):122-129.

(155) Hebert R, Leclerc G, Bravo G, Girouard D, Lefrancois R. Efficacy of a support groupprogramme for caregivers for demented patients in the community: a randomizedcontrolled trial. Arch Gerontol Geriatr 1994; 18:1-14.

(156) Hebert R, Girouard D, Leclerc G, Bravo G, Lefrancois R. The impact of a support groupprogramme for care-givers on the institutionalization of demented patients. ArchGerontol Geriatr 1995; 20:129-134.

(157) Marriott A, Donaldson C, Tarrier N, Burns A. Effectiveness of cognitive-behaviouralfamily intervention in reducing the burden of care in carers of patients with Alzheimer'sdisease. Br J Psychiatry 2000; 176:557-562.

(158) Brodaty H, Green A, Koschera A. Meta-analysis of psychosocial interventions forcaregivers of people with dementia. J Am Geriatr Soc 2003; 51(5):657-664.

(159) Mittelman MS, Ferris SH, Shulman E, Steinberg G, Levin B. A family intervention todelay nursing home placement of patients with Alzheimer disease. A randomizedcontrolled trial. JAMA 1996; 276(21):1725-1731.

(160) Brodaty H, Gresham M, Luscombe G. The Prince Henry Hospital dementia caregivers'training programme. Int J Geriatr Psychiatry 1997; 12(2):183-192.

cmi dementia guidelines - kaiser...

Documents