t************************** · 2014. 4. 9. · brigitte m. duffy. health program office of...

DOCUMENT RESUME

ED 334 505 CG 023 520

AUTHOR Gluck, Michael E.; And OtherS

TITLE The Use of Preventive Services by the Elderly. Paper2.

INSTITUTION Congress of the U.S., Washington, D.C. Office ofTechnology Assessment.

PUB DATE Jan 89NOTE 73p.

PUB TYPE Reports - Descriptive (141)

EDRS PRICE MF01/PC03 Plus Postage.DESCRIPTORS Government Role; Health Education; *Health Services;

*Older Adults; Prevention; Public Policy; UseStudies

IDENTIFIERS *Medicare

ABSTRACTThis study examined the implications of potential

Medicare coverage for the use of preventive services by analyzingcurrent use and and the determinants of that use by the elderly.Trends in available data suggested that the use of preventiveservicee varies widely according to the type of service and that therates of use of specific services show a high level of consistencyacross studies, despite methodological differences. Data alsoindicated that if an elderly person receives any preventive services,he or she is likely to receive multiple services. An ana..ysis of datafor the over 65 population in the 1982 National Health InterviewSurvey found that the probability that an elderly person used each offive selected preventive services (glaucoma screening, eyeexaminations, blood pressure measurement, breast examinations, andPap smears) was consistently related to: (1) being male (for the

three services avatlable for men and women); (2) being younger; (3)

having more education; (4) having greater family income; (5) havinghealth insurance in addition to Medicare; (6) living in ametropolitan area; and (7) having spent more days in bed during theprevious year. The implications of this study suggest that reducingpatients' out-of-pocket expenses for preventive services throughMedicare would probably increase the percentage of elderly receivingpreventive care; Medicare coverage of preventive services mayindirectly increase the use of preventive services; and expansion ofMedicare to cover preventive procedures will represent an immediateboost in the program's financial obligations. (LLL)

********************************************t*************************** Reproductions supplied by EDRS are the best that can be made *

* from the original document. *

***********************************************************************

WA,

.

64AtIfte44,4KfilSrs':

'441111101111ier

0gt4

a

C.)CV

U II MPAPITMINT OF EDUCATIONMei ot Educations/ Research and Improvement

InEDUCATIONAL RESOURCES INFORMATION

CF NTE R (ERIC)

141rN

/This document hal been reproduced asreciirod horn Ulu meson or organization

CVOriginating it

r Milo/ changes have been made to Improve

0 reproduCt ion quality

Points ot vie* of opinions stated in thidockrmint do not necessarily reprefunt Otfictal

4.4PERI position or policy

FOREWORD

Interest in health promotion and disease prevention strategies for the elderly has grown inthe last ten years, partly as a result of advancing knowledge in these areas and partly due to thesearch for ways to moderate the rising cost of health care in this growing segment of the popu-lation. Reflecting this interest, the House Committee on Ways and Means requested that OTAanalyze the effectiveness and costs of providing selected preventive health services to the elderlyunder the Medicare program. The Senate Labor and Human Resources committee has also re-quested that OTA provide information on the value of preventive services for the Americanpeople.

This Staff Paper, The Use of Preventive Services by the Elderly, is the second in a series ofpapers being prepared in response to these requests. Understanding the use of preventive ser-vices by the elderly is an important component of assessing their effectiveness as Medicarebenefits. In this paper we review both new and previously published data on the proportions ofelderly currently receiving a variety of preventive health services; we examine factors associatedwith whether the elderly receive these services; and we analyze the likely implications forMedicare if preventive health services were offered as covered benefits.

The first paper in this series on "Preventive Health Services Under Medicare" examinedglaucoma screening as a potential. Medicare benefit. Subsequent papers will assess screening forcholesterol, cervical cancer, and colorectal cancer, and will analyze broad issues related toMedicare financing of preventive services for the elderly.

JOHN H. GIBBONSDirector

The Use of Preventive Services by the Elderly

by

Michael E. Gluck

Judith L. WagnerBrigitte M. Duffy

Health ProgramOffice of Technology Assessment

Congress of the United StatesWashington, D.C. 20510-8025

January 1989

A Staff Paperin OTA's Series on

Preventive Health Services Under Medicare

Carol Guntow prepared this Staff Paper for publication.

The views expressed in this Staff Paper do notnecessarily represent those of the TechnologyAssessment Board, the Technology AssessmentAdvisory Council, or their individual members.

CONTENTS

Chapter Page1. Summary 1

2. Introduction 5Types of Preventive Services 5

Preventive Services and Medicare 6

3. Recommendations of Professional and Expert Groups for the Use ofPreventive Services by Older Adults 7

4. What Percentage of the Elderly Use Preventive Services? 11

Sources of Data 11

Estimates of Use 14Use of Multiple Services 14Time Trends 15

5. What Factors Promote or Inhibit Elderly People's Use ofPreventive Services? 17

Theoretical Approaches 17Evidence on Patient Behavior 18Evidence From Studies of Health Care Provider Behavior 24

6. Implications for Policy 29Potential Medicare Coverage 29Dflivery of Preventive Services for the Elderly 30

AppendixA. Advisory Panel--Project on Preventive Health S.:rvices Under Medicare 33B. Acknowledgments 34C. Empirical Studies of the Use and/or Determinants of Use of Preventive

Services by the Elderly 35D. Empirical Studies of the Use and/or Determinants of Use of Preventive

Services by the Non-elderly 38E. OTA Analysis ot Preventive Service Use By the Elderly Using Data from

the 1982 National Health Interview Survey 46F. Analysis of Preventive Service Use By Older Adults in a Health

Maintenance Organization 55

Acronyms 59References 60

TablesTable Page

I. Selected Potential Clinical Preventive Services for the Elderly 62. Published Recommendations for the Use of Selected Preventive Services

by Older Adults 8

3. Percent of Liderly People Receiving Preventive Services Within SpecifiedPeriods of Time 12

4. Percent of Persons Over 65 Using Multiple Preventive Services fromthe 1982 National Health Interview Survey 15

ii 3

CONTENTS (cont'd)

5. Trends in the Percent of Adults or Ok:er Adults Ever HavingReceived Selected Preventive Services 16

6. Selected Factors Hypothesized to Affect Use of Preventive Services bythe Elderly 18

7. Significant Predictors of Use from OTA Multivariate Logit Analysis of1982 National Health Interview Survey 19

8. Periods of Time Used by OTA and Two Studies to Measure Older Adults' Useof Preventive Services 47

9. 1982 National Health Interview Survey: Selected Descriptive Statisticsfor Persons Over 65 49

10. Elderly Use of Five Screening Services: Logistic Regression Results 50

11. Correlation Matrix f -I- Variables in Logistic Regression Models 51

12. Effect of Statistically Significant Binary Variables in LogisticRegressions on Elderly Use of Five Screening Services: PredictedProbabilities 53

13. Sample Sizes for Each Measurement Period in OTA's Analysis of PreventiveService Use in One HMO 57

14. Percents of Continuously Enrolled HMO Members Receiving Eight PreventiveServices During Specified Periods of Time 57

FiguresFigure Page

1. Effect of Age on Use: Predicted Probabilities 54

2. Effect of Income on Use: Predicted Probabilities 54

3. Effect of Education on Use: Predicted Probabilities 54

4. Effect of Bed Days on Use: Predicted Probabilities 54

1. SUMMARY

In recent years, clinicians, academics,and policymakers have begun to examine thepotential benefits of services to promotehealth or to prevent disease, disability, ordeath in the elderly. Although Medicare, theFederal program responsible for paying thebulk of the nonilstitutionalized elderly'shealth care bills, currently pays for few pre-ventive services, Congress has severalproposals pending to expand Medicarecoverage of these procedures. In this paper,OTA examines the implications of potentialMedicate coverage for the use of preventiveservices by analyzing current use and thedeterminants of that use.

How Many Elderly Use PreventiveServices?

OTA found few sources that measure theuse of preventive services by the elderly.The data that are available (summarized intable ".) in the text of the paper) suggest twomain conclusions:

The use of preventive services by theelderly varies accurding to the type ofservice from a low of 20 to 30 percentfor routine fecal occult blood testing insome sites to a high of 93 percent forblood pressure measurement. These dif-ferences cannot be explained by dif-ferences in the periods of time overwhich use is measured.Rates of use of specific services show ahigh ovel of consistency actoss studies,despite differences in methods.

Trends in available data suggest that theuse of these procedures has increased over thelast 15 gears. Data also indicate that if anelderly person receives any preventive ser-vices, he or she is likely to receive multipleservices.

Which Elderly Use PreventiveServices?

Studies to isolate factors associated withthe use of preventive services fall into twocategories:

those that focus on the behaviot ofpatients, andthose that focus on the behavior ofhealth care providers and organizations.

Most of the studies in both of these categori-es examine preventive service use among thenonelderly. An analysis of data for the over-65 population in the 1982 National Health In-terview Survey (NHIS) found that, controllingfor other factors, the probability that anelderly person used each of five selected pre-ventive services--glaucoma screening, eye ex-ams, blood pressure measurement, breast ex-ams, and Pap smears--was consistently relatedto:

being male (for the three services avail-able for both men and women),being younger (although still over-65),having more education,having greater family income,

a having some health insurance in additionto Medicare,living in a metropolitan area, andhaving spent more days in bed duringthe previous year.

OTA found that receiving health care througha prepaid health plan was not related to theuse of any preventive service. However, sofew people in the study sample belonged toprepaid plans that it may not have been pos-sible to find a statistically significant effect.Race, living alone, and having some limita-tion in activity had no clear or consistent ef-fect on the use of the five services studied.

1

2 The Use of Preventive Services by the Elderly

Other studies of the relationship betweenpatient characteristics and the use of pre-ventive services have had similar findings.

Among health care providers, physiciansplay a key role in the provision of preventiveservices. The evidence suggests that gaps ex-ist between physicians' knowledge and ex-perts' recommendations on the use of pre-ventive services as well as betweenphysicians' knowledge or beliefs and actualpractice. These gaps may be more prominentin relation to elderly patients. While theymay suggest a shortcoming in physicians' per-formance, they could also indicate thatphysicians take individual patients' situationsinto account when ordering preventive ser-vices.

Other insights into the importance ofhealth care providers in determining whetherthe elderly receive preventive services comefrom trials designed to improve compliancewith expert recommendatiow. These studiesindicate that health care organizations canorganize themselves to affect the percentageof individuals receiving such services.Strategies suggested in the literature worthyof further study include:

targeting groups in need of prevention,using non-physician medical profes-sionals to deliver services, andgenerating reminders to physicians andpatients about the periodic need for pre-ventive services (especially with the aidof computerized record-keeping sys-tems).

Al,hough OTA's analysis of preventive ser-vice use showed that health maintenance or-ganizations (HMOs) had no discernible effecton elderly enrollees' pre entive activities, thereview of the literature on provider behaviorindicates that HMOs and other group prac-tices with centralized administration andrecord-keeping may have potential ft.,r in-creasing the use of such services.

Implications of MedicareCoverage for the Use ofPreventive Services by the Elderly

The findings of this study have threemain implications for potential Medicarecoverage of preventive services:

Reducing patients' out-of-pocket ex-penses for preventive services throughMedicare would probably increase thepercentage of elderly receiving pre-ventive care. For four of the five ser-vices examined in detail by OTA, hav-ing some insurance coverage beyondMedicare is associated with about a 10percent increase in the likelihood ofreceiving each service.

However, there are three caveats to thisfinding:

(1) OTA's analysis measured thepresence of insurance thatreduced patients' total out-of-pocket health care expenditures,not direct coverage of preventiveservices. The effect of directcoverage on use may be differentfrom the effect observed inOTA's analysis.

(2) The association between insuranceand use may not alw.ys reflect adirect cause and effect. Rather,some people may be likely both tobuy supplemental insurance anduse preventive services out ofconcern for their own health.

(3) OTA's analysis suggests that insur-ance coverage alone would not besufficient to induce many elderlyto avail themselves of preventiveservices.

The Use of Preventive Services by the Elderly 3

Medicare coverage of preventive servicesmay indirectly increase the use of pre-ventive services by raising interest Inpreventive care among non-Medicarehealth care consumers, providers, andpayors. Such coverage would, in effect,place the authority of the Federal Gov-ernment behind the covered services.OTA found no existing data to estimatethe existence or magnitude of thispotential effect.Because large numbers of elderly peoplealready use preventive services, expan-sion of Medicare to cover preventiveprocedures will represent an immediateboost in the program's financial obliga-tions even if increases in use are mini-mal or nonexistent. While Medicaremay already pay for some screening ser-vices incorrectly labeled as diagnosticprocedures, Medicare coverage wouldstill transfer a large portion of the cur-rent costs of preventive services frompatients or other payors to the FederalGovernment.

Other Implications for Policy

Among other factors important indetermining whether the elderly receive pre-ventive services, a few such as gender, age,education, income, rural or urban residence,and bed days could be useful in helpingpolicymakers target educational efforts on the

need for prevehtive services to those elderlyat highest risk of not complying with expertrecommendations. The relationship betweenuse and educational level suggests thatpolicyrnakers should carefully consider themedia they employ to promote preventiveservice recommendations, benefits, and otherprograms they undertake. Pamphlets or othermaterials that rely heavily on the writtenword are not as effective for the less well-educated who also have a relatively higherrisk of not receiving preventive procedures.Policymakers could consider using visualmedia to communicate their messages to suchgroups.

Some of the factors important to theelderly's use of preventive services, such asincome or educational level, are unlikely tobe the focus of policy efforts designed solelyto increase the use of preventive services.However, changing these factors for someother purpose might result in increases in use.

The analysis in this paper concentrateson those services most often raised in con-gressional discussions of prevention underMedicare--screening and immunizations. Theconclusions presented above may have limitedapplicability to consideration of other pre-ventive services such as health risk appraisals,health education, counseling services, or pre-vention of disability among elderly sufferingfrom chronic disease.

2. INTRODUCTION

In 1984, personal health care expendi-tures for the 28 million Americans over theage of 65 totaled $120 billion, nearly all ofwhich went toward the treatment of existingconditions rather than to screening for orpreventing health problems (85). Recently,however, policymakers, health advocates, andmedical practitioners have begun to focusgreater attention upon the potential of pre-ventive medicine for the elderly. As theelderly population has grown, physicians anddecisionmakers have looked to preventive ser-vices as a possible means of extending life,reducing morbidity and disability, and con-trolling health care costs (67,50). Congresshas recently mandated studies of community-based preventive health service programs forthe elderly and expanded Medical% coverageof certain services, including screening mam-mography and some immunizations (34).

This paper has three purposes:

to summarize existing professionalrecommendations for older adults' use ofpreventive health services,to estimate the percentage of elderlywho currently u.,e such services, andto identify the factors related to elderlyindividuals' use of preventive care withparticular attention to the potential ef-fects of Medicare coverage.

The information brought together in thispaper has two major policy implications.First, in order to estimate the impact ofMedicare coverage of preventive services onMedicare program expenditures, one must!snow the number of potential users. Whilecurrent rates of use alone may not adequatelypredict use under expanded thi.d-partyfinancing of preventive services, examinationof existing literature and data provides in-sight into factors associated with use. In par-ticular, such ana;ysis reveals the relative im-portance of Medicare coverage in removing

barriers to use for elderly Americans.

Planners and administrators of diseaseprevention for the elderly also benefit froman analysis of current use. By understandingthose factors that affect whether older peopleaccept and receive preventive services, Con-gress may be able to target initiatives wherethey will be most effective or most needed.Where supported by the evidence, this paperpoints out such implications for public policy.

Types of Preventive Services

The traditional taxonomy of preventiondistinguishes among piimary, secondary, andtertiary prevention (38,56). Primary pre-vention refers to activities designed to avoiddisease or other conditions that adversely af-fect health. Immunizations are one exampleof primary prevention. Secondary preventionincludes efforts to identify existing conditionsthat could cause illness and disability beforethe appearance of clinical symptoms, or tominimize the progression of disease. Diseasescreening is one form of secondary pre-vention. Tertiary prevention refers to effortsto control irreversible chronic conditions innrder to avoid disability or death. Kane, etal., have suggested that this typology does notadequately distinguish among preventive ser-vices, especially those targeted toward thechronic conditions common among the elder-ly. For example, while diet change can be ameans of primary prevention of hypertension,treatment of existing hypertension is also pri-mary prevention of stroke.

To avoid such ambiguities, this papersimply distinguishes among immunizations,disease screening, and educegional or counsel-ing services. Table 1 lists specific examplesof each category of prevention. While the listof services in table I is not exhaustive of allpreventive services applicable, it does includethe procedures examined in this paper.

5


Table 1.--Selected Potential ClinicalPreventive Services for the Elderly

ImmunizationsInfluenza

TetanusPneumococcueHepatitis 111'

ScreeningCancer screening:-Breast cancer (clinical examination;mammography)

-Colorectal cancer (occult blood stool;sigmoidoscopy)

-Cervical and uterine cancer (clinicalexamination; Pap smear; endometrial biopsy)-Prostate cancer (clinical examination;ultrasound)

-Skin cancer (clinical examination)Blood pressure measurementVision examinationGlaucoma screeningHearing testCholesterol measurementMental status/dementiaOsteoporosis (standard x-ray; quantitative CT;other radiological tochniques)Diabetes screeningAsymptomatic coronary artery disease (exercisestress test)Dental health assessmentMultiple health risks appraisal/assessmentFunctional status assessmentDepression screeningScreening for hyperthyroidism

Education and CounselingNutritionWeight controlSmoking cessationHome safety/injury preventionStress managementAppropriate use of medicationsAlcohol useExercise

Abbreviation: CT = computed tomography.aCurrently covered by Medicare.bCurrently covered by Medicare for high riskpatients.

SOURCE: Office of Technology Assessment, 1989.

Preventive Services and Medicare

In defining preventive services andmeasuring their use, this paper focuses on theimplications for their potential coverage un-der Medicare.

This perspective limits the preventive inter-ventions analyzed to personal health servicesoffered to individuals. This review does notexamine mass media education programstargeted toward the elderly.

As enacted in 1965, Medicare covered nopreventive services. It paid for procedures ona "diagnostic" basis only--that is, when thepatient has a symptom or a previous diagnosisfor a condition. However, because treatmentof most diagnosed conditions is covered,Medicare does pay for much tertiary pre-vention designed to control existing chronicconditions. In addition, some physiciansprobably receive payment for screening ser-vices they incorrectly label as "diagnostic."The extent of this de facto coverage of pre-vention has gone unmeasured.

In recent years, however, Congress hasincrementally added coverage of some im-munizations and screening. services. Theseinclude hepatitis B immunizations for benefi-ciaries at high risk of contracting the diseaseand pneumococcal pneumonia vaccinationsfor all beneficiaries. The MedicareCatastrophic Coverage Act of 1988 (PublicLaw 100-360) includes coverage of up to $50for biannual screening mammographies be-ginning in 1990. In addition, Congress hasmandated that the Health Care Financing Ad-ministration (HCFA) fund demonstrations ofinfluenza immunization coverage, and oftherapemtic shoes for diabetics, and severalcommunity-based demonstration projects toanalyze health outcomes and costs associatedwith the provision of screening, health riskappraisals, education, and counseling toMedicare beneficiaries.

Additional proposals brought before the100th Congress included coverage of Papsmear screening for cervical cancer and aphysical examination with medical historyupon enrollment in Medicare or on a pe:iodicbasis.

3. RECOMMENDATIONS OF PROFESSIONAL AND EXPERTGROUPS FOR THE USE OF PREVENTIVE SERVICES

BY OLDER ADULTS

One way to measure the use of pre-ventive services is to compare the actual be-havior of individuals with the frequency of'use recommended by expert groups.Numerous groups have provided recom-mendations about the periodicity with whichthe elderly should receive particular im-munizations and screening services. In inter-preting the medical evidence on frequency ofuse, these expert groups vary in the criteriathey employ in developing recommendations.

Table 2 summarizes several selected setsof recommendations made by professional orexpert groups for older adults, primarily forthose over 65 years old. The summary is notcomprehensive; rather it includes a range ofviews on the use of preventive services ana-lyzed in this paper.

The most comprehensive guidelines comefrom the two governmental task forces. Overthe last ten years, the Canadian Task Forceon the Periodic Health Examination has ana-lyzed medical evidence about the effective-ness of preventive services and made recom-mendations for Canadian citizens (18). TheU.S. Preventive Services Task Force hasengaged in a similar exercise and recentlypublished some of its findings. The taskforce will publish its full report in 1989(39,43,79).

Among other U.S. governmental organi-zations, individual institutes within the Na-tional Institutes of Health (NIH) have maderecommendations for cancer and coronaryheart disease screening (37,77,78). Some ofthese recommendations result from intramuralefforts within NIH, while others are the pro-duct of consensus development conferencesthat bring together experts and interested or-ganizations. Additional guidelines come fromprofessional societies such as the AmericanCollege of Physicians (4,5), the AmericanMedical Association (68), the AmericanAcademy of Ophthalmology (2), the Amer-ican Optometric Association (9), the Amer-ican College of Obstetrics and Gynecology(42), and the American College of Radiology(6) as well as health consumer organizationssuch as the American Cancer Society (3), theAmerican Society to Prevent Blindness (10),and the American Heart Association (8). Astable 2 indicates, there is nearly completeagreement among the included groups makingrecommendations for immunizations for theelderly. For screening services there is a highdegree of' consistency among groups, butsome disagreemer:t c;)es exist.

7

Table 2.--Published Recommendstions for the Use of Selected Preventive Services by Older Adults

Preventiveservice

Tetanus

Omiunization

COCa ACPb HNC CTFd USPSTFa

Professional

societiesfConsumer

organizations9

Booster every 10years if primary yearsseries has beendone

Booster every 10 Booster every 10 Booster every

years 10 years

Pnetsnococcaimmunization

Over age 65--once Over age 65--once High risk Over age 65--patients--once once

Influenza

immunizationOver age 65--every year

Over age 65--every year

Occult blood instool


NCI: over age50--every year

ACS: over age50--every year

Sigmoidoscopy NCI: over age50--every 3-5years

ACS: over age50--every 3-5years after 2

negative tests

Digital rectalexam


Not reccerendedfor prostatecancer; norecommendationfor enlargedprostatescreening

Clinical breastexamination

Considered inconjunction withmammography

NCI: over age50--evry year

Every year from Over age 40--

age !) to 59 every year


ACR: over age 35--every year (withmonthly breast self-examination)ACOG: advisesfollowing ACS

guidelines

ACS: over age40every year(with monthlybreast self-

examination)

mammography Says screeningwith mammographyis effective;does not specifyfrequency or whento start; saysscreening womenaged 50 to 59saves lives


Between ages 50and 59--everyyear


ACR: over ego 50--every yearACOG: adVisesfollowing ACSguidelinesAMA: between ages 40and 49--every one totwo years; age 50and over--every year


Table 2.--Published Recommendations for the Use of Selected Preventive Services by Older Adults (Cont'd)

Preventiveservice CDC ACPb CTFd USPSTFe

ProfessionalsocietiesT

Consumer

organizations0

Pap smear NCI: over ageor if sexuallyactive--3consecutiveannual Pap smearsand pelvic examswith negativeresults, thenless frequentlyat discvetion ofphysician

Every 5 yearsfrom age 35 toage 60; screeningshould continueif prior smearshave beenabnormal

AC00, ANA, ANA,AAFP, AND AMWA:support NCIguidelines

ACS: supportsNCI guidelines

Cholesterolscreening

NHLBI: over age20--every 5 years

AHA: supportsNHLBI

recommendations

Serum glucose Not recommendedwithout familyhistory ofdiabetes orprevious

circulatoryproblems

Blood pressure NHLBI: over age18--at least

every 2 years,dependingprevious reading

Over age 65--every 2 years

ADA: people atrisk should be

screened (nofrequency

specified)AHA: 4very 5years from age20 to 75;

optional after

age 75 ifbaselines arewell-documented

AMA: every 5years starting

at age 20

EKG Recommended forsymptomatic

adults only

AHA: at ages 20,40, and 60

Visionexaminationincludingglaucoma

screening bytonometry

4

Not recommended AOA: over age40--every year

AAO: over age40--every 2 to 5

years

ASPB: over age35--every 2 years

I 0

Table 2.--Published Recommendations for the Use of Selected Preventive Servicea by Older Adults (Cont'd)

Preventiveservice CDCe

ProfessionqlACPb NIHc CTFd USPSTF° societiesT

Consunerorgenizations9

Physical

examinationACS: over

age--40 everyyear

AMA: every 5years from age20 to 60;

every 2.5years from age61 to 75; over75--every year

Abbreviations: AAFP = American Academy of Family Physicians; AAO American Academy of Ophthalmology; ACOG * American College of Obstetricians andGynecologists; ACP = American College of Physicians; ACR American College of Radiologists; ACS = American Cancer Society; ADA AmericanDiabetes Association; AMA = American Heart Association; AMA = American Medical Association; AMWA American Medical Women's Association;ANA = American Nurses Association; ADA = American Optometric Association; CDC = Centers for Disease Control; CTF = Canadian Task Force;EKG electrocardiogram; NCI = National Cancer Institute; SIM = National Institutes of Health; USPSTF United States Preventive ServicesTask Force.

°Centers for Disease Control, Public Health Service, U.S. Deportment of Health and Human Services, Adult Immunizations: Recommendations of the ImmunizationPractices Committee, undated.

blmmunizations: American College of Physicians, Committee on Immunizations, Guide For Adult Immunization Philadelphia, PA, 1985; clinical breastexamination and mammography: American College of Physicians, Health and Public Policy Committee, "The Use of Diagnostic Tests for Screening and EvaluatingBreast Lesions," Annals of internal Medicine 103:143-146, 1985.

cCancer: Early Detection Branch, DivisiOn of Cancer Prevention and Control, National Cancer Institute, National Institutes of Health, U.S. Deportment ofHealth and Human Services, "Working Guidelines for Early Cancer Detection: Rationale and Supporting Evidence to Decrease Mortality," (Bethesda, MD:Decemter 1987); cholesterol: National Ctmlesterol Education Program, National Heart, Lung, and Blood Institute, National Institutes of Health, U.S.Department of Health and Human Service..., "Report of the Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults,"(Bethesda, MD: Octot,- 1987); blood pressure: 1988 Joint National Committee, "The 1988 Report of the Joint National Committee on Detection, Evaluation,and Treatment of High Blood Pressure," Arch. Intern. Med. 148(5):1023-1038, 1988.

dCanadian Periodic Health Examination Task Force, "The Periodic Health Examination," Canadian Medical Association Journal 121(9):1193-1254, 1979; and130(10):1276-1292, 1984.

ePreventive Services Task Force, U.S. Department of Health and Human Services, "Recommendations for Breast Cancer Screening," J.A.N.A. 257(16):2196, 1987;F.M. LaForce, "U.S. Preventive Services Task Force: Immunizations, Immunoprophylaxis, and Chemoprophylaxis To Prevent Selected Infections," J.A.M.A.257(18):24642470, 1987.TClinical breast examination: American College of Radiologists, Policy Statement, "Guidelines for Mammogrephy," edupted Sept. 22, 1982; American College ofOtatetricians and Gynecologists, personal communication with Lynne Lawrence, Government Relations Representative, Washington, DC, Oct. 31, 1988;mammography: American College of Radiologists, Policy Statement "Guidelines for Mammography," adopted Sept. 22, 1982; American College of Obstetriciansend Gynecologists, personal ccerunication with Lynne Lawrence, Government Relations Representative, Washington, CC, Oct. 31, 1988; American MedicalAssociation, "Women Age 40 to 49 Should Receive a Nomogram Every One to Two Years, AMA Delegates Recommend," Technology Remibursement Reports 4(28):6,1988; Pap smear: ACOG, ACS, NCI, AMA, ANA, AAFP, and AMWA, "Annual Pap Tests for at Least Three Years Advised by NCI; Guidelines Developed by CancerSociety," Health Policy and Biomedical News of the Week 31(3):6-7, 1988; vision examination: American Optometric Association, "Optometry and the Nation'sHealth: Recommendations for the Implementation of Congress' National Health Priorities," a working paper prepared by the National Health Division,February 19787 American Academy of Ophthalmology, Policy Statement, "Frequency of Ocular Examinations," approved Feb. 6, 1983.

ilAmerican Cancer Society, "Summary of Current Guidelines for the Cancer-Related Checkup: Recommeaaations," (New York: ACS Professional EducationPublication, 1988); American Diabetes Association, "A.D.A. Policy on Screening for Hyperglycenia," June 1983; American Heart Association: (cholesterolscreening) "Public Screening Strategies for Measuring Blood Cholesterol in AdultsIssues for Special Concern," October 1987; (serun glucose, bloodpressure, EKG, and physical examination) S.M. Grundy, P. Greenland, A. Herd, et al., "Cardiovascular and Risk Factor Evaluation of Healthy AmericanAdults," Circulation 75(6):1340A-1362A, 1987; American society for the Prevention of Blindness, "Facts on Blindness and Prevention," February 1988.


8

4. WHAT PERCENTAGE OF THE ELDERLY USEPREVENTIVE SERVICES?

While a large empirical literature existson the use of medical services in general, fewstudies concentrate on preventive services andfewer still analyze use of these services bythe elderly.' Apart from the analyses pre-sented for the first time in this paper, onlynine studies offer empirical evidence aboutthe use of preventive services among olderadults. Appendix C summarizes the scopeand methods of each of these studies.

Because of the small volume of researchexamining preventive service use by theelderly, this paper also draws upon empiricalinvestigations of use by the non-elderly. Ap-pendix D describes 35 studies in thiscategory. Several of these studies examinehow age affects the use of preventive servicesand offer insight into the behavior of olderpatients in seeking out such care.

Sources of Data

Table 3 presents comparative estimates ofthe percentages of elderly people using 17preventive services within specified periodsof time. Three of the seven sources containnational estimates. OTA analyzed data fromthe 1982 National Health Interview Survey(NIIIS). (See appendix E for a description ofthe NHIS.) Results of this analysis, showingthe percentage of the elderly who havereceived five preventive services--glaucomascreening, eye exams, blood pressuremeasurement, breast exams, and Pap smears--within periods of time roughly similar to theintervals suggested by expert groups areshown in the sixth column of the table.

1 Many studies have been carried out on the use andcorrelates of use of non-preventive medical ser-vices and dental care by the elderly. However,

because the purpose, nature, and likelydeterminants of use of these services differmarkedly from those of preventive health services,such studies are not reviewed in this paper.

The second source of national estimatesin table 3 comes from a survey conducted bythe Gallup Organization every 3 or 4 yearsfor the American Cancer Society (ACS). Thishousehold, mail survey examines individuals'knowledge of cancer risk factors and the fre-quency with which they receive certainscreening tests (28). Gallup publishes resultsby gender, age, and selectRd demographicvariables. Although the stuuy does not pre-sent findings for Medicare-eligible respon-dents as a separate group, it does give resultsfor individuals over 50 years old.

These two studies rely on respondents'self-reported behavior, which may affect theaccuracy of the estimates. The direction ofthis potential bias is unclear. On the onehand, lack of familiarity with medical ser-vices may cause respondents not to know thatthey had received a given service, and hence,to underreport use. On the other hand,respondents may perceive preventive behaviorto be socially desirable and may inflate theuse they report to the interviewer. The rela-tive importance of each of these biases in af-fecting the estimates is unknown.

The third national data source is the U.S.Immunization Survey conducted annually bythe Centers for Disease Control (CDC) until1985. A household survey, it provides dataon the percentages of individuals immu-nized against influenza and pneumococcalpneumonia, broken down by age (includingpeople over 65) (26).

Another set of estimates, found in the3eventh column of table 3, comes frcm alarge, urban "closed-panel"2 health main-tenance organization (HMO). This HMO pro-

2 In a "closed panel" HMO, enrollees must receivehealth care from a physician employed directly bythe HMO usually in a clinic run by the organiza-tion.

11


Table 3.--Percent of Elderly People Receiving PreventiveServices Within Specified Periods of Time

Service

Studve

ACS Brownc CDC Chaoe Lazarof OTAg OTAh !Uncle& Winewed

Blood pressureMenWomen

93

93 k(1 yr)

91

92(2 yr)

Breast examWomen 48 53

(1 yr) (<2 yr)

Breast self-examWomenMonthly or more frequentlyLess frequently

3739

CholesterolMen 73

Women 73

(5yr)

Complete check-up By age, not sex:Men 62 60-74=45 49Women 67 75+=24 51

(<1 yr) (N/A) (1 yr)

Eye examinationMen 73 72

Women 76 75

(<3yr) (2 yr)

Fecal occult bloodMen 20 30 49Women 19 29 52 70-80

Total (1 yr) (1 yr) (1 yr) (N/A)

Glaucoma screening

Men 64Women 70

(<3 yr)

Influenza vaccine

Men 23 58

Women (both sexes; 57

1 yr) (1 yr)

MammographyWomen 6 11

(ever) (1 yr)

Pap smear

Women 60 63 50 71

(<=3 yr) (1 yr) (<4 yr) (3 yr)

Pneummococcal vaccineMen 11 38Women (both sexes; 30

1 yr) (ever)

(Cont'd)

The Use of Preventive Services by the Elder!, 13

Table 3.--Percent of Elderly People Receiving PreventiveServices Within Specifed Periods of Time (Cont'd)

Service

$ tudy&

ACS Brownc CDCa Chao° Lazarof OTAg OTAh Rundall Winaweri

Procto exam

Men 15

Women 12

(1 yr)

Rectal exam

Men 28

Women 28(1 yr)

SigmoidoscopyTotal

95

(N/A)

Swine flu vaccineSample eutimete

72

Population est.63(N/A)

Tetanus vaccine

Men30

Women26

(10 yr)

Abbreviations: HMO = health maintenance organization; NH1S = National Health Interview Survey.

a Full descriptions of methodology of each study car. lund in table 4.bGallup Organization, "The 1987 Survey of Public 4SS and Use of Cancer Detection Tests: Summary of Find-

ings," Conducted for the American Cancer Society (Frinceton, NJ: Gallup Organization, January 1988); n=952;

age=50+.C.J.T. Brown and B.S. Hulka, "Screening Mammography in the Elderly: A CaseControl Study," J. Gen.Intern.h4edicine

3:126-131, 1988; n=309; age=60+.dD.S. Fedson, "Influenza and Pneumococcal Immunization Strategies for Phfsicians," Chest 91:436-443, 1987; n=not

given; age=65+.e A. Chao, A. Paganini-Hill, R.K. Ross, et al., "Use of Preventive Care by the Elderly," Preventive edicir,

16:71U-72, 1987; n=11,888; age range=48--100; mean age=74.4.fC.M. Lazaro, D.N. Logsdon, and R. Meier, "Utilization of Preventive Health Services by the Elderly," Insure

Project, Lifecycle Preventive Health S-rvices, New York, NY, presentation to the American Pyschological Asso-

ciation Convention, Aug. 31, 1987, New York, NY; n=713; age=60+. Use ratec from Lazaro studY are proportions

of all persons invited to receive checkup who actually received the service. Proportions of persons accerOUlthe

invitation who actually received the checkup are as '''ollows: ages 60.74=.65; ages 75+=.53.

gOffice of Technology Assessment/NHIS, 1988a; n=114!4; age=65+.hOffice of Technology Assessment/HMO, 1988b; n=5394 for checkup, influenza, and fecal occult blood; 3371 for eye

exams and Pap smears; 2322 for cholesterol; and 894 for pneumococcal and tetanus; age=65+.

iT.G. Rurdall and J.R.C. Wheeler, "Factors Associated kithUtilization of the Shine Flu Vaccination Program Among

Senior Citizens in Tompkins County," Medis...alfar.s. 17:191-200, 1979; sample n=232: population N=5000; age=65+.

S.J. Winawer, M. Baldwin, E. Herbert, et al., "Screening Experience With Fecal Occult Alood Testi;va a a Fum-

tion of Age," in Prospectives on Prevention andTreatment ofCancer in the Elderly, R. Yancik (New York. NY:

Raven Press, 1983); n=21,961; age=40+.kNotation in parentheses indicates period of time over which use was measured.

Key for parenthetical notations:yr = year or years

4 = up to but not including<= = up to and includingN/A= not applicable; study is a sinle trial conducted over a finite period.



vided OTA with data on the percentages ofadults in various age categories who receivedeach of nine preventive services within peri-ods of time specified in table 8. (See appen-dix F for a more complete discussion of thedata and estimation methods.)

Unlike the three national surveys, theseestimates come directly from the provider'srecords, thus avoiding the potential in-accuracies of self-reported data. However,the population from which the HMO data aredrawn is probably not representative of thenational experience or even of other HMOs.The elderly enrolled in this single prepaidplan may be different from the total elderlypopulation in the HMO's market area as wellas the elderly population of other areas. Inaddition, HMOs in general tend to providebetter coverage of preventive services than doother insurance plans (46). This HMO inparticular engaged in activities to promote theuse of some preventive procedures. All ofthese potential distortions suggest thatestimates from this HMO are probably in-dicative of the upper bound of use attainableunder Medicare coverage rather than nationalestimates of current use.

In the four remaining sets of data pre-sented in table 3, estimating use was not theauthors' primary objective. One paper was acase-control study of breast cancer in elderlywomen (15). Another looked at the rela-tionship between screening and disease prog-nosis for colorectal cancer (89). The thirdexamined factors associated with swine fluvaccination during the predicted epidemic of1977 and 1978 (59), and the fourth presentsself-reported data from a retirement com-munity about respondents' most recent use offive preventive services (19).

Estimates of Use

Because of some overlap in the servicesexamined in the seven studies discussedabove, one can compare different estimates ofuse of the same services. These proceduresare general exr.--.inations, fecal occult blood

screening, mammography, breast examina-tions, Pap smears, eye examinations, andblood pressure checks. For four services, theestimates of use are consistent across datasources. About 92 percent of the elderlyreport having their blood pressure checkedwithin a 1-year period and 74 percent reporteye examinations within the previous 2 years.Although estimates for Pap smear use show abit more variation across studies, the rangeruns only from about 50 percent of elderlypeople in the NHIS sample to 71 percent inthe HMO data.

Differences in the periods of time overwhich researchers measure use do not accountfor the variation in estimates that does exist.For example, the ACS estimate of Pap smearuse within a 3-year period is actually higherthan the NHIS estimate that examines a peri-od of up to 4 years. Hence, these differencesreflect either different populations or dif-ferent survey methods.

Despite some consistency across studiesfor the same service, there is little similarityin rates of use across different services. Forexample, while less than 15 percent of theelderly report having had annual rectal ex-ams, 92 percent report an annual blood pres-sure check. Estimates for the remaining ser-vices fall within this wide range. Thesedrastic differences in rates of use suggest thatpreventive services are more different fromone another than they are alike. Severalstudies discussed later in this paper have ex-amined these differences.

Use of Multiple Services

Measuring the percentage of elderly in-dividuals who receive multiple preventiveservices provides a slightly different profileof individuals' preventive behavior than isrevealed by examining one service at a time.As indicated in table 4, a majority of elderlypersons report receiving all three services thatboth sexes can receive (glaucoma screening,eye exams, and blood pressure measurement).One-quarter of men and one-fifth of women

f) ')4.


report receiving one or fewer of the threeservices. The extremely small percentageswho report using no services reflect the al-most universal measurement of blood pres-sure. Looking only at women and includingthe two additional services they can receive(Pap smears and breast exams) reveals thatonly 13 percent report using none or one ser-vice. However, only 30 percent report usingall five services. These data indicate a greatdeal of variation in the number of serviceselderly people receive. In the only otherstudy to examine multiple preventive serviceuse, Calnan found that among middle-agedwomen, the probability of using one servicedoes not predict whether an individual usesothers (17).

Time Trends

The ACS and NH1S data allow examina-tion of time trends in self-reported use ofseveral services over the period from 1973through 1987. As shown in table 5, the per-centage of older Americans who report everhaving received these preventive servicesgrew over the periods measured. Usingidentical questionnaires, the NHIS showedsubstantial increases in the use of seven ser-vices between 1973 and 1982 (72,75).

The trends in the ACS data are not quiteas dramatic (28). Some procedures show littlechange between 1980 and 1983 with five ser-vices showing a decline in use. The declinesbetween 1980 and 1983 most likely reflectsampling error. While all of the tests exceptdigital rectal exams for women increased be-tween 1980 and 1987, the jumps are lessdramatic than those suggested by the NHISdata.

The differences in trends between theACS and NHIS data sources have several pos-sible explanations:

Only Pap smears and breast exams over-lap the NHIS and ACS surveys. Thedifferences between the two data sourcescould be due to different trends in theparticular services each survey exam-ined.NHIS estimates are for individuals over65 years old, while the ACS data are forwider age ranges. If the trend in theNHIS applies only to the elderly, the in-clusion of non-elderly people in theACS samples might obscure this trend.

Table 4.--Percent of Persons Over 65 UsingMultiple Preventive Services (From the 1982

National Health Interview Survey)a

Number ofservices

Glaucoma, Breast exam, Pap smear

eye exam and glaucoma, eye exak

blood pressures and blood Pressure

Men Women Women

ZerooneTwoThreeFour

Five

Total

4 3

22 19

16 13

58 65

N/A N/A

N/A N/A

100 100

310

11

26

21

30

100

Abbreviation: N/A = Not applicable.aFor men and women, table presents proportions ofthe noninstitutionalized, civilian, over-65 pop-ulation using none, one, two, or three of thefollowing services--glaucoma screening, eye ex-amination, and blood pressure measurementwithinthe periods of time listed in table 8.

b For women only, the table also presents the pro-portions of this same population using none, one,two, three, four, or five of the followingservices-glaucoma screening, eye examination,blood pressure measurement, Pap smears, and breastexaminationwithin the periods of time listed in

table 8.


16 es The Use of Preventive Services by the Elderly

Table 5.--Some Trends in the Pe.cent of Adults or Older AdultsEver Having Received Selected Preventive Services°

National Healthb

Interview SurveyAmerican Cancer Society/GallupOrganization survevc

ear YearService 1973 1982 1976 1980 1983 1987

Check-up 45 42 46 47(annual)a (All adults)

Breast exam(women 58 82 74 79 89 81only) (Ages 65+) (All adult women)

Pap smear 54 79 79 86 84 87(womenonly)

(Ages 65+) (All adult women)

EKG 67 82

(Ages 65+)

Eye exam 94 98(Ages 65+)

Glaucoma 56 83(Ages 65+)

Fecal occult bloodMen 17 29 43Women 20 27 47

(Ages 50+)

mammography43 41 62

(women only) (Ages 50+)

ProctosigmoidoscopyMen 37 32 43Women 35 31 42

(Ages 50+)

Rectal examMen 54 56 53Women 52 47 58

(Ages 40+)

Abbreviation: EKG = electrocardiogram

aWith the exception of the medical check-up, this table presents data on the proportion of individuals whoreport ever having received each service. For medical check-ups, the statistics refer to the proportionwho report having a regular, annual exam.

bNational Center for Health Statistics, U.S. Department of Health, Education, and Welfare, "Use of SelectedPreventive Services U.S.--1973," Vitaland Health Statistics, Series 10, No. 110 (Washington, DC: U.S. Gov-ernment Printing Office, March 1977); and National Center for Health Statistics, U.S. Department of Healthand Human Services, "Use of Selected Preventive Services U.S.--I982," Vitaland Health Statistics, Series 10,No. 157 (Washington, DC: U.S. Government Printing Office, August 1986).

cGallup Organization, "The 1987 Survey of Public Awareness and Use of Cancer Detection Tests: Summary ofFindings," conducted for the American Cancer Society (Princeton, NJ: Gallup Organization, January 1988).


0 .?4.1

5. WHAT FACTORS PROMOTE OR INHIBITELDERLY PEOPLE'S USE OF PREVENTIVE SERVICES?

Theoretical Approaches

In addition to providing estimates of thepercentage of the population receiving pre-ventive services, the literature laid out in ap-pendixes C and D offers insight into factorsassociated with use. These studies representat least two theoretical approaches to explain-ing the use of preventive services: 1) an ap-proach that emphasizes patient behavior, and2) an approach that emphasizes provider be-havior.

Patient Behavior

Underlying this approach is the assurnp-tion that the decision to use a preventivehealth service is made by the recipient.Receipt of these services results from factorsthat influence the decision to seek preventivecare and the patient's ability to carry out thatdecision. There are two main versions of thisapproach: I) a model of medical serviceutilization first proposed by Andersen hiscolleagues (11,12,36), and 2) the Health BeliefModel (57,58).

The Andersen Model.--According to thismodel, three types of factors determine anindividual's probability of using medical ser-vices as well as the volume of use:

Predisposing variables includedemographic factors and the individual'sbeliefs about the services.Enabling variables that affect thepatient's ability to gain access to servicesinclude the individual's financialresources, the availability of the servicesin the individual's community, and in-surance coverage.Need variables include practitioners' andpatients' own perceptions of the patient'shealth status. Poor health status may in-dicate a need for better health care, in-cluding preventive services. Alterna-tively, variables that measure healthstatus may actually be proxies for theneed for nonpreventive health services.

To the extent that the need for theseother services increases contact with thehealth care system, individuals may bemore likely to receive preventive ser-vices that require some health care in-tervention. Hence, health status vari-ables may enable or predispose individ-uals to receive preventive services byincreasing their contact with the health..:are system.

The Health Belief Model.--This behav-ioral model arose from an attempt by medicalsociologists during the early 1970s to under-stand patterns of preventive health and healthmaintenance (48). It is similar to Andersen'smodel in its focus on the patient. However,it posits that patient beliefs and attitudes arethe most direct determinants of the decisionto receive preventive care. Sociodemographicfactors, characteristics of the health care sys-tem, and other exogenous variables (such aspublic education or illness of a family mem-ber) all indirectly affect preventive behaviorby influencing the individual's beliefs and at-titudes (57,58).

These attitudinal factors include:

the patient's perceived susceptibility to agiven disease or condition;the perceived potential severity of thatdisease;the perceived benefit of preventive actionin reducing susceptibility or severity;cues to taking the action such as publiceducation programs, reminders andphysician recommendations; andthe perceived barriers to taking the ac-tion including cost, inconvenience, andembarrassment.

One major limitation of the Health BeliefModel in explaining the us (. of preventiveservices is the lack of data measuring indi-vidual attitudes and perceptions. Only datasets constructed specifically for Health BeliefModel analyses are likely to contain the req-uisite information (23,31,54).

4, 17


However, some researchers have used the un-derlying relationships suggested by the HealthBelief Model to design experiments to im-prove preventive behavior among patients(70).

Provider Behavior

While patient behavior models focus onthe consumers of health, provider behaviormodels focus upon providers of such services(86). They suggest that patients receive pre-ventive services as the result of their pro-viders' decisions to offer, encourage, andenable their uptake. While the patient behav-ior models see patients as active decision-makers, the provider behavior approach seespatients as more passive, less important thantheir providers.

Explanatory studies that use the providerbehavior approach examine the effects on useof health care organization, patient contactwith the health care system, or with differenttypes of health personnel, and providers'knowledge of preventive services (55,92).This approach also underlies experiments anddemonstration projects that try to determineLow the manner in which services are pro-vided can maximize their use. Mass screen-ing programs or trials that employ physicianeducation are examples of these types ofstudies (20,24,62,74).

Combining the Patient and ProviderApproaches

The patient and provider approachesneed not be mutually exclusive. At leastthree studies nave attempted to combine theprovider and patient approaches into a singlemodel (21,29,86). Although each approachplaces emphasis upon different groups ofpotential determinants of preventive behavior,they may be valid in explaining differentparts of the variation in use. In addition,there is some overlap among the two ap-proaches. Andersen's enabling variablesrepresent the same basic ideas that the pro-vider behavior models focus upon. However,in the patient behavior models, characteristics

of the health care system affect individualpatient decisionmaking. Provider behaviorstudies implicitly assume a more passive rolefor patients who respond largely to actions ofhealth providers.

Evidence on Patient BehaviorThis section describes the results of

OTA's analysis of the 1982 NHIS data set (seeappendix E for detailed discussion of meth-ods) and examines how these results comparewith results of studies listed in appendixes Cand D. While many of the studies in appen-dix D are limited in their implications forelderly use of prevention, they provide ageneral context within which studies ofelderly populations can be interpreted.

OTA's Analysis of the 1982 N111S

Although the 1982 NHIS does not containall of the variables described in Andersen'sapproach and in the Health Belief Model, itis the most comprehensive existing source ofinformation about the determinants of pre-ventive behavior among the noninstitutional-ized elderly. Table 6 lists potential explana-tory variables included in the Nuns data set.

Table 6.--Selected Factors HypothesizedTo Affect Use of Preventive Services

by the Elderly

Predisposing factorsGender

AgeRace (whiteinon.white)Education

Enabling factorsFamily income

Having health insurance in addition toMedicareaReceiving health care through a prepaid planLiving aloneLiving in a metropolitan area

Measures of health statusBed days in the previous 12 months

Having some limitation on activity

aPotential insurance coverage for the elderly inaddition to Medicare includes privately purchasedhealth insurance, Veterans Administration or mil-itary health insurance, or means tested publicassistance health benefits including Medicaid.


The Use of Preventive Services by the Elderly is 19

The major category of variables missing fromthese data are the health belief variables:patient perceptions of disease threat, per-ceived benefits of preventive services, andperceived barriers to their use.

OTA used weighted logit models toestimate the independent effects of each of

Table 7.--SignIficant Predictors of Use FromOTA Multivariate Logit Analysis of 1982

National Health Interview Survey'

Glaucoma Eye Blood Pap Breast

Variable screening exam pressure smear exam

predisposing factors1. Sex (mate)2. Age3. Race (non-white)4. Education

Enabling factors5. Family income6. Having health

insurance inaddition toMedicare

7. Receiving healthcare through

prepaid plan8. Living in a

metropolitan area9. Living alone

Measures of health10. Bed days in

the previous12 months

11. Having somelimitation

on activity

+ +

+ +

++ ++

NS NS

+ + ++

NS NS

NS

NS++

status

+ + ++

NS

N/A N/A+ +

NS NS NS++ ++

NS ++ ++

++ ++ ++

NS NS NS

NS ++

NS NS ++

+ + ++ ++

NS NS

Abbreviations: NS = Estimated coefficient on vari-

- able not statistically significant;N/A = Not applicable; variable notincluded in model.

aUse measured according to standards described inappendix E and in table 8.

KeY to symbols:+ =Difference in proportions using service sig-

nificnt at 0.05 level (2 tailed), variablepositively associated with use=Difference in proportions using service sig-nificant at 0.01 level (2 tailed), variablepositively associated with use=Diffelz-tce in proportions using service sig-nificant at 0.05 level (2 tailed), variablenegatively associated with use=Difference in proportions using service sig-nificant at 0.01 level (2 tailed), variablenegatively associated with use


4. 4.

the variables listed in table 6 on the use ofeach screening test or examination includedin the NHIS. These models posit that theprobability that an individual uses a pre-ventive service within a specified period oftime is a function of the variables listed inthe table. Appendix E describes each modelspecification in greater detail and presentsactual estimates. Table 7 summarizes thestatistically significant predictors of use.

Despite the substantial variation that ex-ists across the five services in the percentageof elderly receiving the specified ievels ofprevention, the estimated models show a greatdeal of consistency across services in the sig-nificant predictors. Only the use of bloodpressure checks appears different. The anal-ysis suggests that fewer variables are impor-tant in predicting blood pressure checks thanin predicting the other services. This is con-sistent with the relative lack of variation inthe use of this service; over 90 percent of theelderly report having had their blood pressuremeasured within the previous 2-year period.Almost every medical visit includes bloodrressure measurement, and individuals canuse machines found in many supermarketsand restaurants to screen themselves for hy-pertension.

As expected, OTA's analysis found thatyounger age, more education, and higher in-come are all consistently associated with ahigher probability of using the five pre-ventive services measured in the NHIS. Forthe three services applicable to both genders,men over 65 are less likely than women ofsimilar age to receive them.

More bed days are consistently related touse of the five services, suggesting that sickerindividuals have greater contact with thehealth care system, and hence, a greater op-portunity to be offered preventive services.The analysis showed no relationship betweenlimitation in activity and the.use of any ofthe preventive services except for eye exams,where the direction of the association is neg-


ative, and blood pressure measuremPnt, wherethe direction is positive.1

For all services except blood pressuremeasurement, living in one of the CensusBureau's Standard Metropolitan StatisticalAreas (SMSAs), which are defined by geog-raphy and population density, is positivelyand significantly related to the use of pre-ventive services. In urban communities with

.a large number of health facilities and per-sonnel, individuals are likely to live closerand have more ways to get to appropriatehealth facilities than their rural counterparts.

Having some health insurance in additionto Medicare is also associated with use. Al-though such "Medigap" coverage (held by 71percent of the elderly in the NHIS sample) isunlikely to pay for preventive services, itdoes lower patients' out-of-pocket expensesfor medical care, thus making preventive ser-vices more affordable. In addition, apatient's willingness to buy such insurancemay indicate a certain concern for his or herown health also found in individuals likely touse preventive services.

Among variables not associated with theuse of preventive services, membership in anHMO or other prepaid health plan is the mostunexpected. Published literature indicatesthat such health care providers are more like-ly than others to offer preventive services to

It is possible that bed days and limitations aremeasuring health status in the same way. If thetwo variables are measuring the exact same idea,the logit estimation procedure would not be able todistinguish the independent effects of each vari-able. This could lead to the insignificant coef-ficients found for the limitation variable. Totest for this potential problem (multi-collinearity), we examined the correlation betweenthe bed days and limitations variables. We found acorrelation coefficient of 0.31 suggesting thatwhile the estimated standaniernmm of the two logitcoefficients may be somewhat biased downward (thuscreating potential non.significance), the two vari-ables largely measure different notions of healthstatus.

their patients in hopes of lowering treatmentexpenses (46,54). While OTA's analysis sug-gest that this relationship may not exist, it isalso possible that the small number of elderlyNHIS respondents enrolled in HMOs did notprovide enough statistical power to detect anactual effect of prepaid membership. In 1982only 2.3 percelit (or about 573,000) of theelderly belonged to HM0s,2 and the NHISsample reflects this relatively small number.

The remaining two variables in the logitmodels are not consistently associated withuse: race is positively associated with glau-coma screening; living kzlone is positively as-sociated with breast examinations.

Summary of Evidence on the Determinants ofUse of Preventive Services

Age.--Age has generally been found tobe a negative predictor of the elderly's use ofpreventive services (16,19,69). Studies ofyounger adult populations have also foundsuch an association (32,44,80,81,81,88). Inthe OTA analysis, all services except bloodpressure followed this pattern.

Twl. studies have examined the rela-tionship between age and immunization be-havior; neither found any strong associationbetween age and swine flu immunization(22,59). Of the other studies that look at age,one found a positive correlation with thelikelihood and volume of preventive visits toa single HMO (40). Another study thatlooked only at fecal occult blood screeningwithin a well-defined trial also found no ageeffect (86).

2 Since 1982, the number of Medicare beneficiarieshas grown due to risk- and cost-based Medicaredemonstration programs. Under contract with theHealth Care Financing Administration, each HMOparticipating in these programs agrees to provideMedicare benefits to eligible enrollees. As ofJanuary 1988, over 1.7 million elderly were partic-ipating in Medicare/HMO demonstration programs(71).

The Use of Preventive Services by the Elderly s 21

Despite these exceptions, the bulk ofavailable evidence suggests that use of pre-ventive services falls with age, especiallyamong the elderly. There are at least threeexplanations for this observation:

As an individual gets older, he or shemay perceive fewer benefits and morebarriers to receiving commonly recom-mended levels of prevention.The observed difference may reflect atendency of individuals who were elder-ly in the early 1980s not to use pre-vention--a tendency that will disappearamong future groups of elderly.s Thebenefits and availability of most pre-ventive services emphasized today havebeen known for only the past genera-tion. By the time that these services be-came widespread, older individuals mayalready have established patterns ofhealth care that did not include pre-vention. According to this reasoning,succeeding generations may have moreuniform rates of preventive care overthe adult age spectrum.As individuals age, they are more likelyto visit the doctor for diagnostic andtherapeutic services. While they stillmay receive preventive services, theymay not remember that the clinicianperformed these procedures. Pleventionbecomes obscured by treatment.

Education.--All studies of use that haveexamined education as an explanatory vari-able have found it to be a statistically sig-nificant predictor of the use of preventiveservices (22,40,61,81). The more education aperson has, the more likely he or she is to usepreventive services. OTA's analysis of theNHIS data set conforms to this observation.There are two possible explanations for theassociation between educatien and use:

Education may affect the decision to usepreventive services by altering patients'

3 This is sometimes called a "cohort effect." The

trend data presented earlier in this paper areconsistent with this hypothesis.

perceptions about disease and potentialservices. It increases their generalknowledge and ability to evaluate healthrisks and the net benefits of prevention.In addition, it may increase their knowl-edge of specific diseases, recommenda-tions for preventive services, andsources of care.Education and prevention are both in-vestments with expected future payoffs(27). Individuals with more educationmay be more oriented toward the futurethan less educated people. Hence, thesepeople tend to seek both education andprevention.

Gender.--The evidence on the effect ofgender on the use of preventive health ser-vices is conflicting. Several studies, includingOTA's analysis, found a strong significant as-sociation between being female and engagingin preventive health measures (31,40,41). Ananalysis of swine flu vaccinations, however,indicates only a weak correlation betweenbeing a woman and use (22). On the basis oftwo services for the elderly (from OTA'sanalysis) and a few other studies of the wholeadult population, one cannot conclude thatelderly women have a consistently greaterpredisposition toward the use of preventiveservices than do their male counterparts.Even if such a conclusion were empiricallyjustified, no explanation for this finding isreadily apparent.

Race.--The relationship between raceand the use of preventive services by theelderly or other adults is ambiguous. OTA'smultivariate logit analysis revealed that elder-ly whites are more likely to receive glaucomascreening than are elderly members of otherracial groups. However, race was not a sig-nificant predictor of any of the other servicesstudied.

Other studies that have looked at the re-lationship between race and the use of pre-ventive services by the non-elderly presentinconsistent results. Four studies found astatistically significant negative association

0 (I


between being black or non-white and usingpreventive services (60,62,66,82), but threeother studies found no significant relationship(15,31,44). In one study, the results variedaccording to the preventive service (91). In areview of studies of participation in fecal oc-cult blood screening, Blalock and colleaguesreport similarly inconsistent results of the ef-fect of ethnicity on the use of this one cancerscreening test and draw no conclusions (14).

Income.--Income is a reflection of theavailability of financial resources to purchasehealth services. Economic theory suggeststhat the consumption of most goods rispl withincome. Preventive health services for theelderly may be particularly sensitive to in-come for two reasons. First, unlike acute ill-ness care and even some types of chronic ill-ness care, preventive care can be put offwithout short-term consequences. Thus, pre-ventive care may receive a lower prioritythan other types of health care or othernecessary consumption. Second, Medicare,the primary health insurer for the elderly,does not cover most preventive services.Hence, to use such services, the elderly musthave private health insurance (discussed be-low), wealth, or income to pay the out-of-pocket expenses.

Almost all multivariate studies of pre-ventive use, including the OTA analysis,found that income has a significant positiveeffect in predicting the use of preventive ser-vices (17,2231,32,40,41,44). Studies examin-ing only the bivariate relationship betweenprevention and income have also consistentlyfound such an association between preventionand income (54,81,91).

Using Michigan survey data, Rundall andWheeler examined the relationship betweenincome and the use of preventive visits ingreater detail. Their analysis indicated thatalthough income has little direct effect onpreventive use, it indirectly increases thelikelihood of use by altering perceptionsabout health and susceptibility tc: illness, andby increasing the probability that individualshave a regular source of care (60).

Insurance and Price.--Except for OTA'sanalysis, which found that insurance coveragebeyond Medicare had a consistently positivesignificant effect on the use of preventiveservices, the effect of insurance has not beenstudied in elderly populations. The publishedliterature on adults' use of preventive servicesgenerally supports the contention that theout-of-pocket price is a significant negativepredictor of use (16,32,47,61,63,86,91).However, no study has examined the rela-tionship between actual cost to the patientand the use of services.

The potential impact of insurancecoverage on use has important policy implica-tions for consideration of Medicare coverageof preventive services. In assessing the bene-fits and costs of such a decision, one wouldwant to know the number of new users ofcovered services as well as the total numberof users. Th e OTA analysis and otherstudies (16,45) suggest that while insurancecoverage does increase use, a substantial per-centage of individuals do not receive recom-mended levels of preventive care, even in thepresence of generous health insurance. Arecent study that compared the use of threepreventive services--blood pressure measure-ment, breast exams, and Pap smears--in Can-ada, where preventive services are covered bynational h-art insurance, and in the UnitedStates foui ... little difference in rates of useby elderly individuals in the two countries(76). Only breast exams were used withstatistically significantly greater frequency inCanada.

Enrollment in Prepaid Plans.--Exceptfor OTA's analysis, which found no evidencethat enrollment in HMOs increases the use ofpreventive services by the elderly, only oneother study has compared preventive care inHMOs with that of traditional insurance plans(66). The researchers in that study foundthat employed adults in a prepaid grouppractice had utilization rates for preventiveservices no different from those of similarindividuals in a Blue Cross plan.

3 u

The Use of Preventive Services by the Elderly i 23

Living Arrangements and Logistical Bar-riers.- -In addition to the financial costs ofpreventive services, these procedures alsoentail time and transportation costs. Onewould expect the use of preventive servicesto decline as the distance between servicesand an individual's home or job increases.Two analyses of adult demand for preventivecare in an HMO found that distance to asource of medical care was negatively (butnonsignificantly) related to the probability ofuse (40,41). Two other multivariate analysesthat included indexes of time, transportation,and perceived difficulties in obtaining pre-ventive procedures also found no significantrelationship between these logistical barriersand the likelihood of using preventive ser-vices (16,22).

A less perfect measure of logistical bar-riers to access that may be especially relevantfor the elderly is whether or not the individ-ual lives alone. Living with another personcould either raise or lower the logistical bar-riers to preventive services. An additionalhousehold member could assist an individualin overcoming immobility or distance; on theother hand, if the additional householdmember is in some way limited in mobility orfunction, the effect on the healthy member's'use of preventive services may be negativesince it may be difficult to leave a dependentpartner to receive preventive care. In OTA'sanalysis of the NHIS data, living alone was astatistically significant, positive predictor ofuse for breast exams only.

Geographical Location.--The communityin which an older person lives may affect hisor her access to prevention. In a multivariateanalysis of the effect of geographic locationon use based on the same data set that OTAused (i.e., the 1982 NHIS), Woolhandler andcolleagues found that among middle-agedwomen, nonrural residence had a negative ef-fect on the likelihood of having a glaucomatest but had no effect on the use of hyper-tension screening, Pap smears, and clinicalbreast examinations (91). These results con-flict with OTA's analysis which found that

people in urban communities use more pre-ventive care than do those in non-urbancommunities. The differences between thetwo studies may be due to several factors:

Woolhandler, et al., estimated a logitmodel with fewer explanatory variablesand a slightly different distinction be-tween urban and rural residence fromthat used by OTA;Woo !handler, et al., used a less sophisti-cated method of estimating variancesfrom the complex NHIS sample designthan OTA did;living in a rural area may be less of abarrier for middle-aged women than forthe elderly population in obtaining pre-ventive services.

Health Status.--The evidence on the ef-fect of health status on preventive health ser-vice use is equivocal. Most multivariateanalyses have found no significant effect ofhealth status on use (19,40,41,44,91).4 Exceptfor OTA's analysis, which found a strongpositive significrnt relationship betweennumber of bed tl..,ability days and use, onlyRundall and Wheeler found that reportingrelatively poor health has a direct positive ef-fect on the likelihood of receiving preventhecare (60). However, variation in measures ofhealth status, model specifications, andsamples make it difficult to draw conclusionsfrom these studies.

A few other researchers have measuredhealth status by the presence of chronic dis-ease. Blalock, et al., reported that having achronic condition increases the likelihood ofreceiving fecal occult blood screening forcolorectal cancer (14). Warnecke, et aL,found a similar association with the probabil-ity of a regular check-up in Illinois adults(86).

4 In their multivariate models of the use of pre-

ventive services by middle-aged women using the

1982 NHIS data, Woolhandler, et al., found that

being healthy was significantly and positively re-

lated only to blood pressure screening.

31


Limitations of the Patient Behavior Analysesof Preventive Service Use

The literature and analysis reviewed inthis section suggest at least two major limita-tions of the patient behavior approach to un-derstanding the use of preventive services:

Existing studies do not account for manyfactors specific to a particular pre-ventive intervelijn such as patients'perceptions of pain, discomfort, embar-rassment, or complexity of administra-tion that may impinge on the willingnessto use of a particular procedure. Inclu-sion of variables that measure these per-ceptions would be consistent with theHealth Belief Model.The models of use examined in this sec-tion assume that patients themselves de-cide whether to receive preventive care.OTA's analysis and most of the pub-lished patient behavior literature do notdirectly examine the role of the primarycare physician and the health care orga-nization in the decision to use services.

Evidence From Studies of HealthCare Provider Behavior

The Physician

Physicians must perform, supervise, orprescribe most preventive services in orderfor a patient to receive them. In fact, manyadults may depend on their primary carephysician to tell them what types of pre-vention they should receive and how often(92). The literature examining the role ofphysicians in determining the use of pre-ventive services includes three types of anal-ysis:

cc ,arisons of physicians' knowledgea) Iut appropriate prevention with pub-lished sets of recommendations;analyses of actual physician perfor-mance; andexperiments to increase physician com-pliance with recommended procedures.

Because almost all of the trials designed tonarrow the gap between published recom-mendations and actual practice focus onchanges in health care organization or man-agement rather than just physician behavior,this paper considers studies that fall into thethird category in the section on health careorganization below. None of the publishedstudies analyzed the elderly as a group sepa-rate from the general patient population.

Physician Knowledge and Actual Prac-tice.--Woo and her colleagues askrei 83physicians in a hospital-based teachinb am-bulatory care practice about the frequencywith which they recommend 16 screeningprocedures to different age groups (92).Across all patient age groups and procedures,physicians with less training recommendedwith greater frequency. Doctors with a his-tory of cancer in their families recommendedmore frequent sigmoidoscopies and mam-mograms. The mean physician recommenda-tion fo preventive use was more frequentthan the mean of published guidelines in 48situations5 and less frequent in 18 situations.The researchers found close agreement amongthe respondents on Pap smears, blood pres-sure checks, physical exams, and medical his-tories, but wide variation in glucose andcholesterol measurement and mdmmography.

Almost half the physicians reported thatthey knew they recommended preventive ser-vices more frequently than publishedguidelines and cited as reasons patient desiresand the belief that the guidelines are insuffi-cient. Woo suggests that despite recom-mendations published by the Canadian TaskForce and others, the physicians in this studymay believe it better to err on the conserva-tive side by recommending services for whichthe supporting medical evidence of effective-ness is inconclusive.

5 A "situation° is a particular screening servicefor a particular age group.

3 2


An analysis of patient records in thesame study found a higher level of physiciancompliance with recommended guidelines forthose services that doctors could order otherhealth personnel to perform, such as labora-tory or radiological tests, than with servicesthat require labor by the primary carephysician, such as sigmoidoscopy and breastexams. Woo and her colleagues infer that be-cause the patients they surveyed desirescreening with appropriate regularity, the gapin performance must be due to doctors failingto offer prevention according to their own orpublished recommendations. They also sug-gest that rates of use are partially related tothe ease with which physicians can provideth^m. Other studies support this hypothesisLat variation in the use of different pre-ventive services is a function of character-istics of the services themselves (e.g., patients'pain, discomfort, and embarrassment) (57,63,65).

Another study compared the preventivecare recommendations of 31 physicians prac-ticing general internal medicine in NorthCarolina with three sets of publishedguidelines (55). The participating doctorslisted procedures they considered essential toa periodic examination for three age groups(30-39; ages 40-49; and ages 50-59). Amongthe procedures recommended in publishedguidelines but not chosen as essential orroutinely recorded in the medical record byat least two-thirds of the sample physicianswere hearing exams, vision exams, fecal oc-cult blood tests, lipid profiles, mammographyfor women over 50, immunizations, and ex-ams for hypothyroidism. The North Carolinaphysicians also recommended services notrecommended in the published guidelines, in-cluding thorough examinations of the majororgan systems, measurement of blood ureanitrogen, white blood counts, chest x-rays,and microscopic urinalysis.

Examining the records of 334 patientsvisiting the 31 North Carolina internists forgeneral examinations, the researchers foundthat, on average, 59 percent of the procedures

recommended by expert groups were found inthe record. Compliance was greater foryounger patients, a result consistent with theestimates of use of several services reviewedearlier in this paper. The researchers alsofound that compliance was greater forlaboratory and physical examination proce-dures than for medical history and counselingservices and was inversely related to thenumber of expert groups recommending eachprocedure.

Noting that the lowest compliance oc-curred with procedures identified by the Ca-nadian Task Force on the Period Health Ex-amination, one of the recommending bodies,as having strong scientific validity, Rommand colleagues suggest that improving com-pliance requires physician education. In dis-cussing ways to improve compliance, they didnot consider the possibility that physiciansmay take into account individual patientcharacteristics and circumstances in decidingnot to provide recommended preventive ser-vices.

In another study, McPhee, et al., foundthat discrepancies exist between AmericanCancer Society (ACS) recommendations forthe use of seven preventive services andphysirian performance and that physicianstend to overestimate their own provision ofthese procedures (51). The researchers reportthat physicians cite four reasons most fre-quently for not providing recommended ser-vices: forgetfulness, lack of time, in-convenience or logistical difficulties, andpatient discomfort or refusal.

One study suggests that physicians maydiffer by specialty in their performance ofsome preventive services. In a study of Papsmear use by physicians in Maryland, Teitel-baum, et al., found that specialists in ob-stetrics and gynecology (OB/GYN) were morelikely than general practitioners and interniststo encourage patients to receive Pap smears,to remind patients by mail or telephone to geta Pap smear, and to achieve compliance withtheir recommendations (69).


Finally, one study currently underwaymay shed additional light on the role andmotivations of the physician in providingpreventive services. Schwartz and Lewis incooperation with the American College ofPhysicians (ACP) recently surveyed ACPmembers about prevontive practices (64). Inaddition to examining the frequency withwhich physicians say they perform a numberof services, Schwartz and Lewis will look forrelationships between preventive practices forpatients and demographic characteristics, thephysicians' experience, and the physicians'preventive practices for themselves.

Health Care Organization

Because physicians wiirk within a largerhealth care system with other practitionersand administrators, it is possible that healthpersonnel and characteristics of the office,hospital, clinic, group practice, or HMOproviding patients' care could affect whetheror not older individuals receive preventiveservices. This section consioers the rela-tionship between the use of preventive proce-dures and the health care organizations thatprovide them.

Health Maintenance Organizations(HMOs) and Other Prepaid Plans.--Some re-searchers have claimed that HMOs, in gener-al, may promote the use of preventive ser-vices (46). To the extent that HMOs stand togain from potential savings in health carecosts resulting from preventive services, theseorganizations would have an economic incen-tive to offer more preventive services. In ad-dition, since visits to HMOs are either free orvery inexpensive, HMO patients may demandmore such visits.

Data from a sit.gle HMO that providedOTA with estimates of preventive service usesuggests that these health care providers mayhave the ability to organize themselves toprovide more preventive services tnan is nowgenerally received by patients. As table 3above indicates, the rates of use in the HMOwere at least as high as or higher than corn-

parable rates of use reported in other studies.Not only has this HMO made preventive carea stated organizational goil, but it has devel-opee management tools to achieve compliancewith some preventive recommendations, in-cluding computer-generated reminders toboth patients and physicians for immuniza-tions. However, these relatively high rates ofuse may be achieved by recruiting patientswho already have characteristics that makethem more likely to use preventive care. Ifthe high levels of use found in the singleHMO for which OTA obtained data are re-lated to its organization, it is not clearwhether less centralized prepaid health planssuch as independent practice associations(IPAs) would be able to do the same.

OTA's multivariate analysis of the 1982NHIS found that the elderly enrolled in anHMO are no more likely to use preventionthan their unenrolled counterparts. Given thesmall number of elderly in HMOs, however,the NHIS sample may not have been largeenough to detect an actual difference. Withrespect to preventive services other thanscreening, Riddiough, et al., reported mixedevidence about the relative likelihood ofHMOs to provide immunizations (54).

Organizational Factors Related to Use.--Other characteristics of health providers mayalso affect the use of preventive procedures.For example, the use of non-physician per-sonnel, cues to compliance such as remindersand media, or health fairs are all organiza-tional strategies that have been employed toincrease the use of preventive services.

One potential mechanism for increasingthe use of preventive services is the healthscreening fair in which participants canreceive selected procedures at a publicizedtime and place. Irk an analysis of the cost-effectiveness of this screening method, Ber-wick concluded that fairs work best when thetarget population is clearly defined, thescreening tests are appro 'ately chosen, reli-able and accurate, and the fair provides ap-propriate guidelines for abnormal results,

The Use of Preventive Services by the Elderly s 27

follow-up, referral, and treatment (13). Al-though fairs may increase the ability to detectand prevent illness, significant risks may existif screening in the fair setting ig relativelyinsensitive or unspecific or does not provideadequate follow-up.

Among interventions designed to increasepatient compliance with recommended ser-vices, Thompson, et al., studied the value ofcombining talks by physicians and nurses,postcard reminders to patients, and phonecalls to comply with a fecal occult blood test(70). While those receiving any one of theseinterventions had an average compliance rateof 89 percent versus 68 percent among thecontrol group, the reminder postcard was es-pecially cost-effective, raising complianceabout 25 percent to an overall rate of 93 per-cent at a relatively small cost. The talk byhealth personnel, which was somewhat morelabor intensive, increased compliance about13 percent.

McDonald and colleagues found similarresults in another randomized trial. Internsand resident physicians who received com-puter-generated reminders provided 49 per-cent of the preventive services suggested tothem, while physicians in the control groupprovided oniy 29 percent of the services (49).Among physicians who received the reminderintervention., the researchers found that over-all attitudes tow4ra the reminder system andwhether the physician read and signed thereminder were statisf(cally significant predic-tors of use, while yeers of training and facul-ty assessments of the physicir is were not.The researchers conclude that noncompliancewith recommendations is an error of omissionthat can be mitigated by technological aids.

In another study, Satarino and colleaguesretrospectively asked patients receiving freebreast cancer screening at two clinics in NewYork City how they learned of the serviceand their need to be screened (62). Mostblack screenees with less than a high schooleducation learnea of the clinic through tele-vision ads followed by word-of-mouth and

private physician referrals. While this papersuggegts that television may be a useful cuein promoting use, it does not indicate howone might reach individuals who were notscreened at one of the two clinics.

Two studies have tested strategies to in-duce the provision of preventive services topatients. In one study, four clinics were ran-domly assigned either to participate in a pro-gram that combined physician education witha checklist of services due each patient on themedical record or to a control group. Over a4-month period, the researchers measuredrates of mammography and influenza andpneumococcal immunizations among eligiblepatients in each group. The interventiongroup had significant increases in the use ofthese services, ranging from 2 to 40 percentover the control group. The researchers alsofound significant increases in tests ofphysician knowledge and attitudes about pre-vention among physic:ans (20).

r; 'mother randomized t.ial, nurses whoalreac routinely reviewed patient charts in auniversity-based internal medicine practicereminded physicians when a patient was duefor particular screening services and im-munizations (24). For uatients receiving thenurse-reminder intervention, the researchersfound statistically significant increases inrates of use for fecal occult blood screening(32 to 47 percent), breast exams (29 to 46percent) and influenza immunizations (18 to40 percent), but not for Pap smears (13 to 14percent). The study represents the only ran-domized experiment to analyze the role ofnon-physician personnel in providing orboosting the use of preventive services.

Physicians believe that such organiza-tional strategies would improve the use ofpreventive services. In two studies ofphysician attitudes, researchers conclude thatwhile most physicians see themselves as in-effective in improving patient compliancewith their recommendations for preventivecare, they believe that they could be muchmore effective if they had more resources at

35


thei, disposal including better training forthemselves and their support staff, improvedreanbursemeut for preventive services, andbetter educational materials for patients(84,87).

Other research currently underway mayenhance owe understanding of the impact oforganization on use. The Health Services Re-search Center at the University of NorthCArolina, with funds from the National Cen-te for Health Services Research and HealthCare Technology Assessment (NCHSR), hasrecently surveyed administrators, medicaldirectors, and staff physicians in 150 laisemedical practices of different types (25).While most of the study seeks to identify or-ganizational characteristics that contribute tophysician satisfaction with his or her workenvironment, it will also focus on the partici-pating organizations' preventive care prac-tices.

Limitations of Evidence About ProviderBehavior

Compared to studies of pat:e.p:. behavior,there are relatively few studies of the struc-;ire of the health care system or the role ofphysicians in the use of preventive services(54), and none focuses exclusively on pre-vention for an elderly population. A. newstudy of preventive practices among physi-cians currently underway may shed morelight on these issues (64). However, no cur-rent or completed studies to date have exam-ined the role of other potentially importantfactors, including:

potential revenue obtainable from pre-ventive services,potential liability associated with offer-ing or withholding preventive services,andthe degree of management control withinthe health care organization.

The evidence about whether or notHMOs provide greater levels of prevention isambiguous. While HMO enrollees may re-ceive more preventive care than enrollees ofother health plans, other predisposing andenabling factors such as gender, education,and income may explain this differential.Controlling for these factors, OTA's analysisof the NHIS data revealed no effect of HMOmembership on the probability of using fivepreventive services. The data gathered byOTA from one closed-panel HMO suggestthat prepaid health plans may have thepotential to increase the use of preventiveservices among older adults. No data existabout whether other prepaid plans achieve alevel of preventive care comparable to theone OTA examined.

The literature contains several studies ofinterventions within clinics, ambulatory carepractices, and HMOs designed to promote theuse of preventive services. They suggest thatprovider-based strategies can increase the useof preventive strategies. However, the nar-rowness of these studies indicates the needfor more research. In particular, the existingliterature does not adequately address:

the generalizability of particular inter-ventions to other settings,the most effective means of informingthe public about the need for andavability of screening programs,tf. role of non-physician personnel inaffecting patient use of preventive ser-vices, andthe potential of technological advances(e.g., the introduction of computerizedmedical records and new screening tech-nology for the physicians' offices) in af-fecting patient use of preventive ser-vices.

6. IMPLICATIONS FOR POLICY

Potential Medicare Coverage

Medicare represents the Federal Govern-ment's major financial and policy commit-ment to health care. In 1986, Medicarerepresented 58 cents of each Federal dollarspent on health care and was the source ofpayment for 29 percent of all expendituresfor hospital care and 21 percent of expendi-tures for physician services (73). The use andcorrelates of use of preventive services forthe elderly have several important implica-tions for the Medicare program.

First, covering preventive services underthe Medicare program would probably bringabout increases in the percentage of elderlyreceiving preventive care. However, currentevidence suggests that reducing out-of-pocket expenses for patients is not sufficientto assure compliance with published pre-ventive recommendations. OTA's analysis offive preventive services suggested thepresence of insurance beyond Medicare is as-sociated with about a 10-percent increase inthe percentage of elderly receiving each ser-vice during the period of time examined.With the exception of blood pressuremeasurement, which almost all elderly alreadyreceive on a routine basis, substantial portionsof the elderly with additional coverage thatdefrays out-of-pocket expenses do not useeach of the preventive services OTA exam-ined. In addition, because the additional in-surance coverage held by Medicare recipientsin most instances excludes preventive ser-vices, the OTA analysis is not a direct test ofthe impact of coverage of specific procedureson the rates of use of t;.ese services.

Other factors enter into the physician'sdecision about whether to offer or providethe service and the patient's decision aboutwhether to seek or use it. Some of these fac-tors may be amenable to change through pub-lic policy, while other characteristics describegroups of elderly patients at relatively high orlow risk of not receiving adequate preventivecare.

Second, coverage of preventive servkesfor Medicare beneficiaries could affect pre-ventive use beyond the Medicare populationitself. Medicare payment may raise interestin preventive care among health care pro-viders and payers by placing the authority ofthe Federal Government behind it. In addi-tion, consumers of health care may put moreweight on preventive services in managingtheir own health because of the public dis-cussion and attention focused on Medicarecoverage. None of the data or literature cur-rently available allows OTA to estimate theexistence or magnitude of this potential in-direct effect. A recent analysis of preventiveservices in Canada where such procedures arepaid for by the government revealed rates ofuse comparable to those in the United States,suggesting that both the direct an i indirecteffects of government coverage ma l. be small(76). The Medicare Catastrophic CoverageAct of 1988 (Public Law 100-360) may offeran opportunity to assess the full impact ofMedicare coverage of breast cancer screeningfor the elderly by analyzing trends if the useof this service is monitored.

Third, expansion of Medicare to coverpreventive procedures will represent an im-mediate boost in the program's financial ob-ligations even if increases in use are minimalor nonexistent. Although gaps exist betweenexperts' recommendations and current levelsof use, substantial numbers of elderly stillreceive a variety of preventive services atrecommended frequencies. OTA's analysisindicated that at least one-half of the non-institutionalized elderly receives each of thefive services examined on a regular basis.Fcr three of these services (glaucoma screen-ing, eye examinations, and blood pressuremeasurement), rates of use were even higher.Estimates of costs attributable to expandedMedicare coverage of prevemtive servicesmust take account of the program's obligationto pay for procedures whose costs are cur-rently borne by other payers. However, someportion of these services are for patients witha related medical history or symptoms and are

,3 7

29


"diagnostic" and already reimbursable underMedicare. It is also probable that somephysicians may categorize some examinationsand screening services as "diagnostic" so thatthe procedure will be covered by Medicare.No existing data from published literature orthe records of the Health Care Financing Ad-ministration (HCFA), the Federal agency thatadministers Medicare, indicate what portionof all procedures reimbursed by Medicare areactually for screening purposes.

Fourth, the conclusions of this paperabout the use of medical services such asscreening and immunizations may havelimited applicability for policymakers consid-ering Medicare coverage of other preventiveservices such as health risk appraisals, edu-cation, counseling services, or tertiary pre-vention of disability among elderly sufferingfrom chronic disease. Many of the serviceslisted in table I could rely on non-physicianpersonnel to a greater degree than do screen-ing and immunization, and patients couldreceive them in a wider array of settings thanthey receive most medical services. Thesecharacteristics suggest that use of preventiveservices not examined in this paper may bemarkedly different from those explored here.Hence, an understanding of the implicationsof Medicare coverage of services other thanscreening and immunizations use would re-quire additional study.

Delivery of Preventive Servicesfor the Elderly

Although one of the major focuses ofthis paper is the potential impact of insurancecoverage on the use of preventive services bythe elderly, the data and literature reviewedin this paper suggest that other factors arestrongly related to use. This informationwould be useful to public policymaktm whoseek strategies for altering the elderly's use ofpreventive services. Of the patient and pro-vider characteristics related to use, a few areimmutable, some are amenable to changethrough policy, and others are theoreticallyamenable to change, although the policy in-

terventions to accomplish these changes areunlikely to prove cost-effective.

Public policy cannot affect age, gender,rural versus urban residence, and usually,days spent in bed during the previous year,three factors correlated with the use of pre-ventive services. However, these demo-graphic and health status characteristics doidentify segments of the po;Julation particu-larly at risk of not receiving adequate screen-ing or immunizations. Knowing that onaverage more women than men receive suchcare or that recommended prevention appearsto decline with age and good health may helppolicymakers target some of their preventivecare efforts toward the more vulnerablegroups. Because Medicare is an entitlementprogram available to all persons over 65 whoreceive Social Security, it is an unlikelymeans of focusing efforts on demographicallydefined subsets of the elderly population.Nevertheless, other government investmentsin prevention such as mass media campaignsand screening fairs may be able to narrowtheir target. Policymakers who want to bol-ster use among elderly groups unlikely toreceive preventive services may wish to studythe potential costs and effectiveness of suchprograms in detail.

Other factors related to the elderly's useof preventive care do seem amenable te,policy interventions. Studies that examineinfluences on physician behavior suggest thatbetter or more frequent physician educationmay bring about better compliance with pre-ventive recommendations. In addition, evi-dence suggests that record-keeping systemsand reminders to physicians (possibly aidedby computer technology) have positive effectson use. This paper has discussed the poten-tial for insurance coverage to increase some-what the percentage of elderly receiving pre-vention. Government and providers coulddesign policies to bring about these changeswhere they do not already exist.

The relationship between use and pro-vider characteristics is not clear. For exam-


ple, conflicting published literature and thesmall number of elderly enrolled in HMOsmakes it difficult to determine if the enroll-ment of older adults in prepaid health plansincreases the amount of preventive care theyreceive. Additional research is also needed toestablish the potential of nurses and othernon-physician health professionals in provid-ing or promoting the appropriate use of pre-ventive services. Previous research suggeststhe substantial contact they have with patientsin an ambulatory care setting, the availabilityof new screening technologies (e.g., instru-ments that can measure cholesterol in aphysician's office from a finger prick), andthe growth in "health fairs" that provide somepreventive services in alternative settings mayenhance the role of these professionals.

The remaining variables affecting olderindividuals' use of screening and immun-izations- -educational level and familyincome--are also potentially susceptible togovernment interventions. However, publicpolicies designed to change these character-istics are so much more broadly construedthat they would never be implemented simplyto affect the use of preventive care. If, how-ever, the government decides for some otherreason to promote education or supplementincome among the elderly, long-term in-creases in the use of preventive proceduresmay be an aaditional benefit.

Appendix A: ADVISORY PANEL

Gordon De Friese, Panel ChairHealth Services Research Center

Urliv.33ity of North Carolina, Chapel Hill, NC

Marianne C. FahsDepartment of Health EconomicsMt. Sinai Medical CenterNew York, NY

John FrankDepartment of Preventive Medicine &Biostatistics

University of TorontoOntario, Canada

Gary D. FriedmanEpidemiology and Biostatistics DivisionPermanente Medical Group, Inc.Oakland, CA

Lawrence GottliebClinical Guidelines ProgramHarvard Community Health PlanBrookline Village, MA

Mary KnappJohn Whitman and AssociatesPhiladelphia, PA

Risa Lavizzo-MoureyGeriatrics ProgramUniversity of PennsylvaniaPhiladelphia, PA

M. Cristina LeskeDepartment of Preventive MedicineSUNY at Stony BrookStony Brook, NY

Donald LogsdonINSURE ProjectNew York, NY

Mildred B. McCauleyAmerican Association ofRetired Persons

Washington, DC

Peter McMenaminChevy Chase, MD

Meredith MinklerCenter on AgingUniversity of California, BerkeleyBerkeley, CA

Marilyn MoonPublic Policy IrstituteAmerican Associationof Retired Persons

Washington, DC

George MorleyDepartment of Obstetrics/GynecologyUniversity of Michigan Medical CenterAnn Arbor, MI

Gilbert OmennDean, School of Public Health &Community Medicine

University of WashingtonSeattle, WA

George PickettDepartment of Public Health PolicySchool of Public HealthUniversity of MichiganAnn Arbor, MI

Donald ShepardDepartment of Health Policyand Management

Harvard School of Public HealthBoston, MA

Barry StultsDivision of General Internal MedicineUniversity of Utah Medical CenterSalt Lake City, UT

Advisory Panel members provide valuable guidance during the preparation of OTA reports.However, the presence of an individual on the Advisory Panel does not mean that individualagrees with or endorses the conclusions of this particular paper.

33

4 t)

Appendix B: ACKNOWLEDGMENTS

The OTA staff would like to express our appreciation of the following people for theirvaluable guidance. This acknowledgment should not be construed in any way to imply that theindividuals agree with or endorse the conclusions of this paper.

Ronald M. AndersenThe University of ChicagoChicago, IL

Robert BurackWayne State UniversityDetroit, MI

Morris F. Col lenKaiser-Permanante Medical Care ProgramOakland, CA

Alan GarberNational Bureau of Economic Research, Inc.Stanford, CA

Pearl S. GermanThe Johns Hopkins UniversityBaltimore, MD

Robert L. KaneUniversity of MinnesotaMinneapolis, MN

Mary Grace KovarNational Center for Health StatisticsWashington, DC

David R. LairsonHealth Science CenterHouston, TX

A ngela M ickalideU.S. Preventive Services Task ForceWashington, DC

Steven MooreOffice of the Surgeon GeneralRockville, MD

34

Charlotte MullerMt. Sinai Medical CenterNew York, NY

Annlia Pajanini-HillUniversity of Southern CaliforniaLos Angeles, CA

Penelope PollardNational Health Policy ForumThe George Washington UniversityWashington, DC

John A. SawyerUniversity of TexasHealth Science CenterSan Antonio, TX

F. Douglas ScutchfieldAmerican College of Preventive

MedicineWashington, DC

Stephen J. WilliamsSan Diego State UniversitySan Diego, CA

Steven H. WoolfU.S. Preventive Services Task ForceWashington, DC

Steffie WoolhandlerThe Cambridge HospitalCambridge, MA

Appendix C: EMPIRICAL STUDIES OF THE USE AND/ORDETERMINANTS OF USE OF PREVENTIVE SERVICESBY THE ELDERLY

35

Study

Services

examinedPeriod of time in

which use is measured

Sample descriptionand design Analyses

American CancerSociety/Gatlup,1937a

Fecal occult bloodProctosigmoidoscopyNect.:'. exam

MammographyBreast examPap smear

Measuresfrequency ofregular use

Representative national random sample Percent having knowledge of and using cancer

of 1549 adults over 18 years old screening tests by age; trends from previous

ACS surveys

Brown and Hulka,1988d

Mammography Ever 109 cases of women over 60 withmetastatic breast cancer from asingle hospital; 211 control womenwith similar characteristics withcancers of other sites drawn fromtumor registry

Chao et al.,

1987c

Blood pressureFecal occult bloodPap smearMammographyBreast self-exam

Fedson, 1987d Influenza vaccinePnemococca I vaccine

Last use: lastyear, beforelast year,

never; breastself-exam on amonthly basis

Case-control study of whether elderly womenwith metastatic breast cancer were screenedless than women with other types of cancer

11,888 residents of a southernCalifornia retirement community;mailed survey

Use of screening tests by sex, age, medicalcondition and health habits

1-year periodslifetime

Data from national probabilitysample: the U.S. Immunization Surveyconducted annually as part of Bureauof Census Household Survey from late

1960s to 1985

Estimates of annual use of influenza vaccineamong elderly, and prevalence estimates forpneumococcal vaccinations; additionalsources of evidence combined to discussstrategies to increase physicianimmunization of high risk patients

Lazaro, Logsdon,and Meier, 1987e

General preventiveexam; health risk

counseling

Not applicable 713 individuals over age 60 invitedto receive preventive services fromparticipating physician at no cost;

drawn from geographically separateareas; sample represents a 69 percentresponse rate to baseline survey ofcharacteristics and attitudes

Bivariate and multivariate analyses of useduring experimental period as a function ofdemographic and psychosocialcharacteristics, health status, and

attitudinal indices

National Center Blood presst.

for Health Breast examinationStatistics, 1988f Pap smear

Most recent use U.S. data are from 1985 NationalHealth Interview Survey of HealthPromotion and Disease Prevention;Canadian data are from 1985 CanadianHealth Promotion Survey carried out

by Statistics Canada for Health andWelfare Canada

Comparison of national rates of use ofpreventive services in Canada and U.S. (inaddition to preventive services, surveysalso ask about other patient preventivehealth practices)

OTA, 1988g Glaucoma screeningEye examBlood pressurePap smearBreast exam

<3 years

<3 years<2 years

<4 years<2 years

All individuals over 65 from 1982

National Health Interview Survey(11,434 people); two-stage national

random sample

Percent receiving services; relationshipbetween use and predisposing, enabling, andhealth status variables (logit and bivariatemodels)

StudyServicesexamined

Period of time inwhich use is measured


OTA, 1988n CholesterolFecal occult bloodPap smearEye examMammographyPhysical examPneumococcal vaccineInfluenza vaccineTetanus vaccine

5 years1 year

3 years2 years1 year1 year

Lifetime1 year

10 years

Enrollees over 40 in o closed panel Descriptive analyses of percentbges

urban HMO in the northeast; data from receiving services by oge and sexHMO computer system

Radecki, et al.,19881

34 categories ofdiagnostic testsincluding specificscreening procedures

Individual

visits

28,265 visits to internists, family,and general practitioners

Biveriate and multivariate analyses of theeffect of specialty and patient age on ratewith which physicians order proceduresincluding routine screening tests; rotes ofuse compared with efficacy of procedures forolder patients

Rundall andWheeler, 1979J

Swine flu vaccine Lifetime (thoughonly offered fora limitedperiod)

232 responses to questionnaire mailedto random sample of 500 seniorcitizen residents of Tomkins County,New York

Bivariate and multivariate (logit) analysesrelationships between vaccine use end healthbelief (attitudinal) variables

Winawer, et al., Fecal occult blood Not applicable1983k Sigmoidoscopy

Memorial Sloan Kettering CancerCenter-Strang Clinic trial of 21,961patients offered fecal occult blood

test followed by sigmoidoscopy

Compliance with screening recommendationsend rotes of detected cancer as a functionof age; use data gained as part of largertest for effectiveness of screening inreducing mortality

Abbreviations: ACS = American Cancer Society; HMO = health maintenance organization.eGallup Organization, "The 1987 Survey of Public Awareness and Use of Cancer Detection Tests: Summary of Findings," cordUcted for the AmericanCancer Society (Princeton, NJ: Gallup Organization, January 1988).

bJ.T. Brown and B.S. Hulka, "Screening Mammography in the Elderly: A Case-Control Study," J. Gen. Intern, Medicine 3:126-131, 1988.cA. Chao, A. Paganini-Hill, R.K. Ross, et al., "Use of Preventive Care by the Elderly," Preventive Medicine 16:710-722, 1987.dD.S. Fedson, "Influenza and Pneumococcal Immunization Strategies for Physicians," Chest 91:436-443, 1987.eC.M. Lazaro, D.N. Logsdon, and R. Meier, "Utilization of Preventive Health Services by the Elderly," Insure Project, Lifecycle Preventive HealthServices, New York, NY, presentation to the American Psychological Association Convention, Aug. 31, 1987.

fNational Center for Health Statistics, U.S. Department of Health and Human Services, "Adult Health Practices in the United States and Canada,"Vital and Health Statistics, Series 5, No. 3 (Washington, DC: U.S. Government Printing Office, May 1988).gOffice of Technology Assessment, primary analysis of data from the National Health Interview Survey, presented later in this paper, not study

published elsewhere..10ffice of Technology Assessment, primary analysis of data from an HMO presented later in this paper, not s study published elsewhere.

1S.E. Radecki, R.L. Kane, D.H. Solomon, et al., "Are Physicians Sensitive to the Special Problems of Older Patients?" Journal of the AmericanGeriatric Society 36:719-725, 1988.JT.G. Rurdall and J.R.C. Wheeler, "Factors Associated With Utilization of the Swine Flu Vaccination Program Among Senior Citizens in TompkinsCounty," Medical Care 17:191-200, 1979; and T.G. Rundall and J.R.C. Wheeler, "The Effect of Income on Use of Preventive Care: An Evaluation of

Alternative Explanations," Journal of Health and Social Behavior 20:397-406, 1979.kSJ. Winawer, M. Baldwin, E. Herbert, et al., "Screening Experience With Fecal Occult Blood Testing as Function of Age," in Prospectives on

Prevention and Treatment of Cancer in the Elderly, R. Yancik (ed.) (New York, NY: Raven Press, 1983).


4

Appendix D: EMPIRICAL STUDIES OF THE USE AND/ORDETERMINANTS OF USE OF PREVENTIVESERVICES BY THE NON-ELDERLY




American College ofPhysicians/Schwartzand Lewis,

forthcoming1

Patient historyChest x-rayEKG

CholesterolSigmoidoscopyFecal occult bloodBreast examPap smearHepatitis B vaccineInfluenza vaccinePneumococcal vaccineRisk counseling

Measures date oflast use

Survey mailed to 2000 ACPmembers asking about physician'spreventive behavior for self andfor patients

Relatialships between self-reportedphysician preventive care for petientsand practice characteristics, patientrisk factors, demographics, medicalknowledge, medical opinion andpreventive care for self

Adams and Kerner,19822

Breast cancer screening(mammography and exam)

Not applicable Retrospective survey of 2300screenees in free NYC clinic

Demographic characteristics of usersand source of knowledge aboutavailability of service, recommendeduse

Blalock. DeVellis, and Fecal occult bloodSandler, 19873

Not applicable Not applicable Review article of six studies ofcompliance with FOB screening offer;focus on relationships betweenparticipation and age, otherdemographics, physical health status,family history, other preventivehealth practices, and health beliefs

Burack and Liang,19874

Pap smearMammographyFecal occult bloodSiymoidosccpYBreast examPelvic examRectal exam

Not applicable 221 users of inner-city teachinghospital ambulatory care clinic

RaLes uf acceptance and compliancewith offer of cervi;a; relationshiptetween use and dearraphic, ennbling.and health belief variables (bivarieceand multivariate)

Calnan, 19855 Breast screeningCervical screeningDental checkupDietary practiceExerciseSmoking behaviorSafety belt use

Study measures",regular" use ofall preventiveitems exceptdietary practiceand smokingbehavior

Interview surveys with 2084

women 40 to 64 years drawnrandomly from general

practitimer registers in threeEnglish cities

Examined regular use of sevenindivickal and multiple preventivebehaviors; the relationships betweenuse and sociodemographiccharacteristics, health status,

attitudes, and social support/networks

Cohen, Littenberg, etal.. 19826

Periodic health exam(defined by American CancerSociety and Canadian TaskForce recommendations)

Recent past Experiment to increasepreventive services given byphysicians to patients from

outpatient clinic of urbanteachirw hospital

Compared physician preventiveknowledge, attitudes and compliancewith preventive health recaneendationsbetween experimental and controlgroups; experimental group received

checklists and education

4 G

Appendix D.--cont'd

Study

Services

examinedPeriod of time in

which use is measuredSample description

and design Analyses

Cummirnis et al.,

19797

Swine flu vaccine Lifetime (albeitonly offeredwithin finite

period)

Telephone survey of 286 sodultsrandomly drawn from allhouseholds in Oakland County,Michigan

David and Boldt, 19844 None; examined patientattituric: toward preventive

care

Davidson et al.,19849

Harris and Guten,197910

Pap smearFecal occutt bloodBreast examInfluenza ierunization

Not applicable

Uptake of swine flu vaccine as a functionof Health Belief Model vamiables usingeegression and path analyses

92 responses to moiled survey of Descriptive statistics of patient

randan 10% of active patients of attitudes toward prevention, physician

University of Kentucky Medical preventive function, time and cost of

Center preventive care; hive:Hate relationshiFsbetween attitudes and social pcmition

3 years1 year

1 year

1 )aar (for 65+)(or 1 year forall services:not clear)

Patients receiving outpatient Effect of nuise-initiated reminder systemcare through university based of preventive rirvices due; outcome

general internal medicine measure is the difference in rate ofpractice during two 1-year compliance with reccminendations between

perioe0; randomized trial; n=450 experimental and control periods

Health protectiveactivities includingregular medical and dentalcheckups

Not applicable Interview with 842 randomlyselected adults from Cleveland,Ohio area

Re.ationships between constructed scalesof health protective behavior and HealthBelief Model variables (bivariate)

Hayward et al.,1981311

Pap smear

Breast examinationMammographyPeriodic health exam

Howard, 197912

Larison and Swint,1976, 197813

Mamnography

1 year and 3-5

years1 year

1 year

1 year

4659 women from random national Estimates of proportions of women

telephone :..4rvey (1986 Access to receiving cancer screening within

Care Study) recommended periods of time; bivariateand multivariate analysis of correlatesof use

Date c: last L.Ie Not applicable

Preventive and non-preventive physician visits

1 year

Reviews four studies on use ofmammgraphy and three studies of majordeterrents to physician uue ofmaarrography

5% random sample of 3892individuals enrolled in largeprepaid plan (Kaiser Portland)

Compared likelihood and volume of use ofpreventive and nonpreventive vis:ts in an

HMO as a function of health status,demographic, insurance, and other

economic variables

McDonald et al.,198415

variety of screeningservices, imeunizations,and weight reduction

Not applicable 115 resident physicians in ateaching Impital generalmedicine practice (61 study and54 control, randomly assigned)studied over a 2-year period

Estimated the effectiveness of computer-generateo reminder system in achievingcompliance with medical indications forpreventive and other procedures;analyzed the correlates of physician useand reasons for nonuse

Appendix D.--cont'd

stodyServicesexamined



McPhee, et al.,198615

Fecal occult bloodRectal examSigmoidoscopyPap smearPelvic examBreast examMammography

From 1980 ACSrecommendations:

1 year for allservices exceptsigmoidoscopy andPap smears whichare 3 years

Stratified random sample of 525adults from 52 providers in auniversity general internal

medicine practice

Comparison of rates of uSe (based on ACSrecommendations) with actual physicianperformance and physicians' perceptions

of thoil pas fcmance; also examinescorrelates with use and of each service

Physician EmloymentProject/ DeFriese,Madison, Konrad, etal., ongoing16

General preventive careGeneral immunizationsTobacco risk counselingColorectal cancer screeningBreast cancer screening

Not applicable Cross-sectional telephone surveyof members of 119 large medicalgroup practices followed bymailed survey of 6000 physiciansfrom these practices

Descriptive statistics of organizational

policies toward selected preventiveservices, aggregate preventive physicianvisits, and physician attitudes;bivariate and multivariate analyses to

test effect of individual,organizational, and environmental factors

on these same outccae measures

MMWR, 1988a17 Pap smear

Breast exam

Measures date oflast use

Int-rviews with 603 adult womenrandomly selected in four-stageprocess from a 36 county area ofrural Kentucky

Knowledge and use of Pap smears andbreast exams by age and other contact

with the health care system

KMWR, 1988b18

MMWR, 1988c19

MammographyBreast examPhysician visit

1 year 852 Rhode Island women over 40 Use of three preventive services by age,

randomly selected from telephone income, and education

numbers

Pap smearBreast examMammographyDigital rectal examFecal occult bloodProctoscopy

Date of last use: 1987 National Health Interview

<1 year or >=1 Survey Supplement on Cancer

year Control (1st quarter data:

January-March)

Estimates of proportions of non-institutionalized adults over 40 yearsold with knowledge of and using eachservice broken down by gender and race

Rand, 1986 and 198720 Tetanus vaccinePneumococcal vaccineInfluenza vaccinePap smear

MammogramFecal occult bloodSignoidoscopyChest x-ray

3 years for ail

services

2276 enrotlees in Rand healthinsurance experiment between 17and 65 years; (separate analysisfor children's use of preventivecare)

Use of preventive services as a functionof health insurance experiment cost.sharing plans; estimated efficacy ofpreventive care in terms of healthstatus and costs; estimated cost ofbringing use up to recommended levels

Riddiough et al.,198121

Diphtheria vaccineTetanus vaccinePolio vaccineSmallpox vaccineAsian flu vaccineInfluenza B vaccineSwine flu vaccine

Not applicable 2 national random surveys ofOpinion Research Company in 1977and 1978; 2006 and 2084 adultsin respective samples; alsoreviews some existing

literataure

Baseline public data on attitudinal

health belief model variables(likelihood and seriousness of disease,

safey and effectiveness of vaccines,social factors, demographics, cost)

Appendix 0.--contod

Study

Servicesexamined



Romm, Fletcher, andHulks, 198122

Risk counselingMedical historyPap smearCholesterolBlood glucoseEKGUrinalysisFecal occult bloodLipid profileTetanus vaccineDiphtheria vaccine

Used three sets

of sets of timeperiods eachbroken down byage based onpublished expertrecomnendations

Sample of 31 North Carolina Comparison of expert recomnendations,

physicians and the medical surveyed physician opinion and actual

records of 334 of their patients physician use of preventive service,counseling, and medical history taking

Rundall and Wheeler,19791oR4

Preventive visits Within a usualyear

Cluster sample of 781 completed

household interview in WashtenawCounty, Michigan

Path analysis of impact of income onvolume of preventive visits; tests threehypotheses of income effect: direct

(financial constraints), indirect viabeliefs (culture of poverty), indirect viausual source of care (system barriers)

Salkever, 197624 General physical exam 1 year Household samples of 4296 edultsfrom Baltimore, Maryland,

northern Vermont, andSaskatchewan, Canada collected aspart of World Health OrganizationInteroational Collaborative Study

of Medical Care Utilization

Using multivariate regression analysis,examined the effects of economic variablesincluding income, insurance coverage, andtime costs on the likelihood adultsreceive a preventive examination

Satarino, Schwartz,and Swanson, 19822D

Cervical cancer (Pap smearand pelvic exam)

Not applicable 34,135 women screened in MetroDetroit Cancer Control Program

over 2.5-year period

Retrospective comparison of proportion ofscreenees seen in outreach clinics,standing clinics, and employee programs;estimated relative effectiveness of eachclinic type in terms of rate of detectedcancer by age and race

Sawyer et al., 1988,

unpublishedmanuscripte6

Pap smear 3 years Interview survey of 149 black

women 16-75 years old from ruralNorth Carolina solicited door-to-door, referrals from socialworkers, other participants

Use of Pap smears as a function of healthbeliefs, provider, cues, income,education, and health insurance (bivariate

and multivariate)

slesinger, Tessler,

and mechanic 197621

General checkupre skin testBlood pressureUrinalysisComplete blood countPap smearSickle cell (blacks only)

1 year 408 Blue Cross end 506 prepaidparticipant famil!es drawn fromtwo large indust .al firms in

Midwestern city

Use and volume of preventive servics as afunction of demographics, insurance type,

and attitudes

A r,Li .

Appendix D.--cont,d




Teitelbaum et al., Pap smears198828

Asks whetherpatients receivedsmear in each of10 previous years

Thompson et al.,

19866Fecal occult blood Not applicable

(experiment)

Telephone surveys of 1200 womenover age 45 drawn randomly and400 physicians representingthree specialties (generalmedicine, OBGYN, and internists)from the State of Maryland

Physicians surveyed about theirrecommendations for frequency of Papsmear use; broken down by age of patientand specialty of physician; women

surveyed about their own use of Papsmears in previous 10 years; broken downby age and specialty of regular source ofcare

Factorial design random trial ofdifferent cues to use FOB basedon Health Belief Model; 500members of HMO over 45 withexisting physical examappointment

Estime.ed effectiveness and cost-effectiveness of cues: physician talk,nurse talk, phone reminder, postcards,and combinations

Valente et al.,198630

Variety of health behaviors Not applicable Data free' mailed survey of 1040randomly sampled Marylandphysicians

Estimates of physician beliefs andproportions of physicians reporting thatthey gather information from patients on11 health behaviors broken down by careerage and specialty

Warnecke, Havlicek,Manfredi, 198331

Physical examProstate examPap smearBectal/procto examBreast self-exam

within the 3.5-year study:

annual, <annual,never

968 low-income adults over 35drawn from stratified randomsample of illinois phone numbers

Multivariate model of determinants of useas a function of health belief, economic,demographic, and health knowledge factors

Wechsler et el.,198332

Variety of health behaviors Hot applicable

Williamson et al., Screening health exam198833

2 years

Data from survey of 490 randomlysampled Massachusetts of general

medicine, internal medicine, andfamily practice physicians

1788 adults frail 47 family

physicians, offices within asingle State

Estimates of physician attitudes andbeliefs and proportions reporting thatthey gather information ond counselpeients on health behaviors

'Awes of proportions of patientsreceiving exam within previous 2 years;proportion* of patients who say theywould accept physicien recommendations toreceive ach of 10 preventive services ifoffered (analyzed by age); patientknowledge of recommended screening andhealth practices

Windsor and morris,

198434

Hypothetical ccemmityhealth promotion programincluding diet control andsmoking cessation

Not applicable Survey of 400 randomly selectedresidents of urban Mobile,

Alabama, age 20-69

Willingness to pay estimates by age forhypothetical program, projected usederived from self-reported risk factors,willingness to pay, and assumptions

Appendix D.--cait'd

Study

Servicesexamined



Woo et al., 1985-12 History/physical examBlood pressureCholesterolFecal occult bloodPap smearMammogramGlaucomaSigmoidoscopyEKG

Serum glucoseRectal examBreast examUrinalysisPelvic examChest x-ray

Rather than use a

set period oftime, authorscalculate meansof published andphysicianrecommendations,generate expectednumber patientsreceiving eachservice.for eachphysician, andcompare actualnumbers toexpected numbers

83 physicians and 188 patientsfraw an urban teaching hospitalambulatory care practice; actual

use fran patient records

Compares published recommendations,

expected physician use, actual physicianuse, and patient desires for 16

preventive Services

Abbreviations: ACP = American College of Physicians; ACS = American Cancer Society; EKG = electrocardiogram; FOB = fecal occult blood;

HMO a health maintenance organization; MUM = Morbidity and Mortality Weekly Report; OBGYN = obstetrician/gynecologist.

1S. Schwartz, University of Pennsylvania School of Medicine, personal communication, Philadelphia PA, Mar. 16, 1988.

2M. Adams and J.F. Kerner, "Evaluation of Promotional Strategies To Solve the Problem of Underutilization of Breast Examination/Education Center in a

New York City Black Community," in Issues in Cancer Screening and Communications (New York, NY: Alan R. Liss, Inc., 1982).

3S.J. Blalock, B.M. De Vellis, and R.S. Sandler, "Participation in Fecal Occult Blood Screening: A Critical Review," preventive Medicine 16:9-18, 1987.

4R.C. Burack and J. Liang, "The Early Detection of Cancer in the Primary-Care Setting: Factors Associated With the Acceptance and Completion of

.Recommended Procedures," Preventive Medicine 16:739-751, 1987.504. Csinan, "Patterns in Preventive Behaviour: A Study of Women in Middle Age," Soc. Sci Med. 20(3):263-268, 1985.

60.1. Cohen, B. Littenberg, C. Wetzel, et al., "Improving Physician Compliance With Preventive Medicine Guidelines," Medical Care 20(10):1040-1045,

1982.7K.M. Cummings, A.M. Jette, B.M. Brock, et al., "Psychosocial Determinants of Immunization Behavior in a Swine Influenza Campaign," redical Care

17(6):639-649, 1979.8A.K. David and J.S. Boldt, "A Study of Preventive Health Attitudes and Behaviors in a Family Practice Setting," The Journal of _Family Practice

11(1):77-84, 1980.9R.A. Davidson, S.W. Fletcher, S. Retchin, et al., "A Nurse-Initiated Reminder System for the Periodic Health Examination," Arch, Intern. Med. 144:2167-

2170, 1984.10D.M. Harris and S. Guten, "Health-Protective Behavior: An Exploratory Study," Journal of Health and Social_ Behevior 20:17-29, 1979.

11R.A. Hayward, M.F. Shapiro, H.E. Freeman, et al., "Who Gets Screened for Cervical and Breast Cancer?" Arch. Intern. Med. 148:1177-1181.

12J. Howard, "Using Mammography for Cancer Control: An Unrealized Potential," CA-A Cancer Journal for Clinicians 37(1):3348, 1987.

130.R. Lairson, and J.M. Swint, "A Multivariate Analysis of the Likelihood and Volumeof Preventive Visit Demand in a Prepaid Group Practice," Mediceq.

CALI 16(9):730-739, 1978. D.R. Lairson, and J.M. Swint, "Estimates of Preventive Versus Nonpreventive Medical Care Demand in an HMO," Health Services

Research 14(1):33-43, 1979.14C.J. MoDoeald, S.L. Hui, D.M. Smith, et al., "Reminders to Physicians From an Introspective Computer Medical Record," Annals Of Internet Medicine

100:130-138, 1984.15S.J. McPhee, R.J. Richard, and S.N. Solkowitz, "Performance of Cancer Screening in a University General Internal Medicine Practice: Comparison With

the 1980 Cancer Society nuidelines," eralnterrediciJournalofGen 1:275-281, 1986.

16G. DeFriese, Director, Health Services Research Center, University of Me:th Carolina at Chapel Hill, personal communication, Washington, DC, May 20,

1988.17Public Health Service, U.S. Department of Health and Human Services, "Screening for Cervical and Breast Cancer -- Southeastern Kentucky," Morbidity

and Mortality Weekly Report 36(52):845-849, 1988.

Appendix D.--cont'd

18Public Health Service, U.S. Department of Health and Human Services, "Use of Mammography for Breast Cancer Screening -- Rhode Island, 1987," Morbidity

and Mortality Weekly Report 37(23):357-360, 1988.19Public Health Service, U.S. Department of Health and Human Services, "Provisional Estimates From the National Health Interview Survey Supplement on

Cancer Control -- United States, January-March 1987," Morbidity and Mortality Weekly Revert 37(27):417-425, 1988.

201.A. lillard, W.G. Nanning, C. Peterson, et al., Preventive Medical Care: Stenderds. Usage. end Efficecv (Santa Monica, CA: The Rand Corporation,

.1986); and N. Lurie, W.G. Manning, C. Peterson, et al., "Preventive Care: Do We Practice What We Preach," Am. J. Public Health 77(7):801-804, 1987.

21M.A. Riddiough, J.S. Willems, C.R. Sanders, et al., "Factors Affecting the Use of Vaccines: Considerations for Immunization Program Planners," Public

Health Reports 96(6):528-535, 1981.22F.J. Romm, S.W. Fletcher, and B. Hulka, "The Periodic Health Examination: Comparison of Recommendetions and Internists' Performance," Southern Medical

Journal 74(3):265-271, 1981.23T.G. Mundell and J.R.C. Wheeler, "The Effect of Income on Use of Preventive Care: An Evaluation of Alternative Explanations," Journal' of Health and

Social Behavior 20:397-406, 1979.24D.S. Salkever, "Accessibility and the Demand for Preventive Care," Social Science and Medicine 10:469-475, 1976.

25W.A. Satariano, A.G. Schwartz, end G.M. Swanson, "Screening for Cervical Cancer: Results From Several Intervention Strategies," in Issues in Cancer

Screening ano Communications (New York, NY: Alan R. Liss, Inc., 1982).26J.A. Sawyer, J. Earp, R.H. Fletcher, et al., "Pap Smears in Rural Black Women," from the Department of Medicine and School of Public Health,

University of North Carolina at Chapel Hil,, Chapel Hill, NC, unpublished manuscript.27D.P. Slesinger, R.C. Tessler, and D. Mechanic, "The Effects of Social Characteristics on the Utilization of Preventive Medical Services in Contrasting

Health Care Programs Medical Care 14(5):392-404, 1976.28N.A. Teitelbaum, C.S. Weisman, A.C. Klassen, et al., "Pap Testing Intervals: Specialty Differences in Physicians' Recommendations fn Relation to

Women's Pap Testing Behavior," Medical Care 26(6):607-618, 1988.29R.S. Thompson, M.E. Mfchnich, J. Gray, et al., "Maximizing Compliance With Hemoccult Screening for Colon Cancer in Clinical Practice," Medical Care

24(10):904-914, 1986.38C.M. Valente, J. Sobel, H.L. Muncie, et al., "Health Promotion: Physicians' Beliefs, Attitudes, and Practices," Am. J. Preventive Medicine 2(2):82-88,

1986.231R.B. Warnecke, P.L. Havlicek, and C. Manfredi, "Awareness and Use of Screening by Older-Age Persons," in Prospectives on Prevention and Treatment of

Cancer in the Elderly, R. Yancik (ed.) (New York, NY: Raven Press, 1983).

32H. Wechsler, S. Levine, R.K. ldelson, et al., "The Physician's Role in Health Promotion -- A Survey of Primary-Care Practitioners," M. Enol. J. Med.

308(2):97-100, 1983.33P.S. Williamson, C.E. Driscoll, L.D. Dvorak, et al., "Health Screening Examinations: The Pati,nt's Perspective," The Journal of Family Practice

27(2):187-192, 1988.34R.A. Windsor and J. Morris, "Consumer Participation and Payment for Health Promotion Programs at an Urben Primary Care Center," Tte_aletrajalmei_of

Medical Sciences 21(2):209-213, 1984.358. Woo, B. Woo, E.F. Cook, et al., "Screening Procedures in the Asymptomatic Adult: Comparisons of Physicians' Recommendations, Patients' Desires,

Published Guidelines, and Actual Practice," J.A.M.A. 254(11):148D-1484, 1985.

SOuRCE: Office bf Technology Assessment, 1989.

Appendix E: OTA ANALYSiS OF PREVENTIVE SERVICE USE BYTHE ELDERLY WITH DATA FROM THE 1982NATIONAL HEALTH INTERVIEW SURVEY

MethodsThe Data

The National Health Interview Survey(NHIS) is a representative household surveyconducted annually since 1957 by the U.S.Department of Health and Human Service'sNational Center for Health Statistics (NCHS).In addition to a core questionnaire thatmeasures the self-reported prevalence of var-ious medical conditions, the use of healthservices, general health status, disability, anddemographic characteristics, NCHS adds sup-plemental questionnaires on specific topicsthat vary from year to year. In 1982, theNHIS contained supplemental questionnaireson the use of preventive health services andthe types and degree of health insurancecoverage. Although NCHS has publishedsome data from the preventive services sup-plement in tabular form (75), no publishedwork to date has attempted to use these datato understand what factors are associated withthe use of preventive services by the elderly.OTA decided to conduct such an analysis.

OTA obtained magnetic tapes of the coreand suoplemental questionnaire data for the1982 WHIS from NCHS. These data filescontained 103,923 observations reflectingrespondents of all ages with 11,434 observa-tions for individuals 65 years or older. Inaddition to responses to survey questionnaires,each observation contained a uniqueidentifier, variables identifying a stratum andcluster from which it was drawn, and theweights necessary to produce representativeestimates.

Among the variables on the data set,OTA was interested in:

the amount of time elapsed since therespondent last received each of fivescreening services (glaucoma screening,blood pressure measurement, eye exam-ination, breast examination, and Papsmear); and

46

factors potentially associated with theuse or nonuse of these five services.

OTA converted each of the five variablesmeasuring elapsed time since use of pre-ventive services to a binary variable thatmeasures whether or not the individual usedthe service within a specified period of time.

These periods of time are based on therecommendations of expert groups presentedin table 2 in the text of this paper. Becausethere is some variation across the diffeNntsets of expert guidelines listed in table 2,OTA summarized the published recommenda-tions in the composite measures presented intable 8. These composite measures do notrepresent a set of recommendations them-selves; rather, they are merely one benchmarkfor comparing actual use to what is generallyconsidered adequate by recommendinggroups. Where there is disagreement amongrecommending groups, the compositemeasures tend toward longer intervals be-tween screenings in order to measure com-pliance with minimal recommended levels ofprevention. These composites of expertrecommendations pertain only to the primaryanalyses conducted by OTA. As a source ofcomparison, table 8 also includes the periodsused by two other studies of the use of pre-ventive procedures. In one of these papers,the authors measured recommended peri-odicity as the mean of published recom-mendations (92). The other paper formed aconsensus based on their own review ofrelevant literature (45).

Because of the coding scheme of theNH1S, use within an "x" year period reallymeans that the individual had used the ser-vice within a period of less than but not in-cluding "x+1" years. For example, considerthe case of breast examinations. Table 2 sug-gests that one should measure use within theprevious year. However, under the NHIScoding scheme, one would consider an elderly

The Use ol Preventive Services by the Elderly i 47

woman who had her last breast examinatiol21 months ago to have been adequatelyscreened; a woman whose last breast exam-ination was 24 months ago would not. Table9 lists all variables used in the analysis andalso includes appropriately weighted de-scriptive statistics.

Methods of Analysis

OTA used PC-CARPR software (on an80286 personal computer) to analyze the data.PC-CARPR was developed by the StatisticalLaboratory at Iowa State University especiallyfor analyzing data from surveys with complexdesigns like the NHIS. It makes use of thesampling information in the data files to pro-duce appropriate point and variance esti-mates. OTA performed two separate analyses

on the data set:

a descriptive summary of the perc:At-ages of elderly individuals who reportedusing each of the five screening serviceswithin the specified time; and

a multivariate weighted logistic regres-sion analysis of the use or nonuse ofthese services.

In addition, OTA:

examined whether observations droppedfrom the multivariate analysis becausethey contained some missing informationdiffer in any important ways from ob-servations included in the analysis,potentially biasing our estimated para-meters;

Table 8.- -Periods of Time Used by OTA and Two Studies to MeasureOlder Adults' Use of Preventive Services

Period of time to measure use employed by:

Service OTAa Lillard, et al., 1986b Woo, et al., 1985c

Initial or periodic physical exam 1 year 1.4 years

Blood cholesterol level 5 years 4.5 years

Fecal occult blood test 1 year 1 year 1.0 year

Pap smear 3 years* 3 years 4.1 years

Glaucoma screeni g 2 years*

Optometry/ophthalmology examPneumecoccal immunization

2 years*Lifetime Lifetime -

Influenza immunization 1 year 1 year

Tetanus immunization 10 years 10 years

Hypertension screening 1 year* 1.4 years

Breast examination 1 year*

Abbreviation: HMO ft health maintenance organization.aIntervals listed in this column represent composites of the expert recommendations summarized in table 2.

bL.A. Lillard, W.G. Manning, C. Peterson, et al., PreventiveidedicalCare: Standards Usage and Efficacy (Santa

Monica, CA: The Rand Corporation, 1986).cB. Woo, B. Woo, S.F. Cook, et al., "Screening Procedures in the Asymptomatic Adult: Comparisons of

Physicians' Recommendations, Patients' Desires, Published Guidelines, and Actual Practice," .14004.A.

254(11):1480-1484, 1985.

*As described in greater detail in appendices E and F, OTA estimated the use of preventive services among

the elderly with two different data sourcesa single HMO and the 1982 National Health Interview Survey

(NHIS). The asterisk indicates services included in the NH1S analysis. Because of the coding scheme of

the NHIS, use within an "x" year period really means that the individUal has used the service within a

period of less than but not including "x+1" years. Under this scheme, we would consider an elderly woman

who had her last breast examination 21 months ago to have been adequately screened; a woman whose last

breast examination was 24 months ago would not. For the HMO data, use witnin "x" years carries a literal

definition.

Sources: Office of Technology Assessment, 1989.

48 IN The Use of Preventive Services by the Elderly

considered whether multicollinearity inthe models might reduce precision; andexamined how the probabilities of usepredicted by the models varied witheach significant variable.

The multivariate analysis uses Taylorseries techniques to estimate a weightedlogistic regression model of the form:

P- = 1 / [1 + e"( aj +Di (5j +Hiyj ij)]

where

Pu is the probability of elderly person iusing service j. Use is measured bybinary variable Yu (=1 if person ihad service j within the specifiedperiod of time; Yu 0 otherwise).

Di is a vector of predisposing charac-teristics describing elderly person i.

. is a vector of enabling characteristicsdescribing elderly person i.

Hi is a vector of health status charac-teristics describing elderly person i.

j is an estimated parameter, and 6116 ., are vectors of estimatedpajrameters for service j.

is an individual, service specific er-ror term.

PC-CARPR produces estimated coefficientsthat are consistent and appropriatelyweighted. Estimates of asymptotic variancesalso appropriately reflect the complex surveydesign.

PC-CARPR requires that no observationin the data matrix contair missing values.For the 1o8istic regression analysis, OTA usedSPSS-PC+K to create two data files. OTApurged both files in a listwise fashion of ob-servations containing missing data on any

variable in the models.1 OTA used one ofthe data sets to estimate models for pre-ventive services potentially used by bothsexesglaucoma screening, eye examinations,and blood pressure measurement. OTA usedthe other data set, which contained only theobservations for women, to estimate modelsfor breast examinations and Pap smears.

The data set for both sexes contained ex-actly 9,000 out of the original 11,434 obser-vations. The remaining 2,434 observations,which had missing data, represented aweighted 21.5 percent of the over-65 popula-tion. The single vatiable with the most miss-ing observations was family income. Thisvariable alone had missing observationsrepresenting 15.3 weighted percent of theelderly population. Of each of the other var-iables containing missing values, none lackeddata on observations representing more than 4weighted percent of individuals over-65. Thedata set containing only women had 5,040observations out of a possible 5,655. The1,615 observations with missing datarepresented 19.6 weighted percent of allwomen over 65.

ResultsThe estimates of the national proportions

of elderly using each of the five screeningservices within the specified time are pre-sented in table 3 in the text of the paper.Additional descriptive statistics are presentedin table 9. Table 10 below presents the para-meters of the estimateo logistic regressionsthat attempt to explain the use or nonuse ofeach service. Table 7 summarizes theseresults, and the text of the paper discussestheir significance.

i To estimate the descriptive statistics presentedin table 9 and the national percentages of elderlyreceiving the five screening services, OTA useddata sets that contained a(l Observations for whichany data is available for the particular variablein question. The listerilmtdektios of missing valuesdescribed here only applies to the logisticregression analysis.

The Use of Preventive Services by the Elderly a 49

Table 9.--1982 National Health Interview Survey:Selected Descriptive Statistics for Persons Over 65

Number of observations:Weighted number of observations:

11,43425,391,023

Weighted Means and Standard Deviations for Continuous Variabtes

Variable Mean Standard deviation

AGE 73.39 6.63

INCOME 15,217.97 13,853.56

Weighted Frequency Distributions for Categorical Variables

Variable Proportion

GLAUCOMA1=screened for glaucoma within

previous 2 years, 11 months0=otherwise

0.660.34

EYE

1=received eye examination withinprevious 2 years, 11 months

0=otherwise

Variable Proportion

EDLEVEL Highest educational level attained:0=none or kindergarten 0.02

1=1 to 8 years (elementary) 0.38

2=9 to 11 years (some high school) 0.16

3=12 years (high school graduate) 0.26

4=1 to 3 years college 0.09

5=college graduate 0.05

6=post-graduate education 0.04

1=resides in a Census Bureau Standard0.75 Metropolitan Statistical Area (urban area) 0.64

0.25 0=otherwise 0.36

BLOOOP1=had blood pressure measured within

previous 1 year, 11 months 0.93

0=otherwise 0.07

BREASI1=had clinical breast examination

within previous 1 year, 11 months 0.50

0=otherwise 0.50

PAP

1=had Pap smear within previous 3years, 11 months

0=otherwise

0.52

0.48

MALE1=male

0=female

PREPAID1=enrolled in HMO or some other prepaid

health planOmotherwise

0.020.98

HLTHINSR

1=has some health insurance coverageor health benefits beyond Medicareincluding prepaid, Veterans', military,or means tested public assistancehealth benefit 0.78

0=otherwise 0.22

ALONE

1=lives alone0=otherwise

0.30

0.70

BEDDAYS days in bed during previous

0.41 12 months:

0.59 0=none1=1 to 7 days2=8 to 30 days3=31 to 180 days4=181 to 365 days

NWHITE1=nonwhite0=white

0.650.16

0.120.05

0.02

LIMITED

0.10 1=limited in some activity0.90 0=otherwise

0.70

0.30



Table 10.--Eiderly Use of Five Screening Services: Logistic Regression Results

Independent Meanvariable value

GLAUCOMA BREAST PAPEstimatedcoefficients'

Estimated

coefficients'Estimatedcoefficients'

Mean Estimated Estimatedvalue coefficients' coefficients'

INTERCEPT 1.0000000 0.4160073 -0.3433352 0.8399042 1.0000000 -0.0166337 2.1542170(0.2927390) (0.7538013) (0.2963410)0* (0.3960973) (0.3889979)**

Predisoosing factors:MALE 0.4162490 -0.3009192 -0.3259628 -0.2439690

(0.0534922)** (0.0957841)** (0.5874069)** --

AGE 73.2137000 -0.0091917 0.0264826 -0.0061867 73.4815000 -0.0167258 -0.0400128(0.0038157)* (0.0095720)** (0.0039844) (0.0051317)** (0.0049593)**

NONWHITE 0.0945106 -0.3735570 -0.0368144 -0.0887188 0.C937475 0.0263338 0.1506788(0.0838741)** (0.1647689) (0.0923361) (0.0982817) (0.1056188)

EDLEVEL 2.3429000 0.1601070 0.1038493 0.1147214 2.3367600 0.1132536 0.0962961(0.0223536)** (0.0436750)* (0.0229877)** (0.0255963)** (0.0249885)**

Enabling factors:INCOME 15276.8000000 0.0000114 0.0000106 0.0000066 14059000000 0.0000119 0.0000102

(0.0000026)** (0.0000058) (0.0000028)* (0.0000027)**(0.0000028)** (0.0000028)0*SMSA 0.6336510 0.2167839 0.0064913 0.2054843 0.6364580 0.3110662 0.1555873

(0.0571966)** (0.1049219) (0.0589708)** (0.0694007)** moncearw,PREPAID 0.0229561 0.2349249 0.6391572 0.1612398 0.0199337 0.3845290 0.4433711

(0.2200813) (0.5023571) (0.2554439) (0.2405225) (0.2723756) (0.2723756)HLTHINSR 0.7896040 0.4306881 0.5616778 0.3853976 0.7834470 0.3761862 0.3581028

(0.0625027)** (0.1248533)** (0.0628222)** (0.0884009)00(0.0861343)** (0.0861346)**ALONE 0.3067190 0.0062441 -0.1136007 0.0406232 0.4099680 0.1971576 -0.0397508

(0.0589261) (0.0944282) (0.0638674) (0.0743910)**(0.0665113)0*(0.0651125)Health status measures:BEDDAYS 0.6316600 0.1227310 0.8908110 0.0776005 0.6405870 0.3337425 0.1943092

(0.0304799)** (0.1050414)** (0.0281800)** (0.0327194)**(0.0335717)**(0.0333572)**LIMITED 0.2992070 0.0003354 0.3276151 -0.1233437 0.3089910 0.0448802 0.0331326

(0.0585122) (0.1317948)e (0.5909522)* (0.0703522) (0.0762676) (0.0762676)

Model statistics:M ,9000.0 9000.0 9000.0 5040.0 5040.0Dependent variable mean') 0.6635 0.9319 0.7432 0.4985 0.5243F-statistic 32.41** 15.14** 15.12** 23.09** 25.87**Average design effecte 1.32 1.53 1.28 1.20 1.23

'Asymptotic standard errors are in parentheses below each estimated coefficient.

bVariable means were calculated from data matrices used to estimate logit modelsobservations with missing values).

cAverage "effect" ot complex survey design on variances of estimated coefficientsas the number of times greater the variance from the complex design is than therandom design.

*Estimated parameter significant at the 0.05 level, two-tailed test.**Estimated parameter significant at the 0.01 level, two-tailed test.

yariable Key:

Depsocisnt Variables:GLAUCOMA --1=screened for glaucoma within previous 2 years, 11 months; OmotherwiseEYE --1=received eye examination within previous 2 years, 11 months; 0=otherwiseBLOODP --1=had blood pressure measured within previous 1 year, 11 months; 0=otherwiseBREAST -tithed clinical breast examination within previous 1 year, 11 months; 0=otherwisePAP --1=had Pap smear within previous 3 years, 11 months; -'crwise

Isoclent Variables:MALE --1=male; 0=femaleAGE --respondent's age in yearsNONWHITE --1=nonwhite; 0=whiteEDLEVEL -highest educational level attained; 0=none or kindergarten; 1=1 to 8 years (elementary); 2=9

to 11 years (some high school); 3=12 years (high school graduate); 4=1 to 3 years college;5=college gradUate; 6=post-graduate education

INCOME --family income in dollarsSmSA --1=resides in a Census Bureau Standard Metropolitan Statistical Area (urban area); 0=otherwisePREPAID --1=enrolled in HMO or some other prepaid health plan; 0=otherwiseHLTHINSR --1=has some health insurance coverage or health benefits beyond Medicare including prepaid,

Veterans', military, or means tested public assistance health benefit; 0=otherwiseALONE --1=lives alone; 0=otherwiseBEDDAYS --days in bed during previous 12 months; 0=none; 1=1 to 7 days; 2=8 to 30 days; 3=31 to 180

days; 4=181 to 365 days--1=limited in some activity; 0=otherwise

(i.e., purged of

. This "effect" is measuredvariance from a simple

LIMITED

SCURCE: Office of Technology Assessment, 1989.

The Use of Preventive Services by the Elderly SI

Table 11.- -Correlation Matrix for Variables in Logistic Regression Models5

GLAUCOMA BLOCOP EYE BREAST PAP KALE AGE

NON-

WHITE

ED-

LEVEL INCOME SMSA PREPAID

HLTH-

INSR ALONE BEDDAYS LIMITED

GLAUCOMA 1.00XX

BLOODP 0.19 1.00

XX XX

EYE 0.69 0.18 1.00

XX XX XX

BREAST XX XX XX XX

XX XX XX 1.00

PAP XX XX XX XX XX

XX XX XX 0.49 1.00

MALE -0.06 -0.04 -0.05 XX XX 1.00

XX XX XX XX XX 1.00

AGE -0.05 0.03 -0.03 Xx XX -0.06 1.00

Xx XX XX -0.06 -0.15 XX 1.00

NONWHITE -0.09 -0.02 -0.03 XX XX 0.00 -0.01 1.00

XX XX XX -0.02 0.00 XX -0.01 1.00

EDLEVEL 0.15 0.05 0.10 XX XX 0.01 -0.14 -0.16 1.00

XX XX XX 0.11 0.11 XX -0.14 -0.18 1.00

INCOME 0.12 0.04 0.07 XX XX 0.10 -0.11 -0.13 0.39 1.00

XX XX XX 0.09 0.10 XX -0.09 -0.13 0.32 1.00

SMSA 0.07 0.01 0.06 XX XX -0.01 -0.00 0.04 0.09 0.12 1.00

XX XX XX 0.09 0.05 XX 0.01 0.02 0.08 0.11 1.00

PREPAID 0.03 0.02 0.02 XX XX 0.03 -0.03 -0.00 0.04 0.03 0.10 1.00

XX XX XX 0.04 0.04 XX -0.03 -0.01 0.03 0.01 0.08 1.00

HLTHINSR 0.12 0.07 0.09 XX XX 0.02 -0.09 -0.12 0.18 0.11 0.01 0.08 1.00

XX XX XX 0.10 0.09 XX -0.08 -0.11 0.18 0.07 0.00 0.08 1.00

ALONE -0.01 -0.01 0.01 XX XX -0.26 0.18 0.01 -0.02 -0.31 -0.00 .0.04 -0.01 1.00

XX XX XX 0.01 -0.06 XX 0.21 -0.01 -0.00 -0.36 -0.01 -0.03 0.03 1.00

BEDDAYS 0.04 0.13 0.02 XX XX -0.02 0.06 0.04 -0.05 -0.03 -0.01 -0.02 -0.01 0.01 1.00

XX XX XX 0.15 0.08 XX 0.06 0.04 -0.05 -0.04 -0.00 0.00 -0.02 0.01 1.00

LIMITED -0.00 0.06 -0.03 XX XX -0.04 -0.07 0.06 -0.12 -0.07 -0.04 -0.00 -0.02 -0.03 0.31 1.00

XX XX XX 0.04 0.03 XX -0.04 0.07 -0.12 -0.05 -0.03 0.01 -0.01 -0.04 0.45 1.00

Pearz.on correlation coefficients. First row in each cell gives correlation in data set used to estimate

models for GLAUCOMA, EYE and BL000P'(n=9000). Second row gives correlation for data set used to estimate

models for BREAST and PAP (n=5040).

SYmbol Key: XX=Not applicable (Both variables not contained on that data set)

Variable Kev:

Dependent Variables:GLAUCOMA --1=screened for glaucoma within previous 2 years, 11 months; 0=otherwise

EYE --1=received eye examination within previous 2 years, 11 months; 0=otherwise

BLOODP --1=had blood pressure measured within previous 1 year, 11 months; 0=otherwise

BREAST --1=had clinical breast examination within previous 1 year, 11 months; 0:otherwise

PAP ..Ishad Pap smear within previous 3 years, 11 months; 0:otherwise

Independent Variable.:MALE --1=male; 0=female

AGE --respondent's age in years

NONWHITE --1=n0nwhite; 0=white

EDLEVEL --highest edUcatiomai level attained; 0=nene or kindergarten; 1=1 to 8 years (elementary); 2=9 to

11 years (some high school); 3=12 years (high school graduate); 4=1 to 3 years college;

5=college graduate; 6=post-graduate education

INCOME --family income in dollars

SMSA --1=resides in a Census Bureau Standard Metropolitan Statistical Area (urban area); 0=otherwise

PREPAID -.1:enrolled in MO or some other prepaid health plan; 0:otherwise

HLTHINSR --1=hes some health insurance coverage or health benefits beyond Medicare including prepaid,

Veterans', military, or means tested pubLic assistance health benefit; 0=otherwise

ALONE --1=lives *lone; 0=otherwisa

BEDDAYS --days in bed during previous 12 months; 0=none; 1=1 to 7 days; 2=8 to 30 days; 3=31 to 180 days;

4=181 to 365 days

LIMITED --1=limited in some activity; 0=otherwise


1 1

52 T/te Use of Preventive Services by the Elderly

To examine the possibility that multi-collinearity among the independent variablesin the model might preclude precise estima-tion, OTA estimated the weighted first-orderPearson correlation matrices for the two datasets. Table 11 presents the correlationstatistics. Only three pairs of independentvariables had correlations greater than 0.25(or less than -0.25): EDLEVEL and IN-COME (0.39 in the two-gender data set and0.32 in the women only data set), ALONEand INCOME (-0.31 and -0.36 respectively),and the two measures of health status,LIMITED and BEDDAYS (0.31 and 0.43respectively). In addition, ALONE and AGEhave correlations of 0.18 and 0.21 respective-ly in the two data sets. However, despite thepotential effect of this collinearity on theestimated variances, the conclusions are un-likely to change. In all models except bloodpressure measurement, EDLEVEL and IN-COME are both already significant predictorsof preventive service use. ALONE is sig-nificant in three out the five (with bloodpressure measurement and Pap smears beingthe exceptions). Although there is a high de-gree of correlation between BEDDAYS andLIMITED, at least one of them is statisticallysignificant in all of the models except glau-coma screening, thus supporting the notionthat health status is associated with preventiveservice use among the elderly.2

OTA excluded a substantial proportion ofobservations because data were missing forone or more variables in the model. In orderto examine if these exclusions could havebiased the results of the multivariate models,OTA compared the characteristics of the in-cluded and excluded groups. In both datasets, the included respondents were sig-nificantly different from those el:ainated be-

2 However, because multicollinearity reduces theprecision of the estimator, the tandard error ofthese two variables' estimated coefficients may bebiased in the glaucoma screening model.

cause of missing data for only two variables:HLTHINSR (the presence of any health in-surance beyond Medicare) and INCOME.3For each of the other variables (including thedependent variables), the mean for the obser-vations with missing data did not differstatistically from the mean for observationsincluded in our analyses. This analysis sug-gests that the exclusion of observations withmissing data is unlikely to introduce bias intothe multivariate models, but OTA cannot ruleout the possibility.

In order to examine the effect of eachsignificant variable in the estimated models,OTA simulated, one independent variable at atime, how the probability of using eachscreening service varied with each possiblevalue of the independent variables. In thesesimulations, all independent variables, exceptthe one whose effect was being simulated, as-sumed their mean values.

Table 12 and figures 1 through 4 present9.1 results of this analysis for each significantval.aule in our models. Among the indepen-dent variables, holding other factors constant,age, education, and health insurance appearto have the greatest overall effect on theprobability of receiving each of these ser-vices. This analysis also supports the notionthat blood pressure measurement is differentfrom other services. Since almost everyonereceives it, there is less variation to explain.Hence, the variables in the model appear lessimportant in predicting its use than they dofor the oiher services.

3 The group of observations excluded from theanalysis had a lower mean income ($14,475 versus$15,276 in the two gender data set; p<0.01) and wasless likely to have any insurance coverage beyondMedicare (0.70 versus 0.79 in the two gender dataset; p<0.05) than was the included group. The in-come statistic may not accurately reflect theentire group of observations with missing datasince three-quarters of the observations m: iing

any data a. all did not have income data.


Table 12.--Effect of Statistically Significant Binary Variables* in Logistic Regressions onElderly Use of Five Screening Services: Predicted Probabilitiesb

MALE= 0

= 1

NONWHITE= 0

= 1

SMSA= 0

= 1

HLTHINSR= 0

= 1

ALONE= 0

= 1

LIMITED= 0

4 1

GLAUCOMA BLOOOP BREAST APPredictedprobability

Predictedprobability

--__=Predictedprobability



0.70 0.96 0.77 XX XX

0.63 0.94 0.72 XX XX

0.68 -- -- -- --

0.59 -- -- --

0.64 .. 0.73 0.45 0.50

0.77 .. 0.77 0.53 0.54

0.59 0.92 0.69 0.43 0.46

0.69 0.96 0.77 0.52 0.55

-- 0.48.. -- 0.53

-- 0.95 0.76 -- --

0;96 0.74

aEffect of significant non-binary variables shown in figures 1 through 4.

bPredicted probability is estimated as 11(1 e ) where - is the vector of

estimated coefficients and X is the vector of individual characteristics. Of these

characteristics (all independent variables included in the estimated model), eachtakes on its mean value except the one designated in that row of the table above;it takes on the value shown in the row header.

Symbol Key:XX=Independent variable not included in model--=Estimated coefficient on independent variable not significant at 0.05 level, two-tailed test

Variable Kev:Dependent VGLAUCOMAEYE -

BLOODPBREASTPAP

&Asides:1=screened for glaucoma within previous 2 years, 11 months; 0=otherwise

-1=received eye examination within previous 2 years, 11 months; 0=otherwise-1=had blood pressure measured within previous 1 year, 11 months; 0=otherwise

-1=had clinical breast examination within previous 1 year, 11 months; 0=otherwise

1=had Pap smear within previous 3 years, 11 months; 0=otherwise

IminandeutMALE -

NONWHITE -

SMSA -PREPAIDHLTHINSR

ALONE -LIMITED

Variables:-1=male; 0=female-1=nonwhite; 0=white1=resides in a Census Bureau Standard Metropolitan Statistical Area (urban area); 0=otherwise

1=enrolled in HMO or some other prepaid health plan; 0=otherwise-1=has some health insurance coverage or :.ealth benefits beyond Medicare including prepaid,Veterans', military, or means tested public assistance health benefit; 0=otherwise

1=lives alone; 0=otherwise-1=limited in some activity; 0=otherwise


6 t)

Figure 1.--Effect of Age on UsePredicted Probabilities Predicted Probabilities

Figure 2.--Effect of Income on Use

Probability ol Screeningif . - 4.- 4+4 - --6--

oLOt -. - -- 4.4 -

0.6 14-

464.4-b 0.8

Probability 01 Screening0.9

0.8

0.7

--4-0--*--4--.---..-- -4- ..-.- ---.---0--.--4--.---.--..--.---4,--4- /

...---_-

.0-411-41' Sit it it 42-is .0 is 0 0 0 0- (1-0-0 0 41 4.1-441-il_--rl

OA/

0.21, t . 1......1.-.1.. t .1 I. ...L...1 . f . ..../ -.3 . 1.- 1 1-1 -1-1.- 1-1-1--1- 1

05 70 76 80

..............

86

,.......

90

........,_.

950.41 7777: .....4..T.-.1.+17..i........1_.-41-47..11:14P-1.1._i __,.._f ...l

.*-1 -0:001000 6600 15500 22600 47500Age

Glaucoma Screening -.- BlOOd Pressure -°- Eye ESOP, Olescome Screening

-4* 8reeet team -.- Pap Smeate *0-- Breast Poem

Figure 3.--Effect of Education on UsePredicted Probabilities

Probatally of Straten.ng

09;

08;

0

0 5-

.

-

0 4None Elementary Soma ffS HO Oral Some Coll Coll Grad Poal-Col

Level ot Educahon Achieved

alouconA Scripning Blood Pressure Ey. Sava

Orseat Euro Pep Smer

Income

Eye BM%

Pap (imesr

Figure 4.--Effect of Bed Days on UsePredicted Probabilities

Probst:misty ol Scretening

08;I.

-7 -

0 7 f

0 el-1

0 --A

0 4' . . . -Nons 1-7 treys 830 Days 31-180 Days 1f55-305

Bed Days in Last 12 Months

Olsuco.a Screening " blood Proolore Eye Beam

breset BoaM Pep Smell?

Key to Figure 3: Elementary = 1 to 8 years education; Some HS = 9 to 11 years education; MS Grad = 12 years education;Some Coll = 1 to 3 years college; Coll Grad = colleye gladuate; Post-Col = post-graduate education.


APPENDIX F: ANALYSIS OF PREVENTIVE SERVICE USEBY OLDER ADULTS IN A HEALTH

MAINTENANCE ORGANIZATION

Methods

The Data

OTA contracted with a health main-tenance organization (HMO) to provide dataon the use of eight preventive services bytheir over-65 year old enrollees:

Check-up visit,Cholesterol measurement,Eye examination,Fecal occult blood test,Pap smear,Influenza immunization,Pneumococcal immunization, andTetanus immunization.

OTA chose these services in consultation withthe HMO to meet the following criteria:

they are services often included amongdiscussions or recommendations forelderly preventive health; andthe HMO's data system routinely recordstheir use as distinct services.

To examine how use varies with age, theHMO also provided comparable data for en-rollees between the ages of 40 and 64. TheHMO measured the proportions of enrolleesusing each service within the periods of timepresented in table 8.

The HMO is a large, urban, staff modelhealth maintenance organization located inthe Northeastern United States. It serves en-rollees through private employers, govern-ment agencies, and individual accounts.Since January 1976, the HMO has servedMedicare beneficiaries, initially under a planwhere the HMO billed Medicare forMedicare-covered procedures on a fee-for-service basis. The HMO provided non-covered procedures, including the preventiveservices examined in this study, through a"wraparound" or "Medi-gap" policy purchasedby or for the enrollee.

Beginning in July 1985, the HMO enter-ed into a Medicare demonstration risk con-tract with over 80 percent of its 2500 existingMedicare enrollees transferring into this planwithin the first three months. All of the ser-vices covered under the HMO's basic benefitpackage, including preventive services, wereincluded in the risk contract plan.

The HMO has traditionally encouragedthe use of preventive services by at-risk pop-ulations through clinical guidelines for pre-ventive care and coverage of regular check-ups. Before October 1987, the monitoring ofcompliance with these guidelines was limitedto pediatric screening and immunization,prenatal screening, and influenza immuniza-tion. Since that date, the HMO has adopted aprogram to monitor and inform clinicians ateach visit of a patient's compliance with theHMO's preventive guidelines. Since OTAbelieved that this program is not typical ofmost HMOs, this HMO used October 1987 asthe endpoint for measuring rates of use foreach preventive services studied. Hence,during the periods of time examined, onlyinfluenza immunizations reflect any monitor-ing by the HMO, and for that service,clinicians only received information on ag-gregate rates of complian^e among all enrol-lees.

Methods of Analysis

The base population for this study is allpresent and former HMO enrollees who wereage 40 or older as of October 1, 1987. TheHMO identified the base population througha computerized search of enrollment recordsand separaterl the population into four sub-groups on the basis of age:

40 to 49 years old,50 to 64 years old,65 to 74 years old, and75 years and older.

6255


Because the HMO calculated age at the endof the observation period, some of the enroll-ees in each group fell below the low agethreshold at the time they actually used aspecific service.

Through computer searches of this basepopulation, the HMO defined a "denominatorpopulation" for each age group and observa-tion period over which the use of a specificpreventive service was to be measured. Each"denominator population" consisted of all per-sons of appropriate age continuously enrolledin the HMO during the observation period.Since two of the HMO's ten clinics did nothave computerized records at the level ofspecific clinical services, enrollees from thesesites were excluded from the analysis. Enroll-ees exclude from the denominator files be-cause they came from one of the non-computerized sites or because they were notcontinuous members represented 20 percentof the base population in each age group.

The "denominator" file for the over-65age groups consists of all continuously en-rolled individuals from the eight sites. Thedenominator population for the 40 to 64 agegroups were so large that the HMO used arandom sample of these groups for the analy-sis. They chose a 10-percent random samplefor all but the 10-year observation period,where they chose a 20-percent randomsample. Table 13 presents the number of ob-servations in each "denominator" file used tocalculate the rates of use.

In order to measure the use of each ser-vice for each age group, the HMO searchedthe base population to form "numerator" files

consisting of persons who met both of thefollowing criteria:

the individual was enrolled in one of theeight sites at the time the analysis wasconducted (June through September1988); andthe individual received the specific pre-ventive service within the observationperiod.

To calculate rates of use, each "numer-ator" file was compared to its corresponding"denominator" file. Individuals in the numer-ator file who did not appear in the denomi-nator file were discarded. Stratifying bygender, the HMO tallied the number of indi-viduals remaining in each "numerator" fileand divided that number by the number inthe corresponding "denominator" file to calcu-late a rate of use for each service and age-gender group. Table 14 presents the resultsof' this analysis.

It is possible that a few continuously en-rolled members transferred from one of thetwo excluded sites to one of the eight in-cluded sites before October 1, 1987. Whilesuch individuals would be included in the"denominator" files, they would not appear inthe "numerator" file if they received a pre-ventive service at the excluded site. Thiswould deflate the use rate. However, becausethe two excluded sites serve geographicallydistinct communities with most membersliving in close proximity to the clinic, trans-fer to another site is relatively rare. There-fore, OTA and the HMO concluded that thepotential undercounting in the use rates isminimal.

Table 13.--Sample Sizes for Each Measurement Period in OTA's Analysis ofPreventive Service Use in One 11103

Age10 yr. period*(10/77-10/88)

5 yr. period**(10/82-10/87)

3 yr. and 2 yr.**(10/84-10/87)

and (10/85-10/87)1 yr. period**(10/86-10/87)

Male Female Total Male Female Total Male Female Total Male Female Total

40-49* 250 260 510 455 507 962 725 797 1522 995 1068 206350-64* 246 227 473 282 331 613 466 518 984 654 784 143865-74** 329 307 636 849 .956 1805 1265 1440 2705 1902 2219 4128

75+** 113 145 258 204 313 517 271 395 666 514 752 1266

e n for 40-49 an d 50-66 age groups represent a 20 percent sample of members continuously enrolled during each period at the eight sitesstudied. The 65-74 end 75+ age groups repreaent all members.

**The n for 40-49 and 50-64 age groups represent a 10 percent sample of members continuously enrolled during each period at the eight sitesstudied. The 65-74 and 75+ age groups represent All members.

TabLe 14.--Percents of Continuously Enrolled Members Receiving Eight Preventive ServicesDuring Specified Periods of Time

Check-up visitAge (1 year period)

Cholesterol

(5 year period)Eye exam

(2 year period)Fecal occult blood(1 year period)

Pap smear(3 year period)

Iriftuenzi vaccine

(1 year period)

Pneumococcal

vaccine(lifetime)

Tetanus vaccine(10 year period)

Male Female Total Male Female Total Male Female Total Male Female Total Male Female Total Male Female Total Male Female Total Male Female Total

40-49 yrs. 24% 39% 32%50-64 yrs. 38 49 44

Subtotal40-64 29 43 37

65-74 yrs. 48 52 50

75+ yrs. 52 47 49

Subtotal65+ 49 51 50

68% 67% 67%80 77 78

73 71 72

77 76 7669 64 66

75 73 74

39% 50% 44%s1 55 53

44 52 48

70 74 72

80 79 79

72 75 73

21% 32% 27%36 46 41

27 38 33

49 53 51

51 46 48

49 52 50

N/A 80% N/AN/A 73 N/A

N/A 77 N/A

N/A 75 N/AN/A 60 N/A

N/A 71 N/A

4% 5% 4%14 12 13

8 8 8

55 55 55

67 63 64

58 57 57

1% 1% 1%7 5 6

3 2 3

29 27 28

47 42 44

33 30 31

20% 17 18%

13 13 13

17 15 16

47 40 44

40 35 37

45 38 42

Abbreviation: N/A = Not applicable.


ACRONYMS

AAO --American Academy of OphthalmologyACOG --American College of Obstetrics and GynecologyACP --American College of PhysiciansACR --American College of RadiologistsACS --American Cancer SocietyADA --American Diabetes AssociationAHA --American Heart AssociationAMA --American Medical AsociationAMWA --American Medical Women's AssociationANA --American Nurses AssociationAOA --American Optometric AssociationASPB --American Society to Prevent BlindnessCDC --Centers for Disease ControlCPS --Current Population SurveyCTF --Canadian Task Force on the Periodic Health ExaminationEKG --ElectrocardiogramHCFA --Health Care Financing AdministrationHMO --Health maintenance organizationIPA --Independent practice associationMMWR --Morbidity and Mortality Weekly ReportNCHS --National Center for Health StatisticsNCHSR --Nationai Center for Health Services Research and Health Technology AssessmentNCI --National Cancer InstituteNHIS --National Health Ititerview SurveyNHLBI --National Heart, Lung, and Blood InstituteNIH --National Institutes of HealthOTA --Office of Technology AssessmentSMSA --Standard Metropolitan Statistical AreaUSPSTF --United States Preventive Services Task Force

59

C5

REFERENCES

1. Adams, M., and Kerner, J.F., "Evaluation of Promotional Strategies To Solve the Problemof Underutilization of Breast Examination/Education Center in a New York City BlackCommunity," in Issues in Cancer Screening and Communications (New York, NY: Alan R.Liss, Inc., 1982).

2. American Academy of Ophthalmology, Policy Statement, "Frequency of Ocular Examina-tions," approved Feb. 6, 1983.

3. American Cancer Society, Summary of Current Guidelines for the Cancer-Related Check-UM Recommendations (New York, NY: ACS Professional Education Publication, 1988).

4. American College of Physicians, Committee on Immunizations, Guide for Adult Immuniza-tion, Philadelphia, PA, 1985.

5. American College of Physicians, Health and Public Policy Committee, "The Use of Diag-nostic Tests for Screening and Evaluating Breast Lesions," Annals of Internal Medicine103:143-146, 1985.

6. American College of Radiology. Policy Statement, "Guidelines for Mammography," adoptedSept. 22, 1982.

7. American Diabetes Association, "A.D.A. Policy on Scteening for Hyperglycemia," June1983.

8. American Heart Association, "Public Screening Strategies for Measuring Blood Cholesterolin Adults--Issues for Special Concern," October 1987.

9. American Optometric Association, National Health Division, "Optometry and the Nation'sHealth: Recommendations for the Implementation of Congress' National Health Priorities,"February 1978.

10. American Society for the Prevention of Blindness, "Facts on Blindness and Prevention,"February 1988.

11. Andersen, R., and Aday, L.A., "Access to Medical Care in the U.S.: Realized and Poten-tial," Medical Care 16:533-546, 1978.

12. Andersen, R., and Newman, J., "Societal and Individual Determinants of Medical CareUtilization," Milbank Mem, Fund 0,, 51:95-125, 1973.

13. Berwick, D., "Screening in Health Fairs: A Critical Review of Benefits, Risks and Costs,"Office of Disease Prevention and Health Promotion Monograph Series (Washington, DC:U.S. Government Printing Office, 1985).

14. Blalock, S.J., De Vellis, B.M., and Sandler, R.S., "Participation in Fecal Occult BloodScrec.:ning: A Critical Review," Preventive Medicine 16:9-18, 1987.

15. Brown, J.T., and Hulka, B.S., "Screening Mammography in the Elderly: A Case-ControlStudy," J, Gen. Intern, Medicine 3:126-131, 1988.

60

The Use of Preventive Services by the Elderly. 61

16. Burack, R.C., and Liang, J., "The Early Detection of Cancer in the Primary-Care Setting:

Factors Associated With the Acceptance and Completion of Recommended Procedures,"

Preventive Medicine 16:739-751, 1987.

17. Calnan, M., "Patterns in Preventive Behaviour: A Study of Women in Middle Age,"

Sci. Med. 20(3):263-268, 1985.

18. Canadian Periodic Health Examination Task Force, "The Periodic Health Examination,"

Can. Med. Assoc. J. 121(9):1193-1254, 1979; 130(10):1276-1292, 1984; 134(7):724-727,

1986; 138(7):618-626, 1988.

19. Chao, A., Paganini-Hill, A., Ross, R.K., et al., "Use of Preventive Care by the Elderly,"Preventive Medicine 16:710-722, 1987.

20. Cohen, D.I., Littenberg, B., Wetzel, C., et al., "Improving Physician Compliance With Pre-

ventive Medicine Guidelines," Medical Care 20(10):1040-1045, 1982.

21. Cummings, K.M., Becker, M.H., and Maile, M.C., "Bringing the Models Together: An

Empirical Approach to Combining Variables Used to Explain Health Actions," Journal of

Behavioral Medicine 3(2):123-145, 1980.

22. Cummings, K.M., Jette, A.M., Brock, B.M., et al., "Psychosocial Determinants of Im-munization Behavior in a Swine Influenza Campaign," Medical Care 17(6):639-649, 1979.

23. David, A.K., and Boldt, J.S., "A Study of Preventive Health Attitudes and Behaviors in a

Family Practice Setting," The Journal of Family Practice 11(1):77-84, 1980.

24. Davidson, R.A., Fletcher, S.W., Retchin, S., et al., "A Nurse-Initiated Reminder System for

the Periodic Health Examination," Arch. Intern. Med. 144:2167-2170, 1984.

25. DeFriese, G., Director, Health Services Research Center, University of North Carolina at

Chapel Hill, personal communication, Washington, DC, May 20, 1988.

26. Fedson, D.S., "Influenza and Pneumococcal Immunization Strategies for Physicians," Chest

91:436-443, 1987.

27. Fuchs, V., _H_ow We Live (Cambridge, MA: Harvard University Press, 1983).

28. Gallup Organization, "The 1987 Survey of Public Awareness and Use of Cancer DetectionTests: Summary of Findings," conducted for the American Cancer Society (Princeton, NJ:

Gallup Organization, January 1988).

29. Green, L.W., Kreuter, M.W., Deeds, S.G., et al., "Health Education Today and thePRECEDE Framework," in Health Education Planning: A Diagnostic Approach (Palo Alto,

CA: Mayfield Publishing Company, 1980).

30. Grundy, S.M., Greenland, P., Herd, A., et al., "Cardiovascular and Risk Factor Evaluation

of Healthy American Adults," Circulation 75(6):1340A-1362A, 1987.

31. Harris, D.M., and Guten, S., "Health-Protective Behavior: An Exploratory Study," Journal

of Health and Social Behavior 20:17-29, 1979.

32. Hayward, R.A., Shapiro, M.F., Freeman, H.E., et al., "Who Gets Screened for Cervical and

Breast Cancer?" Arch, Intern. Med. 148:1177-1181, 1988.

67

62 IN The Use of Preventive Services by the Elderly

33. Health Policy and Biomedical News of the Week, "Annual Pap Tests for at Least ThreeYears Advised by NCI; Guidelines Developed by Cancer Society," Health Policy andBiomedical News of the Week 31(3):6-7, 1988.

34. Herdman, R., Behney, C.J., and Gluck, M.G., "Legislative Interests in Geriatric HealthPromotion," in Proceedings of the Surgeon QIneral's Workshop on Health Promotion andAging (Washington, DC: U.S. Government Printing Office, 1988).

35. Howard, J., "Using Mammography for Cancer Control: An Unrealized Potential," CA-ACancer Journal for Clinicians 37(1):33-48, 1987.

36. Hulka, B.S., and Wheat, J.R., "Patterns of Utilization: The Patient Perspective," MedicalCare 23(5):438-460, 1985.

37. Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure,"The 1988 Report of the Joint National Committee on Detection, Evaluation, and Treat-ment of High Blood Pressure," Arch. Intern. Med. 148(5):1023-1038, 1988.

38. Kane, R.L., Kane, R.A., and Arnold, S.B., "Prevention in the Elderly: Risk Factors,"Health Services Research 19(6,II):946-1006, 1985.

39. LaForce, F.M., "U.S. Preventive Services Task Force: Immunizations, Immunoprophylaxis,and Chemoprophylaxis To Prevent Selected Infections," LAMA,. 257:2464-2470, 1987.

40. Lairson, D.R., and Swint, J.M., "A Multivariate Analysis of the Likelihood and Volume ofPreventive Visit Demand in a Prepaid Group Practice," Medical Care 16(9):730-739, 1978.

41. Lairson, D.R., and Swint, J.M., "Estimates of Preventive Versus Nonpreventive MedicalCare Demand in an HMO," litalth_SgaiggLasgauh 14(1):33-43, 1979.

42. Lawrence, L., Government Relations Representative, American College of Obstetriciansand Gynecologists, personal communication, Washington, DC, Oct. 31, 1988.

43. Lawrence, R.S., and Mickelide, A., "Preventive Services in Clinical Practice: Designing thePeriodic Health Examination," >LAMA 257:2205-2207, 1987.

44. Lazaro, C.M., Logsdon, D.N., and Meier, R., "Utilization of Preventive Health Services bythe Elderly," Insure Project, Lifecycle Preventive Health Services, New York, NY,presentation to the American Psychological Association Convention, Aug. 31, 1987.

45. Lillard, L.A., Manning, W.G., Peterson, C., et al., Preventive Medical Care: Standards,Usage. and Efficacy (Santa Monica, CA: The Rand Corporation, 1986).

46. Luft, M.S., Health Maintenance Qtganizations: Dimensions of Performance (New York,NY: John Wiley & Sons, Inc., 1981).

47. Lurie, N., Manning, W.G., Peterson, C., et al. "Preventive Care: Do We Practice What WePreach," Am. J. Public Health 77(7):801-804, 1987.

48. Maiman, L.A., and Becker, M.H., "The Health Belief Model: Origins and Correlates," inThe Health Belief Model ane. Personal Health Behavior, M.H. Becker (ed.), Health Educa-tion Monograph, vol. 2, No. 4 (Baltimore, MD: Johns Hopkins University, winter 1974).

i3S


49. McDonald, C.J., Hui, S.L., Smith, D.M., et al., "Reminders to Physicians From an Intro-

spective Computer Medical Record," Annals of Internal Medicine 100:130-138, 1984.

50. McGinnis, M., "The Tithonus Syndrome: Health and Aging in America," in Health Promo-

tion and Disease Prevention in the Elderly, R. Chernoff and D.A. Lipschitz (eds.) (New

York, NY: Raven Press, Ltd., 1988).

51. McPhee, S.J., Richard, R.J., and Solkowitz, S.N., "Performance of Cancer Screening in a

University General Internal Medicine Practice: Comparison With the 1980 Cancer Society

Guidelines," journal of General Internal Medicine 1:275-281, 1986.

52. National Society for the Prevention of Blindness, "Glaucoma Patient Guide" (New York,

NY: April 1984).

53. Radecki, S.E., Kane, RI., Solomon, D.H., et al., "Are Physicians Sensitive to the Special

Problems of Older Patients?" Journal of the American Geriatric Society 36:719-725, 1988.

54. Riddiough, M.A., Willems, J.S., Sanders, C.R., et al., "Factors Affecting the Use of Vac-cines: Considerations for Immunization Program Planners," Public Health Reports96(6):528-535, 1981.

55. Romm, F.J., Fletcher, S.W., and Hulka, B., "The Periodic Health Examination: Comparison

of Recommendations and Internists' Performance," Southern Medical Journal 74(3):265-271,

1981.

56. Rosenau, M.J., and Maxey, K.F., Maxcy-Rosenau Publig_Health and Preventive Medicine,

11 th ed., J.M. Last (ed.) (New York, NY: Appleton-Century-Crofts, 1980).

57. Rosenstock, I.M., "Historical Origins of the Health Belief Model," in The Health Belief

Model and Personal Health Behavior, M.H. Becker (ed.), Health Education Monograph, vol.

2, No. 4 (Baltimore, MD: Johns Hopkins University, winter 1974).

58. Rosenstock, I.M., "The Health Belief Model and Preventive Health Behavior," in TheHealth Belief Model and Personal Health Behav iQ r , M.H. Becker (ed.), Health EducationMonograph, vol. 2, No. 4 (Baltimore, MD: Johns Hopkins University, winter 1974).

59. Rundall, T.G., and Wheeler, J.R.C., "Factors Associated With Utilization of the Swine HuVaccination Program Among Senior Citizens in Tompkins County," Medical Care 17:191-

200, 1979.

60. Rundall, T.G., and Wheeler, J.R.C., "The Effect of Income on Use of Preventive Care: An

Evaluation of Alternative Explanations," Journal of Health and Social Behavior 20:397-406,1979.

61. Salkever, D.S., "Accessibility and the Demand for Preventive Care," Social Science andMedicine 10:469-475, 1976.

62. Satariano, W.A., Schwartz, A.G., and Swanson, G.M., "Screening for Cervical Cancer:Results From Several Intervention Strategies," in Issues in Cances Screening and Communi-cations (New York, NY: Alan R. Liss, Inc., 1982).

63. Sawyer, J.A., Earp, J., Fletcher, R.H., et al., "Pap Smears in Rural Black Women," from

the Department of Medicine and School of Public Health, University of North Carolina atChapel Hill, Chapel Hill, NC, unpublished manuscript.


64. Schwartz, S., University of Pennsylvania School of Medicine, personal communication,Philadelphia, PA, Mar. 16, 1988.

65. Selby, J.V., and G.B. Friedman, "Sigmoidoscopy in the Periodic Health Examination ofAsymptomatic Adults," report for the U.S. Preventive Services Task Force, Oct. 7, 1987.

66. Slesinser, D.P., Tessler, R.C., and Mechanic, D., "The Effects of Social Characteristics onthe Utilization of Preventive Medical Services in Contrasting Health Care Programs,"Medical Care 14(5):392-404, 1976.

67. Stu Its, B., "Preventive Care for the Elderly," Western Journal of Mediciac 141(6):832-845,1984.

68. Technology Reimbursement Reports, "Women Age 40 to 49 Should Receive a MammogramEvery One to Two Years, AMA Delegates Recommend," Technolmtv ReimbursementReports 4(28):6, 1988.

69. Teitelbaum, M.A., Weisman, C.S., Klassen, A.C., et al., "Pap Testing Intervals: SpecialtyDifferences in Physicians' Recommendations in Relation to Women's Pap Testing Behav-ior," Medical Car 26(6):607-618, 1988.

70. Thompson, R.S., Michnich, M.E., Gray, J., et al., "Maximizing Compliance With Hemoc-cult Screening for Colon Cancer in Clinical Practice," Medical Care 24(10):904-914, 1986.

71. U.S. Congress, Congressional Research Service, "Medicare: Risk Contracts With HealthMaintenance Organizations and Competitive Medical Plans," 88-138 EPW, 1988.

72. U.S. Department of Health, Education, and Welfare, National Center for Health Statistics,"Use of Selected Preventive Services U.S.--I973," Vital and Health Statistics, Series 10, No.110 (Washington, DC: U.S. Government Printing Office, March 1977).

73. U.S. Department of Health and Human Services, Health Care Financing Administration,Office of the Actuary, Division of National Cost Estimates, "National Health Expenditures,1986-2000," liga Care R v* w 8(4):I-36, 1987.

74. U.S. Department of Health and Human Services, National Center for Health Statistics, "TheNational Health Interview Survey Design, 1973-1984, and Procedures, 1975-1983," Vitaland Health Statistics, Serie., I, No. 18 (Washington, DC: U.S. Government Printing Office,August 1985).

75. U.S. Department of Health and Human Services, National Center for Health Statistics, "Useof Selected Preventive Services U.S.--1982," Vital and Health Statistics, Series 10, No. 157(Washington, DC: U.S. Government Printing Office, August 1986).

76. U.S. Department of Health and Human Services, National Center for Health Statistics,"Adult Health Practices in the United States and Canada," Vital and Health Statistics, Series5, No. 3 (Washington, DC: U.S. Government Printing Office, May 1988).

77. U.S. Department of Health and Human Services, National Institutes of Health, NationalCancer Institute, Division of Cancer Prevention and Control, Early Detection Branch,"Working Guidelines for Early Cancer Detection: Rationale and Supporting Evidence toDecrease Mortality" (Bethesda, MD: December 1987).

7 o

The Use of Preventive Services by the Elderly. 65

78. U.S. Department of Health and Human Services, National Institutes of Health, National

Heart, Lung and Blood Institute, National Cholesterol Education Program, "Report of the

Expert Panel on Detection, Evaluation and Treatment of High Blood Cholesterol in Adults"

(Bethesda, MD: October 1987).

79. U.S. Department of Health and Human Services, Preventive Services Task Force, "Recom-

mendations for Breast Cancer Screening," j.A.M.A. 257:2196, 1987.

80. U.S. Department of Health and Human Services, Public Health Service, "Screening for Cer-

vical and Breast Cancer -- Southeastern Kentucky," Morbidity and Mortality WeeklyReport 36(52):845-849, 1988.

81. U.S. Department of Health and Human Services, Public Health Service, "Use of Mam-mography for Breast Cancer Screening -- Rhode Island, 1987," Morbidity and Mortality

Weekly Report 37(23):357-360, 1988.

82. U.S. Department of Health and Human Services, Public Health Service, "ProvisionalEstimates From the National Health Interview Survey Supplement on Cancer Control --

United States, January-March 1987," Morbidity and Mortality Weekly Report 37(27):417-425, 1988.

83. U.S. Department of Health and Human Services, Public Health Service, Centers for Disease

Contrcl, Adult Immunizations: Recommendations of the Immunization Practices Com-

mittee, undated.

84. Valentv, C.M., Sobal, J., Muncie, H.L., et al., "Health Promotion: Physicians' Beliefs, Atti-tudes, and Practices," Am. J. Preventive Medicine 2(2):82-88, 1986.

85. Waldo, D.R., and Lazenby H.C., "Demographic Characteristics and Health Care Use and

Expenditures by the Aged in the U.S.: 1977-1984," Hgans_aril_a_gain v w 6(1):1-29, fall 1984.

86. Warnecke, R.B., Havlicek, P.L., and Manfredi, C., "Awareness and Use of Screening by

Older-Age Persons," in Prowectives on Prevention and Treatment of Cancer in the Elderly,R. Yancik (ed.) (New York, NY: Raven Press, 1983).

87. Wechsler, H., Levine, S., Idelson, R.K., et al., "The Physician's Role in Health Promotion

-- A Survey of Primary-Care Practitioners," N. Engl, J. Med. 308(2):97-100, 1983.

88. Williamson, P.S., Driscoll, C.E., Dvorak, L.D., et al., "Health Screening Examinations: ThePatient's Perspective," The Journal of Family Practice 27(2):187-192, 1988.

89. Winawer, S.J., Baldwin, M., Herbert, E, et al., "Screening Experience With Fecal Occult

Blood Testing as a Function of Age," in Prospectives on Prevention and Treatment of Can-

cer in the Elderly, R. Yancik (ed.) (New York, NY: Raven Press, 1983).

90. Windsor, R.A., Morris, J., "Consumer Participation and Payment for Health PromotionPrograms at an Urban Primary Care Center," The Alabama Journal of Medical Science$.

21(2):209-211, 1984.

9l. Woolhandler, S., and Himmelstein, D.U., "Reverse Targeting of Prever 've Care Due toLack of Health Insurance," J.A,M.A. 259(19):2872-2874, 1988.

9t,

66 is The Use of Preventive Services by the Elderly

92. Woo, B., Woo, B., Cook, E.F., et al., "Screening Procedures in the Asymptomatic Adult:Comparisons of Physicians' Recommendations, Patients' Desires, Published Guidelines, andActual Practice," j.A.M.A. 254(11):1480-1484, 1985.

0

For further information on this paper or others inOTA's series on Preventive Health Services UnderMedicare, contact: Judith Wagner, Health Program,Office of Technology Assessment, U.S. Congress,

Washington, D.C. 20510-8025

73

t************************** · 2014. 4. 9. · brigitte m. duffy. health program office of...

Documents