targeting cardiovascular medication adherence interventions · journal of the american pharmacists...

23
Journal of the American Pharmacists Association www.japha.org May/J un 2012 • 52:3 • JAPhA 381 REVIEWS Abstract Objectives: To determine whether adherence interventions should be adminis- tered to all medication takers or targeted to nonadherers. Data sources and study selection: Systematic search (Medline and Embase, 1966–2009) of randomized controlled trials of interventions to improve adherence to medications for preventing or treating cardiovascular disease or diabetes. Data extraction: Articles were classified as (1) broad interventions (targeted all medication takers), (2) focused interventions (targeted nonadherers), or (3) dynamic interventions (administered to all medication takers; real-time adherence informa- tion targets nonadherers as intervention proceeds). Cohen’s d effect sizes were cal- culated. Data synthesis: We identified 7,190 articles; 59 met inclusion criteria. Broad interventions were less likely (18%) to show medium or large effects compared with focused (25%) or dynamic (32%) interventions. Of the 33 dynamic interventions, 6 used externally generated adherence data to target nonadherers. Those with exter- nally generated data were less likely to have a medium or large effect (20% vs. 34.8% self-generated data). Conclusion: Adherence interventions targeting nonadherers are heterogeneous but may have advantages over broad interventions. Dynamic interventions show promise and require further study. Keywords: Medication adherence, cardiovascular disease, diabetes. J Am Pharm Assoc. 2012;52:381–397. doi: 10.1331/JAPhA.2012.10211 Targeting cardiovascular medication adherence interventions Sarah L. Cutrona, Niteesh K. Choudhry, Michael A. Fischer, Amber D. Servi, Margaret Stedman, Joshua N. Liberman, Troyen A. Brennan, and William H. Shrank Received November 23, 2010, and in revised form March 8, 2011. Accepted for publication April 6, 2011. Sarah L. Cutrona, MD, MPH, was a research associate, Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Wom- en’s Hospital, Harvard Medical School, Boston, at the time this study was conducted; she is cur- rently Assistant Professor of Medicine, Division of General Medicine/Primary Care, University of Massachusetts Medical School, Worcester. Niteesh K. Choudhry, MD, PhD, is Assistant Professor of Medicine; Michael A. Fischer, MD, MS, is Assistant Professor of Medicine; and Amber D. Servi, BA, is a research assistant, Di- vision of Pharmacoepidemiology and Pharma- coeconomics, Brigham and Women’s Hospital, Harvard Medical School, Boston. Margaret Stedman, PhD, MPH, is a postdoctoral research fellow, Orthopedics and Arthritis Center for Out- comes Research, Department of Orthopedics, Brigham and Women’s Hospital, Boston. Josh- ua N. Liberman, PhD, is Vice President, Strate- gic Research; and Troyen A. Brennan, MD, JD, is Chief Medical Officer and Executive Vice Pres- ident, CVS Caremark, Hunt Valley, MD. William H. Shrank, MD, MSHS, is Assistant Professor of Medicine, Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Wom- en’s Hospital, Harvard Medical School, Boston, and Associate Faculty, Center for American Political Studies, Faculty of Arts and Sciences, Harvard University, Boston. Correspondence: Sarah L. Cutrona, MD, MPH, University of Massachusetts Medical School, 377 Plantation St., Biotech 4, Suite 315, Worces- ter, MA 01605. Fax: 508-856-5024. E-mail: sarah. [email protected] Disclosure: Dr. Cutrona is supported by award no. KL2RR031981 from the National Center for Research Resources (NCRR). Dr. Stedman is supported by National Institutes of Health grant NRSA/T32 AR055885-03. Drs. Liberman and Brennan are employees of CVS Caremark and participated in manuscript preparation and re- view. Dr. Shrank is supported by a career devel- opment award from the National Heart, Lung, and Blood Institute (HL-090505). The authors declare no conflicts of interest or financial in- terests in any product or service mentioned in this article, including grants, employment, gifts, stock holdings, or honoraria. Funding: Research grant from CVS Caremark. All data analysis and evaluation occurred at Brigham and Women’s Hospital. CVS Caremark did not play a role in the design and conduct of the study; the collection, management, analy- sis, or interpretation of the data; or the prepa- ration, review, or approval of the manuscript. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NCRR or the National Insti- tutes of Health.

Upload: ngotruc

Post on 22-Jul-2018

214 views

Category:

Documents


0 download

TRANSCRIPT

J o u r n a l o f t h e A m e r i c a n P h a r m a c i s t s A s s o c i a t i o n www.japha.org M ay /J u n 2012 • 52:3 • JAPhA • 381

Reviews

Abstract

Objectives: To determine whether adherence interventions should be adminis-tered to all medication takers or targeted to nonadherers.

Data sources and study selection: Systematic search (Medline and Embase, 1966–2009) of randomized controlled trials of interventions to improve adherence to medications for preventing or treating cardiovascular disease or diabetes.

Data extraction: Articles were classified as (1) broad interventions (targeted all medication takers), (2) focused interventions (targeted nonadherers), or (3) dynamic interventions (administered to all medication takers; real-time adherence informa-tion targets nonadherers as intervention proceeds). Cohen’s d effect sizes were cal-culated.

Data synthesis: We identified 7,190 articles; 59 met inclusion criteria. Broad interventions were less likely (18%) to show medium or large effects compared with focused (25%) or dynamic (32%) interventions. Of the 33 dynamic interventions, 6 used externally generated adherence data to target nonadherers. Those with exter-nally generated data were less likely to have a medium or large effect (20% vs. 34.8% self-generated data).

Conclusion: Adherence interventions targeting nonadherers are heterogeneous but may have advantages over broad interventions. Dynamic interventions show promise and require further study.

Keywords: Medication adherence, cardiovascular disease, diabetes.J Am Pharm Assoc. 2012;52:381–397.

doi: 10.1331/JAPhA.2012.10211

Targeting cardiovascular medication adherence interventionssarah L. Cutrona, Niteesh K. Choudhry, Michael A. Fischer, Amber D. servi, Margaret stedman, Joshua N. Liberman, Troyen A. Brennan, and william H. shrank

Received November 23, 2010, and in revised form March 8, 2011. Accepted for publication April 6, 2011.

Sarah L. Cutrona, MD, MPH, was a research associate, Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Wom-en’s Hospital, Harvard Medical School, Boston, at the time this study was conducted; she is cur-rently Assistant Professor of Medicine, Division of General Medicine/Primary Care, University of Massachusetts Medical School, Worcester. Niteesh K. Choudhry, MD, PhD, is Assistant Professor of Medicine; Michael A. Fischer, MD, MS, is Assistant Professor of Medicine; and Amber D. Servi, BA, is a research assistant, Di-vision of Pharmacoepidemiology and Pharma-coeconomics, Brigham and Women’s Hospital, Harvard Medical School, Boston. Margaret Stedman, PhD, MPH, is a postdoctoral research fellow, Orthopedics and Arthritis Center for Out-comes Research, Department of Orthopedics, Brigham and Women’s Hospital, Boston. Josh-ua N. Liberman, PhD, is Vice President, Strate-gic Research; and Troyen A. Brennan, MD, JD, is Chief Medical Officer and Executive Vice Pres-ident, CVS Caremark, Hunt Valley, MD. William H. Shrank, MD, MSHS, is Assistant Professor of Medicine, Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Wom-en’s Hospital, Harvard Medical School, Boston, and Associate Faculty, Center for American Political Studies, Faculty of Arts and Sciences, Harvard University, Boston.

Correspondence: Sarah L. Cutrona, MD, MPH, University of Massachusetts Medical School, 377 Plantation St., Biotech 4, Suite 315, Worces-ter, MA 01605. Fax: 508-856-5024. E-mail: [email protected]

Disclosure: Dr. Cutrona is supported by award no. KL2RR031981 from the National Center for Research Resources (NCRR). Dr. Stedman is supported by National Institutes of Health grant NRSA/T32 AR055885-03. Drs. Liberman and Brennan are employees of CVS Caremark and participated in manuscript preparation and re-view. Dr. Shrank is supported by a career devel-opment award from the National Heart, Lung, and Blood Institute (HL-090505). The authors declare no conflicts of interest or financial in-terests in any product or service mentioned in this article, including grants, employment, gifts, stock holdings, or honoraria.

Funding: Research grant from CVS Caremark.

All data analysis and evaluation occurred at Brigham and Women’s Hospital. CVS Caremark did not play a role in the design and conduct of the study; the collection, management, analy-sis, or interpretation of the data; or the prepa-ration, review, or approval of the manuscript. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NCRR or the National Insti-tutes of Health.

J o u r n a l o f t h e A m e r i c a n P h a r m a c i s t s A s s o c i a t i o nwww.japha.org382 • JAPhA • 52:3 • M ay /J u n 2012

Reviews MEDICATION ADHERENCE INTERVENTIONS

Medications for preventing and treating cardiovascular disease can reduce morbidity and mortality, but nonad-herence limits their benefits. Nonadherence is widely

recognized as a major public health concern that contributes to patient morbidity, mortality, and health care costs.1,2 Recent estimates indicate that nonadherence to essential chronic med-ications may contribute as much as $290 billion in excess costs to the U.S. health care system annually.3 Improving adherence to therapy should be a priority for our health care system.

Despite broad recognition of the importance of medication adherence, little consensus exists about how to best change behavior and support appropriate use.4 Studies show that mul-tifactorial interventions tend to be more effective than simple ones4; however, the best manner in which to target these inter-ventions remains unknown. Specifically, whether interventions should be targeted to nonadherers only or to all medication tak-ers is unclear. Nonadherence may be improved more effectively if the patients most likely to benefit are targeted. Alternatively, waiting for nonadherence to occur may introduce missed op-portunities if the optimal moment for intervention has already passed.

ObjectiveWe conducted a systematic review of interventions to improve adherence to cardiovascular and diabetes medications, in order

to explore what is known about how to target interventions. We evaluated existing evidence regarding delivery of interventions (1) exclusively to nonadherent patients, (2) to all patients re-gardless of adherence behavior, or (3) to all patients but us-ing real-time adherence information during the intervention to identify nonadherers and thereby target resources.

MethodsA systematic search of peer-reviewed journals between 1966 and 2009 was performed using Medline and Embase. We lim-ited our search to randomized controlled trials. Our search terms related to the type of study (randomized controlled trial), adherence (i.e., adherence, compliance, medication adherence, treatment adherence), prescription drugs (i.e. drug, medica-tion, antihypertensive, antihyperlipidemic, hypoglycemic), and cardiovascular disease and diabetes (i.e., myocardial infarc-tion, coronary heart disease, heart failure, hypertension, dys-lipidemia, diabetes.) Articles with at least one search term in three of the main categories (study type and adherence and ei-ther drug or disease) met criteria for review. Search terms and parameters were adjusted for both databases (Medline and Em-base) while maintaining a common overall architecture. Search results then were screened for duplicate entries.

study selectionStudies were included if they reported results of randomized controlled trials studying interventions to improve adherence to medications used for preventing or treating cardiovascular disease or diabetes. Studies were limited to adult patients (age ≥18 years). We included only studies that reported long-term outpatient medication adherence. Studies were excluded if they described an intervention characterized by regimen simplifica-tion (either unit-of-use packaging or changes in dose frequency or formulation), as they could not be placed into one of our pre-specified study strata (described below), and previous studies have demonstrated their effectiveness.4 Non-English studies and those with a follow-up period of less than 24 weeks also were excluded.

study classificationAfter exclusions, 59 articles (Figure 1) were classified based on the target of the main intervention as (1) focused interventions (targeted exclusively to nonadherers), (2) broad interventions (targeted to the entire population of medication takers), or (3) dynamic interventions (administered to all medication takers but using real-time adherence information to identify and target nonadherers). To meet criteria for classification as a dynamic intervention, interventions were required to report that infor-mation was gathered on adherence and acted on before the con-clusion of the study in a way that would differentiate adherers from nonadherers (i.e., an adherence feedback loop). Ongoing measurements of clinical outcomes (e.g., blood pressure) were not considered substitutes for real-time measurements of ad-herence. However, we considered patient self-reported adher-ence to be an acceptable measure (as in a situation where a patient discussed adherence challenges with a pharmacist in the absence of calculations of a numerical adherence outcome).

At a GlanceSynopsis: Medline and Embase were searched for

articles on medication adherence interventions clas-sified as broad interventions (targeting all medication takers), focused (targeting nonadherers), or dynamic (administered to all medication takers, with real-time adherence information targeting nonadherers as in-tervention proceeded). Broad interventions were less likely (18%) to show medium or large effects com-pared with focused (25%) or dynamic (32%) interven-tions. Targeting nonadherent patients may lead to bet-ter adherence; however, data are limited and studies in the literature are highly heterogeneous.

Analysis: Focused interventions allow limited re-sources to be directed toward fewer, higher-risk pa-tients, and dynamic interventions share this advantage when the more costly portion of the intervention is re-served for identified nonadherers. However, attention must be paid to the method of identifying nonadher-ers. None of the focused interventions identified here used pharmacy claims data to identify nonadherence. Dynamic interventions were overwhelmingly depen-dent on self-generated adherence data (often requiring intensive interaction with a health provider), and very few used any form of external data. The accuracy, cost, and reproducibility of methods for identifying target populations must be a central consideration in future studies.

J o u r n a l o f t h e A m e r i c a n P h a r m a c i s t s A s s o c i a t i o n www.japha.org M ay /J u n 2012 • 52:3 • JAPhA • 383

MEDICATION ADHERENCE INTERVENTIONS Reviews

We further classified dynamic studies based on the type of adherence data used in the adherence feedback loop: (1) self-generated data alone or (2) external data (either alone or as a supplement to self-generated data). Self-generated data feed-back loops were those in which the intervention was tailored on the basis of patient self-reported nonadherence. For example, in a pharmacist counseling session in which the pharmacist does not have access to pharmacy records, a patient could re-port nonadherence and that would stimulate additional patient contact to support appropriate use. External data feedback loops were those in which information such as medication pos-session ratio derived from pharmacy records or other external sources was used to tailor the intervention. Authors were re-quired to explicitly state that external data were accessed in real time before a study was characterized as using an exter-nal data feedback loop. We did not consider patient-controlled sources such as medication diary cards to be external data but did consider a pill count conducted by someone other than the patient to meet this criterion.

This classification was often distinct from the way adher-ence was ultimately measured as the study outcome. For ex-ample, patients might describe themselves as nonadherent in a pharmacist counseling session and receive appropriate in-

terventions, while adherence outcomes were measured with claims data. In this scenario, we would classify the feedback loop as using self-generated adherence information.

Data extractionData were extracted by two investigators (S.L.C. and W.H.S.) with disagreements resolved by consensus. We assessed a number of variables related to the organization and outcome of studies, including study design, setting, characteristics of study population, number of participants, mean age (or age range) of participants, characteristics of intervention, meth-ods used to measure medication adherence, clinical outcomes, medication adherence outcomes, and source of funding. CIs are reported when available and P values when no CIs were available. The methodological quality of studies was assessed using the five-point Jadad scale.5 Because the overwhelming majority of adherence interventions identified in this study were not able to be double blinded, Jadad scores tended to be low. For this reason, we chose not to use a cut-off value and did not ultimately consider Jadad scores to be a useful means of ascertaining study quality. Jadad scores and funding sources for all studies are presented in Appendix 1 (electronic version of this article, available online at www.japha.org).

We identified randomized controlled trials in which means

Figure 1. Included and excluded articles

7,008 excluded

123 excluded

7,190 articles found 3,471 Embase 3,719 Medline

- 6,516 did not meet inclusion criteria for title and abstract - 492 citations overlapped in both databases

182 articles considered for inclusion

- 5 duplicate citation - 63 did not meet criteria: 19 excluded due to study design 19 excluded due to medication adherence outcome incomplete 5 excluded due to non-English 9 excluded due to wrong participants - 8 participants <18 years of age - 1 participant no cardiovascular disease 3 excluded due to different intervention (not designed to improve medication adherence) 8 excluded due to no empirical results reported - 29 duration <6 months - 26 used regimen simplification

59 included

J o u r n a l o f t h e A m e r i c a n P h a r m a c i s t s A s s o c i a t i o nwww.japha.org384 • JAPhA • 52:3 • M ay /J u n 2012

Reviews MEDICATION ADHERENCE INTERVENTIONS

and SDs for medication adherence outcomes were presented. A wide range of adherence outcome measures were observed, in-cluding binary (e.g., survey responses or predefined adherence cutoffs) and continuous measures (e.g., proportion of days cov-ered), making interpretation of absolute changes difficult. To compare studies with differing outcomes, we used Cohen’s d statistics, which can be calculated for outcomes that are either binary or continuous.6,7 The effect sizes (ESs) compare the dif-ference in effect between the study groups divided by the SD of this difference.8 When SDs were not reported, we derived them from the P value or t test statistic.

Using standard methods, we considered an ES less than 0.2 to be very small, 0.2 to less than 0.5 small, 0.5 to less than 0.8 medium, and 0.8 or greater large. We used this categoriza-tion to simplify interpretability for readers so that magnitude of effect is more intuitive. We assumed that the estimated Co-hen’s d statistics were independent of scale, sample size, and the SD of the outcome studied.

We attempted a fixed-effects meta-analysis in which ESs were pooled. We performed an analysis of statistical heteroge-neity in the intervention groups with at least 10 studies (broad intervention, dynamic self-generated intervention), and after finding high statistical heterogeneity in these groups (broad heterogeneity = 4.1, dynamic heterogeneity = 2.6), we removed outlying and influential studies and performed the analysis again. Heterogeneity measures remained high (broad hetero-geneity = 1.7, dynamic heterogeneity = 1.8). Heterogeneity statistics above 1.5 indicate high heterogeneity.9 Based on this finding and the clinical heterogeneity of the identified studies, we did not feel that presenting summary estimates was appro-priate.

ResultsFocused interventionsWe found only four focused interventions10–13 (Table 1), with mean participant ages ranging from 62 to 67 years. Adherence was measured differently in each study, and none of the studies made use of pharmacy claims data to determine inclusion cri-teria. One study showed a medium ES; the other three showed small ESs.

Haynes et al.10 identified 39 nonadherent hypertensive pa-tients from an initial group of 245 patients found to be hyper-tensive during workplace screening and advised home blood pressure self-checks, biweekly home visits by research assis-tants, and tailoring of the regimen. They found a medium ES by pill count at 6 months (0.73 [CI 0.07–1.39]).

Of an initial group of 79 patients with diabetes, Rosen et al.11 selected 33 patients on metformin with poor adherence measured with electronic pill bottles. These patients were ran-domized to 16 weeks of programmable electronic pill caps ver-sus nonprogrammable electronic caps. Although Rosen et al. described significantly improved adherence at 16 weeks based on medication possession ratio (80% intervention vs. 60% con-trol, P = 0.017; ES 0.43 [CI –0.27 to 1.14]), the study provides only a graphic representation of outcomes at 28 weeks (both groups declined). Saunders et al.12 identified a group of hyper-

tensive “infrequent attenders” to a medical clinic in Soweto, South Africa, where medications were dispensed at the clinic appointment. They identified 109 nonadherent patients, 72 of whom were included in the analysis. Appointment reminders, patient-retained records, and targeted home visits yield signifi-cant improvement in the intervention group (68% vs. 37% con-trol, P = 0.009) with a small calculated ES.

Taylor et al.13 randomized 81 patients at high risk for medi-cation events and analyzed 69 of them. Inclusion criteria in-cluded (but were not limited to) patient- or physician-reported nonadherence. Intervention patients received 20-minute coun-seling session at the pharmacy before physician office visits. Baseline data for the study indicated that 84.9% of the inter-vention patients and 88.9% of control patients had self-re-ported mean adherence scores between 80% and 100%. At 12 months, self-reported mean adherence scores showed a small ES difference between the intervention (100% adherence) and control (88.9%) groups. High initial adherence rates (or artifi-cially inflated self-reports of adherence) may render this study less representative of a focused intervention than the previous three.

Broad interventionsWe identified 25 broad interventions14–38 (Table 2) and cal-culated ESs in all but three cases.22,25,32 Mean ages of partici-pants ranged from 46 to 76 years. We found medium to large effects on adherence in 18.2% of studies; 68.2% had very small or small effects and 13.6% had no effect or negative effects. The vast majority of interventions (19 of 25 studies) examined hypertensive patients; others addressed patients with diabe-tes (2 studies), dyslipidemia (1), congestive heart failure (1), myocardial infarction (1), and both cardiac and noncardiac dis-eases (1).

A total of 13 studies described interventions dependent on involvement of a health professional: physician medi-ated,14,15,17,25,29,32,38 pharmacist mediated,20,22 or nurse medi-ated.19,22,23,31 Among these studies, ESs could be calculated for 10 studies, with 7 showing small or very small ESs, 1 a me-dium effect, and 2 negative or null effects. Yilmaz et al.,38 who showed an ES in the medium range, implemented a primarily educational intervention. The study involved comprehensive education on statins but allowed for consultation as needed with a physician and showed that intervention patients were more likely to adhere to statins (odds ratio [OR] 1.98) than control patients, based on self-report. The study with an ES of zero (Hamet et al.19) involved nurse counseling by phone com-bined with reminder letters and mailed education brochures, with adherence measured using a single-question self-report.

Six studies described the introduction of an electronic re-source including computerized decision aid,18 electronic moni-tor with reminder,16 or home blood pressure monitor.21,26,27,37 We calculated ESs for all six of these studies, and all but one showed very small or small effects on adherence. Johnson et al.21 conducted a 2 × 2 factorial study combining self-record-ed blood pressure values with monthly home visits, with a calculated ES of 1.51 (CI 0.96–2.05). We identified six stud-

J o u r n a l o f t h e A m e r i c a n P h a r m a c i s t s A s s o c i a t i o n www.japha.org M ay /J u n 2012 • 52:3 • JAPhA • 385

MEDICATION ADHERENCE INTERVENTIONS ReviewsTa

ble

1. Fo

cuse

d in

terv

entio

ns: T

arge

ted

to n

onad

here

rs

Aut

hor,

year

, site

Parti

cipa

nts;

dur

atio

naIn

terv

entio

nb

Met

hod

of ta

rget

ing

inte

rven

tion

grou

p no

nadh

erer

sA

dher

ence

mea

-su

res

Out

com

es, C

ohen

’s d

ES

(95%

CI)c

Hayn

es, 1

976,

Ca

nada

39 h

yper

tens

ive

patie

nts

iden

tified

as n

onad

here

nt

(pill

cou

nt <

80%

) afte

r 6

mon

ths o

f tre

atm

ent,

also

no

t at g

oal d

iast

olic

BP;

12

mon

ths

I: Ho

me

BP se

lf-ch

ecks

, m

edic

atio

n an

d BP

cha

rting

re

view

ed a

t 2-w

eek i

nter

vals

by

rese

arch

ass

ista

nt d

urin

g ho

me

visi

t, ta

ilorin

g of

regi

men

Posi

tive

rein

forc

emen

t for

per

fect

ad

here

nce,

reas

ons f

or n

onad

-he

renc

e so

ught

and

atte

mpt

s at

prob

lem

solv

ing

thro

ugh

tailo

ring

of

regi

men

for n

onad

here

nce

Prop

ortio

n of

pr

escr

ibed

pill

s re

mov

ed fr

om c

on-

tain

ers.

Mea

n ±

SD: I

, 65.

8% ±

8.2%

, w

ithin

-gro

up P

= 0.

001;

C:

43.2

% ±

10.1

%, w

ithin

-gro

up

P =

NS;

bet

wee

n-gr

oup

P =

0.02

5; E

S 0.

73 (0

.07–

1.39

)

Rose

n, 20

04,

Conn

ectic

ut

33 d

iabe

tes p

atie

nts o

n m

et-

form

in w

ith <

80%

adh

eren

ce

afte

r 4-w

eek b

asel

ine;

28

wee

ks

I: Cu

e-do

se tr

aini

ng:

give

n el

ectro

nic

pill c

aps p

ro-

gram

med

to b

eep,

inst

ruct

ion

on o

ther

cue

s

Adhe

renc

e da

ta g

iven

mon

thly

to

prov

ider

s; p

rovi

ders

con

tact

ed b

e-fo

re a

ppoi

ntm

ents

, urg

ed to

dis

cuss

ad

here

nce

data

Mea

n M

PR (d

oses

ne

eded

to b

e ta

ken

on ti

me)

; ele

ctro

nic

pill b

ottle

s

16 w

eeks

of i

nter

vent

ion:

I,

80%

; C, 6

0%, P

= 0.

017;

no

num

bers

giv

en fo

r 28 w

eeks

(g

raph

onl

y); E

S 0.

43 (–

0.27

to

1.14

)Sa

unde

rs, 1

991,

So

wet

o, S

. Afri

ca

Hype

rtens

ive

patie

nts d

i-vi

ded

into

new

ly tr

eate

d (1

15,

not s

how

n he

re) a

nd n

onad

-he

rent

(109

); 6 m

onth

s

I: Ap

poin

tmen

t rem

inde

rs

sent

; pat

ient

-ret

aine

d BP

and

m

edic

atio

n re

cord

s; m

edic

a-tio

ns o

btai

ned

at c

linic

visi

ts

Targ

eted

fiel

dwor

ker h

ome

visi

t fo

llow

-up

if no

resp

onse

to tw

o ca

ll-ba

ck le

tters

MPR

bas

ed o

n cl

inic

in

stru

ctio

ns; a

dher

-en

t if 8

0% c

onsu

med

Non

adhe

rent

: I, 6

8%; C

, 37

%, P

= 0.

009;

ES

0.39

(–0.

20

to 0.

97)

Tayl

or, 2

003,

Al

abam

a

81 p

atie

ntsd a

t hig

h ris

k for

m

edic

atio

n ev

ents

due

to

poly

phar

mac

y, n

onad

her-

ence

(per

med

ical

reco

rd,

self-

repo

rt, o

r refi

ll his

tory

), ot

her;

1 yea

r

I: Ph

arm

acy c

are:

20-m

inut

e in

-per

son

mee

tings

bef

ore

phys

icia

n of

fice

visi

ts, in

-cl

uded

med

ical

his

tory

, med

i-ca

tion

revi

ew, e

duca

tion,

and

m

onito

ring

Phar

mac

ist r

evie

w, a

dvic

e to

phy

si-

cian

(spo

ken

or vi

a m

edic

al re

cord

) an

d to

pat

ient

; sug

gest

ions

for n

on-

adhe

rers

incl

uded

sim

plifi

ed re

gi-

men

, pill

box

es, m

inim

izing

adv

erse

ef

fect

s, c

onso

lidat

ing

regi

men

s,

devi

sing

rem

inde

rs

Estim

ated

MPR

: sel

f-re

port;

mea

n ad

her-

ence

scor

e ca

lcu-

late

d fro

m sc

ores

for

each

med

icat

ion

12 m

onth

: I, 1

00%

; C, 8

8.9%

±

6.3%

, P =

0.11

5; E

S 0.

38

(–0.

11 to

0.87

)

Abbr

evia

tions

use

d: B

P, b

lood

pre

ssur

e; C

, con

trol g

roup

; ES,

effe

ct si

ze; I

, inte

rven

tion

grou

p; M

PR, m

edic

atio

n po

sses

sion

ratio

; NS,

non

sign

ifica

nt.

a Dura

tion

indi

cate

s tim

e un

til la

st fo

llow

-up

at w

hich

adh

eren

ce w

as m

easu

red.

b Co

ntro

l pat

ient

s rec

eive

d us

ual c

are

unle

ss o

ther

wis

e sp

ecifi

ed.

c 95%

CI u

nles

s oth

erw

ise

spec

ified

. Fo

r all s

tudi

es w

here

mea

ns (±

SD) f

or a

dher

ence

out

com

es w

ere

avai

labl

e, C

ohen

’s d

stat

istic

s wer

e ca

lcul

ated

. The

ESs

com

pare

the

diffe

renc

e in

effe

ct b

etw

een

the

stud

y gro

ups d

ivid

ed b

y the

SD

of th

is d

iffer

ence

. We

cons

ider

ed a

n ES

<0.

2 to

be ve

ry sm

all, 0

.2–0

.5 sm

all, 0

.5–0

.8 m

ediu

m, a

nd >

0.8 l

arge

. d Al

l pat

ient

s in

this

stud

y had

thre

e or

mor

e di

seas

es, w

ith h

yper

tens

ion,

dys

lipid

emia

, and

dia

bete

s bei

ng m

ost c

omm

on.

J o u r n a l o f t h e A m e r i c a n P h a r m a c i s t s A s s o c i a t i o nwww.japha.org386 • JAPhA • 52:3 • M ay /J u n 2012

Reviews MEDICATION ADHERENCE INTERVENTIONSTa

ble

2. B

road

inte

rven

tions

(tar

gete

d to

all m

edic

atio

n ta

kers

) des

igne

d to

impr

ove

adhe

renc

e to

car

diov

ascu

lar m

edic

atio

nsA

utho

r, ye

ar, s

itePa

rtici

pant

s; d

urat

iona

Inte

rven

tionb

Adh

eren

ce m

easu

res

Out

com

es, C

ohen

’s d

ESs

(95%

CI)c

Avan

zini, 2

002,

Ita

ly

1,77

1 tre

ated

hyp

erte

nsiv

e pa

tient

s; 1

year

Pa

tient

s fol

low

ed b

y phy

sici

ans w

ho w

rote

gu

idel

ines

for h

yper

tens

ion

man

agem

ent

% o

f pat

ient

s with

poo

r adh

er-

ence

; sel

f-rep

ort (

not d

efine

d fu

rther

)

Poor

adh

eren

ce: I

, 3.8

%; C

, 9.5

%, P

=

0.00

4; E

S 0.

20 (0

.13–

0.27

)

Birtw

hist

le, 2

004,

ur

ban

and

rura

l Ca

nada

614 h

yper

tens

ive

patie

nts;

36

mon

ths

3- vs

. 6-m

onth

phy

sici

an fo

llow

-up

M

oris

ky sc

ale

ques

tions

, incl

ud-

ing

ever

forg

et b

lood

pre

ssur

e pi

lls; s

elf-r

epor

t

I: 3 m

onth

, 30%

; I: 6

mon

th, 2

7%; d

iffer

-en

ce: 2

.96%

± 3.

92%

(mea

n ±

SE);

90%

CI

(–3.

48 to

9.41

); ES

0.24

(0.0

7–0.

42)

Chris

tens

en, 2

010,

Po

land

784 h

yper

tens

ive

patie

nts o

n te

lmis

arta

n; 1

year

(6 m

onth

s pe

r cro

ssov

er a

rm)

Elec

troni

c ad

here

nce

mon

itorin

g w

ith a

u-di

ovis

ual r

emin

der d

evic

e

Self-

repo

rt: n

umbe

r of d

ays i

n pa

st w

eek t

akin

g m

edic

atio

n as

pr

escr

ibed

; mea

n fo

r pop

ulat

ion

give

n as

per

cent

12 m

onth

s: I,

86.3

%; C

, 88.

4%, P

= 0.

812;

ES

0.06

(0.0

1–0.

12)

Düsi

ng, 2

009,

m

ultip

le si

tes i

n Ge

rman

y

206 h

yper

tens

ive

patie

nts

new

ly d

iagn

osed

or u

ntre

ated

fo

r 1 ye

ar; 3

4 wee

ks

Stru

ctur

ed p

hysi

cian

–pat

ient

inte

ract

ion,

pr

inte

d hy

perte

nsio

n in

form

atio

n, re

min

der

stic

kers

, hom

e tim

er, a

nd B

P m

easu

re-

men

ts

Daily

inta

ke o

f med

icat

ion

at c

or-

rect

tim

e; M

EMS

For o

vera

ll stu

dy d

urat

ion:

adh

eren

ce

for I

is g

reat

er th

an C

; P =

0.18

6; E

S 0.

08

(0.2

1–0.

37)

Emm

ett,

2005

, Br

isto

l, U.K

.

217 n

ewly

dia

gnos

ed h

yper

-te

nsiv

e pa

tient

s, p

rimar

y car

e pr

actic

es; 3

year

s

(1) I

n-pe

rson

adm

inis

tratio

n of

dec

isio

n ai

d on

hyp

erte

nsio

n, c

ardi

ovas

cula

r ris

k; (2

) vi

deo

and

leafl

et; (

3) d

ecis

ion

anal

ysis

and

vi

deo,

leafl

et vs

. usu

al c

are

Prop

ortio

n of

pat

ient

s who

repo

rt ta

king

all t

heir

med

icat

ions

Deci

sion

ana

lysi

s: 90

%, a

djus

ted

OR 1.

56

(0.4

9–4.

96),

P =

0.45

; vid

eo p

lus l

eafle

t: 94

%, a

djus

ted

OR 0.

53 (0

.15–

1.84

), P

= 0.

32; E

S 0.

15 (–

0.12

to 0.

42)

Ham

et, 2

003,

Ca

nada

4,86

4 pat

ient

s with

ess

entia

l hy

perte

nsio

n on

irbe

sarta

n;

12 m

onth

s

Beha

vior

al m

odifi

catio

n pr

ogra

m: p

hone

nu

rse

coun

sel, r

emin

der l

ette

rs, B

P di

arie

s,

mai

led

educ

atio

n br

ochu

res

“Are

you

taki

ng yo

ur A

vapr

o ev

ery d

ay?”

(no

= no

nadh

eren

t);

self-

repo

rt

% n

onad

here

nt p

atie

nts:

I, 25

.4%

(23.

7–27

.2);

C, 25

.5%

(23.

8–27

.3);

betw

een-

grou

p di

ffere

nce,

–0.

1% (–

2.6 t

o 2.

3), P

=

0.94

; ES

0 (–0

.61 t

o 0.

62)

Haw

kins

, 197

9,

Texa

s 20

0 hyp

erte

nsiv

e pa

tient

s on

a di

uret

ic w

ith/w

ithou

t met

hyl-

dopa

; 29 m

onth

s

Clin

ical

pha

rmac

ist m

anag

ed h

yper

tens

ion

in p

lace

of p

hysi

cian

%

of a

dher

ent p

atie

nts (

MPR

>0

.80 c

onsi

dere

d ad

here

nt);

phar

mac

y rec

ords

Diur

etic

onl

y: I,

60.5

%; C

, 52.

9%, P

≤ 0.

7;

diur

etic

plu

s met

hyld

opa:

I, 84

.6%

; C,

65.4

%, P

≤ 0.

2; E

S 0.

45 (–

0.07

to 0.

97)

John

son,

1978

, Ha

milt

on, C

anad

a

140 p

eopl

e w

ith p

ersi

s-te

ntly

ele

vate

d di

asto

lic B

P; 6

m

onth

s

2 × 2

fact

oria

l: (1)

self-

reco

rdin

g BP

, (2)

m

onth

ly h

ome

visi

ts, (

3) se

lf-re

cord

ing

and

hom

e vi

sits

; con

trol (

neith

er)

% a

dher

ence

est

imat

ed b

y co

mpa

ring

pills

on

hand

with

pr

escr

iptio

n re

cord

s; se

lf-re

port,

pi

ll cou

nt

Mea

n Δ

adhe

renc

e (±

SD):

I1, s

elf-

reco

rdin

g BP

, 11.

8% ±

4.5%

; I2,

mon

thly

ho

me

visi

ts: 2

.2%

± 5.

6%; I

3, se

lf-re

cord

-in

g +

hom

e vi

sits

, 10.

1% ±

4.9%

; C, 1

.0%

±

7.0%

, P =

NS;

ES

1.51

(0.9

6–2.

05)

Kirs

cht,

1981

, Te

cum

seh,

MI

417 p

atie

nts w

ith h

yper

ten-

sion

; 3 ye

ars

Assi

gned

to fo

ur in

terv

entio

ns in

a fa

ctor

ial

desi

gn, 3

× 2

× 3 ×

2: (1

) prin

ted

mes

sage

s,

(2) n

urse

pho

ne re

min

der a

nd re

info

rce-

men

t, (3

) sel

f-mon

itorin

g w

ith c

harts

, (4)

nu

rse

wor

ked

with

supp

ort p

erso

n

MPR

; pha

rmac

y rec

ords

; ave

r-ag

ed o

ver t

he se

t of h

yper

ten-

sion

med

icat

ions

Prin

ted

info

rmat

ion:

I, 0.

689;

C, 0

.684

, be

twee

n-gr

oup

P =

NS;

nur

se p

hone

co

ntac

t: I,

0.74

9; C

, 0.6

90, b

etw

een-

grou

p P

< 0.

05; s

elf-m

onito

ring:

cha

rts, 0

.683

; BP

, 0.6

65; C

, 0.6

65, b

etw

een-

grou

p P

= N

S; n

urse

supp

ort:

I, 0.

654;

C, 0

.545

, be

twee

n-gr

oup

P <

0.05

; ES

unab

le to

ca

lcul

ate

J o u r n a l o f t h e A m e r i c a n P h a r m a c i s t s A s s o c i a t i o n www.japha.org M ay /J u n 2012 • 52:3 • JAPhA • 387

MEDICATION ADHERENCE INTERVENTIONS Reviews

Tabl

e 2.

Bro

ad in

terv

entio

ns (t

arge

ted

to a

ll med

icat

ion

take

rs) d

esig

ned

to im

prov

e ad

here

nce

to c

ardi

ovas

cula

r med

icat

ions

Tabl

e 2 c

ontin

ued

Loga

n, 19

79, T

o-ro

nto,

Can

ada

457 h

yper

tens

ive

patie

nts s

e-le

cted

from

vario

us w

orks

ites;

6 m

onth

s

Wor

ksite

car

e: n

urse

wor

king

und

er p

hysi

-ci

an su

perv

isio

n m

anag

ed a

ll asp

ects

of

hype

rtens

ion

Adhe

renc

e: ≥

80%

of p

resc

ribed

m

edic

atio

ns w

ere

cons

umed

(p

ill c

ount

) and

pat

ient

cla

imed

to

be

taki

ng th

e m

edic

atio

n as

in

stru

cted

(sel

f-rep

ort)

% o

f adh

eren

t pat

ient

s: I,

67.6

%; C

, 49

.1%

, P <

0.00

5; E

S 0.

38 (0

.13–

0.63

)

Lope

z Cab

ezas

, 20

06, B

arce

lona

, Sp

ain

134 h

ospi

taliz

ed p

atie

nts w

ith

CHF;

1 ye

ar

Phar

mac

ist p

rogr

am: e

duca

tiona

l inte

rvie

w

with

pat

ient

and

car

egiv

er a

t dis

char

ge,

follo

w-u

p ph

one

calls

Adhe

rent

: 95–

100%

of p

resc

ribed

do

ses t

aken

%

of a

dher

ent p

atie

nts a

t 1 ye

ar: I

, 85.

0%;

C, 73

.9%

, P =

NS;

ES

0.28

(–0.

26 to

0.81

)

Man

n, 20

09, N

ew

York

15

0 dia

bete

s pat

ient

s fro

m p

ri-m

ary c

are

cent

ers;

6 m

onth

s In

-per

son

revi

ew o

f sta

tin ri

sks a

nd b

enefi

ts

with

prim

ary c

are

prov

ider

, usi

ng st

atin

de

cisi

on a

id a

t pro

vide

r’s d

iscr

etio

n

Adhe

renc

e as

sess

ed vi

a 8-

item

M

oris

ky a

dher

ence

scal

e; d

oes

not d

efine

“goo

d ad

here

nce”

80%

of p

artic

ipan

ts re

porte

d go

od a

d-he

renc

e in

bot

h gr

oups

; P =

NS;

ES

un-

able

to c

alcu

late

Mar

quez

-Con

tre-

ras,

2006

, Spa

in

250 h

yper

tens

ive

patie

nts

from

prim

ary c

are

cent

ers;

6

mon

ths

Hom

e au

tom

atic

BP

mon

itorin

g

MPR

(exp

ress

ed a

s %);

adhe

rent

is

>80

%; M

EMS

% o

f adh

eren

t pat

ient

s (m

ean

± SD

): I,

92 ±

14.2

); C,

74 ±

18.1

, P =

0.00

07; E

S 0.

29

(0.0

0–0.

58)

Meh

os, 2

000,

Co

lora

do

36 h

yper

tens

ive

patie

nts (

with

ph

arm

acis

t pro

vidi

ng d

irect

cl

inic

al se

rvic

es);

6 mon

ths

Hom

e BP

mon

itors

, dia

ry fo

r BP

and

mis

sed

dose

s; p

harm

acis

t eva

luat

ed B

P by

pho

ne

mon

thly

MPR

(mea

n), e

xpre

ssed

as %

; ph

arm

acy r

efill d

ata

I, 82

%; C

, 89%

, P =

0.29

; ES

0.35

(–0.

32 to

1.

02)

Mor

isky

, 198

5,

Balti

mor

e, M

D

290 h

yper

tens

ive

patie

nts;

18

mon

ths

Fam

ily su

ppor

t: ho

me

inte

rvie

w, t

rain

ing

with

fam

ily m

embe

r Se

lf-re

port:

scal

e 0–

4 (1 p

oint

per

ye

s; 4

= lo

w a

dher

ence

) In

terv

entio

n gr

oup

had

impr

oved

sc

ores

(0.8

76 vs

. 1.9

32, P

< 0.

01);

ES 0.

87

(0.6

3–1.

11)

Mul

lan,

2009

, M

inne

sota

85

dia

bete

s pat

ient

s (ra

ndom

-ize

d 40

clin

icia

ns);

6 mon

ths

Deci

sion

aid

revi

ewin

g 5 d

iabe

tes d

rugs

us

ed d

urin

g in

-per

son

clin

icia

n vi

sit v

s.

cont

rol: e

duca

tiona

l pam

phle

t

Prop

ortio

n of

day

s cov

ered

,%;

phar

mac

y rec

ords

%

(ran

ge):

I, 97

.5 (0

.0–1

00);

C, 10

0 (73

.9–

100)

; adj

uste

d m

ean

diffe

renc

e, –

8.88

(–

13.6

to –

4.14

); ES

–0.

09 (–

0.14

to –

0.04

)Po

wel

l, 199

5, m

id-

wes

t U.S

. 4,

246 p

atie

nts o

n be

nzap

ril,

met

opro

lol, s

imva

stat

in, o

r es

troge

n; 9

mon

ths

Mai

led

1 of 4

edu

catio

nal v

ideo

tape

s

MPR

; pha

rmac

y cla

ims

N

o si

gnifi

cant

bet

wee

n-gr

oup

diffe

renc

-es

in m

ean

MPR

s; E

S 0.

04 (–

0.02

to 0.

10)

Rudd

, 200

4, C

ali-

forn

ia

150 p

atie

nts o

n m

edic

atio

n fo

r hy

perte

nsio

n; 6

mon

ths

Phys

icia

n-di

rect

ed, n

urse

-man

aged

, al

gorit

hm-b

ased

hom

e hy

perte

nsio

n m

an-

agem

ent,

base

d on

self-

BP c

heck

s

Adhe

renc

e: a

vera

ge %

of d

ays

on w

hich

the

corr

ect n

umbe

r of

dose

s wer

e ta

ken;

ele

ctro

nic

pill

mon

itors

Mea

n ±

SD: I

, 80.

5 ± 23

.0%

; C, 6

9.2%

±

31.1

%, P

= 0.

03; E

S 0.

41 (0

.07–

0.76

)

Sack

ett,

1975

, Ha

milt

on, C

anad

a

230 C

anad

ian

stee

lwor

kers

w

ith h

yper

tens

ion

dete

cted

on

scre

enin

g; 6

mon

ths

AC h

yper

tens

ion

treat

ed b

y wor

ksite

phy

si-

cian

; AE:

pro

gram

on

hype

rtens

ion,

pill

-ta

king

rem

inde

rs; C

: no

AC, n

o AE

; I1:

AC,

no

AE; I

2: n

o AC

, AE;

I3, A

C, A

E

Adhe

renc

e: M

PR in

6th

mon

th;

pill c

ount

; adh

eren

t is M

PR ≥

0.8

% a

dher

ent a

t 6 m

onth

s: A

C, 54

%; n

o AC

: 51

%; A

E, 50

%; n

o AE

, 56%

; AE:

with

AC,

48

%; w

ithou

t AC,

53%

; no

AE: w

ith A

C,

62%

; with

out A

C, 48

%; n

onsi

gnifi

cant

di

ffere

nce;

ES

unab

le to

cal

cula

teSc

lar,

1991

, Del

a-w

are,

Tex

as, a

nd

Wis

cons

in

453 h

yper

tens

ive

patie

nts o

n at

enol

ol; 1

80 d

ays

On re

fill, e

duca

tiona

l mat

eria

l giv

en; p

hone

co

ntac

t, re

fill r

emin

der,

mai

lings

(4 a

rms,

ea

ch g

roup

div

ided

: pre

viou

sly t

reat

ed/

new

ly d

iagn

osed

)

MPR

: mul

tiplie

d th

e nu

mbe

r of

requ

este

d at

enol

ol re

fills

by 3

0 an

d di

vide

d by

180

MPR

(mea

n ±

SD):

prev

ious

ly tr

eate

d:

C, 0.

48 ±

0.06

; I, 0

.82 ±

0.04

; new

ly d

iag-

nose

d: C

, 0.5

2 ± 0.

06; I

, 0.9

3 ± 0.

05, P

<

0.05

; ES

7.42

(6.3

4–8.

51)

Smith

, 200

8, U

.S.

urba

n ce

nter

s

907 p

atie

nts a

t hos

pita

l di

scha

rge

post

-MI,

with

be

ta b

lock

er p

resc

riptio

ns; 9

m

onth

s

Two

mai

lings

to p

atie

nts,

PCP

s add

ress

ing

impo

rtanc

e of

bet

a bl

ocke

rs, g

uide

lines

% o

f pat

ient

s with

≥80

% o

f day

s co

vere

d in

the

9 mon

ths a

fter 1

st

mai

ling;

pha

rmac

y cla

ims a

nd

othe

r ele

ctro

nic

data

Trea

tmen

t pat

ient

s wer

e 17

% m

ore

likel

y to

hav

e 80

% o

f day

s cov

ered

(RR

1.17

[1

.02–

1.29

]); E

S 0.

09 (–

0.23

to 0.

42)

J o u r n a l o f t h e A m e r i c a n P h a r m a c i s t s A s s o c i a t i o nwww.japha.org388 • JAPhA • 52:3 • M ay /J u n 2012

Reviews MEDICATION ADHERENCE INTERVENTIONS

ies28,30,33–36 that assessed the impact of family support and edu-cation interventions, all of which had calculable ESs. As with previous groups of broad interventions, the majority were clas-sified as having very small or small effects (three studies). This is a particularly hard group about which to generalize because an additional two studies yielded large effects28,33 and one showed a negative effect.35

Although broad interventions did not incorporate ad-herence feedback loops, more than one-third of these stud-ies15–17,19,20,22,24,25,29,31,35 gathered adherence data at multiple time points. Several of these studies considered avoidance of a feedback loop to be an intentional part of the design. Chris-tensen et al.,16 who used electronic adherence monitoring with an audio-visual reminder device, explicitly stated that “treating physicians had no access to the questionnaire data during the study to avoid social desirability bias.” Rudd et al.,31 describ-ing an intervention in which project staff downloaded data from electronic medication monitors at 3 and 6 month clinic visits, stated that they “provided no feedback on drug adherence to patients, physicians, or nurse care managers.” For many oth-ers, the lack of a feedback loop appeared to be mediated by a separation between the person administering the intervention and those gathering data.

Dynamic interventionsWe found 30 dynamic interventions39–68 (Table 3), 28 of which had ESs that could be calculated.47,68 Mean ages of partici-pants ranged from 49 to 78 years. Of those with calculated ESs, 32.1% of studies had medium or large effects, 50.0% very small or small effects, and 17.9% no effect or negative effects. When examining the percentage of studies yielding medium or large ESs across all groups, dynamic interventions (32.1%) compared favorably with broad (18.2%) or focused (25%) in-terventions.

The majority of interventions (12 of 30) examined hyper-tensive patients; others addressed patients with diabetes (3), dyslipidemia (5), congestive heart failure (6), cardiac disease (3), and mixed diseases (1).

Use of self-generated data in adherence feedback loop. We identified 24 articles (Table 3) that described adher-ence feedback loops based on self-generated data, and all but 1 had calculable ESs. Of those with calculated ESs, 34.8% yield-ed medium or large effects, 43.5% very small or small effects, and 21.7% negative or no effects.

All eight studies39–41,43,53,54,57,58 found to have a medium or large effect were dependent on the involvement of health pro-fessionals, and in all but one case,39 the professional was a pharmacist. Antonicelli et al.39 studied the effect of home tele-monitoring managed by a specialized heart failure team and did not specify the training of the person making the phone calls. Adherence feedback loops were initiated in these studies when a pharmacist (or in the case of Antonicelli et al., a caller whose training was not specified) inquired about medication adherence either in person or by phone. This often occurred multiple times during the intervention. When a patient self-identified as nonadherent, an individualized intervention was Ta

ble

2. B

road

inte

rven

tions

(tar

gete

d to

all m

edic

atio

n ta

kers

) des

igne

d to

impr

ove

adhe

renc

e to

car

diov

ascu

lar m

edic

atio

ns

Tabl

e 2 c

ontin

ued

Stew

art,

2005

, Jo

hann

esbu

rg, S

. Af

rica

83 h

yper

tens

ive

patie

nts,

ma-

jorit

y ind

igen

t; 36

wee

ks

Supp

ort o

f phy

siot

hera

pist

and

fam

ily m

em-

ber b

y pho

ne

Self-

desc

ribed

as a

dher

ent t

o m

edic

atio

ns

I, 82

.4%

; C, 8

6.7%

, P =

0.56

; ES

–0.1

2 (–

0.85

to 0.

61)

Taka

la, 1

983,

so

uthw

est F

inla

nd

147 u

ntre

ated

hyp

erte

nsiv

e pa

tient

s; 2

year

s M

aile

d in

form

atio

n on

hyp

erte

nsio

n

2 yea

rs a

fter i

nter

vent

ion,

ask

ed

if “s

till u

nder

trea

tmen

t afte

r 2

year

s”

Adhe

rent

: I, 9

0%; C

, 79%

, P =

NS;

ES

0.28

(–

0.14

to 0.

69)

van

Onze

noor

t, 20

09, t

he N

ethe

r-la

nds

228 h

yper

tens

ive

patie

nts;

1

year

Ho

me

self-

BP m

easu

rem

ent

%

of d

ays w

ith c

orre

ct d

osin

g;

MEM

S M

edia

n (IQ

R): I

, 92.

3% (8

6.9–

94.4

); C,

90

.9%

(82.

9–93

.7),

P =

0.04

3; E

S 0.

07

(0.0

4–0.

18)

Yilm

az, 2

005,

An-

kara

, Tur

key

202 p

atie

nts o

n st

atin

for s

ec-

onda

ry p

reve

ntio

n; 15

mon

ths

Educ

atio

n in

clud

ing

phys

icia

n co

nver

satio

n on

stat

ins

Odds

of b

eing

on

cont

inu-

ous s

tatin

(afte

r med

ian

of 15

m

onth

s); s

elf-r

epor

t

I, m

ore

likel

y to

be o

n co

ntin

uous

stat

in,

OR 1.

977 (

1.12

7–3.

468)

; ES

0.53

(0.2

5, 0.

82)

Abbr

evia

tions

use

d: A

C, a

ugm

ente

d co

nven

ienc

e; A

E, a

dditi

onal

edu

catio

n; B

P, b

lood

pre

ssur

e; C

, con

trol g

roup

; CHF

, con

gest

ive

hear

t fai

lure

; ES,

effe

ct si

ze; I

, inte

rven

tion

grou

p; IQ

R, in

terq

uarti

le ra

nge;

MEM

S, m

edic

atio

n ev

ent m

onito

ring

syst

em; M

I, m

yoca

rdia

l infa

rctio

n; M

PR, m

edic

atio

n po

sses

sion

ratio

; NS,

non

sign

ifica

nt; O

R, o

dds r

atio

; PCP

, prim

ary c

are

prov

ider

; RR,

rate

ratio

. a Du

ratio

n in

dica

tes t

ime

until

last

follo

w-u

p at

whi

ch a

dher

ence

was

mea

sure

d.

b Cont

rol p

atie

nts r

ecei

ved

usua

l car

e un

less

oth

erw

ise

spec

ified

c 95

% C

I unl

ess o

ther

wis

e sp

ecifi

ed.

For a

ll stu

dies

whe

re m

eans

(±SD

) for

adh

eren

ce o

utco

mes

wer

e av

aila

ble,

Coh

en’s

d st

atis

tics w

ere

calc

ulat

ed. T

he E

Ss c

ompa

re th

e di

ffere

nce

in e

ffect

bet

wee

n th

e st

udy g

roup

s div

ided

by t

he S

D of

this

diff

eren

ce. W

e co

n-si

dere

d an

ES

<0.2

to b

e ve

ry sm

all, 0

.2–0

.5 sm

all, 0

.5–0

.8 m

ediu

m, a

nd >

0.8 l

arge

.

J o u r n a l o f t h e A m e r i c a n P h a r m a c i s t s A s s o c i a t i o n www.japha.org M ay /J u n 2012 • 52:3 • JAPhA • 389

MEDICATION ADHERENCE INTERVENTIONS ReviewsTa

ble

3. D

ynam

ica m

edic

atio

n ad

here

nce

inte

rven

tions

: Cla

ssifi

ed b

y typ

e of

adh

eren

ce d

ata

used

in a

dher

ence

feed

back

loop

b

Aut

hor,

year

, site

Pa

rtici

pant

s,

dura

tionc

Inte

rven

tiond

Met

hod

of ta

rget

ing

inte

rven

tion

grou

p no

nadh

erer

sA

dher

ence

mea

sure

se O

utco

mes

, Coh

en’s

d E

Ss

(95%

CI)f

Self-

gene

rate

d ad

-he

renc

e da

ta o

nly

Anto

nice

lli, 2

008,

Ita

ly

57 h

ospi

taliz

ed C

HF

patie

nts,

age

>70

ye

ars;

12 m

onth

s

I: ho

me

tele

mon

itorin

g m

anag

ed

by sp

ecia

lized

CHF

team

Wee

kly c

alls

by C

HF te

am c

ol-

lect

ing

info

rmat

ion

on sy

mpt

oms,

ad

here

nce.

Reg

imen

adj

ustm

ent

base

d on

feed

back

% a

dher

ent p

atie

nts (

no

furth

er d

efini

tion)

I: 91

%; C

: 46%

, P <

0.03

; ES

1.12

(0.5

2–1.

69)

Blen

kins

opp,

2000

, U.

K.

282 h

yper

tens

ive

patie

nts f

rom

com

-m

unity

pha

rmac

ies;

6 m

onth

s

I: Ph

arm

acis

t cou

nsel

: stru

ctur

ed

ques

tions

, med

icat

ion

advi

ce,

hype

rtens

ion

teac

hing

, eve

ry 2

m

onth

s

Phar

mac

ist g

ave

advi

ce (v

erba

l or

writ

ten)

; pha

rmac

ist m

ight

re

fer t

o GP

or s

peak

dire

ctly

with

GP

% a

dher

ent p

atie

nts;

mod

i-fie

d ve

rsio

n of

Hor

ne’s

m

edic

atio

n ad

here

nce

repo

rt sc

ale

used

I: 62

.9%

; C: 5

0.0%

, P <

0.05

; ES

0.56

(0.2

9–0.

84)

Bouv

y, 20

03, t

he

Net

herla

nds

152 C

HF p

atie

nts,

in

patie

nt a

nd o

utpa

-tie

nt; 6

mon

ths

I: Co

mm

unity

pha

rmac

ist:

stru

c-tu

red

inte

rvie

w u

sing

com

pute

r-ize

d m

edic

atio

n re

cord

at i

nitia

l en

coun

ter,

disc

usse

d m

edic

a-tio

ns, n

onad

here

nce

Mon

thly

pha

rmac

ist c

onta

ct; d

is-

cuss

ion

of re

ason

s for

non

adhe

r-en

ce a

nd re

info

rced

adh

eren

ce

Adhe

renc

e ba

sed

on %

of

day

s with

out o

peni

ng

pill b

ottle

; non

adhe

renc

e de

fined

as <

80%

of d

ays;

M

EMS

% n

onad

here

nt p

atie

nts;

I:

0%; C

: 14%

; RR

0.5;

CI

0.4–

0.6;

ES

0.57

(0.1

4–1.

00)

Edw

orth

y, 20

07, C

al-

gary

, Can

ada

2,64

3 car

diac

pa-

tient

s afte

r hos

pita

l-iza

tion;

19 m

onth

s

I: In

-hos

pita

l indi

vidu

al a

nd g

roup

co

unse

ling

on m

edic

atio

ns a

nd

med

ical

con

ditio

ns; v

ideo

s,

prin

ted

mat

eria

ls; d

evel

oped

long

-te

rm m

edic

atio

n pl

ans;

follo

w-u

p co

ntac

t by p

harm

acis

t onc

e an

d m

onth

ly b

y nur

se

Nur

ses a

nd p

harm

acis

ts id

enti-

fied

and

addr

esse

d m

edic

atio

n pr

oble

ms;

com

mun

ity p

harm

a-ci

sts c

ouns

eled

inte

rven

tion

patie

nts

% o

f adh

eren

t pat

ient

s (no

t fu

rther

defi

ned)

; sel

f-rep

ort

data

on

med

icat

ion

use

colle

cted

by:

I: n

urse

s; C

: no

nmed

ical

staf

f

Beta

-blo

cker

s: I:

89%

; C:

80%

, P <

0.01

; lipi

d-lo

wer

-in

g ag

ents

: I: 8

3%; C

: 78%

, P

< 0.

05; E

S 0.

04 (–

0.07

to 0.

14)

Faul

kner

, 200

0,

Omah

a, N

E

30 p

atie

nts p

ost-

CABG

, PTC

A, o

r bo

th (7

–30 d

ays)

; 2

year

s

I: W

eekl

y pha

rmac

ist c

alls

(12

wee

ks);

all r

ecei

ved

lova

stat

in

daily

and

col

estip

ol tw

ice

daily

Wee

kly p

hone

cal

l to

chec

k on

adhe

renc

e, m

edic

atio

n is

sues

, an

d co

sts;

ask

ed a

bout

spec

ific

reas

ons f

or n

onad

here

nce.

Patie

nts r

etur

ning

mor

e th

an 20

% o

f pre

scrib

ed

pills

and

thos

e fa

iling

to

fill 8

0% o

r mor

e of

scrip

ts

at 1

and

2 yea

rs w

ere

cons

ider

ed n

onad

here

nt;

phar

mac

y rec

ords

% a

dher

ent p

atie

nts:

lo-

vast

atin

: I: 1

year

, 67%

; 2

year

s: 60

%; C

: 1 ye

ar: 3

3%;

2 yea

rs: 2

7%, P

< 0.

05 fo

r all

valu

es (c

oles

tipol

find

ings

si

mila

r, no

t sho

wn

here

); ES

0.54

(–0.

19 to

1.27

)Fr

iedm

an, 1

996,

Bo

ston

299 h

yper

tens

ive

patie

nts;

6 m

onth

s

I: In

tera

ctiv

e co

mpu

ter-

base

d ho

me

mon

itorin

g. P

atie

nt se

lf-BP

ch

ecks

, wee

kly c

alls

to c

ouns

el

on a

dher

ence

Auto

mat

ed p

hone

cou

nsel

ing

on

adhe

renc

e; d

ata

colle

cted

wee

kly

and

trans

mitt

ed to

PCP

MPR

(exp

ress

ed a

s per

-ce

nt);

hom

e pi

ll cou

nt;

adhe

rers

: MPR

≥80

%

Mea

n Δ

adhe

renc

e, u

nad-

just

ed: I

: +2.

4%; C

: –0.

4%,

P =

0.29

; adj

uste

d fo

r bas

e-lin

e ad

here

nce:

I: 17

.7%

; C:

11.7

%, P

= 0.

03; E

S 0.

13

(–0.

12 to

0.37

)Gu

thrie

, 200

1, O

hio

13,1

00 p

atie

nts w

ith

elev

ated

tota

l cho

-le

ster

ol; 6

mon

ths

I: Po

stal

, pho

ne re

min

ders

abo

ut

coro

nary

risk

redu

ctio

n an

d m

edi-

catio

n ad

here

nce

Dire

ctly

repo

rted

prav

asta

tin a

d-he

renc

e to

thei

r phy

sici

ans (

alon

g w

ith lif

esty

le c

hang

es, a

dver

se

even

ts) a

t 3 m

onth

s

Taki

ng p

rava

stat

in a

s pr

escr

ibed

per

self-

repo

rt:

yes/

no

Taki

ng a

s pre

scrib

ed: I

: 79

.7%

; C: 7

7.4%

; aut

hors

co

nclu

de “n

o m

eani

ngfu

l di

ffere

nce”

; ES

0.06

(–0.

02

to 0.

13)

J o u r n a l o f t h e A m e r i c a n P h a r m a c i s t s A s s o c i a t i o nwww.japha.org390 • JAPhA • 52:3 • M ay /J u n 2012

Reviews MEDICATION ADHERENCE INTERVENTIONS

Tabl

e 3.

Dyn

amic

a med

icat

ion

adhe

renc

e in

terv

entio

ns: C

lass

ified

by t

ype

of a

dher

ence

dat

a us

ed in

adh

eren

ce fe

edba

ck lo

opb

Tabl

e 3 c

ontin

ued

Hunt

, 200

8, O

rego

n

463 u

ncon

trolle

d hy

perte

nsiv

e pa

-tie

nts;

12 m

onth

s

I: Co

mm

unity

pha

rmac

ists

man

-ag

ed h

yper

tens

ion

in P

CP o

ffice

(h

ad P

CP in

put)

Indi

vidu

alize

d ph

arm

acis

t cou

n-se

ling

incl

udin

g id

entifi

catio

n of

ad

here

nce

barr

iers

; fol

low

-up

inte

rval

varie

d

% w

ith h

igh

adhe

renc

e,

cate

goriz

ed b

y Mor

isky

sc

ale;

self-

repo

rt

I: 67

%; C

: 69%

, P =

0.77

; ch

ange

in I:

P =

0.08

; ch

ange

in C

: P =

0.52

; ES

–0.0

4 (–0

.29 t

o 0.

20)

Jaffr

ay, 2

007,

U.K

.

1,49

3 CAD

pat

ient

s fro

m p

rimar

y car

e or

gani

zatio

ns; 1

2 m

onth

s

Com

mun

ity p

harm

acis

t ass

esse

d th

erap

y, a

dher

ence

, life

styl

e, so

-ci

al su

ppor

t

All p

artic

ipan

ts g

ot in

itial

con

sult;

fu

rther

con

sults

at p

harm

acis

t di

scre

tion

base

d on

ass

esse

d ad

here

nce

and

othe

r nee

ds

Adhe

renc

e sc

ore

(12–

60)

base

d on

12 q

uest

ions

; se

lf-re

port

I: 59

(IQR

57–6

0); C

: 59

(57–

60);

OR 1.

0 (95

% C

I 0.

61–1

.65)

, P =

0.99

; ES:

un

able

to c

alcu

late

Kran

tz, 2

008,

Den

ver,

CO

64 C

HF p

atie

nts w

ith

ejec

tion

fract

ion

<40%

; 6 m

onth

s

Preh

ospi

tal d

isch

arge

car

vedi

lol

initi

atio

n an

d nu

rse

coun

selin

g w

ith o

utpa

tient

nur

se m

anag

e-m

ent

2 wee

ks a

fter d

isch

arge

, met

w

ith n

urse

man

ager

, the

n ev

ery 2

w

eeks

unt

il dee

med

stab

le; c

oun-

selin

g, in

clud

ing

on a

dher

ence

Beta

-blo

cker

util

izatio

n;

on m

edic

atio

n (y

es/n

o); p

ill

coun

t

Beta

-blo

cker

util

izatio

n:

at d

isch

arge

: C: 9

.4%

; I:

96.9

%; 6

mon

ths:

C: 4

7.8%

; I:

96.2

%; u

tiliza

tion

sig-

nific

antly

hig

her f

or I a

t all

time

perio

ds (P

< 0.

001)

; ES

0.30

(–0.

29 to

0.89

)Lo

gan,

1983

, Tor

onto

, Ca

nada

194 h

yper

tens

ive

patie

nts;

1 ye

ar

I: W

orks

ite c

are

by p

hysi

cian

plu

s nu

rse

Non

adhe

rent

pat

ient

s cou

nsel

ed

on m

edic

atio

n di

ary,

tailo

red

regi

-m

en, h

ome

BP; in

crea

sed

visi

t fre

quen

cy; n

urse

hom

e ph

one

call

for m

isse

d vi

sits

% a

dher

ent p

atie

nts (

≥80%

of

pre

scrib

ed m

edic

atio

n ta

ken)

; hom

e pi

ll cou

nt

I: 55

.4%

; C: 5

5.7%

, P =

NS;

ES

–0.

01 (–

0.31

to 0.

29)

Odeg

ard,

2005

, Se-

attle

, WA

77 p

artic

ipan

ts w

ith

A1C

≥9%

taki

ng

diab

etes

med

ica-

tion;

12 m

onth

s

I: Pr

imar

y car

e ph

arm

acis

t de-

velo

ped

care

pla

n, in

-per

son

or

phon

e m

eetin

gs (w

eekl

y, th

en

mon

thly

)

Adhe

renc

e as

sess

ed a

t bas

elin

e,

6 mon

ths,

and

12 m

onth

s; p

harm

a-ci

st g

ave

regu

lar a

dvic

e to

pat

ient

an

d pr

ovid

er w

here

nee

ded;

ad-

dres

sed

diab

etes

car

e, m

edic

a-tio

n pr

oble

ms

Adhe

renc

e ba

sed

on 2

qu

estio

ns: “

Do yo

u ev

er

find

it di

fficu

lt to

rem

em-

ber t

o ta

ke (m

edic

atio

n na

me)

?” If

yes,

“How

man

y tim

es o

ver t

he la

st 2

wee

ks

have

you

mis

sed

a do

se?”

C sh

owed

bet

ter a

dher

-en

ce th

an I t

hrou

ghou

t st

udy (

P =

0.00

3); E

S –0

.73

(–1.

25 to

–0.

21)

Oged

egbe

, 200

8,

New

Yor

k

190 h

yper

tens

ive

patie

nts,

bla

ck

race

/eth

nici

ty, m

ost

wom

en; 1

2 mon

ths

I: M

otiv

atio

nal in

terv

iew

ing

with

pa

tient

-cen

tere

d co

unse

ling

by

rese

arch

ass

ista

nts w

ho e

licite

d ad

here

nce

barr

iers

, dis

cuss

ed

solu

tions

Asse

ssm

ents

eve

ry 3

mon

ths

base

d on

pat

ient

verb

ally

dis

-cu

ssin

g ad

here

nce

barr

iers

; M

EMS

data

are

dow

nloa

ded

but

not u

sed

at th

ose

visi

ts

MPR

, exp

ress

ed a

s %;

MEM

S

I: 60

%; C

: 47%

, P =

0.05

4;

inte

nt-to

-trea

t ana

lysi

s sh

owed

mod

el-p

redi

cted

ra

tes:

I: 57

%; C

: 43%

, P =

0.

027;

ES

0.13

(–0.

01 to

0.27

)Pi

ette

, 200

0, C

ali-

forn

ia

280 d

iabe

tes p

a-tie

nts o

n hy

pogl

y-ce

mic

med

icat

ions

; in

clud

ed S

pani

sh-

spea

king

pat

ient

s;

12 m

onth

s

I: Bi

wee

kly a

utom

ated

ass

ess-

men

t/edu

catio

n ca

lls: h

iera

rchi

-ca

lly st

ruct

ured

mes

sage

s with

ta

rget

ed n

urse

follo

w-u

p ca

lls

In a

utom

ated

cal

ls, p

atie

nts g

iven

po

sitiv

e fe

edba

ck a

nd re

info

rce-

men

t; no

nadh

erer

s wer

e as

ked

abou

t bar

riers

, giv

en a

utom

ated

ad

vice

with

targ

eted

, prio

ritize

d nu

rse

phon

e fo

llow

-up;

nur

se

calle

d th

ose

who

rare

ly re

spon

d-ed

to a

utom

ated

cal

ls

Abbr

evia

ted

Mor

isky

in-

dex;

pat

ient

s con

side

red

nona

dher

ent i

f the

y som

e-tim

es fo

rgot

or s

topp

ed

med

icat

ion;

pho

ne in

ter-

view

s, se

lf-re

port

I: “S

ubst

antia

lly le

ss lik

ely”

to

repo

rt ad

here

nce

prob

-le

ms (

P =

0.00

3); E

S 0.

38

(0.1

2–0.

63)

J o u r n a l o f t h e A m e r i c a n P h a r m a c i s t s A s s o c i a t i o n www.japha.org M ay /J u n 2012 • 52:3 • JAPhA • 391

MEDICATION ADHERENCE INTERVENTIONS Reviews

Tabl

e 3.

Dyn

amic

a med

icat

ion

adhe

renc

e in

terv

entio

ns: C

lass

ified

by t

ype

of a

dher

ence

dat

a us

ed in

adh

eren

ce fe

edba

ck lo

opb

Tabl

e 3 c

ontin

ued

Plan

as, 2

009,

Okl

a-ho

ma

52 d

iabe

tes a

nd

hype

rtens

ive

man

-ag

ed c

are

patie

nts;

9 m

onth

s

I: M

onth

ly in

-per

son

com

mun

ity

phar

mac

ist c

ouns

elin

g; id

entifi

ed

and

addr

esse

d m

edic

atio

n pr

ob-

lem

s, lif

esty

le c

ouns

elin

g

Phar

mac

ist a

sses

ses a

nd e

n-co

urag

es a

dher

ence

, com

mun

i-ca

tes p

robl

ems t

o PC

P vi

a no

te

Used

MPR

bas

ed o

n pr

escr

iptio

n cl

aim

s dat

a fro

m m

anag

ed c

are

or-

gani

zatio

n; o

nly i

nclu

ded

pres

crip

tions

with

≥3

cons

ecut

ive

refil

ls in

the

9 mon

ths b

efor

e or

afte

r ba

selin

e vi

sit

Mea

n ad

here

nce

(%):

I: 87

.5 (C

I 82.

1–93

.0);

C: 78

.8

(CI 6

9.7–

87.9

); ES

0.54

(0

.13–

1.21

)

Sadi

k, 20

05, A

l-Ain

, Un

ited

Arab

Em

irate

s

221 C

HF p

atie

nts;

12

mon

ths

I: St

ruct

ured

pha

rmac

ist c

ouns

el

(in c

linic

or h

ospi

tal);

CHF

and

m

edic

atio

n ed

ucat

ion,

boo

klet

Phar

mac

ist v

isits

with

cou

nsel

ing

ever

y 3 m

onth

s; se

lf-m

onito

ring:

1-

mon

th c

ard

(pha

rmac

ist r

e-vi

ewed

regu

larly

, tol

d to

brin

g to

PC

P); p

harm

acis

t spo

ke w

ith P

CP

as n

eede

d

“Pat

ient

self-

repo

rt on

m

issi

ng d

ose

or ta

k-in

g ex

tra d

oses

with

out

med

ical

adv

ice”

; no

furth

er

defin

ition

Num

ber o

f adh

eren

t pa-

tient

s: I:

85; C

: 35,

P <

0.05

; ES

1.26

(0.9

9–1.

54)

Sche

ctm

an, 1

994,

M

ilwau

kee,

WI

162 p

atie

nts w

ith

dysl

ipid

emia

; 6

mon

ths

I: W

eekl

y pho

ne c

ouns

el in

1st

mon

th o

f the

rapy

by m

edic

al

assi

stan

t; ea

ch g

roup

als

o ra

n-do

mize

d to

nia

cin

vs. b

ile a

cid

sequ

estra

nts

Info

rmat

ion

on m

edic

atio

n ad

her-

ence

and

pro

blem

s gat

here

d by

ph

one

and

at 2-

mon

th c

linic

visi

ts;

med

ical

ass

ista

nt o

ffere

d ad

vice

on

adv

erse

effe

cts,

arr

ange

d ph

arm

acis

t or p

hysi

cian

pho

ne

cont

act a

s nee

ded

MPR

; pha

rmac

y cla

ims

% a

dher

ence

(mea

n ±

SD):

bile

aci

d se

ques

trant

s (m

ean

± SD

): I:

88 ±

SD

4;

C: 82

± 4,

P =

0.32

; nia

cin:

I:

90 ±

2; C

: 84 ±

3, P

= 0.

07; E

S 0.

41 (–

0.05

to 0.

86)

Schr

oede

r, 20

05,

Avon

, U.K

.

245 h

yper

tens

ive

patie

nts;

6 m

onth

s

I: N

urse

-led

adhe

renc

e su

ppor

t se

ssio

ns

Rein

forc

emen

t con

sult

2 mon

ths

afte

r ran

dom

izatio

n: n

urse

spok

e to

pat

ient

s abo

ut a

dher

ence

ch

alle

nges

but

had

no

acce

ss to

M

EMS

data

at t

hat t

ime

MPR

(exp

ress

ed a

s %);

MEM

S, u

sed

for 1

med

ica-

tion

only

Mea

n ±

SD: I

: 95.

6 ± 16

.4; C

: 95

.6 ±

15.7

, P =

0.76

; ES

0.06

(–

0.24

to 0.

35)

Solo

mon

, 199

8, m

ul-

tiple

site

s

133 h

yper

tens

ive

patie

nts;

6 m

onth

s

I: St

anda

rdize

d ph

arm

acy c

are:

m

onth

ly p

atie

nt a

sses

smen

t, di

s-ea

se m

anag

emen

t, an

d ed

ucat

ion

Mon

thly

pha

rmac

ist a

dher

ence

as

sess

men

t, in

clud

es d

evel

op-

men

t of a

dher

ence

aid

as n

eede

d

Adhe

renc

e sc

ore

base

d on

4-po

int M

oris

ky sc

ale;

se

lf-re

port

Mea

n ±

SD: I

: 0.2

3 ± 0.

054;

C:

0.61

± 0.

094,

P <

0.05

co

mpa

red

with

in a

nd

betw

een

grou

ps; E

S 0.

57

(0.2

1–0.

93)

Sook

anek

nun,

2004

, ur

ban

and

rura

l Th

aila

nd

235 h

yper

tens

ive

patie

nts f

rom

pha

r-m

acy a

nd p

rimar

y ca

re; 6

mon

ths

I: Re

sear

ch p

harm

acis

t con

sult:

as

sess

ed m

edic

atio

n un

ders

tand

-in

g, a

dher

ence

, cou

nsel

ed o

n us

e, lif

esty

le; e

duca

tiona

l leafl

ets,

di

ary

Regu

lar p

harm

acis

t vis

its (a

ppea

r to

be

mon

thly

) to

iden

tify a

nd

addr

ess m

edic

atio

n pr

oble

ms;

re

com

men

datio

ns g

iven

to p

hy-

sici

ans b

y let

ter a

nd b

y not

e in

m

edic

al re

cord

Calc

ulat

ed M

PR, e

x-pr

esse

d as

%; ≥

80%

nec

-es

sary

to b

e co

nsid

ered

ad

here

nt

% o

f adh

eren

t pat

ient

s:

I: 63

.64%

; C: 5

5.56

%, P

=

0.01

4; E

S 0.

60 (0

.33–

0.87

)

J o u r n a l o f t h e A m e r i c a n P h a r m a c i s t s A s s o c i a t i o nwww.japha.org392 • JAPhA • 52:3 • M ay /J u n 2012

Reviews MEDICATION ADHERENCE INTERVENTIONS

Tabl

e 3.

Dyn

amic

a med

icat

ion

adhe

renc

e in

terv

entio

ns: C

lass

ified

by t

ype

of a

dher

ence

dat

a us

ed in

adh

eren

ce fe

edba

ck lo

opb

Tabl

e 3 c

ontin

ued

Stac

y, 20

09, M

assa

-ch

uset

ts

497 n

ew st

atin

us-

ers;

6 m

onth

s

I: In

tera

ctiv

e vo

ice

resp

onse

te

leph

one

tech

nolo

gy p

rovi

ding

ta

ilore

d m

edic

atio

n co

unse

ling;

m

aile

d le

tters

Inte

ract

ive

voic

e re

spon

se p

ro-

vide

d ta

ilore

d m

essa

ges r

einf

orc-

ing

adhe

renc

e ba

sed

on p

atie

nt

resp

onse

s

6-m

onth

poi

nt p

reva

lenc

e pe

rsis

tenc

y: p

osse

ssio

n of

stat

in a

t end

of 1

80-d

ay

perio

d ba

sed

on c

laim

s re

cord

s; a

lso

prov

ide

MPR

Poin

t pre

vale

nce

pers

is-

tenc

y: I:

70.4

%; C

: 60.

7%, P

<

0.05

; MPR

>80

%: I

: 47.

0%;

C: 38

.9%

, P <

0.10

; ES

0.08

(0

.01–

0.17

)Ts

uyuk

i, 200

4,

Cana

da

276 p

atie

nts w

ith

CHF d

isch

arge

d fro

m h

ospi

tal; 6

m

onth

s

I: Ph

one

call a

t 2 w

eeks

, 4 w

eeks

, th

en m

onth

ly; e

duca

tion;

adh

er-

ence

aid

s (or

gani

zer,

sche

dule

), ph

one,

mai

l fol

low

-up;

C: p

amph

let

Rese

arch

coo

rdin

ator

cal

led

patie

nts t

o as

sess

ACE

I use

, re

info

rced

adh

eren

ce; a

dvis

ed

patie

nts t

o co

nsul

t phy

sici

an fo

r AC

EI d

ose

chan

ge, m

edic

atio

n pr

oble

ms

MPR

for A

CE in

hibi

tor,

expr

esse

d as

%; p

harm

acy

reco

rds

% a

dher

ence

(mea

n ±

SD):

I: 83

.5 ±

31.2

; C: 8

6.2 ±

29.0

, P

= 0.

691;

ES

–0.0

9 (–0

.33

to 0.

15)

Varm

a, 19

99, N

orth

-er

n Ire

land

83 e

lder

ly C

HF p

a-tie

nts f

ollo

wed

afte

r ho

spita

l dis

char

ge;

12 m

onth

s

I: In

-per

son

com

mun

ity p

harm

a-ci

st c

ouns

elin

g on

CHF

med

ica-

tions

, adh

eren

ce, li

fest

yle;

dos

e si

mpl

ifica

tion;

sym

ptom

mon

itor-

ing

Ever

y 3 m

onth

s: a

sses

sed

adhe

r-en

ce w

ith p

resc

ribed

dru

gs, in

-cl

udin

g re

view

of d

rug

diar

y car

ds

Adhe

renc

e de

fined

as

80–1

20%

of m

edic

atio

ns

take

n fo

r all C

HF d

rugs

as-

sess

ed; p

harm

acy r

ecor

ds

% o

f adh

eren

t pat

ient

s: I:

76

.9; C

: 30,

P =

0.03

9; E

S 0.

30

(–0.

29 to

0.89

)

Vivi

an, 2

002,

Phi

la-

delp

hia

56 m

ale

hype

r-te

nsiv

e pa

tient

s,

maj

ority

bla

ck; 6

m

onth

s

I: M

onth

ly p

harm

acis

t cou

nsel

ing;

ch

ange

d dr

ugs,

adj

uste

d do

ses

Adhe

renc

e as

sess

men

t and

co

unse

ling

at e

ach

visi

t

Non

adhe

renc

e m

eant

fail-

ure

to re

fill w

ithin

2 w

eeks

of

sche

dule

d re

fill d

ate

or m

issi

ng >

3 dos

es in

1

wee

k; p

harm

acy r

ecor

ds

% o

f adh

eren

t pat

ient

s: I:

85

; C: 9

3, P

> 0.

42; E

S –0

.26

(–0.

81 to

0.29

)

Exte

rnal

adh

eren

ce

data

(alo

ne o

r in

com

bina

tion

with

se

lf-ge

nera

ted

data

)Jo

hnso

n, 20

06,

Mas

sach

uset

ts a

nd

Rhod

e Is

land

404 a

dults

with

dy

slip

idem

ia; 1

8 m

onth

s

I: Po

pula

tion-

base

d, c

ompu

ter-

gene

rate

d in

divi

dual

ized

inte

rven

-tio

n as

sess

ing

stag

e of

cha

nge

(pre

cont

empl

atio

n, c

onte

mpl

a-tio

n, p

repa

ratio

n, a

ctio

n, m

aint

e-na

nce)

via

ques

tionn

aire

Writ

ten

repo

rt m

aile

d to

pat

ient

(w

ithin

1 w

eek o

f ass

essm

ent)

prov

idin

g fe

edba

ck: (

1) a

t bas

e-lin

e (c

ompa

rison

with

oth

ers t

ry-

ing

to c

hang

e ad

here

nce

beha

v-io

r) an

d (2

) at t

wo

follo

w-u

p po

ints

(c

ompa

rison

with

gro

up a

nd w

ith

indi

vidu

al’s

pas

t res

pons

es)

Resp

onse

s to

5 que

stio

ns

(on

Like

rt sc

ale)

sum

med

to

cre

ate

a co

ntin

uous

m

easu

re; c

alcu

late

d od

ds

of a

ppro

pria

te a

dher

ence

; se

lf-re

port

Adhe

renc

e as

con

tinuo

us

mea

sure

: 6-m

onth

OR

2.03

, P

> 0.

05; 1

8-m

onth

OR

2.86

, P

< 0.

05; E

S 0.

18 (–

0.08

to

0.45

)

Mur

ray,

2007

, Ind

ia-

napo

lis, I

N

314 h

yper

tens

ive

patie

nts f

rom

inne

r-ci

ty p

ract

ice;

12

mon

ths

I: Ph

arm

acis

t med

icat

ion

hist

ory,

kn

owle

dge

asse

ssm

ent,

verb

al,

writ

ten

educ

atio

n; b

oth

in-p

erso

n an

d m

onth

ly p

hone

con

tact

Phar

mac

ist a

dher

ence

cou

nsel

-in

g in

clud

ed o

ptio

n to

revi

ew

MEM

S pl

ot w

ith p

atie

nt, e

ngag

e pa

tient

in p

robl

em so

lvin

g

% o

f pre

scrib

ed m

edic

a-tio

n ta

ken;

MEM

S

Durin

g in

terv

entio

n: I:

78

.8%

; C: 6

7.9%

; diff

eren

ce:

10.9

% (C

I 5.0

–16.

7); p

ostin

t-er

vent

ion

diffe

renc

e: 3.

9%

(CI:

–2.8

to 10

.7);

ES 0.

08

(–0.

89 to

1.06

)

J o u r n a l o f t h e A m e r i c a n P h a r m a c i s t s A s s o c i a t i o n www.japha.org M ay /J u n 2012 • 52:3 • JAPhA • 393

MEDICATION ADHERENCE INTERVENTIONS Reviews

Tabl

e 3.

Dyn

amic

a med

icat

ion

adhe

renc

e in

terv

entio

ns: C

lass

ified

by t

ype

of a

dher

ence

dat

a us

ed in

adh

eren

ce fe

edba

ck lo

opb

Tabl

e 3 c

ontin

ued

Phum

ipa-

mor

n,

2008

, Kra

bi P

rovi

nce,

Th

aila

nd

135 M

uslim

di

abet

es p

atie

nts;

8

mon

ths

I: Ph

arm

acis

t mee

ting

day o

f phy

-si

cian

visi

t; vi

sit r

emin

der 3

day

s pr

ior;

give

n re

fills

, life

styl

e an

d m

edic

atio

n ed

ucat

ion

Rese

arch

pha

rmac

ist r

efille

d m

edic

atio

ns, c

heck

ed p

ill c

ount

, an

d ad

vise

d on

med

icat

ions

be

fore

usu

al P

CP vi

sit (

ever

y 4–8

w

eeks

).

MPR

(exp

ress

ed a

s %);

pill

coun

t

Mea

n di

ffere

nce

(CI):

I:

+6.8

% (2

.1–1

1.4)

, P =

0.00

5;

C: –

2.8 (

–7.3

1 to

1.7)

, P =

0.

29; b

etw

een-

grou

p m

ean

diffe

renc

e P

= 0.

004;

ES

0.50

(0.1

5–0.

86)

Robi

nson

, 201

0,

Tam

pa, F

L

376 p

atie

nts o

n hy

perte

nsio

n m

edic

atio

ns w

ith

unco

ntro

lled

BP;

7–12

mon

ths

I: M

onth

ly (o

r mor

e fre

quen

t) in

-per

son

phar

mac

ist c

ouns

el-

ing

incl

udes

lifes

tyle

, med

icat

ion

adhe

renc

e ed

ucat

ion

Phar

mac

ist g

ives

tailo

red

adhe

r-en

ce im

prov

emen

t stra

tegi

es,

base

s fee

dbac

k on

self-

repo

rt an

d us

e of

refil

l his

tory

to o

btai

n es

timat

ed ra

te o

f adh

eren

ce

MPR

(mea

n ad

here

nce

rate

); ph

arm

acy r

ecor

ds

1–6 m

onth

per

iod

(mea

n ±

SD):

I: 0.

91 ±

0.15

; C: 0

.78 ±

0.

30, P

= 0.

02; 7

–12 m

onth

pe

riod:

I: 0.

91 ±

0.15

; C: 0

.83

± 0.

28, P

= 0.

09; E

S 0.

36

(0.1

5–0.

86)

Tam

blyn

, 200

9, M

on-

treal

and

Que

bec,

Ca

nada

2,29

3 pat

ient

s on

lipid

-low

erin

g or

hy

perte

nsio

n m

edi-

catio

ns; 6

mon

ths

I: Co

mpu

teriz

ed c

ompl

ete

drug

pr

ofile

with

gra

phic

dis

play

s,

refil

l adh

eren

ce c

alcu

latio

n, a

nd

adhe

renc

e al

erts

as p

art o

f com

-pu

teriz

ed m

edic

al re

cord

use

d by

pr

imar

y car

e ph

ysic

ians

; C: c

om-

pute

rized

med

icat

ion

list a

lone

(u

sual

car

e)

Gaps

in g

raph

ical

ly d

ispl

ayed

m

edic

atio

n pr

ofile

(or n

umer

ical

da

ta) i

nfor

m p

hysi

cian

of n

on-

adhe

renc

e; if

adh

eren

ce <

80%

, ph

ysic

ian

got a

lert

whe

n op

enin

g m

edic

al c

hart,

told

to c

heck

dru

g pr

ofile

Mea

n re

fill a

dher

ence

: pr

opor

tion

of d

ays c

ov-

ered

; acc

esse

d vi

a da

ily

retri

eval

of c

ompu

teriz

ed

disp

ensi

ng in

form

atio

n;

real

-tim

e up

date

s of r

e-co

rds

I: 73

.5%

; C: 7

2.9%

; Δ m

ean

adhe

renc

e ±

SD: I

: –6.

2 ±

24.1

; C: –

6.4 ±

24.1

, P =

0.90

; ES

0.01

(–0.

08, 0

.09)

Vrije

ns, 2

006,

Bel

-gi

um

392 p

atie

nts w

ith

dysl

ipid

emia

on

ator

vast

atin

; 12

mon

ths

I: Ph

arm

acy p

rogr

am: m

edic

atio

n hi

stor

y edu

catio

nal r

emin

ders

, w

ritte

n in

form

atio

n; C

: writ

ten

info

rmat

ion

Phar

mac

ist s

its w

ith p

atie

nt a

nd

revi

ews e

lect

roni

cally

com

pile

d do

sing

his

tory

of p

ast m

onth

s;

patie

nt ta

ught

to re

ad M

EMS

grap

hics

% o

f day

s tha

t med

icat

ion

cont

aine

r ope

ning

was

re

cord

ed; M

EMS

I: 95

.89%

(CI 9

0.28

–98.

66);

C: 89

.37 (

69.7

0–96

.33)

, P <

0.

001;

ES:

una

ble

to c

al-

cula

te

Abbr

evia

tions

use

d: A

1C, g

lyco

syla

ted

hem

oglo

bin;

ACE

I, an

giot

ensi

n-co

nver

ting

enzy

me

inhi

bito

r; BP

, blo

od p

ress

ure;

C, c

ontro

l gro

up; C

AD, c

oron

ary a

rtery

dis

ease

; CHF

, con

gest

ive

hear

t fai

lure

; ES,

effe

ct si

ze; G

P, g

ener

al

prac

titio

ner;

I, in

terv

entio

n gr

oup;

IQR,

inte

rqua

rtile

rang

e; M

EMS,

med

icat

ion

even

t mon

itorin

g sy

stem

; MPR

, med

icat

ion

poss

essi

on ra

tio; N

S, n

onsi

gnifi

cant

; OR,

odd

s rat

io; P

CP, p

rimar

y car

e pr

ovid

er; P

TCA,

per

cuta

neou

s tra

nslu

min

al c

oron

ary a

ngio

plas

ty; R

R, ra

te ra

tio.

a Dyna

mic

inte

rven

tions

are

firs

t adm

inis

tere

d to

all m

edic

atio

n ta

kers

, the

n re

al-ti

me

adhe

renc

e in

form

atio

n is

use

d to

furth

er ta

rget

non

adhe

rers

. b An

adh

eren

ce fe

edba

ck lo

op is

cre

ated

whe

reby

adh

eren

ce d

ata

are

gene

rate

d an

d th

en fe

d ba

ck in

to th

e in

terv

entio

n. A

dher

ence

feed

back

loop

s use

eith

er (1

) sel

f-gen

erat

ed d

ata

alon

e or

(2) e

xter

nal d

ata

alon

e or

alo

ng w

ith

self-

gene

rate

d da

ta. S

elf-g

ener

ated

dat

a fe

edba

ck lo

ops w

ere

thos

e in

whi

ch th

e ta

ilorin

g of

the

inte

rven

tion

was

bas

ed e

ntire

ly o

n pa

tient

self-

repo

rt of

non

adhe

renc

e. E

xter

nal d

ata

feed

back

loop

s wer

e th

ose

in w

hich

info

rma-

tion

such

as M

PR d

eriv

ed fr

om p

harm

acy r

ecor

ds o

r oth

er e

xter

nal s

ourc

es w

as u

sed

to ta

ilor t

he in

terv

entio

n.

c Dura

tion

indi

cate

s tim

e un

til la

st fo

llow

-up

in w

hich

adh

eren

ce is

mea

sure

d.

d Cont

rol p

atie

nts r

ecei

ved

usua

l car

e un

less

oth

erw

ise

spec

ified

. e M

PR: m

edic

atio

n do

ses t

aken

div

ided

by d

oses

pre

scrib

ed. M

oris

ky sc

ale

has f

our q

uest

ions

(1 p

oint

for e

very

“yes

” res

pons

e): (

1) D

o yo

u ev

er fo

rget

to ta

ke yo

ur m

edic

atio

n?; (

2) A

re yo

u ca

rele

ss a

t tim

es a

bout

taki

ng yo

ur

med

icat

ion?

; (3)

Whe

n yo

u fe

el b

ette

r, do

you

som

etim

es st

op ta

king

your

med

icat

ion?

; (4)

Som

etim

es if

you

feel

wor

se w

hen

you

take

your

med

icat

ion,

do

you

stop

taki

ng it

?f 95

% C

I unl

ess o

ther

wis

e sp

ecifi

ed.

For a

ll stu

dies

whe

re m

eans

(±SD

) for

adh

eren

ce o

utco

mes

wer

e av

aila

ble,

Coh

en’s

d st

atis

tics w

ere

calc

ulat

ed. T

he E

Ss c

ompa

re th

e di

ffere

nce

in e

ffect

bet

wee

n th

e st

udy g

roup

s div

ided

by t

he S

D of

this

diff

eren

ce. W

e co

n-si

dere

d an

ES

<0.2

to b

e ve

ry sm

all, 0

.2–0

.5 sm

all, 0

.5–0

.8 m

ediu

m, a

nd >

0.8 l

arge

.

J o u r n a l o f t h e A m e r i c a n P h a r m a c i s t s A s s o c i a t i o nwww.japha.org394 • JAPhA • 52:3 • M ay /J u n 2012

Reviews MEDICATION ADHERENCE INTERVENTIONS

delivered based on the patient’s unique situation, incorporating components such as regimen adjustment, reinforcement, edu-cation, lifestyle counseling, advice on contacting physicians, and, in some cases, direct contact of primary physicians by the pharmacist.

A total of 10 studies yielded very small or small effects. Of these, four42,48,56,61 relied on health professionals. These four interventions had similar feedback structures to those described above, although nurses played the main role, with only one intervention mediated primarily by a pharmacist.61 Two interventions51,55 were carried out by lay persons including research assistants who conducted motivational interviews51 and medical assistants who provided medication counseling by phone.55 Three interventions44,52,59 made use of electronic systems including interactive computer-based home monitor-ing,44 interactive phone technology with tailored messages,59 and automated calls with structured messages and targeted nurse follow-up.52 One study in this group45 consisted mainly of postal and phone reminders but had patients reporting ad-herence directly to physicians at 3 months. There was no pre-specified physician response to nonadherence; in this case, we assumed that discussion occurred or medication changes were administered if patients reported nonadherence directly to the physician.

Similar to the most effective group, the five studies that showed no improvement in adherence23,46,50,60,62 were almost all conducted by health professionals. Hunt et al.46 reported the effect of pharmacist-managed hypertension, and Logan et al.23 described worksite hypertension management carried out by a nurse with physician support. Both had ESs very close to zero, as did Tsuyuki et al.,60 who examined the effect of lay person (research coordinator) phone calls in which adherence was reinforced and patients were referred to their physicians for questions or adherence concerns.

In contrast, Vivian62 studied monthly pharmacist counsel-ing, including regimen adjustments as needed, and found a small negative effect. Odegard et al.50 described the effect of a diabetes care plan developed by a primary care pharmacist along with follow-up meetings and calls but followed a group in which control patients showed better adherence than interven-tion patients.

Use of external data in adherence feedback loop. We identified six articles63–68 that described adherence feedback loops reliant on externally generated data; for one of these,68 an ES could not be calculated. Of those with calculated ESs, 20% (one study) yielded medium or large effects and 80% (four studies) very small or small effects.

The study of Phumipamorn et al. 65 was the only dynamic in-tervention to yield a medium to large ES while using an external adherence data feedback loop, and this study did not rely on au-tomated adherence data. Research pharmacists meeting with patients with diabetes on the day of their physician visit con-ducted pill counts and provided counseling along with refills.

The four dynamic interventions with very small or small ef-fects included three in which a health professional (pharmacist or physician) had access to electronic adherence data during

the patient interaction. Robinson et al.66 studied the effect of pharmacist adherence counseling using both self-reported ad-herence and pharmacy refill history.

Murray et al.64 described pharmacist adherence counseling in which pharmacists had the option to review electronic pill-box adherence data with patients and engage in problem-solv-ing based on these data. Pillbox data were reported to be avail-able in plot form for ease of communication; however, such a review was at the pharmacist’s discretion and not conducted with every patient. Tamblyn et al.67 described an intervention in which computerized complete drug profiles with graphic displays were made available to primary care physicians, in-corporating refill adherence calculation and adherence alerts as part of an electronic medical record already in use by physi-cians. One of the more complex systems for delivery of external adherence data directly to physicians, this study had a small and likely clinically insignificant ES (0.01 [CI –0.08 to 0.09]), although physicians overwhelmingly requested to have access to the graphics at the conclusion of the trial.

Johnson et al.21 used a computer-generated intervention assessing stage of change with respect to adherence behavior based on a questionnaire administered to patients.21 The exter-nal data in this feedback loop were a computerized interpreta-tion of patients’ readiness to change adherence behaviors that was delivered with patient-appropriate recommendations via a mailed written report to patients. The authors found that the odds of appropriate adherence were improved in the interven-tion group (OR 2.86) with a small ES (0.18 [–0.08 to 0.45]).

DiscussionCompared with broadly administered interventions, those that targeted nonadherers (either exclusively or in a dynamic fash-ion) tended to have a larger effect on medication adherence. In addition to the statistical heterogeneity reflected in our analy-ses, we found considerable clinical heterogeneity among the studies identified. Adherence measures differed between stud-ies both as an outcome and a means of defining a target group. This heterogeneity prevented us from pursuing a meta-analysis and prompted us to interpret all findings with caution. Our find-ings highlight the need for standardized approaches to adher-ence measurement.

With respect to dynamic interventions, our findings sup-port further attention to the unique components of the self-gen-erated dynamic intervention. Possible mechanisms for the rel-ative success we found among individuals participating in self-generated adherence feedback loops included an increased ability to provide tailored medication advice in real time and the requirement that patients have at least a degree of insight into their nonadherent behaviors.

Vrijens et al.68 called for increased use of health informa-tion technology to identify “patients for whom poor medica-tion adherence may undermine clinical goals and patients who could benefit from interventions aimed at achieving optimal medication use.” Despite this emphasis on health information technology, our study did not find any advantage to using ex-ternally generated data for dynamic feedback loops. Although

J o u r n a l o f t h e A m e r i c a n P h a r m a c i s t s A s s o c i a t i o n www.japha.org M ay /J u n 2012 • 52:3 • JAPhA • 395

MEDICATION ADHERENCE INTERVENTIONS Reviews

we acknowledge the heterogeneity of the studies and our dif-ficulty drawing firm conclusions on that basis, our findings do not support widescale implementation of automated electronic adherence feedback at this time. Further study in this regard is needed to demonstrate incremental benefit of this approach.

Focused interventions identify nonadherers before starting the intervention. This group provides very limited data, and the studies that exist do not consistently make use of reproducible, standardized methods for identifying nonadherers. This is an area that would benefit from additional studies, particularly ones in which better methods for identifying the focused popu-lation are presented.

Broad interventions appear to be least effective. Such in-terventions aim to prevent nonadherence by educating and motivating patients to adhere to treatment. Although many of these interventions were effective, their benefits were likely diluted because of the small effect of the intervention on pa-tients already inclined toward adherence. Moreover, without the benefit of identifying patients and their specific barriers to nonadherence, these interventions may have been too general to motivate individual patients to meaningfully change their behavior. In a resource-constrained health care system, broad interventions without a feedback loop may not provide the best return on investment.

The current results have important implications. Broad interventions are the least effective and potentially the most expensive; research is needed to more fully evaluate focused and dynamic interventions. Focused interventions allow limit-ed resources to be directed toward fewer, higher-risk patients, and dynamic interventions share this advantage when the more costly portion of the intervention is reserved for identified non-adherers. Attention must be paid, however, to the method of identifying nonadherers. None of the focused interventions that we identified made use of pharmacy claims data to iden-tify nonadherence. Focused studies were administered after a baseline period in which adherence data were gathered from a larger group of patients, or ascertainment of nonadherent status was based more loosely on a combination of physician documentation, self-report, and refill history. Dynamic inter-ventions were overwhelmingly dependent on self-generated adherence data (often requiring intensive interaction with a health provider), and very few used any form of external data. The accuracy, cost, and reproducibility of methods for identify-ing target populations must be a central consideration in future studies.

Based on our findings, we recommend further investigation of targeted adherence improvement efforts aimed at patients who are nonadherent. Methods for identifying nonadherence should be defined in advance as clearly as possible and evalu-ated for cost, validity, and reproducibility. As electronic phar-macy data become increasingly available, external adherence data should be explored as a means of focusing interventions from the outset and informing dynamic feedback loops; how-ever, we must keep in mind the limited effectiveness demon-strated to date.

LimitationsLimitations of our review included the substantial clinical and statistical heterogeneity of the identified studies, thereby pre-venting meta-analysis estimates from being presented. Other limitations included the variable quality of study methodology, the possibility for publication bias, the paucity of available information on intervention costs, and the lack of standard-ized methods to study this issue. Heterogeneous adherence outcome measures prompted us to translate adherence out-comes into Cohen’s d ESs, allowing for between-study com-parison but possibly providing a less direct measure of adher-ence improvement. Because variability existed in the degree of detail offered on the adherence interventions, we may have misclassified some interventions, although we do not expect that this would have occurred in a biased manner. Our deci-sion to exclude studies characterized by regimen simplifica-tion limits our ability to comment on this important group of adherence interventions. Success rates seen in focused inter-ventions may reflect their exclusion of already-adherent pa-tients, for whom any intervention would yield little additional improvement. Because dynamic and broad interventions mea-sure improvements in adherence for the whole population, we were unable to directly compare the impact of dynamic, broad, and focused interventions on each group’s subset of nonadherent patients. Thus, considerable gaps remain in the existing evidence base for targeting medication adherence in-terventions.

ConclusionTargeting patients who are nonadherent to their cardiovas-cular medications may lead to better adherence; however, data are limited and studies are highly heterogeneous. With cost an ever-present consideration, future efforts to improve adherence may best be directed at patients who demonstrate that they do not adhere to therapy.

References1. Institute of Medicine. 100 initial priority topics for comparative

effectiveness research. Accessed at www.iom.edu/~/media/Files/Report%20Files/2009/ComparativeEffectivenessResearch-Priorities/Stand%20Alone%20List%20of%20100%20CER%20Priorities%20-%20for%20web.ashx, June 8, 2010.

2. Rasmussen JN, Chong A, Alter DA. Relationship between adher-ence to evidence-based pharmacotherapy and long-term mor-tality after acute myocardial infarction. JAMA. 2007;297:177–86.

3. New England Healthcare Institute. Thinking outside the pillbox: a system-wide approach to improving patient medication ad-herence for chronic disease. Accessed at www.nehi.net/pub-lications/44/thinking_outside_the_pillbox_a_systemwide_ap-proach_to_improving_patient_medication_adherence_for_chronic_disease, May 27, 2010.

4. Kripalani S, Yao X, Haynes R. Interventions to enhance medica-tion adherence in chronic medical conditions: a systematic re-view. Arch Intern Med. 2007;167:540–50.

5. Jadad AR, Moore RA, Carroll D, et al. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Con-trol Clin Trials. 1996;17:1–12.

J o u r n a l o f t h e A m e r i c a n P h a r m a c i s t s A s s o c i a t i o nwww.japha.org396 • JAPhA • 52:3 • M ay /J u n 2012

Reviews MEDICATION ADHERENCE INTERVENTIONS

6. Cohen J. Statistical power analysis for the behavioral sciences. New York: Lawrence Erlbaum Associates; 1988.

7. Dunlap W, Cortina J, Vaslow J, Burke M. Meta-analysis of experi-ments with matched groups or repeated measures designs. Psy-chological Methods. 1996;1:170–7.

8. Hedges L, Olkin I. Statistical methods for meta-analysis. San Di-ego, CA: Academic Press; 1985.

9. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med. 2002;21:1539–58.

10. Haynes RB, Sackett DL, Gibson ES, et al. Improvement of medication compliance in uncontrolled hypertension. Lancet. 1976;1:1265–8.

11. Rosen MI, Rigsby MO, Salahi JT, et al. Electronic monitoring and counseling to improve medication adherence. Behav Res Ther. 2004;42:409–22.

12. Saunders LD, Irwig LM, Gear JS, Ramushu DL. A randomized controlled trial of compliance improving strategies in Soweto hypertensives. Med Care. 1991;29:669–78.

13. Taylor CT, Byrd DC, Krueger K. Improving primary care in rural Alabama with a pharmacy initiative. Am J Health Syst Pharm. 2003;60:1123–9.

14. Avanzini F, Corsetti A, Maglione T, et al. Simple, shared guide-lines raise the quality of antihypertensive treatment in routine care. Am Heart J. 2002;144:726–32.

15. Birtwhistle RV, Godwin MS, Delva MD, et al. Randomised equivalence trial comparing three and six months of follow up of patients with hypertension by family practitioners. BMJ. 2004;328:204.

16. Christensen A, Christrupa LL, Fabriciusb PE, et al. The impact of an electronic monitoring and reminder device on patient com-pliance with antihypertensive therapy: a randomized controlled trial. J Hypertens. 2010;28:194–200.

17. Dusing R, Handrock R, Klebs S, et al. Impact of supportive mea-sures on drug adherence in patients with essential hyperten-sion treated with valsartan: the randomized, open-label, parallel group study VALIDATE. J Hypertens. 2009;27:894–901.

18. Emmett CL, Montgomery AA, Peters TJ, Fahey T. Three-year fol-low-up of a factorial randomised controlled trial of two decision aids for newly diagnosed hypertensive patients. Br J Gen Pract. 2005;55:551–3.

19. Hamet P, Campbell N, Curnew G, et al. Avapromise: a random-ized clinical trial for increasing adherence through behav-ioural modification in essential hypertension. Exp Clin Cardiol. 2003;7:165–72.

20. Hawkins DW, Fiedler FP, Douglas HL, Eschbach RC. Evaluation of a clinical pharmacist in caring for hypertensive and diabetic patients. Am J Hosp Pharm. 1979;36:1321–5.

21. Johnson AL, Taylor DW, Sackett DL, et al. Self-recording of blood pressure in the management of hypertension. Can Med Assoc J. 1978;119:1034–9.

22. Kirscht JP, Kirscht JL, Rosenstock IM. A test of interventions to increase adherence to hypertensive medical regimens. Health Education Quarterly. 1981;8:261–72.

23. Logan AG, Milne BJ, Achber C, et al. Work-site treatment of hy-pertension by specially trained nurses: controlled trial. Lancet. 1979;2:1175–8.

24. Lopez Cabezas C, Falces Salvador C, Cubi Quadrada D, et al. Randomized clinical trial of a postdischarge pharmaceutical care program vs regular follow-up in patients with heart failure. Farm Hosp. 2006;30:328–42.

25. Mann DM, Ponieman D, Montori VM, et al. The statin choice decision aid in primary care: a randomized trial. Patient Educ Couns. 2010;80:138–40.

26. Marquez-Contreras E, Martell-Claros N, Gil-Guillen V, et al. Ef-ficacy of a home blood pressure monitoring programme on therapeutic compliance in hypertension: the EAPACUM-HTA study. J Hypertens. 2006;24:169–75.

27. Mehos BM, Saseen JJ, MacLaughlin EJ. Effect of pharmacist intervention and initiation of home blood pressure monitoring in patients with uncontrolled hypertension. Pharmacotherapy. 2000;20:1384–9.

28. Morisky DE, DeMuth NM, Field-Fass M, et al. Evaluation of fami-ly health education to build social support for long-term control of high blood pressure. Health Educ Q. 1985;12:35–50.

29. Mullan RJ, Montori VM, Shah ND, et al. The diabetes mellitus medication choice decision aid: a randomized trial. Arch Intern Med. 2009;169:1560–8.

30. Powell KM, Edgren B. Failure of educational videotapes to im-prove medication compliance in a health maintenance organi-zation. Am J Health Syst Pharm. 1995;52:2196–9.

31. Rudd P, Miller NH, Kaufman J, et al. Nurse management for hy-pertension: a systems approach. Am J Hypertens. 2004;17:921–7.

32. Sackett DL, Gibson ES, Taylor DW, et al. Randomized clinical trial of strategies for improving medication compliance in pri-mary hypertension. Lancet. 1975;1:1205–7.

33. Sclar DA, Chin A, Skaer TL, et al. Effect of health-education in promoting prescription refill compliance among patients with hypertension. Clin Ther. 1991;13:489–95.

34. Smith DH, Kramer JM, Perrin N, et al. A randomized trial of direct-to-patient communication to enhance adherence to beta-blocker therapy following myocardial infarction. Arch Intern Med. 2008;168:477–83.

35. Stewart A, Noakes T, Eales C, et al. Adherence to cardiovascular risk factor modification in patients with hypertension. Cardio-vasc J S Afr. 2005;16:102–7.

36. Takala J. Screening, treatment and adherence to treatment for hypertension. Scand J Prim Health Care. 1983;1:114–9.

37. van Onzenoort HA, Verberk WJ, Kroon AA, et al. Effect of self-measurement of blood pressure on adherence to treatment in patients with mild-to-moderate hypertension. J Hypertens. 2010;28:622–7.

38. Yilmaz MB, Pinar M, Naharci I, et al. Being well-informed about statin is associated with continuous adherence and reaching targets. Cardiovasc Drugs Ther. 2005;19:437–40.

39. Antonicelli R, Testarmata P, Spazzafumo L, et al. Impact of tele-monitoring at home on the management of elderly patients with congestive heart failure. J Telemed Telecare. 2008;14:300–5.

40. Blenkinsopp A, Phelan M, Bourne J, Dakhil N. Extended ad-herence support by community pharmacists for patients with hypertension: a randomised controlled trial. Int J Pharm Pract. 2000;8:165–75.

J o u r n a l o f t h e A m e r i c a n P h a r m a c i s t s A s s o c i a t i o n www.japha.org M ay /J u n 2012 • 52:3 • JAPhA • 397

MEDICATION ADHERENCE INTERVENTIONS Reviews

41. Bouvy ML, Heerdink ER, Urquhart J, et al. Effect of a pharmacist-led intervention on diuretic compliance in heart failure patients: a randomized controlled study. J Card Fail. 2003;9:404–11.

42. Edworthy SM, Baptie B, Galvin D, et al. Effects of an enhanced secondary prevention program for patients with heart disease: a prospective randomized trial. Can J Cardiol. 2007;23:1066–72.

43. Faulkner MA, Wadibia EC, Lucas BD, Hilleman DE. Impact of pharmacy counseling on compliance and effectiveness of com-bination lipid-lowering therapy in patients undergoing coronary artery revascularization: a randomized, controlled trial. Pharma-cotherapy. 2000;20:410–6.

44. Friedman RH, Kazis LE, Jette A, et al. A telecommunications sys-tem for monitoring and counseling patients with hypertension: impact on medication adherence and blood pressure control. Am J Hypertens. 1996;9:285–92.

45. Guthrie RM. The effects of postal and telephone reminders on compliance with pravastatin therapy in a national registry: re-sults of the First Myocardial Infarction Risk Reduction Program. Clin Ther. 2001;23:970–80.

46. Hunt JS, Siemienczuk J, Pape G, et al. A randomized controlled trial of team-based care: impact of physician-pharmacist col-laboration on uncontrolled hypertension. J Gen Intern Med. 2008;23:1966–72.

47. Jaffray M, Bond C, Watson M, et al. The MEDMAN study: a ran-domized controlled trial of community pharmacy-led medicines management for patients with coronary heart disease. Fam Pract. 2007;24:189–200.

48. Krantz MJ, Havranek EP, Haynes DK, et al. Inpatient initiation of beta-blockade plus nurse management in vulnerable heart fail-ure patients: a randomized study. J Card Fail. 2008;14:303–9.

49. Logan AG, Milne BJ, Flanagan PT, Haynes RB. Clinical effective-ness and cost-effectiveness of monitoring blood-pressure of hy-pertensive employees at work. Hypertension. 1983;5:828–36.

50. Odegard PS, Goo A, Hummel J, et al. Caring for poorly controlled diabetes mellitus: a randomized pharmacist intervention. Ann Pharmacother. 2005;39:433–40.

51. Ogedegbe G, Chaplin W, Schoenthaler A, et al. A practice-based trial of motivational interviewing and adherence in hypertensive African Americans. Am J Hypertens. 2008;21:1137–43.

52. Piette JD, Weinberger M, McPhee SJ, et al. Do automated calls with nurse follow-up improve self-care and glycemic con-trol among vulnerable patients with diabetes? Am J Med. 2000;108:20–7.

53. Planas LG, Crosby KM, Mitchell KD, Farmer KC. Evaluation of a hypertension medication therapy management program in pa-tients with diabetes. J Am Pharm Assoc. 2009;49:164–70.

54. Sadik A, Yousif M, McElnay JC. Pharmaceutical care of patients with heart failure. Br J Clin Pharmacol. 2005;60:183–93.

55. Schectman G, Hiatt J, Hartz A. Telephone contacts do not im-prove adherence to niacin or bile-acid sequestrant therapy. Ann Pharmacother. 1994;28:29–35.

56. Schroeder K, Fahey T, Hollinghurst S, Peters TJ. Nurse-led ad-herence support in hypertension: a randomized controlled trial. Fam Pract. 2005;22:144–51.

57. Solomon DK, Portner TS, Bass GE, et al. Clinical and economic outcomes in the hypertension and COPD arms of a multicenter outcomes study. J Am Pharm Assoc. 1998;38:574–85.

58. Sookaneknun P, Richards RM, Sanguansermsri J, Teerasut C. Pharmacist involvement in primary care improves hypertensive patient clinical outcomes. Ann Pharmacother. 2004;38:2023–8.

59. Stacy JN, Schwartz SM, Ershoff D, Shreve MS. Incorporating tailored interactive patient solutions using interactive voice response technology to improve statin adherence: results of a randomized clinical trial in a managed care setting. Popul Health Manag. 2009;12:241–54.

60. Tsuyuki RT, Fradette M, Johnson JA, et al. A multicenter disease management program for hospitalized patients with heart fail-ure. J Card Fail. 2004;10:473–80.

61. Varma S, McElnay JC, Hughes CM, et al. Pharmaceutical care of patients with congestive heart failure: interventions and out-comes. Pharmacotherapy. 1999;19:860–9.

62. Vivian EM. Improving blood pressure control in a pharmacist-managed hypertension clinic. Pharmacotherapy. 2002;22:1533–40.

63. Johnson SS, Driskell MM, Johnson JL, et al. Transtheoretical model intervention for adherence to lipid-lowering drugs. Dis Manag. 2006;9:102–14.

64. Murray MD, Young J, Hoke S, et al. Pharmacist intervention to improve medication adherence in heart failure: a randomized trial. Ann Intern Med. 2007;146:714–25.

65. Phumipamorn S, Pongwecharak J, Soorapan S, Pattharachay-akul S. Effects of the pharmacist’s input on glycaemic control and cardiovascular risks in Muslim diabetes. Prim Care Diabetes. 2008;2:31–7.

66. Robinson JD, Segal R, Lopez LM, Doty RE. Impact of a pharma-ceutical care intervention on blood pressure control in a chain pharmacy practice. Ann Pharmacother. 2010;44:88–96.

67. Tamblyn R, Reidel K, Huang A, et al. Increasing the detection and response to adherence problems with cardiovascular medica-tion in primary care through computerized drug management systems: a randomized controlled trial. Med Decis Making. 2010;30:176–88.

68. Vrijens B, Belmans A, Matthys K, et al. Effect of intervention through a pharmaceutical care program on patient adherence with prescribed once-daily atorvastatin. Pharmacoepidemiol Drug Saf. 2006;15:115–21.

Author,  Year,    Site  

Funding  source   Jadad  Score  

From  Table  1      

Haynes,  RB,  1976  Canada  

Grant  no.  MA-­‐5159  from  Medical  Research  Council  of  Canada,  National  Health  Grant  from  Health  and  Welfare  Canada,  and  a  grant  from  Dominion  Foundries  and  Steel  company  of  Canada.    One  of  authors  is  Physicians  Services  Inc.  foundation  fellow.  

3    

 Rosen,  2004  

Connecticut  

VA  Merit  Review  grant,  an  RO1  and  the  VISN1  .  Not  supported  by  any  company  making  MEMS  or  related  products.  

3  

Saunders,  LD,  1991  

Soweto,  S  Africa  

Supported  by  the  S.  African  Medical  Research  Council.  2    

Taylor,  2003  

Alabama    

ASHP  Research  and  Education  Foundation  2    

From  Table  2      Avanzini,  2002  Italy    

Supported  in  part  by  an  educational  grant  from  Hoechst-­‐Roussel  and  Du  Pont  Pharma  Italia  

2    

Birtwhistle,  2004  

Urban,  rural  Canada  

Canadien  Institute  for  Health  Research  and  McKnight  Fund  for  Queen's  University  

3      

Christensen  ,  2010  Poland  

Bang  &  Olufsen  Medicom  A/S  and  the  Danish  Ministry  of  Science,  Technology  and  Innovation.  

2  

Düsing  R  2009  

multiple  sites,  Germany  

Novartis  Pharma  GmBH,  NÜrnberg,  Germany   2  

Emmett,  CL,  2005  Bristol,  England.  

Royal  College  of  General  Practitioners  Scientific  Foundation  Board,  Training  Fellowship  in  Health  Services  Research  at  MRC  and  Royal  College  of  General  Practitioners  Scientific  Foundation  

Board  

2    

Hamet,  P,  2003  Canada  

 

Bristol-­‐Myers  Squibb  Canda  and  Sanofi  Canada  3    

Hawkins,  1979  

 DHEW  public  service  grant   2  

Johnson,   Health  research  grant  from  Ontario  Ministry  of  Health   2  

1978  Hamilton,  Ontario,  Canada.  

 Kirscht,  JP,  

1981  Tecumseh,  Michigan  

   

Grant  HL18401  from  the  National  Heart  Lung  and  Blood  Institute.  

2  

Logan,  1979  

Toronto,  Canada  

 

Grant  from  Ontario  Ministry  of  Health  2      

Lopez  Cabezas,C,  

2006  Barcelona,  

Spain    

Health  research  Fund  (Fonodo  de  Investigacion  Sanitaria,  FIS)  and  the  European  Regional  Development  Fund  (ERDF).  

2    

Mann,  2009  

New  York  Not  given   1  

Marquez-­‐Contreras,  

2006  Spain  

 

Novartis  Farmaceutica,  Spain  3    

Mehos,  2000  

Colorado  

1998-­‐99  Bristol-­‐Myers  Squibb  Pharmacy  Practice  Hypertension  Program  grant  from  American  Association  of  Colleges  of  

Pharmacy  

3    

Morisky,  1985  

Baltimore,  Maryland.  

NHLBI  grants,  NCHRS  and  BRSG  grant  from  the  Biomedical  Research  Support  Grant  Program,  NIH.  

2    

 Mullan,  2009  

Minnesota  

American  Diabetes  Association.    Novo  Nordisk,  a  maker  of  insulin,  subsidized  the  ADA  grant  program  but  did  not  have  

direct  contact  with  the  investigators  and  idd  not  play  any  role  in  the  awarding  the  grant  to  the  research  team  

3  

Powell,  KM,  1995  

 Ciba  Geigy  Corp,  Merck  (education  grants)   1  

Rudd,  2004  

California  Grant  from  CorSolutions,  Inc  (Buffalo  Grove,  IL)  

3    

Sackett,  1975  

 Hamilton,  Ontario  

 

Not  specified  2    

Sclar,  DA,  1991  

Delaware,  Texas  and  Wisconsin.  

 

grant  from  ICI  Pharmaceuticals,  Wilmington,  Delaware  1    

Smith,  2008  

U.S.  urban  centers  

"ccoperative  agreement"  from  AHRQ  2    

Stewart,  A,  2005  

Johannesburg,  S.  Africa.  

Not  specified   2  

Takala,  J,  1983  

Southwest  Finland  

Not  specified   2  

van  Onzenoort,  2009  The  

Netherlands  

The  Netherlands  Organzation  for  Health  Research  and  Development  (Healthcare  Efficiency  Research  Program;  grant  

945-­‐01-­‐043)  2  

Yilmaz,  MB,  2005  

Ankara,  Turkey.  Not  specified   2  

From  Table  3      Antonicelli,  

2008  Italy  

Grants  from  the  Italian  Ministry  of  Health   2  

Blenkinsopp,  2000  

England  

(England)  Dept  of  Health  as  part  of  its  Community  Pharmacy  Wider  Role  Programme  

2    

Bouvy,  2003  The  

Netherlands  

Unrestricted  research  grant  from  independent  nonprofit  foundation  for  efficient  ues  of  eds  (DGMN)  

3  

Edworthy,  2007  

Calgary,  Alberta.  Canada.  

Supported  by  grant  from  Merck  Frost  Canada  2    

Faulkner,  2000  

Omaha,  Nebraska  

Not  specified  3    

Friedman,  1996  Boston  

 

grant  from  NHLBI   2  

Guthrie,  2001  Ohio  

Bristol-­‐Myers  Squibb  Co,  Princeton,  NJ  2    

Hunt,  2008  

Oregon      

grant  from  Boehringer  Ingelheim   3  

Jaffray,  2007  England  

Dept  Health  for  England  and  Wales  managed  by  collaboration  of  Nat'l  Pharm  Assoc'n,  Poyal  Pharm  Society  of  Great  Britain,  Company  Chemist  Assoc  and  Coop  Pharmacy  Technical  Panel  

3  

Krantz,  2008  

Denver,  Colorado  

Glaxo  Smith  Kline.   3  

Logan,  1983  

Toronto  Ontario  Ministry  of  Health,  Public  Health  grant  CHS-­‐R21   1  

Odegard,  2005  

Seattle,  Washington  

 

grant  from  Academic  and  Managed  care  Forum,  Quality  Care  Research  Fund  

3  

Ogedegbe,  2008  

New  York,  NY.    

NHLBI,  NIH  grants   3  

Piette,  2000  

 

Clinical  research  grants  program  o  f  the  ADA  and  the  Health  Services  Research  and  Devlopment  Service  and  mental  Health  

Strategic  Health  Group,  Dept  of  VA  3  

Planas,  2009  

American  Pharmacists  Association  Foundation,  the  American  Society  of  Health-­‐System  Pharmacists  Foundation,  and  USA  

Drug  Stores    

Sadik,  2005  

Al-­‐Ain,  United  Arab  Emirates  

 

Not  specified   3  

Schectman,  G,   HSR&D  Grant  from  the  VA  and  grant  from  Squibb-­‐Bristol   2  

1994  Milwaukee  

company  

Schroeder,  K,  2005  

Avon,  UK  

Medical  Research  Council  Trianing  Fellowship  in  Health  Services  Research  

3  

Solomon,  1998  

Multiple  sites      

Educational  grant  Novartis  pharmaceuticals  corp   2  

Sookaneknun  2004  

Urban  and  rural  Thailand  

 

Research  grant  from  Chiang  Mai  University,  Thailand   2  

Stacy,  2009  

Not  available   3  

Tsuyuki,  RT,  2004  Canada  

   

Unrestricted  educational  grant  from  Parke  Davis  Canada  (now  Pfizer  Canada)  and  the  U  of  Alberta  Hospital  Foundation  

3  

Varma,  1999  

Northern  Ireland  

 

Not  specified   3  

Vivian,  EM,  2002  

Philadelphia,  Pennsylvania  

 

Supported  by  the  Christian  R  and  Mary  F  Lindback  Foundation   2  

External  adherence  

data  (alone  or  in  

combination  with  self-­‐generated  

data)  

   

Johnson,  SS,  2006  

Massachusetts  and  Rhode  Island  

Grant  from  the  National  Heart.  Lung  and  Blood  Institute  (grant  no.  R44  HL64504)  

2  

Murray,  2007  

National  Institutes  of  Health  grants  R01  AG19105  and  R01  HL  69399  and  AG01799  

3  

Indianapolis,  Indiana  Phumipa-­‐morn,  S,  2008  Krabi  

Province,  Thailand.  

Research  grants  from  Graduate  School,  Prince  of  Songkla  University  and  the  Provincial  Public  Health  Dept  of  Krabi  

Province,  Thailand  3  

Robinson,  2010  

 Tampa,  FL  

Pfizer  unrestricted  grant   1  

Tamblyn,  2009  

Montreal  and  Quebec,  Canada  

Canadien  Institutes  of  Health  Research  and  Pfizer  Canada  Inc   2  

Vrijens,  2006  

Belgium  Contract  grant  sponsor  is  Pfizer  Belgium   1