The Impact of Medication Adherence on Coronary Artery Disease … · The Impact of Medication...

C

WOp

NrC

CLINICAL RESEARCH STUDY

The Impact of Medication Adherence on Coronary ArteryDisease Costs and Outcomes: A Systematic ReviewAsaf Bitton, MD, MPH,a,b Niteesh K. Choudhry, MD, PhD,c Olga S. Matlin, PhD,d Kellie Swanton, BA,c

William H. Shrank, MD, MSc

aDivision of General Medicine, Brigham and Women’s Hospital, Boston, Mass; bDepartment of Health Care Policy and Center for PrimaryCare, Harvard Medical School, Boston, Mass; cDivision of Pharmacoepidemiology, Brigham and Women’s Hospital, Boston, Mass; dCVS

aremark, Northbrook, Ill.

hain Drug Stores

0002-9343/$ -see fhttp://dx.doi.org/10

ABSTRACT

BACKGROUND: Given the huge burden of coronary artery disease and the effectiveness of medicationtherapy, understanding and quantifying known impacts of poor medication adherence for primary andsecondary prevention is crucial. We sought to systematically review the literature on this topic area witha focus on quantified cost and clinical outcomes related to adherence.METHODS: We conducted a systematic review of the literature between 1966 and November 2011 using afixed search strategy, multiple reviewers, and a quality rating scale. We found 2636 articles using thisstrategy, eventually weaning them down to 25 studies that met our inclusion criteria. Three reviewersindependently reviewed the studies and scored them for quality using the Newcastle Ottawa Scoring Scale.RESULTS: We found 5 studies (4 of which focused on statins) that measured the impact of medication adherenceon primary prevention of coronary artery disease and 20 articles that focused on the relationship betweenmedication adherence to costs and outcomes related to secondary prevention of coronary artery disease. Mostof these latter studies focused on antihypertensive medications and aspirin. All controlled for confoundingcomorbidities and sociodemographic characteristics, but few controlled for likelihood of adherent patients to havehealthier behaviors (“healthy adherer effect”). Three studies found that high adherence significantly improves healthoutcomes and reduces annual costs for secondary prevention of coronary artery disease (between $294 and $868 perpatient, equating to 10.1%-17.8% cost reductions between high- and low-adherence groups). The studies were all ofgenerally of high quality on the Newcastle Ottawa Scale (median score 8 of 9).CONCLUSIONS: Increased medication adherence is associated with improved outcomes and reduced costs,but most studies do not control for a “healthy adherer” effect.© 2013 Elsevier Inc. All rights reserved. • The American Journal of Medicine (2013) 126,357.e7-357.e27

KEYWORDS: Coronary artery disease; Costs; Medication adherence; Outcomes; Pharmacoeconomics; Systematicreview

p

Cardiovascular disease, of which coronary artery disease isthe predominant form, is the most common cause of deathin the US and has an annual health care cost of approxi-mately $475 billion.1 Adherence to medications has been

Funding: AB did not receive financial support for work on this study.S, NC, and KS were supported by a research grant from CVS Caremark.M is employed by CVS Caremark. All data analysis and evaluation tooklace at Brigham and Women’s Hospital.

Conflict of Interest: AB declares no conflicts of interests. WS andC have received funding from Aetna and Express Scripts. WS has

eceived funding from Lilly, Teva, and the National Associations of
and has served as a consultant for United Healthcare.
ront matter © 2013 Elsevier Inc. All rights reserved..1016/j.amjmed.2012.09.004

associated with improved outcomes in primary and second-ary coronary artery disease in some early studies.2 Furtherliterature suggests that, as is the case with diabetes,3 im-roved adherence to a number of chronic disease medica-

NC serves as a consultant for Mercer. OM is employed by CVSCaremark.

Authorship: All authors had access to the data and played a role inwriting this manuscript.

Requests for reprints should be addressed to Asaf Bitton, MD, MPH,Division of General Medicine, Brigham and Women’s Hospital, Depart-ment of Health Care Policy, Harvard Medical School, 1620 Tremont St.,
3rd Floor, Rm 3-002P, Boston, MA 02120.
E-mail address: [email protected]

mailto:[email protected]

http://dx.doi.org/10.1016/j.amjmed.2012.09.004

sdteahO

y adh

357.e8 The American Journal of Medicine, Vol 126, No 4, April 2013

tions also may reduce overall health care costs.4 Supportfor new models of patient care, such as the patient-centered medical home and accountable care organiza-tions, is in part premised upon the potential for thesemodels to encourage improved patient self-managementand medication adherence.5

However, a previous system-atic review2 on the association be-tween adherence and coronary ar-tery disease outcomes suggestedthat the existing literature wasproblematic due to heterogeneousmethods and few available studiesthat directly measured outcomes.There is a large amount of litera-ture on factors associated withadherence or nonadherence, butmuch less on what outcomes areimproved through better adher-ence, and what the return on in-vestment in improved adherencecould be.

A further issue exists aroundwhether the association betweenadherence and better outcomes isconfounded by endogenous char-acteristics of the patients who arepredisposed to take medication ina more consistent way. Termedthe “healthy adherer” effect, this association could signifi-cantly amplify the measured relationship between medica-tion adherence and health outcomes.

The aim of this study was to review systematically theliterature on the relationship between medication adherenceand primary and secondary prevention of coronary arterydisease, with a focus on key coronary artery disease-relatedmortality, service utilization, and cost outcomes. We alsocarefully appraised whether these studies controlled for thehealthy adherer effect, and evaluated the strength of theevidence published in the prior decade since the last sys-tematic review of the literature. Finally, we link the resultsof these studies to current discussion of health care deliverysystem reform in the US.

METHODS

Data SourcesWe conducted a systematic search to identify studies ad-dressing the effect of medication adherence on patient out-comes and costs. Initial searches were limited to articlespublished in MEDLINE and EMBASE between 1966 andNovember 30, 2011. We performed our search with theassistance of a professional librarian. The search focused onany term relating to adherence and coronary artery disease(eg, medication adherence AND cardiac ischemia OR myo-cardial infarction OR coronary heart disease). Our full

CLINICAL SIGNIF

● All 25 studies inatic review fountion adherenceprimary and secoronary artery d

● Studies of secodisease preventicost reductions (tient) betweenence patients.

● Significant methity exists acrossnot control foradherent patienthaviors (“health

search strategy can be found in the Appendix. Articles 1

containing at least one search term were included in thereview. Initial articles selected were mined for additionalreferences.

Study SelectionArticles were included if they re-ported original cost or quality out-come results. Studies that only ex-amined predictors of medicationadherence without outcome mea-sures comparing adherent andnonadherent populations of pa-tients were not included. Case re-ports, case series, and simulationand modeling studies were ex-cluded. Articles that did not di-rectly pertain to coronary arterydisease, or those not published inEnglish, also were excluded. Tworeviewers (AB, WS) evaluated thetitles and abstracts of search re-sults to identify potentially rele-vant articles, and 3 reviewers (AB,WS, KS) assessed complete arti-cles for inclusion.

Data ExtractionThree reviewers (AB, WS, KS)extracted data from selected arti-

cles and resolved differences by consensus. Variables as-sessed included the main research questions, patient popu-lation, inclusion criteria, measure of adherence, studydesign (with a focus on how authors accounted for sourcesof unmeasured confounding), and results regarding out-comes and costs (Table 1,6-8,10,30 and Table 29,11-29). Tworeviewers (AB, KS) used the Newcastle-Ottawa Scale toassess the quality of each cohort study on a scale of 1 to 9.Disagreements were adjudicated by a third reviewer (WS).Studies were grouped into 2 main evidence tables: articlespertaining to the effects of medication adherence on primaryprevention of coronary artery disease (Table 1), and thosepertaining to secondary prevention of coronary artery dis-ease (Table 2).

RESULTSOur search strategy yielded a total of 2636 articles, of whichwe selected 25 for inclusion in the analysis based on a prioricriteria above (Table 1 and Table 2). A total of 80% (20/25tudies pertained to secondary prevention of coronary arteryisease, with 20% regarding primary prevention. Amonghe 5 studies of primary prevention of coronary artery dis-ase, 80% focused exclusively on statins. Quality scoring ofll these studies found that generally the studies were ofigh quality (median and mode score 8/9 on the Newcastlettawa Scale). Of the 20 studies on secondary prevention,

CE

d in this system-t higher medica-ficantly improvedry prevention ofe outcomes.

coronary arteryow 10.1%-17.8%and $868 per pa-and low-adher-

gical heterogene-studies; most do

kelihood of moreave healthier be-

erer effect”).

ICAN

cluded thasigni

condaiseas

ndaryon sh$294high-

odolothesethe lis to h

4 (70%) included antihypertensive medications.

Table 1 Adherence Effects on Primary Prevention of Coronary Artery Disease

Study: FirstAuthor, Country

Patient Population &Inclusion Criteria

Coronary ArteryDisease Medications Adherence Measure Design Features Outcomes

NewcastleOttawaRating

Bouchard (2007),Canada6

Cohort of patients collected(n � 20,543) usingQuébec provincial publichealth insurancedatabases. Eligiblepatients were 50-64 yearsold (mean � 58) withoutcardiovascular disease;newly treated withstatins from 1998-2000.

Statins Adherence level waspercentage ofprescribedmedication dosesused over specifiedperiod and classifiedas �90% or �90%.Measure based on thequantity dispensedand number of dayssupplied for eachfilled prescription.

Used a nested case-control design tostudy nonfatal coronary arterydisease. Every case was matchedwith 20 randomly selected controls.Rate ratios of nonfatal coronaryartery disease were calculatedthrough conditional logisticregression adjusted for age, sex,socioeconomic status, diabetes andhypertension. Also conductedsensitivity analyses around differentadherence cut-off levels andestimated the size necessary for anunmeasured confounding variable toalter the results.

Adherence �90% was associatedwith fewer nonfatal coronaryartery disease events (rateratio .81; .67, .97) comparedwith adherence �90%adherence.

8/9

Breekveldt-Postma(2008),Netherlands7

Used a pharmacy recordlinkage system toexamine drug dispensingand hospitalizations forover 2 million patients inthe Netherlands.Included new users ofantihypertensivedrugs � 18 years of agebetween 1993 and 2002.Excluded patient usingantihypertensive drugsfor secondary preventionof coronary arterydisease. 77,193 met theinclusion criteria of thisstudy

All antihypertensivedrugs except loopdiuretics (toexclude heartfailure only)

Permissible gaps ofantihypertensivedrugs therapy weredetermined.Persistence ofantihypertensivedrugs use wasdefined as thenumber of days ofcontinuous use ofany antihypertensivedrugs from the indexdate meeting thosecriteria. Patientswere defined as 2-year persistent ornonpersistentantihypertensivedrugs users.

Patients initially followed for 2 yearsto determine persistence withantihypertensive drugs; then for afurther 2 years or until first hospitaladmission for acute myocardialinfarction or stroke, death, or end ofstudy period. Effect ofnonpersistence with antihypertensivedrugs during first 2 years oftreatment on risk of hospitalizationfor acute myocardial infarction orstroke thereafter was estimated inCox regression model. Adjusted forage, sex, initial prescriber, numberof different antihypertensive drugclasses in first 2 years of treatment,initial antihypertensive drug class,year of start of antihypertensive druguse, and use of other drugs beforestart of outcome follow-up.

Overall percentage ofnonpersistent patients was55%. The lowestnonpersistence rates were forARBs and ACE inhibitors(40%), with the highest ratesfor CCBs, BBs, and diuretics(range 54%-61%).Nonpersistent antihypertensivedrug use associated with anincreased risk of acutemyocardial infarction (RR1.15; 95% CI, 1.00-1.33) andincreased risk of stroke (RR1.28; 95% CI, 1.15-1.45).

8/9

357.e9Bitton

etal

CoronaryArtery

DiseaseOutcom

esand

PatientAdherence

Table 1 Continued

Study: FirstAuthor, Country

Patient Population &Inclusion Criteria

Coronary ArteryDisease Medications Adherence Measure Design Features Outcomes


Corrao (2010),Italy8

Identified 90,832 adultswithout ischemic heartdisease, newly treatedwith statins in 2002 to2003 from a healthservices database

Statins PDC by therapy withstatins, classified asvery low (�25%),low (26%-50%),intermediate (51%-75%), or high(�75%) coverage.

Proportional hazards model to estimateHR and 95% CIs for associationbetween time-dependent categoriesof PDC and time of ischemic heartdisease hospitalization. Controlledfor age, sex, type of statin dispensedat start, Charlson comorbidity index,current use of other drugs (ie,antidiabetics, nitrates,antihypertensive drugs, or othercardiac medications), and whether ornot a patient switched statins.

A total of 59.5% of patients hadlow or very low adherence tostatins. In adjusted models,patients with low,intermediate, or high statincoverage had HR (95% CI) forischemic heart diseasehospitalization of .85 (.72-.98), .82 (.71-.95), and.81 (.71-.94), respectively,compared with patients withvery low coverage.

8/9

Dragomir (2010),Canada30

A total of 55,134 who werenewly treated withstatins were followedfrom public healthinsurance databases inQuebec. The authorsincluded patients aged45-85 withoutcardiovascular diseasewho were initially treatedwith statins between1999 and 2002, with atleast 3 years of follow-up.

Statins Adherence to statinsmeasured using theMPR-, andcategorized as �80%or �80%.

Adjusted OR of cardiovascular eventsbetween high and low adherencegroups was estimated usingpolytomous logistic analysis.Covariates included age, sex,hypertension, overall health status(modified Van Korff chronic diseasescore), social assistance status,previous hospitalization in the yearbefore entry of cohorts, and meannumber of physician visits per year.They analyzed mean costs of directhealth-care services categorized asmedical services, hospitalization, ordrug related.

Mean MPR in high adherencelevel was 96% vs 42% for lowadherence. Patients with lowstatin adherence were morelikely to have coronary arterydisease (OR 1.07; 95% CI1.01-1.13), cerebrovasculardisease (OR 1.13; 95% CI,1.03-1.25), and chronic heartfailure within 3-year period offollow-up (OR 1.13; 95% CI,1.01-1.26). Low adherence tostatins associated withincreased risk ofhospitalization (OR 1.04; 95%CI, 1.01-1.09). Amonghospitalized patients, lowstatin adherence associatedwith statistically significantincrease of hospitalizationcosts by $1060/patient for the3-year period.

8/9

357.e10The

American

Journalof

Medicine,

Vol126,

No4,

April2013

tahctp

cble

1Co

ntin

ued

dy:

Firs

tth

or,

Coun

try

Pati

ent

Popu

lati

on&

Incl

usio

nCr

iter

iaCo

rona

ryAr

tery

Dise

ase

Med

icat

ions

Adhe

renc

eM

easu

reDe

sign

Feat

ures

Outc

omes

New

cast

leOt

taw

aRa

ting

reau

lt(2

009)

,ad

a10

Aco

hort

of11

5,29

0pa

tien

tsw

asco

nstr

ucte

dus

ing

publ

iche

alth

insu

ranc

eda

taba

sein

Québ

ec.

Pati

ents

wer

eag

ed45

to85

year

sw

itho

utca

rdio

vasc

ular

dise

ase,

and

new

lytr

eate

dw

ith

stat

ins

betw

een

1999

and

2004

wer

ein

clud

ed.

Stat

ins

MPR

asm

easu

reof

adhe

renc

e,ca

tego

rize

das

�80

%or

�80

%.

Use

da

nest

edca

se-c

ontr

olde

sign

tost

udy

CAD.

Case

sm

atch

edfo

rag

ean

ddu

rati

onof

follo

w-u

pw

ith

rand

omly

sele

cted

cont

rols

.

Hig

had

here

nce

was

asso

ciat

edw

ith

rela

tive

risk

redu

ctio

nof

18%

(RR

.82;

95%

CI.7

7-.8

7)co

mpa

red

wit

hlo

wer

adhe

renc

ele

vel.

8/9

ARBs

�an

giot

ensi

nre

cept

orbl

ocke

r;AC

E�

angi

oten

sin-

conv

erti

ngen

zym

e;CA

D�

coro

nary

arte

rydi

seas

e;CC

Bs�

calc

ium

chan

nelbl

ocke

rs;

BBs�

beta

-blo

cker

s;CI

�co

nfide

nce

inte

rval

;H

R�

haza

rdti

o;M

PR�

med

icat

ion

poss

essi

onra

tio;

OR�

odds

rati

o;PD

C�

prop

orti

onof

days

cove

red;

RR�

rela

tive

risk

.

357.e11Bitton et al Coronary Artery Disease Outcomes and Patient Adherence

Primary Prevention: StatinsFour studies evaluated adherence to statin medicationsfor primary prevention. Of these studies, 3 came from thesame research group in Canada using a similar retrospec-tive approach to claims data.6,10,30 In 2 studies,6,10 usingvariations of the medication possession ratio (MPR), highadherence (�80% MPR) was associated with adjustedrisk reduction of between 18% and 19%, compared withlow-adherence patients. Dragomir et al (2010)30 found anadjusted odds ratio for incident hospitalization for coro-nary artery disease of 1.07 (95% confidence interval [CI],1.01-1.13) in patients with �80% adherence. Theyestimated that among patients who were hospitalized,low adherence to statins before hospitalization was asso-ciated with a statistically significant increase in hospital-ization costs of Canadian $1060 ($1.00 Canadian �$0.974 USD, December 12, 2011) per patient hospital-ized over 3 years.

Primary Prevention: AntihypertensiveMedicationsOne study from the Netherlands evaluated claims data froma pharmaceutical claims database.7 The investigators foundthat over a 2-year period after treatment initiation, thosepatients with medication nonpersistence had a 15% in-creased risk (95% CI, 1.00-1.33) of hospitalization for myo-cardial infarction.

Secondary Prevention: StatinsThree of the 7 studies about statin adherence in secondaryprevention of coronary artery disease focused exclusivelyon that drug family. Notably, 2 of these 3 studies in-cluded cost end-point measures. Aubert et al12 retrospec-ively analyzed adherence measured as MPR for 2 years,nd found that patients with MPR �80% had 8 moreospitalizations per 100 patients than more adherentounterparts (3.8% absolute difference), leading to a sta-istically significant annual difference in costs of $868er patient ($4908 vs $4040). Pittman9 studied retrospec-

tive claims data in a similar fashion, finding a significantadjusted odds ratio of 1.26 for cardiovascular diseasehospitalization in the lowest- (MPR �60%) comparedwith the highest-adherence group. This translated into anannual per-patient difference of $294 between thesegroups ($2395 vs $2689). A retrospective Scottish cohortstudy measuring proportion of days covered (PDC) witha statin found that, compared with patients with no ad-herence, those with high PDC (�80%) had adjusted haz-ard ratios (HR) of .19 (95% CI, .08-.47) for recurrentmyocardial infarction, and HR .47 (95% CI, .22-.99) forall-cause mortality.29 Shalev26 found, in a similar Israeliohort study, that the adjusted HR for each 10% increase

was .94 (95% CI, .93-.95).Ta Stu

Au Per

Can

ra

Table 2 Adherence Effects on Secondary Prevention of Coronary Artery Disease

Study: FirstAuthor,Country

Patient Population, &Inclusion Criteria

Coronary Artery DiseaseMedications Adherence Measure Design Features Outcomes


Amin (2009),USA11

Consecutive patients admittedto a coronary care unit at alarge urban hospital inChicago with myocardialinfarction during 2003 and2004.

Aspirin, clopidogrel,angiotensin-converting enzymeinhibitor,angiotensin receptorblocker, statin

Proportion of days covered(PDC). Poor compliancewas defined as �80%.The authors constructeda noncompliance score(0,1) for each of 5meds. Patients withscore �4 were labeledas “noncompliant” inmain analysis

Cox proportional hazard modelsadjusting for severity ofcoronary artery disease, leftventricular dysfunction�30%, acuity of myocardialinfarction, diabetes,dyslipidemia, hypertension,race, insurance status,current/past smoking,cocaine or polysubstanceabuse, homelessness, meanfrequency of all 5 classes ofmeds, global registry ofacute coronary events riskscore for predicting deathduring 6 months post-discharge, and number ofoutpatient cardiology follow-up visits.

Main outcome was a compositeof all-cause mortality andreinfarction. Patients non-compliant with �4 meds hadHR 2.83 (95% CI, 1.60-5.01). Kaplan-Meier survivalcurves separated at 6months (log rank P �.001).Noncompliance with 2 or 3meds was statisticallynonsignificant, whereasnoncompliance with 0 or 1meds had a HR of .44 (95%CI, .25-.78).

8/9

Aubert(2010),USA12

Retrospective analysis ofstatin adherence onhospitalization rates anddirect medical costs in10,227 total patients.Patients included in thestudy received first statinprescription in 2001 and2002, had continuoushealth coverage for 6months prior and 3 yearsafter index prescription,and continued therapy afterindex date.

Statin Medication adherencemeasured using the MPRover the first 2 years oftherapy. Nonadherencewas defined as MPR�80%. Patients alsostratified into quartilesof MPR

Cost analyzed using linearregression. Logisticregression model used forhospitalization eventcontrolling for age, sex,chronic disease score,cardiovascular diseases,cardiovascular medications,and other comorbidities

A total of 33 hospitalizationsper 100 patients were foundin the nonadherent group vs25 hospitalizations/100patients in the adherentgroup.

Total of 23 cardiovascular -related conditions innonadherent vs 16 inadherent group. 19.3%nonadherent patientshospitalized vs 15.5%adherent patients hospitalizedduring the study period. Costanalysis showed $4908 annualper-patient direct medicalcosts in nonadherent group vs$4040 in adherent group (allP �.01)

8/9

357.e12The

American

Journalof

Medicine,

Vol126,

No4,

April2013

Table 2 Continued





Chapman(2010),USA13

Retrospective cohort studyevaluating effect of single-pill amlodipine -atorvastatin vs 2-pillcalcium channel-blockeradherence on cardiovascularevents. 19,447 patientsalready taking one of the 2medications, who initiatedsingle combination pill oradded other medicationwere followed.)

Single-pill amlodipine -atorvastatin, calciumchannel blocker,statin

Proportion of dayscovered. NonadherencePDC �80% for 6-monthstudy period.

Cox proportional hazardsmodels adjusting for age,sex, therapy type,geographic region, insurancestatus, related comorbidities,and number of priorantihypertensivemedications.

Adherence to calcium channelblocker and statinmedications was associatedwith lower risk ofcardiovascular events, with anincidence rate of 1.79 vs 2.26in the adherent vsnonadherent groups. Patientsreceiving combination pillwere more likely to beadherent than the 2-pillgroup. (56.5% vs 21.4%)

8/9

Choudhry(2011),USA14

Patients discharged from aftera myocardial infarction whoreceived both medical andprescription drug coveragethrough one health insurer

Angiotensin-convertingenzyme inhibitor,angiotensin receptorblocker, beta-blocker

Statin

Medication adherencemeasured using the MPRmultiplied by 100 togenerate absoluteadherence percentages.Nonadherenceconsidered MPR �80%

Randomized controlled trialthat assigned patients to fullprescription drug coverage orusual coverage. Authorsanalyzed effect of drug costson medication adherence asa secondary outcome.Utilized an identity linkfunction with normallydistributed errors to compareMPRs between 2 randomizedgroups, and a logit linkfunction with binarydistributed errors to comparerates of full adherence

Rates of adherence increased by5.6% (CI 3.4-7.7) forangiotensin-convertingenzyme inhibitor orangiotensin receptor blockers,by 4.4% (CI 2.3-6.5) for beta-blockers; by 6.2% (CI 3.9-8.5)for statins; and by 5.4% (CI3.6-7.2) for all medicationclasses. (P �.001 for allcomparisons). Odds of fulladherence to study medsincreased by 31% to 41%(P �.001). Out-of-pocketspending was reduced forboth prescription drugs(relative spending, .70; 95%CI .65-.75; P �.001) and fornondrug medical services(relative spending, .82; 95%CI .72-.94; P � .005). Meantotal spending was $66,008 infull-coverage group and$71,778 in usual-coveragegroup (relative spending, .89;95% CI, .50-1.56; P � .68).

N/A

357.e13Bitton

etal

CoronaryArtery

DiseaseOutcom

esand

PatientAdherence

Table 2 Continued





Corrao(2011),Italy15

Population-based prospectivecohort study of 242,594patients in a nationalItalian database newlytreated for hypertensionduring 2000-2001, analyzedthrough 2007).

BP meds (all) Proportion of dayscovered. Four levels ofadherence: very low(�25%), low (26-50%),intermediate (51-75%)and high (�75%)

Cox proportional hazardregression model adjustedfor sex, age, Charlson index,antihypertensive drugsprescribed (both type andnumber), and use ofnonhypertensive drugs Twotypes of sensitivity analyses:varying the definition ofadherence to treatment. andevaluating confounders onthe results (ie, depression)

Compared with patients withvery low adherence,reduction in risk was 20%for intermediate adherence,and 25% for high adherence(P �.05). Significantreduction in HR withincreasing adherence wasstill observed afteraccounting for confounderssuch as age or length offollow-up

8/9

Gallagher(1993),USA16

Post hoc analysis ofrandomized controlled trialof propranolol followingmyocardial infarction. 505of original 3837 womenincluded in this analysisaged 30-69 years who hadadherence data available

Beta-blocker(propranolol)

At each 3-month follow-upinterval during the RCT,patients asked to returnall unused medications.Adherence defined asproportion of prescribedmeds actually taken.Adherence in eachpatient was the mean ofthe quarterly adherenceestimates, highadherence defined as�75%.

Cox proportional hazard modelsincluded age, smokingstatus, race, education,clinical severity compositescore, high risk myocardialinfarction, sociodemographiccomposite score, myocardialinfarction severity(postevent complications,including heart failure),heart failure at baseline

Overall mortality was mainoutcome. Of 505 womenwith fully available data,adherence �75% wasassociated with 2.4increased RR of mortality(95% CI, 1.1-5.6).Adjustment for any of thecovariates led to RR ofbetween 2.5 and 3.0 (P�.02)

N/A

Gehi (2007),USA17

Prospective evaluation of1015 patients with stablecoronary heart disease.Patients enrolled between2000 and 2002, andfollowed by telephoneannually for average of 3.9yrs.

All meds (specificclasses not specified)

Single survey questionabout overall medicationadherence, using aLikert scale.

Cox proportional hazardscomputed and adjusted forage, sex, race, education,hypertension, diabetes,smoking status, depression,angina, number ofcardiovascular meds, beta-blocker use, statin use,cholesterol levels, leftventricular ejection fraction.

Among 1007 with completefollow-up data, nonadherentpatients had HR ofcardiovascular events of2.3 (95% CI, 1.3-4.3).

8/9

357.e14The

American

Journalof

Medicine,

Vol126,

No4,

April2013

Table 2 Continued





Ho (2006),USA18

Registry, interview, and chartanalysis of 2498 patientswith acute myocardialinfarction in 2003-2004recruited from 19 UShospitals (academiccenters, inner-cityhospitals, several single-payer systems, andnonuniversity hospitals).

Aspirin, beta-blocker,statin

Main adherence measurewas medicationdiscontinuation at 1month followingmyocardial infarction,abstracted from hospitalrecords and telephoneinterviews afterdischarge.

Adjusted Cox proportional hazardmodels for age, sex, race,insurance, marital status,education status, diabetes,hypercholesterolemia,hypertension, heart failure,cerebrovascular disease, renalfailure, chronic lung disease,smoking, depression, priorcoronary disease, myocardialinfarction type, coronaryrevascularization socialsupport cost-relatednonadherence, and site ofenrollment

Patients who discontinued useof all medications were atincreased adjusted risk ofdeath during 1-year follow-up (HR 3.81; 95% CI, 1.88-7.72) Beta-blocker therapydiscontinuation (HR 1.96;95% CI, 1.10-3.45) wasindividually associated withhigher mortality, as wasstatin therapydiscontinuation (HR 2.86;95% CI, 1.47-5.55)

8/9

Ho (2008),USA19

Retrospective cohort study of10,447 patients in KaiserPermanente Colorado withcoronary artery diseaseevaluated the impact ofmedication nonadherenceand therapy intensificationon reaching target BP goals.Modeled 3 clinically relevantSBP groups: 1) patients withcontrolled blood pressurethat remained stable overtime based on systolicpressure �140 mm Hg(normal-normal); 2) patientswho started with high bloodpressure that decreased overtime (hi-normal); and 3)those who started with highblood pressure that remainedhigh (hi-hi).

Blood pressure meds(all)

Medication adherencecalculated frompharmacy records as thePDC for filledprescriptions ofantihypertensivemedications.

Because statins wereprescribed for most patients(85%), evaluated for healthyadherer effect using statinadherence (PDC �.80) as asurrogate. Controlled for age,sex, smoking status, previousrevascularization procedureor myocardial infarction,hypertension,cerebrovascular accident,diabetes, atrial fibrillation,heart failure, obstructivelung disease, cancer, sleepapnea, hyperlipidemia,dementia, depression

Medication nonadherence wassignificantly associated withuncontrolled blood pressure(hi-hi group) compared withhigh blood pressure thatbecame controlled over time(hi-normal group) (OR 1.73;95% CI, 1.34-2.24). Noassociation found betweenmedication nonadherenceand being in the hi-normalgroup compared with thenormal-normal group. Statinadherence associated withBP control (OR .53; 95% CI,.44-.65), even afteradjustment for bloodpressure medicationadherence and therapyintensification (OR .54; 95%CI, .43-.66).

8/9

357.e15Bitton

etal

CoronaryArtery

DiseaseOutcom

esand

PatientAdherence

Table 2 Continued





Ho (2008),USA20

Retrospective cohort study of15,767 patients in KaiserPermanente Colorado withcoronary artery diseaselooking at the impact ofmedication nonadherenceand therapy intensificationon reaching target bloodpressure goals.

Beta-blockers,angiotensin-converting enzymeinhibitors,angiotensin receptorblockers, statin

Medication adherence wascalculated as PDC, forthe following classes:Beta-blockers,angiotensin-convertingenzyme inhibitors,angiotensin receptorblockers, and statinmedications. Patientswere classified as“nonadherent” based ona PDC �.80, calculatedat 180-day intervals

Controlled for age, sex,smoking status, previousrevascularization procedureor myocardial infarction,hypertension,cerebrovascular accident,diabetes, atrial fibrillation,heart failure, obstructivelung disease, cancer, sleepapnea, hyperlipidemia,dementia, depression

Medication nonadherence wassignificantly associated inadjusted models withincreased all-cause mortalityrisk for beta-blockers (HR1.50, 95% CI, 1.33-1.71),angiotensin convertingenzyme inhibitors (HR 1.74,95% CI, 1.52-1.98), andstatins (HR 1.85, 95% CI,1.63-2.09). Nonadherencewas significantly associatedwith higher risk ofcardiovascular mortality forbeta-blockers (HR 1.53, 95%CI, 1.16-2.01), angiotensin-converting enzyme inhibitors(HR 1.66, 95% CI, 1.26-2.20), and statins (HR 1.62,95% CI, 1.124-2.13).

8/9

Irvine(1999),Canada21

1141 patients participated inmulticenter double-blind,placebo-controlled trial ofamiodarone treatment forthe prevention of suddencardiac death in patientswith frequent or repetitiveventricular ectopy after anacute myocardial infarction.

Amiodarone Adherence assessed by pillcount at each clinicvisit, calculated as theproportion of pills notreturned, multiplied by100, averaged over the2 years of follow-up forthe trial

Controlled for age, previousmyocardial infarction, historyof heart failure; heart failureat the time of hospitalizationfor the index acute myocardialinfarction; ventricular ectopyon baseline 24-hour Holtermonitoring record; highventricular ectopy rate,diabetes

In multivariate models, lowadherence associated with asignificantly higher risk oftotal cardiac deaths (RR2.04, 95% CI, 1.12-3.72, P�.02) and all-causemortality (RR 2.25, 95% CI,1.27-3.97, P �.001) in theplacebo group and totalcardiac deaths (RR 2.49,95% CI, 1.32-4.72, P �.01)and all-cause mortality inthe amiodarone treatmentgroup (RR 2.34, 95% CI,1.32-4.17, P �.004).

7/9

357.e16The

American

Journalof

Medicine,

Vol126,

No4,

April2013

Table 2 Continued





Kleiner(2009),USA22

Retrospective study of beta-blocker adherence onmortality and repeatmyocardial infarction usinghealth insurance claimsrecords of all patients whowere discharged alive afterAMI between 1/03 and 6/04, with prescription drugcoverage, and prescribedBBs (n � 3923)

Beta-blocker Calculated adherenceusing claims dataalgorithm, similar toPDC. Adherence split itinto quartiles

Adjusted for age, sex, income,presence of smoker inhousehold, and specialty ofpatient’s primary care physician.Also adjusted comorbidity riskscore (quartiles) measuredbefore the initial event. Createdoverall model controlling forpresence of atherosclerosis,aortic aneurysm, cerebrovasculardisease, depression,dysrhythmias, ischemic heartdisease, liver disease, neoplasm,renal disease, obesity, diabetes,lipid disorders, congestive heartfailure, interventional testing,cardiac testing, carotidendarterectomy, peripheralvascular disease, andmedications taken.

In the adjusted survival modelthe hazard ratio was.37 (95% CI, .25-.54) for themost adherent vs leastadherent patients. Similarresults were found formortality or admission forrepeat myocardial infarction,with an adjusted hazardratio of .43 (95% CI, .33-.57)

8/9

Newby(2006),USA23

Registry study of 31,750living patients withdocumented coronary arterydisease who underwentcardiac procedure at largeacademic medical centersince 1969 and had at least2 consecutive follow-upsurveys completed between1995 and 2002. Stratifiedsubgroups into those withand without heart failure

Aspirin, beta-blocker;statin

Defined 4 patterns of anindividual’s medication usein survey questions: 1)Always using; 2) neverusing; 3) no use to use(positive “converters”); 4)use to no use (negative“converters”). “Consistentuse” was defined asreporting a medication useon �2 consecutiveoccasions and continuinguntil death, withdrawalfrom survey, or end ofstudy period. Patients wereconsidered inconsistentusers if did not meetcriteria for any patterns.

Controlled age, sex, and race,year of entry into database,smoking, diabetes,hyperlipidemia, hypertension,prior documentedcardiovascular events orprocedures cerebrovasculardisease, peripheral vasculardisease, chronic obstructivepulmonary disease, renaldisease, heart failure, pepticulcer disease, systolic anddiastolic BP at entry, eventsduring the study period, andmedications at first follow-up.

Consistent medication use ofthe following medicationswas associated with loweradjusted mortality: aspirinHR .58 and 95% CI, .54-.62;beta-blocker, HR .63 and95% CI, .59-.67; statin, HR.52 and 95% CI, .42-.65; all3 medications, HR .67 and95% CI, .59-.77; aspirin andbeta-blocker, HR .61 and95% CI, .57-.65

7/9

357.e17Bitton

etal

CoronaryArtery

DiseaseOutcom

esand

PatientAdherence

Table 2 Continued





Pittman(2010),USA24

Retrospective analysis of625,620 in a nationalpharmacy benefits databaseof de-identified pharmacyand medical claims amongpatients with hypertension,with 2 or more claims forantihypertensivemedications. Excludedpatients �64 years; thosewith cancer or HIV; thosewithout continuouseligibility for prescriptiondrug coverage; thosewithout data from 2006-2008

Antihypertensivemedications (all)

MPR with antihypertensivemedications 3 adherencecategories: adherent(MPR �80%); moderateadherence (MPR60-79%); low adherence(MPR �60%)

Binary outcome variable ofhigh adherence vs low-to-moderate adherence. Logisticregression analysis used toassess possible predictors ofadherence, which includeddemographics alone andtogether with comorbidities,chronic disease score,antihypertensive class, out-of-pocket cost formedication, and use of mailorder. Logistic regressionused to evaluate associationof level of adherence withCV-related hospitalizationsand emergency departmentvisits. Adjusted generallinear models used todetermine mean cost foreach level of adherence.

After adjustment, low andmoderate adherence patientsmore likely to behospitalized for a CV event(ORs 1.33 [low] and 1.28[moderate]) or have 1 ormore CV-related ED visit(ORs 1.45 [low] and 1.27[moderate]). Patients inhigh adherence group hadsignificantly lower totalhealth care costs ($7182)compared with patients inmoderate ($7560) and low($7995) adherence groups(P �.001 for both)

8/9

Pittman(2011), USA9

Retrospective cohort study of381,422 adults with using astatin medication during2008-2009. Compiled cohortfrom integrated pharmacyand medical claimsdatabase. Assessed relationamong statin adherence,subsequenthospitalizations, and healthcare costs.

Statin MPR with statins. 3adherence categories:adherent (MPR �80%);moderate adherence(MPR 60%-79%); lowadherence (MPR �60%)

Logistic regression analysis tostudy predictors ofadherence controlling forage, sex, comorbidities, out-of-pocket costs formedication chronic diseasescore, and use of mail-orderpharmacy. Generalized linearmodels performed todetermine average costs infollow-up period for eachlevel of adherence, adjustedfor age, sex, comorbidities,first-year total medicationand health costs.

Adjusted OR of cardiovasculardisease-relatedhospitalization (comparedwith fully adherent group):1.26 (low adherence) and1.12 (moderate adherence)(P �.05). Fully adherentpatients had cardiovascular-related medicalcosts � $2395 � 20.5compared with moderate($2583 � $40.4) and low($2689 � $43.9) adherence(P �.001).

8/9

357.e18The

American

Journalof

Medicine,

Vol126,

No4,

April2013

Table 2 Continued





Rasmussen(2007),Canada25

Population-based,observational, longitudinalstudy of 31,455 elderly (66and older) acute myocardialinfarction survivors between1999 and 2003 in Ontario.Patients had to havesurvived �1 year and 3months after initialhospitalization All patientsfilled a prescription forstatins, beta-blockers, orcalcium channel blockers.The use of calcium channelblockers was considered acontrol given absence ofclinical trial-proven survivalbenefits.

Statin; beta-blocker,calcium channelblocker

Claims data used tocalculate the PDC.Adherence subdivided apriori into 3 categories:high (PDC �80%),intermediate (PDC 40%-79%), and low (PDC�40%).

The research team used Ontarioacute myocardial infarctionmortality prediction rule tocontrol for severity ofillness, including age, sex,severity of cardiac disease,and presence of coexistingillnesses (stroke, diabetes,cancer, and acute or chronicrenal disease). They alsoadjusted for whetherpatients receivedrevascularization procedures,admission for ischemic heartdisease, metabolic diseases,mental disorders, stroke,diseases of the respiratorysystem, concomitant use ofother cardiac meds, and thenumber of all-causeadmissions during the studyperiod. To disentanglebiological drug effects fromhealthy adherer effects, theyevaluated impact ofcardiovascular drugadherence on sample ofprespecified outcomes withneither clinical evidence norbiological plausibility (lung,breast, and prostate cancer).

Among statin users, comparedwith their high-adherencecounterparts, low adherencepatients had the highestsignificantly increasedmortality risk (adjusted HR1.25; 95% CI, 1.09-1.42;P � .001). This increased riskwas somewhat attenuatedfor intermediate adherers(adjusted HR 1.12; 95% CI,1.01-1.25; P � .03). Similarbut less pronounced dose-response–type adherence-mortality associationobserved for beta-blockers,in which low adherence HRwas 1.13 (95% CI,1.03-1.25); intermediate was1.01 (95% CI, .93-1.09).Mortality and cancer-relatedadmissions were notassociated with adherence tocalcium channel blockers,outcomes for whichbiological plausibility do notexist.

9/9

357.e19Bitton

etal

CoronaryArtery

DiseaseOutcom

esand

PatientAdherence

Table 2 Continued





Shalev(2009),Israel26

Retrospective cohort study ofadult enrollees in an Israelihealth maintenanceorganization who initiatedstatin treatment from 1998through 2006. Two cohortswere formed: primary(136,052) and secondaryprevention statin users(93,866).

Statin PDC with statins Full multivariate model age,sex, marital status,nationality, SES, place ofresidency, immigrationstatus, nationality, disabilitystatus, presence of chroniccomorbidities, utilization ofhealth services, lipid level,efficacy of the initial statintherapy, and chronicconditions (psychoticdisease, chronic lungdisease, morbid obesity,Alzheimer disease, asthma,diabetes, cancer)

Adjusted HR associated witheach 10% increase in PDCwas .94 (95% CI, .93-.95)and .93 (.93-.94) in theprimary and secondaryprevention cohorts,respectively. Compared withPDC of �10%, PDC of 90%or higher associated with anadjusted HR of .42 (95% CI,.37-.47) and .39 (95% CI,.36-.42) among the primaryprevention and secondaryprevention cohorts.

8/9

Spertus(2006),USA27

Retrospectively collectedregistry data study from19-center study ofmyocardial infarctionpatients. 2003 and 2004,2498 patients prospectivelyscreened and enrolled intoProspective RegistryEvaluating MyocardialInfarction: Events andRecovery (PREMIER) Study,of which 500 patients werestudied in this cohort study

Thienopyridines Patients contacted bytelephone 1 month afterinitial procedure todetermine currentmedical regimen. At 30-day interview, patientswere surveyed aboutadherence tothienopyridines as ayes/no question.

Authors constructedmultivariable logisticregression models to identifypredictors of discontinuingthienopyridine therapy.Covariates included age, sex,race, marital status,education level, avoidance ofhealth care because of cost,depression screen any priorcondition (myocardialinfarction, revascularizationprocedure, stroke, ordocumented peripheralarterial disease); anemia,warfarin use, receipt ofinstructions on dischargemedications; and referral tocardiac rehabilitation.

Adjusting for the propensity todiscontinue, patients whostopped thienopyridinetherapy by 30 days weresignificantly more likely todie during the next 11months (7.5% vs .7%, P�.0001; adjusted HR 9.0;95% CI, 1.3-60.6) and to berehospitalized (23% vs 14%,P � .08; adjusted HR 1.5;95% CI, .78 to 3.0).

8/9

357.e20The

American

Journalof

Medicine,

Vol126,

No4,

April2013

Table 2 Continued





Sun (2008),USA28

National retrospectiveadministrative data analysisof patients discharged fromthe hospital with a primarydiagnosis of myocardialinfarction or heart failurebetween 2003- and 2006.Included patients were �18years at the time ofdiagnosis and followed for1 year after index date viamedical and prescriptionrecords.

Angiotensin-convertingenzyme inhibitors,angiotensin receptorblockers

Medication adherencemeasured using the MPRin the 1 year after theirhospitalization.Adherence, partialadherence, andnonadherence defined ashaving a MPR �.8, .5-.79, or �.5,respectively.Nonpersistence wasdefined as any gap of�30 days in therapy.

The authors constructedlogistic regression modelscontrolling for age, sex, andcomorbidities (measuredthrough the Charlsoncomorbidity index)

Among patients in the myocardialinfarction group, adherentpatients compared withnonadherent, patients were noless likely to be rehospitalizedafter discharge (OR .66; 95%CI, .31-1.41; P�NS). Findingsfor the partially adherent groupalso were not statisticallysignificant (OR 1.10; 95% CI,.50-2.43; P�NS). However,patients persistent withangiotensin-converting enzymeinhibitors or angiotensinreceptor blockers weresignificantly less likely to berehospitalized (OR .50; 95% CI,.27-.95; P �.036).

7/9

Wei (2002),Scotland29

Scottish patients whoexperienced their firstmyocardial infarctionbetween 1990 and 1995identified from Taysidehospital discharge data.Maximum follow-up was 6years.

Statin Adherence to statintreatment calculated asPDC. Adherence dividedinto 4 categories: 0%,�39%, 40-79%, and�80%. For eachprescription for a statinauthors had electronicrecord access to tabletstrength, number oftablets dispensed, andinstructions on howthese should be taken.

Multivariate analyses adjustedfor age, sex, socioeconomicdeprivation, statin dailydose, lipid lowering drugtreatment before myocardialinfarction, serum totalcholesterol concentration,diabetes, othercardiovascular drugprescriptions, and otherhospitalizations (excludingsurgical events) during thefollow-up.

In multivariate analyses,compared with those with 0%adherence, those with �39%adherence had an adjusted HRfor myocardial infarctionrecurrence� .59 (95% CI, .22-1.59); those with 40%-79%adherence had an adjustedHR� .51 (95% CI, .19-1.35);and those with 80%-100%adherence had an adjustedHR� .19 (95% CI, .08-.47).Compared with 0% adherence,80%-100% adherence adjustedHR� .47 for all-cause mortality(95% CI, .22-.99)

9/9

CI � confidence interval; HR � hazard ratio; NS � nonsignificant; MPR � medication possession ratio; BP � blood pressure; OR � odds ratio; PDC � proportion of days covered; CV � cardiovascular;RR � relative risk.

357.e21Bitton

etal

CoronaryArtery

DiseaseOutcom

esand

PatientAdherence

ad.p

scdetccvtspslb

mnfahaw

loai


Secondary Prevention: AntihypertensiveMedicationsTwo studies evaluated the impact of adherence to beta-blockers alone. Gallagher et al (1993)16 performed a posthoc analysis of a randomized controlled trial of propranololuse after myocardial infarction. In a randomized controlledtrial subset of 505 women studied for adherence, thosewhose direct pill counts of unused medications from thetrial showed adherence of �75% had a relative risk formortality of 2.4 (95% CI, 1.1-5.6) compared with those withhigher adherence. Kleiner et al22 retrospectively studieddherence to beta-blockers in a health insurance claimsatabase using PDC, finding an adjusted HR for survival of37 (95% CI, .25-.54) for most- versus least-adherentatients.

Secondary Prevention: Multiple MedicationRegimens and Other MedicationsA number of studies looked at the effect of adherence tovarious combinations of antihypertensive medications andstatins on outcomes. While most evaluated specific combi-nations of medications examined, a number of studies didnot specify regimens clearly. The Canadian study by Ras-mussen et al25 and the randomized controlled trial byChoudhry et al14 are the most sound methodologically.Rasmussen et al25 evaluated the relationship between mor-tality and adherence to essential cardiac medications (statinand beta-blocker) after a myocardial infarction. Notably,they controlled for adherence to another medication, a cal-cium channel blocker, which has not been shown to havemeaningful effects on mortality in postmyocardial infarc-tion patients. They found that patients who were least ad-herent to statins were 25% more likely to die than patientswho were most adherent. Intermediate adherers were 12%more likely to die than the best adherers. There was nosignificant relationship between adherence to calcium chan-nel blockers and mortality. Interestingly, the magnitude ofthe relationship between adherence and health outcomes inthe Rasmussen et al article25 was similar in magnitude to anearlier meta-analysis of medication adherence in coronaryartery disease results from DiMatteo et al.2

Choudhry et al14 randomly assigned 5855 patients whouffered a myocardial infarction to full prescription drugoverage post myocardial infarction for all coronary arteryisease-related medications or usual coverage, and analyzedffects on medication adherence, cardiovascular events, andotal, cardiovascular, pharmaceutical, and nonpharmaceuti-al spending. Adherence in the drug coverage group in-reased between 4% and 6% in each drug class. First majorascular event or revascularization rates did not differ, butotal major vascular events or revascularization rates wereignificantly decreased in the drug coverage group com-ared with the usual coverage group. Pharmaceutical drugpending increased, but nondrug spending was reduced,eading to a nonsignificant overall decrease in total spending
etween the 2 groups.
Pittman et al24 looked at a retrospective cohort of phar-acy beneficiaries in the US with multiple claims for a

umber of different antihypertensive medications. Theyound that those with MPR �60% had 30%-40% higherdjusted odds ratio for emergency department visit andospitalization for a cardiovascular event, resulting in annnual cost difference of $813 ($7995 vs $7182), comparedith those patients with MPR �80%.Two studies examined adherence to medications not for

ipid or blood pressure control, one of which was for thien-pyridines (antiplatelet drugs) and one which examinedntiarrhythmic effects (amiodarone). At 30 days post stent-ng for myocardial infarction, Spertus et al27 found that

those patients who stopped thienopyridines experienced astatistically significantly absolute increase in risk of death of6.8% over the succeeding 11 months, even after adjustingfor a patient’s propensity to discontinue medications. Irvineet al21 examined a post hoc analysis of pill counts for anamiodarone randomized controlled trial, reporting that car-diovascular and total cause mortality were significantly el-evated in the low-adherence amiodarone treatment group(relative risk 2.34-2.49).

DISCUSSIONThis systematic review of studies that link medication ad-herence to coronary artery disease cost and mortality out-comes yielded a mix of observational studies with a fewpost hoc analyses of randomized controlled studies and oneprespecified randomized controlled trial. Most studies wereretrospective in nature and analyzed claims databases fromlarge insurers or pharmacy benefit managers from NorthAmerica, Europe, and Israel. Most studies examined adher-ence to statins or antihypertensive medications, either aloneor in combination. These studies used a variety of adherencemeasures, from medication possession ratio, proportion ofdays covered, direct pill counts, and patient survey ques-tions. They also used a wide mix of techniques to attempt toadjust for potential confounders, primarily focused on rel-evant sociodemographic, insurance, and comorbidity char-acteristics. The quality of studies was generally good per theNewcastle Ottawa Quality Scale.

Overall, our review of the published literature showed aconsistent trend toward benefits in reduced coronary arterydisease-related events, mortality, readmissions, and costsacross a variety of adherence measures and medicationclasses. In general, despite specific method heterogeneity, asimilar overall analytic approach was used: the authorsidentified patients as adherent or nonadherent based on refillclaims data and evaluated the relationship between adher-ence category and health outcomes (hospitalizations, costs,or death). All of the studies reviewed found that adherencesignificantly improves health outcomes, and those that an-alyzed costs found reduced total annual coronary arterydisease costs (consistently between $294 and $868 per pa-tient, equating to 10.1%-17.8% cost reductions between
high- and low-adherence groups).

ssaptht

os

wntamr

iaoieaspamoe

trmd

cgisccs(ttcprpwabr

otoaotebh


Results from one previous systematic review and meta-analysis of studies2 reported findings that were generallyimilar to the newer articles listed here. However, thistudy2 was a systematic review of all types of behaviors forhost of chronic diseases and was conducted before the

ublication of many of the papers we analyzed; includinghe recent papers that directly linked cost analyses to ad-erence, as well as the prespecified randomized controlledrial.

One key problem with this literature base is that, in theverwhelming majority of these studies (except for Rasmus-en et al,25 Ho et al,19 and Choudhry et al14), the authors did

not consider the healthy adherer effect. The magnitude ofthe mortality and hospitalization benefits for better adher-ence is a function of whether the study controlled well forthis potentially confounding effect. Patients who adhere totheir medications also may be more likely to adhere to ahealthier diet, exercise more frequently, and pursue ahealthier lifestyle. Dormuth et al31 found a strong relation-ship between statin adherence and motor vehicle accidents,for which there is little biologic plausibility. This researchraises questions about the validity of the magnitude (but lesslikely the direction) of relationships between adherence andoutcomes in observational studies that do not control forother healthy behaviors. Interestingly, the study on amioda-rone use found large mortality reduction benefits (adjustedodds ratio �2.4) for highly adhering patients, which isnotable considering that a 2009 meta-analysis of amioda-rone trials concluded that amiodarone use after myocardialinfarction does not reduce all-cause deaths.32 Nonetheless,

hile the literature is not ideal, and a definitive study hasot been performed, there is a preponderance of evidencehat suggests the presence of a strong relationship betweendherence to essential coronary artery disease preventionedications and both improved health outcomes and cost

eductions.These results have a number of policy and research

mplications. First, although the direction of the associationsll point to significant coronary artery disease cost andutcome improvement with better adherence, only two stud-es we found directly controlled for the healthy adhererffect (and one indirectly controlled for it by virtue of beingrandomized trial). Thus, there is a critical need for better

tudies with healthy user controls, especially as payers andurchasers are starting to apply large resources towardsdherence improvement and thus need more precise esti-ates for return on investment analyses. One good example

f a recent study that controlled for this effect was Roebuckt al,4 which used advanced econometric techniques to ad-

dress this effect in claims data through elimination of un-measured confounders if they did not change over time.Findings from Roebuck et al4 also were consistent in direc-ion and magnitude with the studies we identified in thiseview, although it focused on cardiovascular and otheretabolic diseases in general and not on coronary artery

isease specifically.

Second, patient-centered medical home and accountableare organization delivery pilots need to have more fine-rained understanding and evaluation of how they mightmprove adherence before they begin targeting it. Studiesuch as those reviewed here can help evaluators and poli-ymakers make better estimates of expected outcome orost improvements that may be achieved in federal sharedavings pilots or private-payer care-delivery demonstrationswe found cost reductions in the range of 10%-18%). Fur-hermore, efforts to explicitly improve adherence will needo determine which domains of adherence to target (eg,ognitive, cost), and how to best measure the putative endoints of these experiments. As Choudhry et al14 showed,educing cost barriers can improve adherence to secondaryrevention medications and potentially reduce some un-anted outcomes. Whether interventions that address other

dherence barriers (such as smart pillboxes for cognitivearriers) can achieve similar improvements remains to beigorously evaluated on a large scale.

Many key delivery system reforms will increasingly relyn projected medication adherence improvements througheam-based provider efforts, financial incentive alignment,r better technical support of patients at home in order tochieve important cost and outcome targets. Given the pacef delivery innovation throughout the US, and the magni-ude of possible benefits from improved medication adher-nce, renewed efforts at the private and national levels toetter understand and support increases in medication ad-erence are urgently needed.33 While the promise of im-

proved adherence around chronic disease such as coronaryartery disease is large, much methodological and implemen-tation work remains before we can conclusively quantifythese benefits.

References1. Heidenreich PA, Trogdon JG, Khavjou OA, et al. Forecasting the

future of cardiovascular disease in the United States: a policy state-ment from the American Heart Association. Circulation. 2011;123(8):933-944.

2. DiMatteo MR, Giordani PJ, Lepper HS, Croghan TW. Patient adher-ence and medical treatment outcomes: a meta-analysis. Med Care.2002;40(9):794-811.

3. Salas M, Hughes D, Zuluaga A, et al. Costs of medication nonadher-ence in patients with diabetes mellitus: a systematic review and criticalanalysis of the literature. Value Health. 2009;12(6):915-922.

4. Roebuck MC, Liberman JN, Gemmill-Toyama M, Brennan TA. Med-ication adherence leads to lower health care use and costs despiteincreased drug spending. Health Aff (Millwood). 2011;30(1):91-99.

5. Bechtel C, Ness DL. If you build it, will they come? Designing trulypatient-centered health care. Health Aff (Millwood). 2010;29(5):914-920.

6. Bouchard MH, Dragomir A, Blais L, et al. Impact of adherence tostatins on coronary artery disease in primary prevention. Br J ClinPharmacol. 2007;63(6):698-708.

7. Breekveldt-Postma NS, Penning-van Beest FJ, Siiskonen SJ, et al. Theeffect of discontinuation of antihypertensives on the risk of acutemyocardial infarction and stroke. Curr Med Res Opin. 2008;24(1):121-127.

8. Corrao G, Conti V, Merlino L, et al. Results of a retrospective database
analysis of adherence to statin therapy and risk of nonfatal ischemic

1

1

1

1

1

1

1

1

1

2


heart disease in daily clinical practice in Italy. Clin Ther.2010;32(2):300-310.

9. Pittman DG, Chen W, Bowlin SJ, Foody JM. Adherence to statins,subsequent healthcare costs, and cardiovascular hospitalizations. Am JCardiol. 2011;107(11):1662-1666.

10. Perreault S, Dragomir A, Blais L, et al. Impact of better adherence tostatin agents in the primary prevention of coronary artery disease. EurJ Clin Pharmacol. 2009;65(10):1013-1024.

1. Amin AP, Mukhopadhyay E, Nathan S, et al. Association of medicalnoncompliance and long-term adverse outcomes, after myocardialinfarction in a minority and uninsured population. Transl Res. 2009;154(2):78-89.

2. Aubert RE, Yao J, Xia F, Garavaglia SB. Is there a relationshipbetween early statin compliance and a reduction in healthcare utiliza-tion? Am J Manag Care. 2010;16(6):459-466.

3. Chapman RH, Yeaw J, Roberts CS. Association between adherence tocalcium-channel blocker and statin medications and likelihood of car-diovascular events among US managed care enrollees. BMC Cardio-vasc Disord. 2010;10:29.

4. Choudhry NK, Avorn J, Glynn RJ, et al. Full coverage for preventivemedications after myocardial infarction. N Engl J Med. 2011;365(22):2088-2097.

5. Corrao G, Parodi A, Nicotra F, et al. Better compliance to antihyper-tensive medications reduces cardiovascular risk. J Hypertens. 2011;29(3):610-618.

6. Gallagher EJ, Viscoli CM, Horwitz RI. The relationship of treatmentadherence to the risk of death after myocardial infarction in women.JAMA. 1993;270(6):742-744.

7. Gehi AK, Ali S, Na B, Whooley MA. Self-reported medication ad-herence and cardiovascular events in patients with stable coronaryheart disease: the heart and soul study. Arch Intern Med. 2007;167(16):1798-1803.

8. Ho PM, Spertus JA, Masoudi FA, et al. Impact of medication therapydiscontinuation on mortality after myocardial infarction. Arch InternMed. 2006;166(17):1842-1847.

9. Ho PM, Magid DJ, Shetterly SM, et al. Importance of therapy inten-sification and medication nonadherence for blood pressure control inpatients with coronary disease. Arch Intern Med. 2008;168(3):271-276.

0. Ho PM, Magid DJ, Shetterly SM, et al. Medication nonadherence isassociated with a broad range of adverse outcomes in patients with
coronary artery disease. Am Heart J. 2008;155(4):772-779.
21. Irvine J, Baker B, Smith J, et al. Poor adherence to placebo oramiodarone therapy predicts mortality: results from the CAMIATstudy. Canadian Amiodarone Myocardial Infarction Arrhythmia Trial.Psychosom Med. 1999;61(4):566-575.

22. Kleiner SA, Vogt WB, Gladowski P, et al. Beta-blocker compliance,mortality, and reinfarction: validation of clinical trial association usinginsurer claims data. Am J Med Qual. 2009;24(6):512-519.

23. Newby LK, LaPointe NM, Chen AY, et al. Long-term adherence toevidence-based secondary prevention therapies in coronary artery dis-ease. Circulation. 2006;113(2):203-212.

24. Pittman DG, Tao Z, Chen W, Stettin GD. Antihypertensive medicationadherence and subsequent healthcare utilization and costs. Am JManag Care. 2010;16(8):568-576.

25. Rasmussen JN, Chong A, Alter DA. Relationship between adherenceto evidence-based pharmacotherapy and long-term mortality afteracute myocardial infarction. JAMA. 2007;297(2):177-186.

26. Shalev V, Chodick G, Silber H, et al. Continuation of statin treatmentand all-cause mortality: a population-based cohort study. Arch InternMed. 2009;169(3):260-268.

27. Spertus JA, Kettelkamp R, Vance C, et al. Prevalence, predictors, andoutcomes of premature discontinuation of thienopyridine therapy afterdrug-eluting stent placement: results from the PREMIER registry.Circulation. 2006;113(24):2803-2809.

28. Sun SX, Ye X, Lee KY, et al. Retrospective claims database analysisto determine relationship between renin-angiotensin system agents,rehospitalization, and health care costs in patients with heart failure ormyocardial infarction. Clin Ther. 2008;30(Pt 2):2217-2227.

29. Wei L, Wang J, Thompson P, et al. Adherence to statin treatment andreadmission of patients after myocardial infarction: a six year followup study. Heart. 2002;88(3):229-233.

30. Dragomir A, Cote R, White M, et al. Relationship between adherencelevel to statins, clinical issues and health-care costs in real-life clinicalsetting. Value Health. 2010;13(1):87-94.

31. Dormuth CR, Patrick AR, Shrank WH, et al. Statin adherence and riskof accidents: a cautionary tale. Circulation. 2009;119(15):2051-2057.

32. Piccini JP, Berger JS, O’Connor CM. Amiodarone for the preventionof sudden cardiac death: a meta-analysis of randomized controlledtrials. Eur Heart J. 2009;30(10):1245-1253.

33. Goldman L, Epstein AM. Improving adherence–money isn’t the only
thing. N Engl J Med. 2011;365(22):2131-2133.


Appendix: Search Terms Used“Epidemiologic Studies” [Mesh] OR “EpidemiologicalStudies” [tw] OR “Epidemiological Study” [tw] OR “Case-Control” [tw] OR “Case-Compeer” [tw] OR “Case-Com-parison” [tw] OR “Case-Referent” [tw] OR “Case Base”[tw] OR “retrospective study” [tw] OR “retrospective stud-ies” [tw] OR “Cohort Study” [tw] OR “Concurrent Studies”[tw] OR “Concurrent Study” [tw] OR “Cohort Analyses” [tw]OR “Cohort Analysis” [tw] OR “Incidence Studies” [tw] OR“Incidence Study” [tw] OR “Longitudinal Study” [tw] OR“Longitudinal Studies” [tw] OR “Longitudinal Survey” [tw]OR “Longitudinal Surveys” [tw] OR “Follow-Up Study” [tw]OR “Follow Up Studies” [tw] OR “Followup Study” [tw] OR“Followup Studies” [tw] OR “Prospective Study” [tw] OR“Prospective Studies” [tw] OR “Cross Sectional Studies” [tw]OR “Cross Sectional Study” [tw] OR “Cross-Sectional Study”[tw] OR “Cross-Sectional Studies” [tw] OR “Disease Fre-quency Surveys” [tw] OR “Disease Frequency Survey” [tw]OR “Cross Sectional Survey” [tw] OR “Cross Sectional Sur-veys” [tw] OR “Cross-Sectional Survey” [tw] OR “Cross-Sectional Survey” [tw] OR “Cross Sectional Analysis” [tw]OR “Cross Sectional Analyses” [tw] OR “Cross-SectionalAnalysis” [tw] OR “Cross-Sectional Analyses” [tw] OR “Prev-alence Study” [tw] OR “Prevalence Studies”

AND (“compliance” [tiab] OR “patient compliance”[mesh] OR “patient adherence” [tw] OR “patients adherence”[tw] OR “patient compliance” [tw] OR “medication adher-ence” [tw] OR “medication compliance” [tw] OR “treatmentcompliance” [tw] OR “treatment adherence” [tw] OR “treat-ment refusal” [mesh] OR “patient dropouts” [mesh] OR “treat-ment refusal” [tw] OR “patient dropout” [tw] OR “patientdropouts” [tw] OR “patient dropped” [tw] OR “drug adher-ence” [tw] OR “drug compliance” [tw] OR “persistence” [tiab]

AND (“drug therapy” [mesh] OR “drug therapy” [sh]OR “Pharmaceutical Services” [mesh] OR “Medication sys-tems” [Mesh] OR “Drug Utilization Review” [mesh] OR“pharmacies” [mesh] OR “Pharmaceutical Preparations”[mesh] OR “prescription drugs” [tw] OR “prescription drug”[tw] OR “drug therapy” [tw] OR “drug treatment” [tw] OR((”drug” [tw] OR “drugs” [tw] OR “medicine” [tw] OR “med-icines” [tw] OR “medication” [tw] OR “medications” [tw])NOT medline[sb]) OR “Cardiovascular Agents/therapeuticuse” [Mesh] OR “Hypoglycemic Agents/therapeutic use”[Mesh] OR “Antihypertensive agents/therapeutic use” [Mesh]OR “Antilipemic Agents/therapeutic use” [Mesh]) OR(”Cardiovascular Agents” [Mesh] OR “CardiovascularAgents “[Pharmacological Action]) OR “Insulin” [Mesh]OR “insulin” [tw] OR “hypoglycemic agents” [Mesh] OR“hypoglycemic agents” [pa] OR “hypoglycemic agent”[tw] OR “antidiabetic” [tw] OR “hypoglycemic drug”[tw] OR “hypoglycemic agents” [tw] OR “antidiabetics”[tw] OR “hypoglycemic drugs“[tw]) OR (”Antihyperten-sive agents” [Mesh] OR “Antihypertensive Agents” [pa])

AND (“Diabetes Mellitus” [Mesh] OR “diabetes” [tw]OR “Blood Pressure” [Mesh]) OR “Cardiovascular Dis-
eases” [Mesh] OR “Heart arrest” [mesh] OR “Heart arrest”
[tw] OR “Heart arrests” [tw] OR “Cardiac Arrest” [tw] OR“Cardiac Arrests” [tw] OR “Asystole” [tw] OR “Asystoles”[tw] OR “Cardiopulmonary Arrest” [tw] OR “Cardiac Sud-den Death” [tw] OR “Sudden Cardiac Death” [tw] OR“circulation arrest” [tw] OR “circulatory arrest” [tw] OR“heart standstill” [tw] OR “Ventricular Fibrillation” [mesh]OR “Ventricular Fibrillation” [tw] OR “Ventricular Fibril-lations” [tw] OR “Myocardial Ischemia” [mesh] OR “Myo-cardial Ischemias OR “Myocardial Ischemia” [tw] OR“Ischemic Heart Disease” [tw] OR “Ischemic Heart Dis-eases” [tw] OR “Acute Coronary Syndromes” [tw] OR“Acute Coronary Syndrome” [tw] OR “Stenocardia” [tw]OR “Stenocardias” [tw] OR “Angor Pectoris” [tw] OR“Angina” [tw] OR “Anginas” [tw] OR “Myocardial Prein-farction” [tw] OR “Cardiac Syndrome X” [tw] OR “Coro-nary Diseases” [tw] OR “Coronary Disease” [tw] OR “Cor-onary Heart Disease” [tw] OR “Coronary Heart Diseases”[tw] OR “Coronary Aneurysms” [tw] OR “Coronary Aneu-rysm” [tw] OR “Coronary Artery Diseases” [tw] OR “Cor-onary Artery Disease” [tw] OR “Arteriosclerosis” [tw] OR“Arterioscleroses” [tw] OR “Coronary Occlusions” [tw] OR“Coronary Occlusion” [tw] OR “Coronary Artery Stenosis”[tw] OR “Coronary Stenosis” [tw] OR “Coronary Stenoses”[tw] OR “Coronary Artery Stenoses” [tw] OR “CoronaryRestenoses” [tw] OR “Coronary Restenosis” [tw] OR “Cor-onary Thromboses” [tw] OR “Coronary Thrombosis” [tw]OR “Coronary Vasospasms” [tw] OR “Coronary Vaso-spasm” [tw] OR “Coronary Artery Vasospasm” [tw] OR“Coronary Artery Vasospasms” [tw] OR “Myocardial In-farctions” [tw] OR “Myocardial Infarction” [tw] OR “Myo-cardial Infarct” [tw] OR “Myocardial Infarcts” [tw] OR“Heart Failure” [mesh] OR “Heart Failure” [tw] OR “Car-diac Failure” [tw] OR “Myocardial Failure” [tw] OR “HeartDecompensation” [tw] OR “Paroxysmal Dyspnea” [tw] OR“Paroxysmal Dyspneas” [tw] OR “Paroxysmal NocturnalDyspnea” [tw] OR “Cardiac Asthma” [tw] OR “CardiacEdemas” [tw] OR “Cardiac Edema” [tw] OR “Heart Fail-ures” [tw] OR “backward failure” [tw] OR “cardiac back-ward failure” [tw] OR “cardiac decompensation” [tw] OR“cardiac incompetence” [tw] OR “cardiac insufficiency”[tw] OR “cardial decompensation” [tw] OR “cardial insuf-ficiency OR “heart insufficiency” [tw] OR “decompensatiocordis” [tw] OR “heart incompetence” [tw] OR “heart in-sufficiency” [tw] OR “insufficientia cardis” [tw] OR “myo-cardial insufficiency” [tw] OR “Ventricular Dysfunction”[mesh] OR “Ventricular Dysfunctions” [tw] OR “Ventric-ular Dysfunction” [tw] OR “ventricle insufficiency” [tw]OR “ventrical failure” [tw] OR “ventricular insufficiency”[tw] OR “ventricular failure” OR “Cerebrovascular Disor-ders” [mesh] OR “Cerebrovascular Disorder” [tw] OR“Cerebrovascular Insufficiency” [tw] OR “CerebrovascularInsufficiencies” [tw] OR “Cerebrovascular Occlusion” [tw]OR “Cerebrovascular Occlusions” [tw] OR “Lenticulostri-ate Vasculopathy” [tw] OR “Lenticulostriate Vasculopa-thies” [tw] OR “Basal Ganglionic Hemorrhage” [tw] OR
“Basal Ganglia Hematoma” [tw] OR “Putamen Hemor-


rhage” [tw] OR “Putaminal Hematoma” [tw] OR “Putami-nal Brain Hemorrhage” [tw] OR “Putaminal Brain Hemor-rhages” [tw] OR “Brain Ischemias” [tw] OR “BrainIschemia” [tw] OR “Ischemic Encephalopathy” [tw] OR“Ischemic Encephalopathies” [tw] OR “Cerebral Ischemia”[tw] OR “Cerebral Ischemias” [tw] OR “Brain Infarctions”[tw] OR “Brain Infarction” [tw] OR “Brain Venous Infarc-tion” [tw] OR “Brain Venous Infarctions” [tw] OR “Lacu-nar Infarction” [tw] OR “Lacunar Infarctions” [tw] OR“Cerebral Circulation Infarction” [tw] OR “Brainstem In-farction” [tw] OR “Brainstem Infarctions” [tw] OR “ClaudeSyndrome” [tw] OR “Weber Syndrome” [tw] OR “MillardGublar” [tw] OR “Top of the Basilar Syndrome” [tw] OR“Benedict Syndrome” [tw] OR “Foville Syndrome” [tw]OR “Lateral Medullary Syndromes” [tw] OR “PosteriorInferior Cerebellar Artery Syndrome” [tw] OR “Wallen-berg’s Syndrome” [tw] OR “Wallenbergs Syndrome” [tw]OR “Dorsolateral Medullary Syndrome” [tw] OR “LateralBulbar Syndrome” [tw] OR “Cerebral Infarctions” [tw] OR“Cerebral Infarction” [tw] OR “Subcortical Infarction” [tw]OR “Subcortical Infarctions” [tw] OR “Cerebral ArteryInfarction” [tw] OR “Multi Infarct Dementias” [tw] OR“Multi Infarct Dementia” [tw] OR “Lacunar Dementia” [tw]OR “Lacunar Dementias” [tw] OR “ACA Infarction” [tw]OR “ACA Infarctions” [tw] OR “Cerebral Artery Circula-tion Infarction” [tw] OR “Cerebral Artery Distribution In-farction” [tw] OR “Heubner Artery Infarction” [tw] OR“Heubner’s Artery Infarction” [tw] OR “Heubners ArteryInfarction” [tw] OR “Cerebral Artery Syndrome” [tw] OR“MCA Infarction” [tw] OR “Cerebral Artery Embolus” [tw]OR “Cerebral Artery Occlusion” [tw] OR “Cerebral ArteryThrombotic Infarction” [tw] OR “Cerebral Artery Throm-bosis” [tw] OR “Cerebral Artery Embolic Infarction” [tw]OR “PCA Infarction” [tw] OR “Brain Hypoxia Ischemias”[tw] OR “Brain Hypoxia Ischemia” [tw] OR “HypoxicIschemic Encephalopathy” [tw] OR “Hypoxic Ischemic En-cephalopathy” [tw] OR “Hypoxic Ischemic Encephalopa-thies” [tw] OR “Ischemic Hypoxic Encephalopathy” [tw]OR “Ischemic Hypoxic Encephalopathies” [tw] OR “An-oxic Ischemic Encephalopathy” [tw] OR “Anoxic IschemicEncephalopathies” [tw] OR “Cerebral Hypoxia-Ischemia”[tw] OR “Cerebral Hypoxia Ischemia” [tw] OR “CerebralIschemia Hypoxia” [tw] OR “Brain Ischemia Anoxia” [tw]OR “Brain Anoxia Ischemia” [tw] OR “Cerebral IschemiaAnoxia” [tw] OR “Cerebral Anoxia Ischemia” [tw] OR“TIA” [tw] OR “TIAs” [tw] OR “Transient Ischemic At-tack” [tw] OR “Transient Ischemic Attacks” [tw] OR “BrainTIA” [tw] OR “Transient Brainstem Ischemia” [tw] OR“Transient Brainstem Ischemias” [tw] OR “Transient Cere-bral Ischemia” [tw] OR “Transient Cerebral Ischemias” [tw]OR “Vertebrobasilar Insufficiencies” [tw] OR “Vertebro-basilar Insufficiency” [tw] OR “Vertebrobasilar Ischemia”[tw] OR “Vertebrobasilar Ischemias” [tw] OR “Vertebro-basilar Dolichoectasia” [tw] OR “Vertebrobasilar Dolicho-ectasias” [tw] OR “Vertebral Artery” [tw] OR “Basilar
Artery” [tw] OR “Subclavian Artery Stenosis” [tw] OR
“Subclavian Artery Stenoses” [tw] OR “Basilar Steal Syn-drome” [tw] OR “Basilar Steal Syndromes” [tw] OR “Bra-chial Basilar Insufficiency Syndrome” [tw] OR “SubclavianSteal” [tw] OR “Brain Vascular Trauma” [tw] OR “BrainVascular Injury” [tw] OR “Vascular Brain Injuries” [tw]OR “Vascular Brain Injury” [tw] OR “Subarachnoid Hem-orrhages” [tw] OR “Subarachnoid Hemorrhage” [tw] OR“Intracranial Arterial Disease” [tw] OR “Intracranial Arte-rial Disorder” [tw] OR “Intracranial Arterial Disorders”[tw] OR “Arterial Brain Disease” [tw] OR “Arterial BrainDiseases” [tw] OR “Arterial Brain Disorder” [tw] OR “Ar-terial Brain Disorders” [tw] OR “Cerebral Arterial Disease”[tw] OR “Cerebral Arterial Diseases” [tw] OR “CerebralArtery Diseases” [tw] OR “Cerebral Artery Disease OR“CADASILM” [tw] OR “Cerebral Amyloid Angiopathies”[tw] OR “Congophilic Angiopathy” [tw] OR “CongophilicAngiopathies” [tw] OR “Cerebral Amyloid Angiopathy”[tw] OR “Icelandic Type Amyloidosis” [tw] OR “ACAInfarction” [tw] OR “ACA Infarctions” [tw] OR “MCAInfarction” [tw] OR “PCA Infarction” [tw] OR “ProgressiveIntracranial Occlusive Arteropathy (Moyamoya)” [tw] OR“Moyamoya Syndrome” [tw] OR “Moya Moya Disease”[tw] OR “Intracranial Aneurysms” [tw] OR “Artery Aneu-rysm” [tw] OR “Berry Aneurysm” [tw] OR “Berry Aneu-rysms” [tw] OR “Brain Aneurysm” [tw] OR “Brain Aneu-rysms” [tw] OR “Cerebral Aneurysm” [tw] OR “CerebralAneurysms” [tw] OR “Intracranial Mycotic Aneurysm”[tw] OR “Intracranial Mycotic Aneurysms” [tw] OR “Intra-cranial Arterioscleroses” [tw] OR “Intracranial Atherosclero-sis” [tw] OR “Intracranial Atheroscleroses” [tw] OR “CerebralArteriosclerosis” [tw] OR “Cerebral Arterioscleroses” [tw] OR“Cerebral Atherosclerosis” [tw] OR “Cerebral Atherosclero-ses” [tw] OR “Vascular Dementias” [tw] OR “Vascular De-mentia” [tw] OR “Arteriosclerotic Dementia” [tw] OR “Arte-riosclerotic Dementias” [tw] OR “Binswanger Disease” [tw]OR “Binswanger Encephalopathy” [tw] OR “Chronic Pro-gressive Subcortical Encephalopathy” [tw] OR “Bins-wanger’s Encephalopathy” [tw] OR “Binswanger’s Dis-ease” [tw] OR “Binswangers Disease” [tw] OR “SubcorticalLeukoencephalopathies” [tw] OR “Subcortical Arterioscle-rotic Encephalopathy” [tw] OR “Subcortical Leukoenceph-alopathy” [tw] OR “Subcortical Arteriosclerotic Ence-phalopathies” [tw] OR “Intracranial Arteriovenous Malfor-mation” [tw] OR “Intracranial Arteriovenous Malforma-tions” [tw] OR “AVM Intracranial” [tw] OR “CerebralArteriovenous Malformation” [tw] OR “Cerebral Arterio-venous Malformations” [tw] OR “Galen MalformationsVeins” [tw] OR “Carotid Artery Thromboses” [tw] OR“Carotid Artery Thromboses” [tw] OR “Carotid Thrombo-sis” [tw] OR “Brain Embolism” [tw] OR “Brain Emboli”[tw] OR “Brain Embolus” [tw] OR “Cerebral Embol-ism” [tw] OR “Cerebral Emboli” [tw] OR “Cerebral Em-bolus” [tw] OR “Intracranial Thromboses” [tw] OR “Intra-cranial Thrombus” [tw] OR “Cerebral Thrombosis” [tw]
OR “Cerebral Thromboses” [tw] OR “Brain Thrombosis”


[tw] OR “Brain Thromboses” [tw] OR “Brain Thrombus”[tw] OR “Intracranial Hemorrhage” [tw] OR “PosteriorFossa Hemorrhage” [tw] OR “Posterior Fossa Hemor-rhages” [tw] OR “Brain Hemorrhage” [tw] OR “Brain Hem-orrhages” [tw] OR “Cerebrum Hemorrhage” [tw] OR “Ce-rebrum Hemorrhages” [tw] OR “Intracerebral Hemorrhage”[tw] OR “Intracerebral Hemorrhages” [tw] OR “CerebralHemorrhages” [tw] OR “Cerebral Hypertensive Hemor-rhage” [tw] OR “Cerebral Hypertensive Hemorrhages” [tw]OR “Traumatic Intracranial Hematoma” [tw] OR “Trau-matic Intracranial Hematomas” [tw] OR “SAH (Subarach-noid Hemorrhage)” [tw] OR “SAHs (Subarachnoid Hemor-rhage)” [tw] OR “Subarachnoid Hemorrhages” [tw] OR“Subarachnoid Hemorrhage” [tw] OR “Strokes” [tw] OR“Stroke” [tw] OR “Brain Vascular Accident” [tw] OR“Brain Vascular Accidents” [tw] OR “CVA (Cerebrovascu-
lar Accident)” [tw] OR “CVAs (Cerebrovascular Acci-
dent)” [tw] OR “Apoplexy” [tw] OR “Cerebrovascular Ac-cident” [tw] OR “Cerebrovascular Accidents” [tw] OR“Susac’s Syndrome” [tw] OR “Susacs Syndrome” [tw] OR“Retinocochleocerebral Vasculopathy” [tw] OR “Retinoco-chleocerebral Vasculopathies” [tw] OR “Peripheral Vascu-lar Disease” [mesh] OR “PAD” [tw] OR “Peripheral Vas-cular Disease” [tw] OR “Peripheral Vascular Disease” [tw]OR “Peripheral Angiopathies” [tw] OR “Peripheral Angi-opathy” [tw] OR “Peripheral Arterial Disease” [tw] OR“Peripheral Arterial Disease” [tw] OR “peripheral arteriopa-thy” [tw] OR “peripheral blood vessel disease” [tw] OR“peripheral vascular disorder” [tw] OR “peripheral vascu-lopathy” [tw] OR “peripheral vessel disease” [tw] OR “Hy-pertension” [mesh] OR “hypertensive” [tw] OR “HighBlood Pressure” [tw] OR “High Blood Pressures” [tw] OR“Hypertension” [tw] OR “Hypertensions” [tw] OR “Gold-
blatt Syndrome” [tw])

Date post:	30-Jun-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

The Impact of Medication Adherence on Coronary Artery Disease … · The Impact of Medication...

Documents