medication noncompliance and patient satisfaction following percutaneous coronary intervention

C©2012, Wiley Periodicals, Inc.DOI: 10.1111/j.1540-8183.2012.00743.x

ORIGINAL INVESTIGATION

Medication Noncompliance and Patient Satisfaction Following PercutaneousCoronary Intervention: Armenia

SEDA AGHABEKYAN, M.D., M.P.H.,1 MICHAEL E. THOMPSON, M.S., DR.P.H.,2

and LUSINE ABRAHAMYAN, M.D., PH.D., M.P.H.3

From the 1Center for Health Services Research and Development, American University of Armenia, Yerevan, Armenia;2Department of PublicHealth Sciences, The University of North Carolina at Charlotte, Charlotte, North Carolina; and 3Institute for Clinical Evaluative Sciences,Toronto, Ontario, Canada

Objectives: This study evaluated the relationship between medication noncompliance and patient satisfaction inpatients with drug-eluting stent revascularization in Armenia.Background: Reasons for medication noncompliance are multifactorial—often related to health system, patient,condition, therapy, and socioeconomic factors.Methods: The analytical cross-sectional survey used a simple random sample of patients aged 18 and over whohad undergone percutaneous coronary artery intervention with drug-eluting stent from 2006 to 2008 (n = 271)at Nork Marash Medical Center, Yerevan, Armenia. Medication noncompliance was assessed using the MoriskyAdherence Scale. Patient satisfaction items were selected from the Patient Satisfaction Questionnaire (PSQ-18).Results: Respondents’ mean age was 57.5 ± 9.8 years. Most (87.8%) were male. Nearly one-third of patients(31.0%) were noncompliant. Most reported good health (91.9%). Respondents most often cited out-of-pocket costsas a reason for noncompliance (19.2%). Age, gender, health status, smoking status, and cost were associated withmedication noncompliance (P < 0.05). Noncompliance was not associated with elapsed time after the interventionor satisfaction (P > 0.05). Medication noncompliance was positively related to cost (odds ratio [OR] = 2.57, 95%CI = 1.33–4.97) and inversely related to health status (OR = 0.46, 95% CI = 0.25–0.85) and age (OR = 0.94,95% CI = 0.91–0.97).Conclusion: Medication noncompliance is a multifactorial problem. Strategies reducing the economic burden onpatients should improve compliance and, thus, treatment outcomes. These findings further efforts to benchmarkperformance in Armenia and other post-Soviet countries against western standards and experiences. (J IntervenCardiol 2012;∗∗:1–7)

Introduction

Coronary heart disease (CHD) was the leading causeof mortality worldwide in 2004 with 7.2 million deaths,accounting for 11.2% of all deaths,1 and continues tobe a major cause of morbidity and mortality through-out the world. Both surgical revascularization (coro-nary artery bypass grafting [CABG]) and percuta-neous coronary intervention (PCI) with stenting are

Address for reprints: Seda Aghabekyan, M.D., M.P.H., Center forHealth Services Research and Development, American Universityof Armenia, 40 Marshal Bagramian, Yerevan 0019, Armenia. Fax:(374 10) 51 25 66; e-mail: [email protected]

accepted treatments for this condition. In Europe, Asia,Canada, and United States, the rates of PCI relativeto CABG are rapidly rising, likely due to introduc-tion of more effective drug-eluting stents (DES) andchanges in patient and provider preferences.2 Stent-ing, however, is associated with a higher risk of com-plications. In one study, for example, stent throm-bosis occurred only in 1.3% of patients at 9-monthfollow-up but accounted for 45% of the postoperativemortality.3

According to American Heart Association (AHA)guidelines, dual antiplatelet therapy consisting of as-pirin and a thienopyridine drug such as clopidogrel

Vol. 00, No. 0, 2012 Journal of Interventional Cardiology 1

AGHABEKYAN, ET AL.

(Plavix) or ticlopidine (Ticlid) should be given for atleast a full year, and perhaps even longer followingcertain types of stent placements for patients with-out major risk for bleeding.4 However, noncompli-ance with such medication regimens can be harmfulto the patient and negatively impact reaching thera-peutic goals and often lead to a deterioration of healthstatus. The premature discontinuation of the dual ther-apy is the single greatest predictor of stent thrombo-sis.4 Dissatisfaction with care predicted discontinua-tion of therapy in patients with kidney disease and withhyperlipidemia.5,6

Medication noncompliance is prevalent among car-diovascular patient population reaching as high as 50–80% in western countries.7 For example, the Aggrenoxversus Aspirin Treatment Evaluation (AGATE) Trialrevealed that nearly 10% of postischemic stroke pa-tients purposefully discontinued taking their prescribedaspirin.8 Given the increased frequency of minor bleed-ing in patients treated with dual antiplatelet therapy,the 10% noncompliance level with aspirin therapy ob-served in the AGATE trial where bleeding was citedas a major factor in noncompliance, suggests that non-compliance will be appreciably higher among patientstreated with ticlopidine and clopidogrel.8 After dis-charge, nearly 1 of 4 patients (24%) with coronaryartery disease are noncompliant within a week of thedischarge,9 and the long-term self-reported complianceto combined aspirin, β-blocker, and lipid-loweringagent may be at most 40%.10 Treatment noncompliancemay be a significant factor in bleeding events and nega-tively impact treatment efficacy,11 leading to increasedhospitalizations, mortality, and costs.7 Therefore, med-ication noncompliance, along with associated factorsincluding severity of illness and current health sta-tus,5,6 should be considered as potential confounderswhen assessing treatment outcomes.

Nork Marash Medical Center (NMMC) is a 90-bedcardiac hospital located in Yerevan, Armenia, that con-ducts an average of 400 CABG and 500 stent revascu-larizations per year with outcomes comparable to itspeers in the region.12 A 2007 study assessed medica-tion adherence following CABG surgery at NMMC.Over 61% of patients were noncompliant,13 as mea-sured by the Morisky Adherence Scale (MAS) scale.14

This study involved NMMC patients that had PCI withDES from 2006 to 2008. This study assessed med-ication noncompliance and its relationship with pa-tient satisfaction and other potential predictors andconfounders.

Methods

A simple random sample of 375 patients whohad percutaneous coronary artery intervention withDES during 2006–2008 at NMMC was drawn fromNMMC’s cardiovascular patient registry and contactedby telephone for study eligibility and consent. The el-igible patient had to be an adult, residing in Armeniaat the time of the survey, and able to give consent tothe telephone interview. Patients with a prior CABGor stent placement were excluded. Oral informed con-sent was obtained from each patient before asking thesurvey questions. This sample (targeted enrollment of270) was powered to detect a 20% difference in sat-isfaction, allowing for refusals and those not meetingeligibility requirements.

The study protocol complies with the Declaration ofHelsinki as reflected in a priori approval by the Institu-tional Review Board (IRB) of the American Universityof Armenia.

Standard of Care. Before PCI with DES, NMMCinterventional cardiologists routinely discuss with pa-tients the need for and the duration of antiplatelet ther-apy, the impact of noncompliance, and the potentialmedication side effects. After discharge, all NMMCPCI patients are assigned follow-up visits to the out-patient clinic monthly for the first 6 months, at the endof the 12th month, and annually thereafter. Prescribedmedications are discussed in detail and educationalbooklets are provided at discharge and as needed ateach follow-up visit with the treating physician. In ad-dition, patients are encouraged to contact the outpatientclinic any time they have questions or concerns.

Study Instrument and Outcomes. The studyinstrument was developed by adapting questionsfrom validated questionnaires for the study purposesand to the NMMC context. The survey questionselicited information about patients’ demographiccharacteristics, health status, medication compliance,and satisfaction. Medication compliance items wereadapted from the MAS with binary response options.14

Compliance was determined based on MAS scoresranging from 0 to 4, with lower scores correspondingto better compliance. Patients who were compliantwith 3 or more of the 4 MAS items (MAS score 0or 1) were classified as compliant. Patients compliantwith fewer than 3 items (MAS score 2, 3, or 4)were classified as noncompliant.15 Respondents alsowere asked to identify the reasons for medicationnoncompliance.

2 Journal of Interventional Cardiology Vol. 00, No. 0, 2012

MEDICATION NONCOMPLIANCE AND PATIENT SATISFACTION

Patient satisfaction items were drawn from threeof the seven subscales of the Patient SatisfactionQuestionnaire (PSQ-18): General Satisfaction, Inter-personal Manner, and Communication.16 Each itemwas accompanied by 5 Likert-type response cate-gories (strongly agree, agree, uncertain, disagree, andstrongly disagree) scored from 1 to 5 with higher scoresreflecting increased satisfaction with medical care. Amean satisfaction score was calculated within eachscale. Patients were dichotomized as “satisfied” (meanscore ≥4) or “not satisfied” (mean score <4). Patients’current health status was assessed using a self-reportedmeasure scaled as “excellent,” “very good,” “good,”“fair,” or “poor” adapted from the SF-36,17 then usedas 3 levels, “good” and collapsing the 2 upper levelsand the 2 lower levels into a single category each.

Data Analysis. Data were entered into an SPSSdataset, cleaned, and recoded for further analysis. Dif-ferences in characteristics between compliant and non-compliant patients were compared using either a stan-dard 2-tailed t-test (for continuous variables) or a chi-square test (for dichotomous variables). Unadjustedodds ratios (ORs) were used to assess the relationshipbetween each of the independent variables and the de-pendent variable (medication noncompliance). Inter-actions between health status and patient satisfactionand between gender and smoking level also were as-sessed. Independent predictors of noncompliance (in-cluding patient satisfaction) were assessed using mul-tiple logistic regression analysis. Backward stepwiseelimination was used to build the model that includedall variables significant at P < 0.05 in the univari-ate analysis. The final model was tested for potentialconfounders and effect modifiers. Model fit was as-sessed using the Hosmer–Lemeshow goodness-of-fitstatistic and area under the receiver operating curve(ROC).18

Results

Descriptive Statistics. Of 375 patients selectedfor interview, 271 complied—a response rate of 72.3%.Of the 104 nonrespondents, 1 refused to participate,1 stopped the interview (incomplete), 7 had died, 10were absent from the country, and 85 were not reach-able (wrong number, no answer, moved, etc.). The av-erage time between stent placement and the interview(elapsed time) was 22.5 months for the patients in-cluded in the study and 24.2 months for the patients

who were not reached. This difference was not statis-tically significant. Demographic and other study par-ticipant characteristics are listed in Table 1. The meanage of participants was 57.5 ± 9.8 years; 238 (87.8%)were male. Among the participants, 102 (37.7%) eval-uated their health “excellent” or “very good,” with147 (54.2%) reporting “good” and 22 (8.1%) report-ing “fair” or “poor” health status. Nearly half (46.9%,n = 127) were former smokers, 83 (30.6%) currentsmokers, and 61 (22.5%) had never smoked.

Patient Satisfaction. As depicted in Table 1,77.5% (95% CI: 72.5–82.5) of respondents were “gen-erally satisfied,” 95.9% (95% CI: 93.6–98.3) were sat-isfied with their providers’ “interpersonal manner,”and 88.2% (95% CI: 84.3–92.1) were satisfied withprovider “communication.” Almost all (97.8%, 95%CI: 96.0–99.5) were satisfied with their discharge in-structions.

Medication Noncompliance. Approximately31.0% of the patients were noncompliant (n = 84)based on their MAS score (Table 2). The results ofresponses to individual MAS items are presented inTable 3. Of all respondents, 62 (22.8%) cited at leastone reason for noncompliance. The most commonlycited reasons for medication noncompliance werecosts/affordability (19.2%, n = 52) and unavailabilityat the pharmacy (3.7%, n = 10) (Table 4). Less-frequently cited reasons included travel distance, poormemory, and perception that the medications wereineffective. No respondent mentioned side effects as areason for medication noncompliance.

Standard 2-Tailed t-Tests and Chi-Square Tests.Noncompliant patients were significantly more likelyto be younger (54.3 vs. 58.9 years, P = 0.000), male(95.2% vs. 84.5%, P = 0.01), and to have difficultieswith the cost of their medications (28.6%, vs. 14.9% P= 0.009) and significantly less likely to never have beensmokers (13.1% vs. 26.7%, P = 0.04) and to report“very good” or “excellent” health status (26.2% vs.42.8%%, P = 0.03). Medication noncompliance wasnot associated with patient satisfaction—neither by theaggregate measure of satisfaction nor for any of itsconstituent domains. Time since stent placement wasnot significantly associated with noncompliance (22.4vs. 22.6 months, P = 0.89). No other variables yieldedsignificant associations with compliance (results notshown).

Logistic Regression. The bivariate logistic regres-sion analysis (unadjusted ORs) identified the same


AGHABEKYAN, ET AL.

Table 1. Participant Demographic, Health, and SatisfactionCharacteristics (n = 271)

Variable Name Result (n, %)

Age (mean ± SD) 57.5 ± 9.8Male 238(87.8)Nationality

Armenian 268(98.9)Russian 3(1.1)

EducationSchool (<10 years) 16(5.9)School (10 years) 60(22.1)Professional/technical education (10–13) 50(18.5)Institute/university 135(49.8)Postgraduate 10(3.7)

Marital statusMarried 261(96.3)Single 3(1.1)Divorced 1(0.4)Widowed 6(2.2)

Employed 143(52.8)If unemployed

Unemployed, looking for work 9(6.9)Unemployed, not looking for work 18(13.9)Permanently disabled 14(10.8)Homemaker 11(8.5)Retired 73(56.1)Self-employed 2(1.5)Farmer 3(2.3)

Household income (last month)∗Less than 25,000 drams 22(8.1)From 25,000 to 50,000 drams 56(20.7)From 51,000 to 100,000 drams 74(27.3)From 101,000 to 250,000 drams 62(22.9)Above 250,000 drams 41(15.1)Don’t know 16(5.9)

Health CharacteristicsHealth statusExcellent 17(6.3)Very good 85(31.4)Good 147(54.2)Fair 21(7.7)Poor 1(0.4)

Smoking statusNever 61(22.5)Former 127(46.9)Current 83(30.6)Diabetes 52(19.2)Hypertension 135(49.8)Current angina pectoris 66(24.4)

Cardiac rehabilitation referralRarely or not at all 10(3.7)Some of the time 49(18.1)Almost always 212(78.2)Elapsed time (months), (mean ± SD) 22.5 ± 6.9

SatisfactionGenerally satisfied 210(77.5)Satisfied with interpersonal manners 260(95.9)Satisfied with communication 239(88.2)Satisfied with discharge instructions 265(97.8)

∗The exchange rate was approximately 375 Armenian Drams/US$ duringthe study period.

Table 2. Participant Medication Adherence Scale (MAS) Score

MAS score N (%) Cumulative%

Compliant 0 118(43.5) (43.5)1 69(25.5) (69.0)

Noncompliant 2 53(19.6) (88.6)3 26(9.6) (98.2)4 5(1.8) (100.0)

Table 3. Medication Adherence Scale (MAS) Items

Items N (%)

Forgot to take medication 104(38.4)Careless about taking medication 95(35.1)When feeling better, stopped taking medicine 58(21.4)When feeling worse, stopped taking medicine 16(5.9)

Table 4. Reasons of Medication Noncompliance

Reasons N (%)

Costs/affordability 52 19.2Unavailability at the pharmacy 10 3.7Travel distance 4 1.5Poor memory 3 1.1Perception that the medications were ineffective 2 0.7

factors for inclusion in the multivariate model (age,gender, health status, smoking status, and cost) asdid the 2-tailed t-tests and chi-squared analyses re-ported earlier. The interaction between health statusand general satisfaction was not statistically significant(Table 5). The interaction between smoking statusand gender was statistically significant in the bivari-ate analysis and, thus, included in the multivariatemodel.

The multiple logistic regression model revealed thatcost, health status, and age were independent predic-tors of medication noncompliance. After adjusting forhealth status and age, the odds of being noncompliantwere 2.57 times higher (95% CI: 1.33–4.97) among thepatients for whom the cost for buying medications wasa problem. Similarly, the odds of being noncompli-ant with prescribed medications decreased by 6% foreach additional year of age (OR = 0.94, 95% CI: 0.91–0.97), and by 54% (OR = 0.46, 95% CI: 0.25–0.85) forpatients reporting “very good” or “excellent” health.The model had reasonable fit (Hosmer–Lemeshow,P = 0.146) and good discrimination (the area underthe ROC was 0.702).18



Table 5. Unadjusted and Adjusted Logistic Regression Analyses of Medication Noncompliance

Crude Odds Ratios Adjusted Odds Ratios

Variable Name Odds ratio 95% CI P value Odds ratio 95% CI P value

Age (years) 0.95 0.92–0.98 0.000 0.94 0.91–0.97 0.000Gender

Female 1.0 – 0.018∗Male 3.67 1.25–10.81

Health status (categorical)Very good /excellent 0.49 0.27–0.87 0.015 0.46 0.25–0.85 0.013Good 1.00 – 1.00Fair/poor 1.23 0.49–3.06 0.66 1.14 0.45–2.90 0.78

Smoking statusNever 1.00 –

Former 2.44 1.10–5.39 0.03∗Current 2.41 1.14–5.09 0.02∗

Comorbidities (yes vs. no)Diabetes 0.91 0.48–1.73 0.77Hypertension 1.22 0.73–2.04 0.46Current “angina pectoris” 0.79 0.44–1.43 0.44

Postoperative considerations (yes vs. no)Cardiac rehabilitation referral 0.71 0.44–1.14 0.15Elapsed time (months) 0.99 0.96–1.04 0.89

Reasons for noncompliance (yes vs. no)Medication affordability/cost problems 2.27 1.22–4.22 0.01 2.57 1.33–4.97 0.005Absence of prescribed medication 1.83 0.38–8.82 0.45

Satisfaction (satisfied vs. not)General satisfaction 0.99 0.54–1.83 0.98Satisfaction with interpersonal manners 2.07 0.44–9.81 0.36Satisfaction with communication 0.62 0.29–1.32 0.21Satisfaction with discharge instructions 0.89 0.16–4.99 0.90

Interaction termsHealth status × general satisfaction 0.63 0.39–1.04 0.07Smoking status × gender 1.42 1.08–1.86 0.013∗

∗These variables were no longer statistically significant in the multiple logistic regression model and omitted from the final fitted model.

Discussion

This study estimated the prevalence of noncompli-ance with medications in patients with stent placementat NMMC in Armenia. Almost one-third (31%) ofour respondents were noncompliant. These findingsare consistent with the published literature,7 but dis-similar to an unpublished study assessing adherence tomedications in patients after CABG surgery conductedat NMMC in 2007. In that study using the same MASscale, 61% were noncompliant.13 This nearly twofolddifference in reported medication noncompliance maybe due to the study populations (PCI vs. CABG pa-tients). Patients undergoing CABG surgery do not relypostoperatively on medications to the extent that PCIpatients do, perhaps making medication compliance

less emphasized in their follow-up treatment. In ad-dition, the PCI and CABG studies involved differentdrug categories with different regimens and potentialside effect profiles that could yield different rates ofnoncompliance.

This study also measured patient satisfaction withcare, discharge instructions, and the physician–patientrelationship. PCI patients at NMMC rated all of theseaspects of satisfaction very high, a finding consistentwith an unpublished study conducted at the NMMCOutpatient Clinic in 2003 among CABG patientswhere 96% of patients were satisfied with nursingcare and 92% with physician care.19 The same studyfound that 94.5% of patients with regular follow-up and 89.7% of those lost to follow-up were satis-fied with services.19 The finding in this study of no


AGHABEKYAN, ET AL.

statistically significant association between noncom-pliance and patient satisfaction is inconsistent with thepublished literature.5,6 In our study, this finding maybe partly explained by highly (positively) skewed sat-isfaction levels that mask the hypothesized associationbetween medication noncompliance and patient sat-isfaction. Elapsed time since stent placement was notassociated with noncompliance, minimizing the threatsconnected with patient recall and treatment changes bydoctor within the sample.

Study patients who were noncompliant with theirprescribed medications differed significantly fromcompliant patients in several ways that are consistentwith the published literature. For example, an asso-ciation was observed between medication noncompli-ance and medication affordability/cost.16,20 In addi-tion, younger patients were more likely to be non-compliant,6 and patients with “very good”/“excellent”health status were significantly less likely to be non-compliant than patients with “good” health status.However, contrary to the literature,20 “fair”/“poor”health status patients were not more likely to be non-compliant than “good” health status patients. Thisnull finding might be explained by the small frac-tion of patients (about 8%) reporting “fair” or “poor”health status and the resulting unstable parameterestimates.

The study design imposed several limitations. Be-cause of possible treatment changes, the study ad-dressed medication noncompliance in general termsrather than focusing specifically on noncompliance todual antiplatelet therapy. Prior stent replacement orCABG (n = 5) was used as an exclusion criterion,as was death before the study (n = 7). These criteriamight have preferentially excluded patients who werenoncompliant in the past, creating a conservative bias.Self-reported information on medication noncompli-ance could be subject to reporting bias as patients oftenreport more desirable answers to questions. Likewise,patients’ health status is a self-reported measure. Al-though a known and reliable indicator, it can deviatefrom objective measures of health status. Respondentsalso were asked questions about the services they re-ceived at the time of the procedure (e.g., dischargeinstructions, satisfaction with services), which couldpotentially introduce recall bias due to the elapsedtime between the intervention and the interview. An-other limitation is that the patients were not askedabout which medication(s) they were prescribed or

which medication was too costly for them. However,this is the first study in this population in Armeniaand future studies are planned to investigate these is-sues prospectively. Likewise, further study is neededto assess the impact of medication noncompliance onhealth outcomes. A qualitative study could prove use-ful in more fully identifying the main obstacles tomedication compliance and opportunities to increasecompliance.

Reasons for medication noncompliance aremultifactorial—often related to health system, pa-tient, condition, therapy, and socioeconomic factors.7

A known obstacle to medication compliance was med-ication affordability.7,13 Due to economic conditionsin Armenia, PCI patients must pay out-of-pocket fortheir medical care, including diagnostic procedures,interventions, and medications. Being unable to affordneeded medical care, patients’ delay seeking medicalcare and diminishing treatment effectiveness. To im-prove compliance and, thus, treatment outcomes, thegovernment, the pharmaceutical industry, the healthsystem, and health advocates should implement strate-gies to reduce the economic burden on patients.

Younger patients, those with “fair” or “poor” per-ceived health status, or those for whom cost presents aproblem are more at-risk for noncompliance. NMMCshould target these groups with quality assurance ac-tivities that address medication noncompliance. In ad-dition to following treatment guidelines, physiciansshould continually reinforce lifestyle advice for sec-ondary prevention of heart attacks (e.g., smoking ces-sation and controlling high blood pressure).

Conclusion

In conclusion, the majority of Armenian patientswere medication compliant after PCI. Medication af-fordability, patients’ health status, and age at the time ofthe interview were the main predictors of compliance.These findings are generally consistent with those ob-served elsewhere. The methods used in this study areapplicable to other drug compliance studies and shouldhelp to improve health outcomes for the population inArmenia and the region. These findings will contributeto ongoing quality assurance activities at NMMC andfurther efforts to benchmark performance in Arme-nia and other post-Soviet countries against westernstandards and experiences.



References

1. The top 10 causes of death. Fact sheet No 310. World HealthOrganization (Online). 2008. Available at: http://www.who.int/mediacentre/factsheets/fs310/en/. Accessed 2009 January18.

2. Hassan A, Newman A, Ko DT, et al. Increasing rates of angio-plasty versus bypass surgery in Canada, 1994–2005. Am HeartJ 2010;160(5):958–965.

3. Iakovou I, Schmidt T, Bonizzoni E, et al. Incidence, predictors,and outcome of thrombosis after successful implantation ofdrug-eluting stents. JAMA 2005;293(17):2126–2130.

4. Grines CL, Bonow RO, Casey DE Jr, et al. Prevention ofpremature discontinuation of dual antiplatelet therapy in pa-tients with coronary artery stents: A science advisory fromthe American Heart Association, American College of Car-diology, Society for Cardiovascular Angiography and In-terventions, American College of Surgeons, and AmericanDental Association, with representation from the Ameri-can College of Physicians. Anesth Prog J 2007;54:161–162.

5. Moreira LB, Fernandes PF, Mota RS, et al. Medica-tion noncompliance in chronic kidney disease. J Nephrol2008;21(3):354–362.

6. Kim YS, Sunwoo S, Lee HR, et al. Korea Post-MarketingSurveillance Research Group. Determinants of non-compliancewith lipid-lowering therapy in hyperlipidemic patients. Phar-macoepidemiol Drug Saf 2002;11(7):593–600.

7. Ho PM, Bryson CL, Rumsfeld JS. Medication adher-ence: Its importance in cardiovascular outcomes. Circulation2009;19(23):3028–3035.

8. Serebruany V, Malinin A, Sane D, et al, for the AGATE TrialInvestigators. The AGgrenox versus aspirin treatment evalua-tion (AGATE) for platelet inhibition after stroke. Circulation2003;108:IV–602.

9. Jackevicius CA, Li P, Tu JV. Prevalence, predictors, and out-comes of primary nonadherence after acute myocardial infarc-tion. Circulation 2008;117:1028–1036.

10. Newby LK, LaPointe NM, Chen AY, et al. Long-termadherence to evidence-based secondary prevention thera-

pies in coronary artery disease. Circulation 2006;113:203–212.

11. Saw J, Steinhubl SR, Berger PB, et al. Lack of adverseclopidogrel-atorvastatin interactions from secondary analysesof a randomized placebo-controlled clopidogrel trial. Circula-tion 2003;108:921–924.

12. Abrahamyan L, Demirchyan A, Thompson ME, HovaguimianH. Determinants of morbidity and intensive care unitstay after coronary surgery. Asian Cardiovasc Thorac Ann2006;14(2):114–118.

13. Nahapetyan A. Relationship between patients’ knowledgeabout post operative risk factors after coronary artery bypasssurgery (CABG) and adherence to medication and lifestylechanges in Armenia (unpublished MPH dissertation). Yerevan,Armenia: College of Health Sciences, American University ofArmenia, 2007. Available at: http://chsr.aua.am/mph2007.php.Accessed 2008 December 11.

14. Morisky DE, Green LW, Levine DM. Concurrent and predictivevalidity of a self-reported measure of medication adherence.Med Care 1986;24:67–74.

15. Vik SA,Maxwell CJ, Hogan DB, et al. Assessing medicationadherence among older persons in community settings. Can JClin Pharmacol 2005;12(1):e152–e164.

16. Marshall GN, Hays RD. The Patient Satisfaction QuestionnaireShort-Form (PSQ-18). Santa Monica, CA: RAND Corpora-tion, 1994.

17. Ware JE, Jr, Sherbourne CD. The MOS 36-item short-formhealth survey (SF-36): I. Conceptual framework and item se-lection. Med Care 1992;30(6):473–483.

18. Hosmer D, Lemeshow S. Applied Logistic regression, 2nd edi-tion, New York: Wiley Series in Probability and Statistics,2000, p. 340.

19. Abrahamyan L. Research on lost to follow-up of patients atthe outpatient clinic of Nork Marash Medical Center (unpub-lished MPH dissertation). Yerevan, Armenia: College of HealthSciences, American University of Armenia, 2003. Availableat: http://chsr.aua.am/mph2003.php. Accessed 2008 December13.

20. Wroth TH, Pathman DE. Primary medication adherence in arural population: The role of the patient-physician relationshipand satisfaction with care. JABFM 2006;19(5):478–486.


medication noncompliance and patient satisfaction following percutaneous coronary intervention

Documents