Table 4.

Characteristics, methods, and results of the included studies

Number, Name of Author, date, country, N, Age and Sex	Type of medication adherence and measurement tools	Determinants found significant	Results
1. Roe et al. [24] USA N = 869 Age = 60 Men = 51 %	Medication compliance, measured by medication possession ratio (MPR), continuation of therapy and dosing, were calculated based on medical and pharmacy claims data from a database containing information on more than 1.1 million Americans belonging to numerous health plans MPR: supply of ACEi/number of days between the first claim and ACEi during the post period and the end of the post period Continuation of therapy: Termination date minus the date of the index prescription Dosing: mean milligrams dispensed per day = the mg per tablet (or capsule) multiplied by the quantity of medication dispensed, divided by the days supply as indicated by pharmacist. Mean mg. dispensed per day was added across prescriptions and divided by the total number of prescriptions, leading to a mean dispensed dose per prescription. Mean percentage of an adequate daily dose dispensed was calculated as the mean milligrams dispensed per day divided by the adequate daily dose for the medication	MPR: 1. Sex (male) 2. Chronic disease score 3. Systolic proxy diagnosis 4. Outpatient visits 5. New user 6. Renal insufficiency 7. Enalapril 8. Lisinopril 9. Switched medication 10. Antihyperlipidemic agents Presumed determinants for which a non-significant (NS) relationship with medication adherence was found: 11. Prior myocardial infarction 12. Other ACE inhibiter medication Continuation of therapy: 1. Sex (male) 2. Outpatient visits 3. New user 4. Renal insufficiency 5. ACE inhibitor Enalapvil 6. Switched medication 7. Digitalis Presumed determinants for which a NS relationship with medication adherence was found: 8. Lisinopril 9. Other ACE inhibiter 10. Other cardiovascular drugs Dosing: 1. Outpatient visits 2. New user 3. Enalapril 4. Lisinopril 5. Other medication 6. Switched medication 7. Other hypertensive agents 8. Beta blockers	MPR: 1. B = 0.047, p < 0.05 2. B = −1.23, p < 0.05 3. B = 0.045, p < 0.05 4. B = 0.132, p < 0.001 5. B = −0.099, p < 0.0001 6. B = −0.159, p < 0.005 7. B = −0.072, p < 0.05 8. B = 0.110, p < 0.0005 9. B = 0.126, p < 0.0001 10. B = 0.048, p < 0.05 Continuation of therapy: 1. β = 0.56, p < 0.005 2. β = 0.46, p < 0.0001 3. β = 2.70, p < 0.0001 4. β = 2.16, p < 0.001 5. β = 1.85, p = 0.05 6. β = 0.25, p < 0.005 7. β = 0.76, p < 0.05 Dosing: 1. B = 0.159, p < 0.01 2. B = −0.196, p < 0.0005 3. B = 0.284, p < 0.0005 4. B = 0.504, p < 0.0001 5. B = 0.767, p < 0.0001 6. B = 0.463, p < 0.0001 7. B = 0.265, p < 0.0005 8. B = 0.133, p < 0.05
2. Bagchi et al. [25] USA N = 45572 Age = unknown Men = 28 %	MPR and persistence were used to measure adherence to therapy. Data extracted from Medicaid files MPR: the number of days a patient was supplied with more than one CHF drug in relation to the patient’s first and last prescription dates Persistence: The number of days of continuous use of CHF medications per month	Determinants of medication possession ratio: 1. Arkansas 2. Indiana 3. New Jersey 4. Age 65–74 year 5. Age 75–84 year 6. Age >85 year 7. Comorbid coronary artery disease 8. Comorbid diabetes mellitus 9. Dually eligible 10. Disabled 11. Arkansas 12. Men 13. Black race 14. Other/unknown race 15. CHF-related hospitalization in 1998 16. Non-CHF related hospitalization in 1998 17. High Chronic Disease and Disability Payment System scores 18. Percentage of generic CHF drugs Determinants of persistence: 1. Indiana 2. New Jersey 3. Arkansas 4. Age 65–74 5. Age 75–84 6. Age >85 7. Dually eligible 8. Disabled 9. Comorbid coronary artery disease 10. Comorbid diabetes mellitus 11. CHF-related hospitalization in 1998 12. Non-CHF related hospitalization in 1998 13. Black race 14. Other/unknown race 15. Non-CHF-related hospitalization 16. Chronic Disease Payment System risk score 17. Percentage of generic CHF drugs	MPR: 1. β = 1.51 (SE 0.433) 2. β = −4.79 (SE 0.422) 3. β = 1.97 (SE 0.400) 4. β = 2.14 (SE 0.489) 5. β = 4.45 (SE 0.563) 6. β = 5.27 (SE 0.644) 7. β = 5.42 (SE 0.307) 8. β = 4.75 (SE 0.310) 9. β = 1.72 (SE 0.388) 10. β = 2.58 (SE 0.411) 11. β = −4.79 (SE 0.422) 12. β = −1.19 (SE 0.314) 13. β = −6.23 (SE 0.337) 14. β = −4.94 (SE 0.385) 15. β = 2.59, (SE 0.285) 16. β = −1.65, (SE 0.289) 17. β = −2.75 (SE 0.174) 18. β = −0.06 (SE 0.004) Persistentce: 1. β = 0.55 (SE 0.130) 2. β = 0.52 (SE 0.120) 3. β = −1.08 (SE 0.127) 4. β = 0.63 (SE 0.147) 5. β = 1.24 (SE 0.169) 6. β = 1.65 (SE 0.193) 7. β = 0.45 (SE 0.116) 8. β = 0.63 (SE 0.123) 9. β = 1.26 (SE 0.092) 10. β = 1.12 (SE 0.093) 11. β = 0.90 (SE 0.086) 12. β = −0.28 (SE 0.086) 13.β = −1.50 (SE 0.101) 14. β = −1.28 (SE 0.116) 15. β = −0.28 (SE 0.087) 16. β = −0.72 (SE 0.052) 17. β = −0.02 (SE 0.001) All p < 0.01, reference groups for state: California
3. Cholowski et al. [26] Australia N = 54 Age = 72 Men = 61 %	Medication compliance was measured with a semi-structured interview. Four compliance behaviors were measured: forgetting to take medication, being careless about taking medication, stopping to take medication when feeling better, stopping to take medications because of feeling worse as a result of taking it	Stopping to take medications as a result of feeling worse: 1. Not complying when feeling worse as a result taking medication was related to number of co morbidities 2. Being careless about taking medication was related to depression 3. Being careless about taking medication was related to perceiving barriers to dietary compliance 4. Men were more likely to be careless about taking medications 5. Total compliance scores (including the four compliance behaviors) were related to beliefs about medication compliance (including both of the scales about perceived benefits and barriers) 6. Total compliance scores were related to the perceived barriers scale (but not the perceived benefit scale) when the scales were assessed separately Presumed determinants for which a NS relationship with medication adherence was found: 7. Number of medications 8. Number of risk factors 9. Proactive coping 10. Reflective coping 11. Strategic planning 12. Preventative coping 13. Instrumental support seeking 14. Avoidant coping 15. Self-regulation 16. Benefits to medication compliance 17. Beliefs about dietary compliance 18. Age	1. r = −0.43, p < 0.05 2. r = −0.31, p < 0.05 3. r = −0.35, p < 0.05 4. t = −2.16, p < 0.05 5. r = −0.33, p < 0.05 6. r = −0.42, p < 0.05
4. Molloy et al. [27] UK N = 147 Age = 80 Men = 57 %	ACE activity measured with serum from clotted blood	Illness beliefs about the following topics: 1. Length of the condition and the cyclical nature of it 2. The consequences of the condition 3. The personal control patients have over their condition 4. That treatments will be effective 5. That the illness makes sense 6. That it will make them emotionally distressed 7. That the illness has symptoms Presumed determinants for which a NS relationship with medication adherence was found: 8. Time-line acute/chronic	Time-line cyclical Consequences Personal control Treatment control Illness coherence Emotional representations Identity (These determinants were found to be significantly related to adherence at p = 0.10. We only regard significant relationships as those with a p value < 0.05)
5. Sayers et al. [28] USA N = 163 Age = 63 Men = 96 %	A four-item questionnaire	1. Emotional support Presumed determinants for which a NS relationship with medication adherence was found: 2. Instrumental support 3. Family involvement	β = −0.41, p < 0.05
6. Schweitzer et al. [29] Australia N = 115 Age = 64 Male = 71 %	The heart failure compliance questionnaire	Presumed determinants for which a NS relationship with medication adherence was found: 1. Age 2. Gender 3. NYHA 4. LVEF 5. Depression 6. Anxiety 7. Self-efficacy
7. Wu et al. [30] USA N = 134 Age = 61 Men = 70 %	The measurement tool used was a medication monitoring system (MEMS): an unobtrusive microelectronic monitoring device in the caps of medication bottles. With this system, medication adherence was indicated with: 1. Dose count: the % of prescribed doses taken 2. Dose days: the % of days that right number of doses were taken 3. Dose time: the % of doses that were taken on schedule	Dose count: 1. Treatment-related barriers 2. Socio economic 3. Perceived social support Presumed determinants for which a NS relationship with medication adherence was found: 4. Gender 5. Age 6. Attitudes 7. Knowledge 8. NYHA 9. Comorbidity 10. Depression 11. Number of pills taken per day 12. Medication frequency 13. Patient-provider relationship 14. Educational level 15. Financial status Dose day: 1. NYHA 2. Barriers 3. Financial status 4. Perceived social support Presumed determinants for which a NS relationship with medication adherence was found: 5. Gender 6. Age 7. Attitudes 8. Knowledge 9. Comorbidity 10. Depression 11. Number of pills taken per day 12. Medication frequency 13. Patient-provider relationship 14. Ethnicity 15. Educational level 16. Financial status 17. Perceived social support Dose time: 1. Treatment-related barriers 2. Financial status Presumed determinants for which a NS relationship with medication adherence was found: 3. Gender 4. Age 5. Attitudes 6. Knowledge 7. NYHA 8. Comorbidity 9. Depression 10. Number of pills taken per day 11. Medication frequency 12. Barriers 13. Patient-provider relationship 14. Ethnicity 15. Educational level 16. Perceived social support	Dose count: 1. β = 0.352, p < 0.001 2. β = −0.208, p = 0.025 3. β = −0.241, p = 0.014 Dose day: 1. β = 0.181, p = 0.049 2. β = 0.349, p < 0.001 3. β = 0.208, p = 0.036 4. β = −0.221, p = 0.026 Dose time: 1. β = 0.268, p = .008 2. β = 0.216, p = .039
8. Evangelista et al. [34] USA N = 82 Age = 54 Men = 38 %	A modified version of the Compliance Questionnaire	1. Age 2. Neuroticism Presumed determinants for which a NS relationship with medication adherence was found: 3. Race 4. Education 5. Marital status 6. Mental health 7. Physical health 8. Health satisfaction	1. Adjusted R2 = 0.185, p = .000 2. Adjusted R2 = 0.252, p = .006
9. Monane et al. [31] USA N = 7247 Age = 77 Men = 21 %	Digoxin filling during 12 months: Nr. of days without therapy was computed and used as a measure of (non)compliance	Estimated number of days without therapy by: 1. Age 2. Race 3. Female gender 4. Institutionalization (hospitalization or nursing home stay) 120 days prior to digoxin prescription 5. Number of pharmacies used 120 days prior to digoxin prescription, 6. Number of non-study medications 120 days prior to digoxin prescription 7. Concurrent congestive HF medications 120 days prior to digoxin prescription Presumed determinants for which a NS relationship with medication adherence was found: 8. Age 75–84 9. Number of physicians seen	Number of days without therapy: 1. Older than 85: −17.0 days (CI −23.7, −10.3) 2. Other (not white or black): 13.6 days (7.3, 19.9) 3. −18.7 days (−24.6, −12.8) 4. −34.4 days (−39.7, 29.1) 5. 16.0 days (9.9, 22.1), 6. 4 to 7 medications −6.4 days (−12.3, −0.5) 8 or more medications 7.4 days (−13,3, −0,7) 7. Yes: −56.3 days (−61.4, 51.2) (p value for all <0.05)
10. Rodgers et al. [32] USA N = 64 Age = 65 Men = 57 %	Medication non-adherence was calculated as follows: Non-adherence by percent acquisition = days supply dispensed/actual days between refills × 100. It is not specified how they had data to make this calculation	1. Age 2. NYHA class 3. Hyperlipidemia 4. Asthma/COPD 5. Number of hospitalizations in the past year Presumed determinants for which a NS relationship with medication adherence was found: 6. Gender 7. Race 8. Number of years with congestive HF 9. Number of visits to primary care physician in the previous year 10. Number of health care professionals seen in previous three months 11. Visits to pharmacist managed outpatient clinics 12. Payment method 13. Number of enalapril doses per day 14. Number of other medications 15. Number of individual doses of all medications per day 16. Notation of adverse effects of enalapril 17. Tobacco or alcohol use	Predictors of non-adherence: 1. Age group 57–64 OR 17.8 Age group 65–72 OR 1.9 Age group 73–89 OR 3.3, 2. NYHA class II OR 0.04 NYHA class III OR 0.08 3. OR 0.09 4. OR 0.09 5. OR 0.16
11. Granger et al. [33] 25 participating countries (CHARM trial) N = 7599 Age = 66 Men = 78 %	Compliance was estimated by patients report, investigators’ inspection of pill bottles and tablet count in case of uncertainty	1. Gender (female) 2. Number of comorbid illnesses 3. Heart rate 4. Presence of pacemaker 5. Number of medications Presumed determinants for which a NS relationship with medication adherence was found: 6. Age 7. NYHA class 8. Ejection fraction 9. Systolic blood pressure 10. Body mass index 11. Smoking (current)	1. β = −0.049 p ≤ 0.001 2. B = −0.041 p = 0.001 3. B = −0.051 p = 0.000 4. B = −0.027 p = 0.019 5. B = 0.030 p = 0.022

NS nonsignificant