TABLE 1.

Studies Included in Systematic Review

Source	Country	Participant Profile		How Adherence Is Assessed	Main Factors Addressed for Primary Nonadherence
		Final Sample Size	Sample Description		Patient Factors	Medication Factors	Health Care Provider Factors	Health Care System Factors	Socioeconomic Factors
Adamson et al. (2017)14	United States	2,496	New dermatology patients with 1 or more medications prescribed	Electronic medical records and pharmacy claims records	Higher PNA among younger patients Age factor differs in men and women	Number of dispensed drug (polypharmacy) Cost, especially for elderly patients	–	Method of prescription (electronic prescription vs. paper prescription) Infrastructure to accommodate the needs of non-English-speaking patients	–
Bauer et al. (2013)22	United States	1,366	Adults aged 30-75 years with type 2 diabetes who were prescribed a new antidepressant during 2006-2010	Pharmacy claims records	Health literacy and race/ethnicity	–	–	–	Patients with health literacy limitations have poorer adherence
Cheetham et al. (2013)15	United States	19,826	Patients aged ≥ 24 years with a new statin prescription (having no statin prescriptions in the previous 12 months) in a large integrated health care delivery system	Electronic medical records	Younger and healthier patients, with fewer comorbid conditions, lower rates of hospitalization, fewer clinic and emergency department visits in previous year	Fewer concurrent prescriptions	–	–	–
da Costa et al. (2015)20	Portugal	375	Patients aged ≥ 15 years with chronic medical conditions and a prescription of at least 1 drug for diabetes, hypertension, or hyperlipidemia	Questionnaire study and data of medication collection from pharmacist	Higher nonadherence among women for antidiabetic medication	Availability of medication at home Higher nonadherence among women for antidiabetic medication	–	–	Financial problems faced by patients
Fallis et al. (2013)33	Canada	232	Patients aged ≥ 66 years, discharged from the general internal medicine service of a hospital	Claims data in Drug Profile Viewer	–	–	More likely among patients discharged to nursing homes	–	–
Harrison et al. (2013)28	United States	98	Patients aged ≥ 24 years, with no record of redeeming a new statin medication within 1 to 2 weeks of being ordered	Phone interview data; self-reported nonadherence	Concerns about taking the medication Patients’ preference for lifestyle modification (e.g., diet and exercise) instead of taking medication Fear of side effects Patients’ perceived redundancy and ineffectiveness of medication	–	Lack of communication between patient and physician	Financial hardships Inadequate health literacy
Jackevicius et al. (2008)19	Canada	4,591	Acute myocardial infarction (AMI) patients aged ≥ 66 years, enrolled in the Enhanced Feedback for Effective Cardiac Treatment (EFFECT) study registry	The EFFECT study registry and the AMI charts it collects from 104 acute care hospitals in Ontario	Patients’ perceived ineffectiveness and redundancy of medication Older patients	Patients with more pre-AMI/baseline prescriptions	Patients who do not receive medication counseling and education after discharge Patients who do not have a cardiologist as the most responsible physician	–	–
Jackson et al. (2014)16	United States	29,238	Adult patients aged ≥ 18 years, with a new electronic prescription for medications intended to treat chronic conditions, as supplied by the Pharmacy Quality Alliance (PQA)	Prescription data from 100 retail pharmacies	Slightly younger in age: 59.42 vs. 59.60 years	PQA-defined drug class (e.g., high nonadherence is observed for antiretrovirals) Higher out-of-pocket costs for medication Prescriptions accompanied by another prescription on the same day	Higher nonadherence when prescriber is neither a physician (both specialist and primary care), physician assistant, or advanced practice nurse Nonadherence more likely to occur in pharmacies with lower prescription volumes	–	Higher nonadherence when prescriptions originate in pharmacies located in neighborhoods with higher household incomes and educational levels
Karter et al. (2010)29	United States	169	Patients with type 2 diabetes receiving a new electronic prescription for insulin—those who are primary adherent and primary nonadherent	Data from computer-assisted interviews and self-administered mailed surveys	Patients’ decision to improve other health behaviors instead of insulin-taking Patients’ perceived negative impact on social and work life Injection phobia Concerns about side effects	–	Lack of provider-patient communication and explanation of the potential risks and benefits associated with insulin Inadequate shared decision making between provider and patient Lack of insulin self-treatment training for patients	–	–
Polinski et al. (2014)25	United States	26	Patients aged ≥ 25 years with PNA for anti-hypertensive medications	Focus group discussions	Patients’ misperception about medication Fear of side effects Patients’ distrust of health care provider Suspicion of provider’s diagnosis and motivation to prescribe	Cost Complexity of medication regimen (polypharmacy)	Poor communication between physician and patient	–	–
Pottegård et al. (2014)8	Denmark	146,959	Patients aged ≥ 18 years with free and direct access to general practitioners	Pharmacy records and prescription registry data	Female Younger patients aged 18-29 years Patients with a diagnosis of ischemic heart disease are less adherent Patients with a diagnosis of chronic obstructive pulmonary disease are more adherent	Polypharmacy	–	–	Highest PNA among patients who earn < 250,000 Danish krone per year
Raebel et al. (2011)7	United States	12,061	Members of an integrated health delivery system, with a newly initiated order for an antihypertensive, antidiabetic, or anti-hyperlipidemic medication	Electronic health records within an integrated system	–	PNA varied by therapeutic class—highest among those who ordered antihyperlipidemic medications and lowest among those who ordered anti-hypertensive medications	PNA lower for antihyperlipidemic medications prescribed by a provider in a nonprimary care department	–	–
Rashid et al. (2017)17	United States	9,050	Patients aged ≥ 18 years with new overactive bladder prescriptions	Electronic medical records	Female Younger patients aged 56.9 years on average, compared with the overall average age of 62.6 years Patients who have fewer comorbid conditions (i.e., generally healthier otherwise) Race other than white	Patients who have fewer concomitant medications and prescriptions dispensed in the past year	–	Patients who have commercial insurance compared with those with Medicaid and Medicare	–
Reynolds et al. (2013)30	United States	8,454	Women aged ≥ 55 years with a new prescription of oral bisphosphonates	Electronic medical records	Older women with prior emergency department visits Patients’ perceived need for and benefits and risks associated with medication	–	Prescriptions written by providers with 10 or more years of experience more likely to be redeemed	–	–
Shin et al. (2012)5	United States	569,095 new prescriptions (398,025 patients)	New prescriptions written for 10 therapeutic drug groups in a 3-month period	Electronic medical records	Black and Hispanic patients Patients naive to therapy and treatment Patients with baseline comorbidities Patients who redeem at least 1 prescription in the previous year or had a prescription for symptomatic disease Younger patients more likely to fill acute medication; less likely for chronic medication	Types of drug group Regimen complexity Cost of medication (only when disease is acute)	–	–	–
Storm et al. (2008)21	Denmark	322	Outpatients of a public hospital dermatology department, who receive a prescription for an initial treatment with a previously untried medication	Electronic pharmacy register	Patients with chronic diseases (e.g, eczema) less adherent than patients with short-term diseases (e.g., infections) Men more adherent than women Elderly patients the most adherent Patients with topical treatment (compared with those with systemic treatmenta) less adherent	–	Better adherence among patients who see specialists rather than junior physicians	–	–
Thengilsdóttir et al. (2015)6	Iceland	10,685	Adult patients aged ≥ 18 years from the capital area in Iceland who received a new statin or antidepressant prescription within the study period	Prescription database records	Vulnerable groups of patients with disabilities prescribed expensive drugs Women and younger patients	Patients prescribed SNRIs and atorvastatin compared with those prescribed SSRIs and simvastatin	–	–	–
Wamala et al. (2007)31	Sweden	31,895	Patients aged 21-84 years who corresponded with a physician at a hospital or primary care center within a 3-month period	Self-reported nonadherence using postal self-administered questionnaire	–	–	–	Sweden’s “care on equal terms” health policies (publicly funded health care system and subsidized medication) less successful among socio-economically disadvantaged elderly patients	Patients placed lower on the socioeconomic index, especially elderly women
Williams et al. (2007)23	United States	1,064	Asthma patients aged 5-56 years, with at least 1 electronic prescription for inhaled corticosteroids and at least 3 months follow-up after the prescription	Electronic prescription information and pharmacy fill data	Low baseline rescue medication use Lower perceived need for medication Race and ethnicity	–	Frequency of contact between patient and physician, especially for African American patients	–	–
Wooldridge et al. (2016)24	United States	341	Adult patients who were hospitalized for cardiovascular events, had new discharge prescriptions to fill post-discharge, and had received study intervention about filling discharge prescriptions	Secondary analysis of data from a randomized, controlled trial evaluating the effect of tailored intervention in adults hospitalized for acute coronary syndromes or acute decompensated heart failure	Single marital status	Polypharmacy (having more than 10 total discharge medications)	Not applicable as patients have undergone tailored intervention of pharmacy-assisted medication reconciliation, discharge counseling, low-literacy adherence aids, and follow-up phone calls	–	Low income: inability to afford medication cost and also faced transportation limitations
Wroth et al. (2006)18	United States	3,926	Adult aged ≥ 18 years, having lived in the southeastern rural community for more than 1 year, and visited a health care provider in the previous year	Phone survey data; self-reported non-adherence	Patients aged < 65 years African American Patients with transportation problems Trust and confidence in patient-physician relationship Patients’ satisfaction with care provided	–	Trust and confidence in patient-physician relationship Patients’ satisfaction with care provided Patient-physician concordance on medication	–	Annual income of < $25,000
Yu et al. (2015)26	United States	430	Women aged ≥ 55 years with an untreated osteoporosis diagnosis (i.e., no claims for osteoporosis-specific medication)	Mail survey data	Concern over side effects of medication Patients’ beliefs about osteoporosis and osteoporosis medication	–	–	Cost of medication: contribution of insurance to medication costs	–

^aUsing substances that travel through the bloodstream, reaching and affecting cells all over the body.

PNA = primary nonadherence; SNRI = serotonin-norepinephrine reuptake inhibitor; SSRI = selective serotonin reuptake inhibitor.