Determinants of Non-Adherence to the Medications for Dyslipidemia: A Systematic Review.

PURPOSE: Dyslipidemia is a major cardiovascular risk factor, and its control leads to less cardiovascular events. Many patients will need some medications to achieve ideal targets. Non-adherence to medications is a complex problem with high impact on their effectiveness. This study aims to identify the determinants of non-adherence to medications in patients with dyslipidemia. PATIENTS AND METHODS: We conducted a systematic review. PubMed and Scopus databases were searched for original articles, published between 2000 and 2020, using the MeSH terms "Dyslipidemias" and "Medication Adherence".
RESULTS: From the initial 3502 identified articles, we selected 46 to include in the final qualitative synthesis. The determinants associated with non-adherence were lower age (≤50 years), female sex, African American ethnicity, smoking habits, being a new user of lipid-lowering medications, reporting or having concerns about lipid-lowering medication side effects and some comorbidities (chronic obstructive pulmonary disease, Alzheimer's disease/dementia, depression and diabetes). On the contrary, adherence is higher in older patients, alcohol drinking habits, taking β-blockers, having a higher number of comorbidities, having a history of cardiovascular events, cardiac interventions or revascularization procedures, having health insurance and having more provider follow-up visits.
CONCLUSION: There are important identifiable determinants of non-adherence in patients with dyslipidemia. These patients benefit from a specific approach to minimize the problem and maximize the potential benefit of the prescription.

Introduction

Dyslipidemia is an important risk factor for developing cardiovascular diseases (CVD), the leading cause of death in the world, both for ischemic heart disease1 and stroke.1 In Europe, CVD account for more than 4 million deaths per year, about 45% of the total.2 In this study, we consider dyslipidemia as a broader term, covering a wide range of lipid abnormalities, and not merely having elevated total cholesterol and low-density lipoprotein.3 Therefore, patients were considered as having dyslipidemia if they had any lipid abnormalities that justified treatment, including secondary prevention of cardiovascular events and laboratory confirmation of hypercholesterolemia, hypertriglyceridemia or both. Medication non-adherence is a serious and complex problem. Adherence is defined as the extent to which the patients’ behavior correspond to the agreed recommendation of their doctors, both in taking the medications, following a diet or executing lifestyle changes.4 It depends on several variables related to the condition itself, the therapeutics, the factors related to the patient, including social and cultural features and the health system-related factors. It develops in an interactive triangle of patient, healthcare provider and healthcare system, where each one insert both barriers and facilitators, complementary and interdependently for the outcome.5

About half of patients do not take the medications accordingly to the doctor’s prescription,6 with a significant impact on treatment’s efficacy. Also, non-adherence presents an important economic impact, with annual costs varying from 3347$ to 19,472$ per patient.7

In the context of CVD, non-adherence to medications is associated with significant increase in CVD events8 and higher economic costs.9 It may justify about 9% of all CVD events.10

It is crucial to take action to improve adherence to medications, namely in patients with dyslipidemia. The risk of non-adherence varies with several factors including patients’ characteristics, the disease itself and constraints of health system organizations.5,11 The capacity to predict this risk is an important help to allow a better adjustment of the personalized approach in the decision-making of the clinical consultation. Patients with a predictably higher risk of non-adherence benefit from better attention and greater intervention to improve it, thus better controlling dyslipidemia and consequently the cardiovascular risk. Therefore, it is essential to perceive and better characterize this group of patients with higher risk of non-adherence.

The aim of our study is to identify the determinants of non-adherence to medications in patients with dyslipidemia.

Patients and Methods

In January 2021, we performed a literature search in the PubMed and Scopus databases for all the records published from January 2000 to December 2020 (including), in English, Portuguese and Spanish, using the MeSH terms “Dyslipidemias” and “Medication Adherence”. To organize this systematic review, we used the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA).12 All the articles and references were managed using the citation manager software EndNote (version 20).

Study Selection and Data Extraction

The process for selecting studies involved two phases: screening by title and abstract; and assessing the eligibility in full text. Data collected from each study and selected for the final qualitative synthesis, were the study identification (author and citation), study design and setting, patient population size for each study, the type of measure of adherence and the determinants associated with non-adherence. Article screening and data collection were conducted by two independent authors. All differences were discussed and decided by mutual agreement.

Inclusion Criteria

We included studies on patients with dyslipidemia that evaluated the adherence to any dyslipidemia medication, including statins, or the predictors of adherence to medications. We did not include studies evaluating adherence to dyslipidemia guidelines, interventions to improve adherence or validation of such interventions, studies for evaluation or validation of methods to measure medication adherence, studies on adherence trends, and studies of cost or outcome analysis.

Exclusion Criteria

We excluded reviews and case reports, studies involving pediatric populations, studies that assessed concomitantly the adherence to medications to other cardiovascular risk factors beyond dyslipidemia (including antidiabetics and antihypertensives), and studies that limited the population to specific groups, like sex or diseases beyond CVD or cardiovascular risk factors beyond dyslipidemia.

Assessment of Study Quality

To assess the quality of the studies, we used an adapted version of the 2018 Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies from the National Heart Lung and Blood Institute.13 We choose 4 questions out of 14 that compose the tool, based on the adequacy for the proposal of this study: “1. Was the research question or objective in this paper clearly stated?”, “2. Was the study population clearly specified and defined?”, “11. Were the outcome measures (dependent variables) clearly defined, valid, reliable, and implemented consistently across all study participants?” and “14. Were key potential confounding variables measured and adjusted statistically for their impact on the relationship between exposure(s) and outcome(s)?”. We classified each question in “Yes” (1) or “No” (0) for each study. The resulting score ranged from 0 to 4. Only studies with 2 or more points were considered with the sufficient quality and therefore accepted for inclusion in the final analysis.

Ethical Concerns

Previous ethical appraisal by the ethics committee was not mandatory due to the review nature of this research. Nevertheless, we assumed the commitment to the ethical principles of the Helsinki Declaration, in the conception, analysis and discussion of our results.

Results

We identified 1954 records from PubMed search database and 1548 from the Scopus search database, in a total sample of 2644 articles, after eliminating the duplicates. The screening by title and abstract excluded a total of 2507 records, and the remaining 137 records were selected for a full-text assessment. From this final assessment, 90 articles were excluded. In this phase, 47 studies remained for the inclusion on the qualitative synthesis. The process is summarized in Figure 1.

Figure 1

PRISMA flow diagram for the review.

Quality Assessment

The results of the quality assessment of each article are described in Table 1. From the 47 studies analyzed: 29 had a score of 4, 16 had a score of 3, 1 had a score of 2 and 1 had a score of 1. Consequently, only one study was excluded, and the remaining 46 were included in the final qualitative synthesis.

Table 1

Quality Assessment of the Articles

Article Identification – Author (Year)	Question 1	Question 2	Question 11	Question 14
Kiortsis et al (2000)14	Yes	Yes	Yes	No
Mansur et al (2001)15	Yes	Yes	Yes	No
O’Donnell et al (2001)16	Yes	Yes	Yes	No
Yang et al (2003)17	Yes	No	Yes	Yes
Stilley et al (2004)18	Yes	Yes	Yes	No
Benner et al (2005)19	Yes	Yes	Yes	Yes
Caspard et al (2005)20	Yes	Yes	Yes	Yes
Di Martino et al (2005)21	Yes	Yes	Yes	Yes
Huser et al (2005)22	Yes	No	No	No
Batal et al (2007)23	Yes	Yes	Yes	Yes
McGinnis et al (2007)24	Yes	Yes	Yes	No
Natarajan et al (2007)25	Yes	Yes	Yes	No
Ye et al (2007)26	Yes	Yes	Yes	Yes
Chodick et al (2008)27	Yes	Yes	Yes	Yes
Vinker et al (2008)28	Yes	Yes	Yes	Yes
Latry et al (2011)29	Yes	Yes	Yes	No
Wong et al (2011)30	Yes	Yes	Yes	Yes
Wong et al (2011)31	Yes	Yes	Yes	Yes
Brogaard et al (2012)32	Yes	Yes	Yes	No
Cheetham et al (2013)33	Yes	Yes	Yes	Yes
Christian et al (2013)34	Yes	Yes	Yes	Yes
Xie et al (2013)35	Yes	Yes	Yes	Yes
Cicero et al (2014)36	Yes	Yes	Yes	Yes
Halava et al (2014)37	Yes	Yes	Yes	Yes
Gaisenok et al (2015)38	Yes	Yes	No	No
Halava et al (2015)39	Yes	Yes	Yes	Yes
Korhonen et al (2015)40	Yes	Yes	Yes	Yes
Warren et al (2015)41	Yes	Yes	Yes	Yes
Korhonen et al (2016)42	Yes	Yes	Yes	Yes
Kronish et al (2016)43	Yes	Yes	Yes	Yes
Tokgözoğlu et al (2016)44	Yes	Yes	Yes	No
Al-Foraih et al (2017)45	Yes	Yes	Yes	Yes
Booth et al (2017)46	Yes	Yes	Yes	No
Devaraj et al (2017)47	Yes	Yes	Yes	Yes
Hickson et al (2017)48	Yes	Yes	Yes	Yes
Wawruch et al (2017)49	Yes	Yes	Yes	No
Chung et al (2018)50	Yes	Yes	Yes	Yes
do Nascimento et al (2018)51	Yes	Yes	Yes	Yes
Haddad et al (2018)52	Yes	Yes	Yes	Yes
Kriegbaum et al (2018)53	Yes	Yes	Yes	Yes
Ofori-Asenso et al (2018)54	Yes	Yes	Yes	No
Chen et al (2019)55	Yes	Yes	Yes	Yes
Phan et al (2019)56	Yes	Yes	Yes	Yes
Waßmuth et al (2019)57	Yes	Yes	Yes	No
Bruckert et al (2020)58	Yes	Yes	Yes	No
Seaman et al (2020)59	Yes	Yes	Yes	Yes
Shakarneh et al (2020)60	Yes	Yes	Yes	No

Notes: Question 1. “Was the research question or objective in this paper clearly stated?”, Question 2. “Was the study population clearly specified and defined?”, Question 11. “Were the outcome measures (dependent variables) clearly defined, valid, reliable, and implemented consistently across all study participants?”, Question 14. “Were key potential confounding variables measured and adjusted statistically for their impact on the relationship between exposure(s) and outcome(s)?”.

Two authors read independently all 46 studies, searching for outcomes of non-adherence. The studies evaluated non-adherence, adherence, discontinuation, or persistence with medication. All articles were interpreted in a case-by-case approach and considered in terms of non-adherence. The outcomes of non-adherence searched preferentially were statistical measures: odds ratio, risk ratio and hazard ratio. Many studies only presented a qualitative assessment of the results that were also considered.

All the variables assessed for each of the 46 articles selected for the final qualitative synthesis are summarized in Table 2. From these 46 observational studies analyzed, 25 were retrospective, 12 cross-sectional studies and 9 prospective cohort studies. The patient’s population in the studies ranged from 117 to 229.918 patients. In terms of measures of adherence, 15 studies used the “Proportion of Days Covered (PDC)”, 7 used the “Medication Possession Ratio (MPR)”, 3 used the “8-item self-report Morisky Medication Adherence Scale (MMAS-8)” and 2 used the “4-item self-report Morisky Medication Adherence Scale (MMAS-4)”. The remaining 18 studies had a more specific measure of adherence, designed for each, individually.

Table 2

Articles Assessed in Qualitative Synthesis

Article	Study Design	Setting (Country)	Patient Population	Adherence Measure	Determinants of Non-Adherence
Kiortsis et al (2000)14	Cross-sectional study, by survey	Clinic (Hôpital Pitié-Salpétrière), Department of Endocrinology and Metabolism, Unit for Prevention of Cardiovascular Diseases (Paris)	193 patients	Compliance was analyzed by the percentage of pills missed during the previous month. Patients were divided into three groups: (i) high compliance (all prescribed pills were taken), (ii) intermediate group (<6% of the prescribed pills were missed) and (iii) low compliance group (patients who missed ≥6% of the prescribed pills)	Increased non-adherence: younger age, smoking habit, perceived high frequency of side effects, higher number of prescription medications, bad doctor–patient relationship, perception that treatment is not effective
Mansur et al (2001)15	Prospective observational study	Coronariopathy Unit at the Heart Institute, Medical School, University of São Paulo (InCor) (Brazil)	207 patients	Nonadherence was classified as 1) lack of understanding by patients of the physician’s instructions for taking the medication, 2) due to side effects, 3) non-utilization of the medication for other reasons	Increased non-adherence: high cost of the medication, lack of proper instruction on how to use medication
O’Donnell et al (2001)16	Cross-sectional study, by survey	Multidisciplinary lipid clinic of an integrated health care system	117 patients	Adherence was determined by the number of days (in 30 days) that patients took their lipid-lowering medications exactly as prescribed. 30 days was classified as adherent and ≤29 was classified as nonadherent	Increased non-adherence: perceived burden of the diseaseDecreased non-adherence: older age, perceived benefits of medication
Yang et al (2003)17	Retrospective cohort study	UK residents enrolled with selected general practitioners (GPs) (UK)	22.408 patients	Discontinuation of the original LLD was identified when a patient did not receive any further LLD prescription within 3 months after the expected last date of any previous LLD prescription	Increased non-adherence: female sex, smoking habit, less follow-up visits, higher number of non-cardiovascular medications (≥4)Decreased non-adherence: older age, diabetes, taking more classes of cardiovascular medications (≥3), statins or fibrates as initial lipid-lowering drug
Stilley et al (2004)18	Prospective study	University of Pittsburgh Medical Center (USA)	158 patients	Medication adherence data was tracked with electronic cap monitors, the Medication Event Monitoring System (MEMS). Percentage of days compliant was used to measure adherence in the primary analyses. Adherence to dose and schedule over the entire period was also assessed. Patients were considered adherent if they had a rate of compliance ≥ 80%	Increased non-adherence: anxiety, depressionDecreased non-adherence: higher conscientiousness, higher estimated IQ, higher visuospatial/constructional ability
Benner et al (2005)19	Retrospective cohort study	Enrollees in a Southeastern managed care plan (USA)	9.510 patients	Proportion of days covered (PDC). Subjects were considered adherent if they had a PDC of ≥80%	Increased non-adherence: more months of therapy duration, higher number of prescription medications (9–43)Decreased non-adherence: younger age, cardiovascular intervention, less change in LDL cholesterol concentration at 3 months of therapy, adherence ≥80% during the first 3 months of therapy
Caspard et al (2005)20	Retrospective observational study	Massachussets health maintenance organization (USA)	4.776 patients	Adherence was measured by Proportion of days covered (PDC). Patients were considered adherent if they had a PDC of ≥80%	Increased non-adherence: younger age (<50 years), female sex, baseline LDL cholesterol ≥190 mg/dLDecreased non-adherence: previous antihyperlipidemia treatment
Di Martino et al (2005)21	Retrospective observational study	Patients attending general practitioners in the Ravenna area (Italy)	4.764 patients	Poor adherence was defined as a Standardized Mean Daily Dose (SDD) <0.8 tablets/day (less than 80% of the Defined Daily Doses (DDD))	Increased non-adherence: absence of previous atherosclerotic disease, absence of concomitant diseases, smoking habit
Batal et al (2007)23	Retrospective cohort study	Denver Health Medical Center (USA)	3.386 patients	Adherence was assessed by an adherence score, calculated as the number of days of drug acquired divided by number of days in the study. High adherence was defined by an adherence score of ≥80% and non-adherent as an adherence score of <80%	Increased non-adherence: female sex, African American ethnicity, Hispanic ethnicityDecreased non-adherence: age (per 10-year increase), 60 versus 30 days supply medication, each additional comorbidity
McGinnis et al (2007)24	Phase 1: retrospective medical record reviewPhase 2: cross-sectional telephone survey of patients for whom a medical record review was completed	Kaiser Permanente Colorado, a health maintenance organization that provides integrated health care services to members at 18 medical offices in the Denver-Boulder metropolitan area (USA)	Phase 1: 435 patientsPhase 2: 242 patients	Proportion of days covered (PDC). Cohorts were divided into 3 PDC groups: ≥80%, 20–79%, and <20%	Increased non-adherence: less provider follow-up visits, less laboratory visits, bad doctor–patient relationship, reported adverse effects, being unsure of the benefits of statins
Natarajan et al (2007)25	Cross-sectional study, by survey	Two academic family practice clinics (Halifax, NS)	284 patients	Level of adherence to statin medications was measured by patients’ self-report using the 4-item Morisky scale, a 4-item adherence measure	Decreased non-adherence: older age (>75 years), higher number of other prescribed medications (4–6), healthier lifestyle (exercise and/or healthy diet)
Ye et al (2007)26	Longitudinal, Retrospective, Cohort Study	Privately insured individuals (USA)	5.548 patients	Adherence was measured by medication possession ratio (MPR). Patients were considered adherent if they had an MPR of ≥80%.	Increased non-adherence: copayment ≥$20, female sex, depression, non-statin lipid lowering drug useDecreased non-adherence: higher age, dyslipidemia
Chodick et al (2008)27	Retrospective cohort study	Maccabi Healthcare Services (MHS) and from death certificates (Israel)	229.918 patients	Proportion of days covered (PDC). Patients were categorized into 3 groups: adherent (PDC≥80%; partially adherent (PDC≥20% and <80%), or nonadherent (PDC <20%)	Increased non-adherence: female sex, less time of residence in Israel, chronic obstructive pulmonary disease, Alzheimer’s disease, lower LDL cholesterol (<130 mg/dL), moderate statin therapeutic efficacy, lowest socioeconomic levelDecreased non-adherence: age (per 10-year increase), more health services utilization (higher number of outpatient visits and hospitalizations), diabetes mellitus, hypertension, taking ACEI, taking β-Blockers, high statin therapeutic efficacy
Vinker et al (2008)28	Retrospective cohort study	Central District of Clalit HealthServices HMO (Israel)	47.680 patients	Adherence was defined as buying at least 80% of the expected number of pills over this a 12 month period.	Increased non-adherence: patients originary from Ethiopia, new immigrants in 1990Decreased non-adherence: older age, female sex, chronic diseases
Latry et al (2011)29	Retrospective cohort study	French health insurance system: the Caisse Nationale d’Assurance Maladie des Travailleurs Salariés (Cnam-TS), from Aquitaine region from southwest France (France)	16.397 patients	Adherence was assessed by four criteria: (i) proportion of days covered (PDC) by statins, (ii) regularity of the treatment over time, (iii) persistence of treatment, and (iv) refill delay. The PDC was estimated using the “Continuous Multiple-interval measures of medication Availability” (CMA) definition. A CMA <80% was considered as unsatisfactory	Decreased non-adherence: associated cardiovascular risk factors
Wong et al (2011)30	Retrospective cohort study	Public, primary care clinics (Hong Kong, China)	11.042 patients	Medication Possession Ratio (MPR). Patients were considered adherent if MPR≥0.8	Decreased non-adherence: older age, provider is a Family Medicine Specialist Clinic, more provider follow-up visits, comorbidities
Wong et al (2011)31	Retrospective cohort study	Primary care clinic in 1 territory of Hong Kong (China)	12.875 patients	Discontinuation was considered as the absence of a refill prescription in all subsequent clinic visits without issuance of another lipid-lowering agent 180 days from the first prescription date	Decreased non-adherence: older age, male sex, fee payers, provider is a Family Medicine Specialist Clinic, residents in rural districts, more provider follow-up visits, comorbidities, taking fibrates
Brogaard et al (2012)32	Retrospective cohort study	Department of Cardiology, Odense University Hospital (Denmark)	Two cohorts:2003: 474 patients2008: 550 patients	Medication Possession Ratio (MPR). Patients were considered adherent if they had an MPR ≥ 80%	Increased non-adherence: younger ageDecreased non-adherence: older age (≥ 80 years)
Cheetham et al (2013)33	Retrospective cohort study	Kaiser Permanente Southern California (KPSC), a managed care organization (MCO), in Southern California (USA)	19.826 patients	Patients were followed for a period of 90 days after the index date. Patients who did not pick up their new statin prescriptions within this 90-day window were considered primary nonadherent.	Increased non-adherence: African American ethnicity, higher provider age, provider is a high prescriber (>7 statin prescriptions during study)Decreased non-adherence: older age, LDL cholesterol >160 mg/dL, higher health care utilization (ED visits, hospitalizations, clinic visits or any Rx, in a year), higher Charlson Comorbidity Index, provider is male
Christian et al (2013)34	Retrospective cohort study	US insurance-carrying individuals (USA)	42.685 patients	Adherence was quantified using medication possession ratio (MPR). Patients were categorized into 4 MPR categories (<0.40, 0.40–0.59, 0.60–0.79, and ≥0.80), where 80% or greater is commonly defined as adherent.	Increased non-adherence: female sex, peripheral vascular disease, chronic pulmonary disease, diabetes, liver disease, rheumatic diseaseDecreased non-adherence: older age, initial treatment drug being niacin + statin, higher prescription drug classes
Xie et al (2013)35	Cross-sectional study	Twelve cities, and one county, entailing 21 province-level hospitals (equivalent to a teaching hospital in Europe) and six county-level hospitals (equivalent to a district hospital in Europe) (China)	1.890 patients	Medication Possession Ratio (MPR). We defined good compliance has an MPR of ≥80%	Decreased non-adherence: unemployed, cover rate of medicine insurance ≥70%, Province/County hospital, being attended in Cardiology department, very high CVD risk, using statins
Cicero et al (2014)36	Retrospective cohort study	Lipid clinics of the University of Bologna and the University of Pavia (Italy)	628 patients	Persistence in medication after 2 years	Decreased non-adherence: female sex, family history of early cardiovascular disease, baseline LDL cholesterol, treatment with nutraceuticals versus statins
Halava et al (2014)37	Prospective cohort study	Data from the Finnish Public Sector Study, a prospective study involving public sector employees in 10 municipalities and 21 hospitals (Finland)	9.285 patients	Proportion of days covered (PDC). Non-adherence was defined as PDC<80%	Increased non-adherence: female sex, younger age (<60 years), single
Gaisenok et al (2015)38	Retrospective observational study	Preventive Pharmacotherapy Department of the Ministry of Healthcare of the Russian Federation (Russia)	274 patients	Compliance with statin therapy was assessed with information on regularity of statin intake and the responses to four questions: (1) if they knew, according to the results of previous exams, that they had elevated cholesterol levels; (2) what method of hypercholesterolemia correction they used; (3) if they were taking any statins; and (4) if yes, what statin preparation and what dose they were taking.	Increased non-adherence: older age (>70 years)Decreased non-adherence: coronary heart disease, history of myocardial infarction, history of cardiovascular intervention
Halava et al (2015)39	Prospective cohort study	Retired, insured, swedish patients taking statins (Sweden)	11.718 patients	Adherence was defined as the Proportion of Days Covered (PDC). Non-adherence was defined as (PDC<80%). Adherence was defined as PDC≥80%	Increased non-adherence: retirement
Korhonen et al (2015)40	Prospective cohort study	Finnish Public Sector (10 towns and 6 hospital districts) (Finland)	1.916 patients	Proportion of days covered (PDC). Non-adherence was defined as PDC<80%	Increased non-adherence: higher number of childhood adversities (only in men), experiencing severe conflicts in the family
Warren et al (2015)41	Linked data from a prospective study	Data from a prospective study of 267,091 Australians with age ≥45 years to national data sets on prescription reimbursements, general practice claims, hospitalizations and deaths (Australia)	36.144 patients	Medication possession ratio (MPR). Patients were considered adherent if MPR≥80%	Increased non-adherence: highest education qualification, language spoken at home is other than English, single, widowed or separated, employed, annual income ≥ $70.000, smoking habit, new statin user, very high psychological distressDecreased non-adherence: older age, private health insurance, obesity, past smoker, self-reported heart disease, higher number of comorbidities, living in more remote areas
Korhonen et al (2016)42	Prospective cohort study	Employees of ten towns and six hospital districts who were targeted by questionnaire surveys in 2/4-year intervals since 2000 (Finland)	1.924 individuals	Adherence was measured using the Proportion of Days Covered (PDC). Non-adherence was defined as PDC<80%	Increased non-adherence: depression, anxiety
Kronish et al (2016)43	Retrospective cohort study	Centers for Medicare and Medicaid Services Chronic Condition Data Warehouse (USA)	6.618 patients	Proportion of days covered (PDC). Patients were categorized as nonadherent (PDC <80%) or adherent (PDC ≥80%)	Increased non-adherence: older age (>75 years), diabetes mellitus, depression, dementiaDecreased non-adherence: being eligible for low-income subsidy
Tokgözoğlu et al (2016)44	National cross-sectional non-interventional observational study	14 centers of family medicine/general practice and cardiology clinics (Turkey)	532 patients	Discontinuation of statin therapy is defined as failure to renew a statin prescription with a ≥30-day gap between the end of a prescription and the start of the next prescription	Increased non-adherence: higher educational status, negative coverage of information about statin treatment and side effects (hepatic, renal and muscular) on television, patients’ lack of sufficient knowledge regarding high cholesterol and related risks
Al-Foraih et al (2017)45	Cross-sectional study study	Twelve polyclinic/general practice clinics across 4 of the 6 governorates in Kuwait (Kuwait)	200 patients	Adherence was assessed using the 8-item self-report Morisky Medication Adherence Scale (MMAS-8)	Increased non-adherence: diabetesDecreased non-adherence: older age
Booth et al (2017)46	Retrospective cohort study	Medicare beneficiaries with a hospitalization for myocardial infarction (USA)	158.795 patients	Proportion of Days Covered (PDC) High persistence was defined as PDC ≥80%	Increased non-adherence: not having a history of statin use, others than non-Hispanic white, cardiologist care prior to myocardial infarction hospitalization, diabetes mellitus, history of chronic kidney disease, heart failure, coronary heart disease, statin intoleranceDecreased non-adherence: male sex, low-income subsidy, higher area-level median income, region of residence, statin use prior to myocardial infarction event, coronary stent insertion, being hospitalized for ≥5 days, using moderate-intensity and high-intensity statins
Devaraj et al (2017)47	Cross-sectional study, by survey	Urban primary care clinic (Malaysia)	398 patients	8-item Morisky Medication Adherence Scale. A patient was considered adherent if had a score≥6	Increased non-adherence: male sex, duration of lipid lowering therapy (>10 years), taking medication(s) at night or at no specific time, more follow-up visits, single follow-up clinic
Hickson et al (2017)48	Retrospective cohort study	Medicare beneficiaries taking statins, with age ≥ 66 years and with an acute myocardial infarction hospitalization in 2008–2010 (USA)	113.296 patients	Proportion of days covered (PDC). Patients were categorized into 3 categories: severely nonadherent (PDC <40%), moderately nonadherent (PDC 40–79.9%), and adherent (PDC ≥80%)	Increased non-adherence: African American ethnicity, Hispanic ethnicity, higher baseline Charlson Comorbidity Index, dementia/Alzheimer’s disease, depression, revascularization/cardiac intervention, ischemic heart disease, rhabdomyolysis/myopathy, higher length of stay in hospital post cardiovascular eventDecreased non-adherence: dual insurance eligibility (Medicare and Medicaid), follow-up care by Cardiologist only
Wawruch et al (2017)49	Retrospective cohort study	The sample for this study was assembled from the database of the General Health Insurance Company (Slovak Republic).	797 patients	Treatment gap was defined as the presence of an at least 6-month period without any statin prescription. Patients with a treatment gap period were designated as ‘non-persistent’, whereas those without such a gap were designated as ‘persistent’ patients	Increased non-adherence: female sexDecreased non-adherence: older age (≥65 years), hypertension, diabetes mellitus, hypercholesterolemia, higher number of medications
Chung et al (2018)50	Prospective, non-interventional, observational study	Hospital Departments of Neurology (Korea)	991 patients	Patient self-reported adherence to statin therapy was assessed using the Morisky Medication Adherence Scale (MMAS-8)	Increased non-adherence: taking high statin dose, higher daily number of medication pillsDecreased non-adherence: patients’ awareness of hyperlipidemia, current drinker
do Nascimento et al (2018)51	Cross-sectional study, by survey	1305 primary healthcare services located in 272 Brazilian cities (Brazil)	8.803 patients	Adherence was assessed by asking patients whether they missed any dose in the 7 days prior to the interview. Patients who reported missing at least one dose in the past 7 days were classified as having poor adherence	Decreased non-adherence: older age, higher education, dyslipidemia and diabetes mellitus, heart diseases, polypharmacy (≥5), alcohol use, not using emergency care
Haddad et al (2018)52	Cross-sectional study	20 community pharmacies from all 5 districts of Lebanon (list provided by the Lebanese Order of Pharmacists) (Lebanon)	247 patients	Adherence was assessed using a 3 questions questionnaire, asking patients about the frequency, percentage, and rating response of their statin use during previous month. Possible responses were divided in 6 categories: 0%, 20%, 40%, 60%, 80% and 100%. A total score was calculated by summing all 3 answers and presented in a percentage.	Increased non-adherence: divorced, widowed, primary and university level of education, smoking 15 cigarettes per day, not alcohol drinkers, duration of dyslipidemiaDecreased non-adherence: married, single, secondary educational level, non-smokers, occasional alcohol drinking, better illness perception, better treatment satisfaction, better quality of life perception
Kriegbaum et al (2018)53	Cross-sectional study, by survey	Danish population participating in “LIFESTAT – Living with Statins Project” (Denmark)	3.050 patients	Discontinuation was based on self-reporting of current statin use in the interview, with 3 possible predefined answers. Discontinuation was defined as those who answered: “I have previously used cholesterol lowering drugs”	Increased non-adherence: experience of side effects, fear of side effects, feelings of uncertainty and confusion regarding information on statins (from both mass media and from general practitioners)
Ofori-Asenso et al (2018)54	Retrospective cohort study	A 10% random sample of the Australian population from data from the Pharmaceutical Benefits Scheme (Australia)	49.380 patients	The number of days on statin were calculated, assuming a dosage of 1 tablet daily. Discontinuation was defined as the first ≥90 days without statin coverage	Increased non-adherence: older age (>75 years), female sex, therapeutic initiated by general practitioner, diabetesDecreased non-adherence: taking atorvastatin or rosuvastatin (VS simvastatin), use of cardiovascular pharmacotherapies (antiplatelet, β-blockers, ACE inhibitors or ARBs)
Chen et al (2019)55	Population-based retrospective cohort study	Taiwan National HealthInsurance claims (Taiwan)	169.624 patients	Medication possession ratio (MPR) and Proportion of days covered (PDC). Good adherence was considered if PDC≥0.80%	Increased non-adherence: older age (>75 years), renal disease, liver disease, depression, chronic obstructive pulmonary disease, taking insulin, digitalis, OADs, ACEIs or diureticsDecreased non-adherence: female sex, previous statin user, revascularization/cardiovascular procedures, hyperlipidemia, taking β-blockers, ARBs or antiplatelet agents, healthcare provider is a Medical Center, statin prescribed by Cardiovascular Medicine or Cardiovascular Surgery
Phan et al (2019)56	Retrospective population-based cohort study	Integrated healthcare system in Southern California (USA)	5.629 patients	Proportion of Days Covered (PDC). Adherence levels were categorized as: high (≥80%), partial (≥40% and <80%), and low (<40%)	Increased non-adherence: heart failureDecreased non-adherence: male sex, Caucasian, hyperlipidemia, higher adherence to β-blocker
Waßmuth et al (2019)57	Monocentric, prospective cohort study	Department of Cardiology and Department of Cardiac Surgery of the University Clinic Halle (Saale) (Germany)	542 patients	Adherence was defined has unchanged if statinsand/or ezetimibe were unchanged, newly prescribed, or increased. Non-adherence was considered if statins and/or ezetimibe were stopped or dosage was decreased	Increased non-adherence: peripheral artery disease, atrial fibrillation, coronary artery bypass surgeryDecreased non-adherence: taking aspirin, P2Y12 inhibitors or β-blockers (at discharge from cardiovascular event)
Bruckert et al (2020)58	Retrospective observational cohort study	The Pharmacoepidemiologic General Research eXtension (PGRx)-acute coronary syndrome (ACS) registry, from cardiology centres (France)	2.695 patients	Adherence was measured using the Proportion of Days Covered (PDC). Patients were considered adherent if they had a PDC ≥80%	Decreased non-adherence: previous cardiovascular disease event
Seaman et al (2020)59	Retrospective observational study	Western Australian Population (Australia)	205.924 patients	Medication possession ratio (MPR), with the threshold of adherence defined as 80%	Increased non-adherence: younger age, living in more remote areas, not having previous ischemic heart disease, not having previous coronary artery revascularization procedure, not having previous statin use, beneficiary status, not taking cardiovascular medication, higher number of other medicationsDecreased non-adherence: older age
Shakarneh et al (2020)60	Cross-sectional study, by survey	Two middle governmental primary healthcare clinics in the cities of Ramallah and Bethlehem (Palestine)	185 patients	Adherence was determined using the 4-item Morisky medication adherence scale (MMAS-4). Total scores were summed (range 0–4), with scores of 0–1 denoting high adherence and 2–4 denoting low adherence	Increased non-adherence: low education (illiterate), polypharmacy (> 4), comorbidities, concerns about side effects

Determinants for Non-Adherence to Medications

We found several relevant determinants associated with non-adherence, summarized in Table 3.

Table 3

Main Determinants Associated with Non-Adherence

Main Identified Determinants Associated with Non-Adherence
	Associated with Higher Non-Adherence	Associated with Lower Non-Adherence
Age	Younger age (≤ 50 years)	Older age
Sex	Female sex	Male sex
Ethnicity	African AmericanHispanic	Caucasian
Marital Status	DivorcedSeparatedWidowedSingle	Married
Habits	Smoking habits	Alcohol drinking habits
Comorbidities	Chronic obstructive pulmonary diseaseAlzheimer disease/dementiaDepressionDiabetesAnxietyHepatic diseaseChronic kidney diseasePeripheral vascular diseaseHeart failure	Higher number of comorbiditiesHigher number of cardiovascular risk factorsPresent heart diseaseHypertension
Medications	New users of lipid-lowering medicationsPatients presenting side effects or concerns about themHigher duration of lipid-lowering treatment	Patients taking β-blockersPatients taking antiplatelets
Cardiovascular diseases and interventions		History of cardiovascular disease eventsHistory of cardiac intervention or revascularization procedure
Healthcare services utilization		Higher healthcare services utilizationPatients treated in cardiologyPatients treated in family medicineHaving health insuranceHigher number of follow-up visits
Other determinants	UnemploymentBad doctor–patient relationship

Sociodemographic Variables

Age

Non-adherence decreases with age.16,26,33,45 For each 10-year increase in age, the risk of non-adherence decreases about 5–11%.23,27 Analyzing particular age groups, we observed the same trend of decreasing non-adherence with higher age. Non-adherence is higher in younger patients (age ≤ 50 years),14,32,37,59 reaching values of 50%.20 It is lower between 50 and 70 years (8–34%),17,19,20,28,30,31,41 and in older patients (age ≥ 70 years),25,30,31,59 who present the risk of non-adherence of 11–65%.17,28,41,49

Sex

In general, females are more non-adherent than males,26,31,34,37,46,56 with an increased risk of 5–31%.17,20,23,27,49,54

Ethnicity

Several studies show an increase of non-adherence in African American patients,33,48 up to 30%.23 Hispanics23,48 are also associated with worst adherence to medications.

Marital Status

Non-adherence is lower in married patients,52 higher in single patients37,41 and increased in divorced, separated or widowed patients.41,52

Employment Status

Non-adherence is worst in unemployed.35,41

Lifestyle Variables

Smoking Status

Non-adherence is higher in current smokers,14,41,52 with an increase of 18–163%.17,21

Alcohol Drinking Status

Non-adherence is lower in active alcohol drinkers,52 showing values of odds ratio between 0.61 and 0.71.50,51

Medication Use

Several conditions regarding medications are associated with non-adherence. In general, non-adherence is higher in new users of lipid-lowering medications,20,46,55,59 justifying up to 47% more risk.41 Also, in patients presenting side effects or expressing concerns about them, non-adherence is higher,14,24,53 with an associated odds ratio of 2.89.60 Non-adherence is also worst in patients having a higher duration of lipid-lowering treatment.19,47 On the other hand, several studies have shown that non-adherence is lower in patients taking β-blockers,55–57 who have approximately less 12% risk of non-adherence27,54 as in patients taking other cardiovascular medications.17,24,59

Comorbidities/Disease Status

Number of Patients’ Comorbidities

Non-adherence decreases with a higher number of comorbidities. For each comorbidity present, the risk of non-adherence decreases by 4%.23 This is apparent since the first comorbidity,30,31,41 leading to 5% improvement.41 This is even more apparent in patients with two or more comorbidities,30,31,41 with a 7–8% improvement in adherence.41 Inversely, the absence of concomitant diseases is associated with a 59% increase in the risk of non-adherence.21

Type of Comorbidity

Several comorbidities are identified as especially associated with higher non-adherence: chronic obstructive pulmonary disease,34,55 with a 7–10% increased risk;27 Alzheimer’s disease or other dementias,48 with a 20% increased risk;27,43 depression,18,26,48,55 with a 9–15% increased risk;42,43 diabetes,34,45,46 with a 7–10% increased risk of non-adherence;43,54 anxiety;18,42 hepatic disease;34,55 chronic kidney disease;46,55 peripheral vascular disease;34,57 and heart failure.46,56

On the other hand, non-adherence is lower in patients with present heart diseases,41,51 in patients with a higher number of cardiovascular risk factors,29,35 in patients with hypertension27,49 and in patients taking antiplatelets.54,55

Cardiovascular Diseases and Interventions

The history of cardiovascular disease events is associated with lower non-adherence.38,58,59 Patients with a previous cardiac intervention or a revascularization procedure, present lower non-adherence,46,59 with odds ratio of 0.19–0.68.19,38,55 Patients without a previous atherosclerotic event present 135% more risk of non-adherence.21

Healthcare Related Variables

Healthcare Services Utilization

Non-adherence is decreased in patients with higher healthcare services utilization.27,33 It is also influenced by providers, with cardiology35,48 and family medicine30,31 contributing for lower levels of non-adherence.

Having Health Insurance

Non-adherence is lower in patients who had health insurance.35,48 These patients have 3–5% less risk of non-adherence.41

Follow-Up Visits

The follow-up visits are associated with lower non-adherence,24,30,31 with an impact on the risk of 60%.17

Physician–Patient Relationship

Having a bad doctor–patient relationship14,24 is associated with increased non-adherence.

Discussion

Adherence is a cornerstone of the effectiveness of the medications. The best therapeutic will never work if the patient does not take it. In this review, we identify a set of determinants related to non-adherence to medications in patients with dyslipidemia. Although the different weight, each characteristic is associated with a significant variation in non-adherence, allowing to adjust and to personalize the therapeutic decision in these patients. The more of these conditions the higher should be our concern about the risk of non-adherence in the patient.

Our review included studies related to dyslipidemia, in a broader approach, to find a wide range of factors related to adherence. Our results are in line with other studies. Ofori-Asenso et al published a systematic review with meta-analysis, focused on the non-adherence to statins in patients aged ≥65 years.61 It included 45 articles and identified the female sex, African American ethnicity, new lipid-lowering drug users, taking lower number of cardiovascular medications and having depression as factors associated with higher non-adherence. Contrary to our findings, this study identified kidney disease as a factor for decreased non-adherence. Mann et al also reviewed 22 cohort studies with meta-analysis, focusing on non-adherence to statins. He described a U-shaped association with age, higher in the younger and in the elderly and lower in middle ages, as in patients with diabetes.62 Unlike other reviews, we focused on patients with dyslipidemia, irrespective of the medication, and tried obtaining practical and objective information that could allow physicians to rapidly consult that information in a context of a consultation, namely on the visit where the first prescription occurs.

A difficulty in our review was the differences observed between the studies: the different study designs; different sample sizes of study population; different tools to measure adherence (“Proportion of Days Covered (PDC)”, “Medication Possession Ratio (MPR)”, the “8-item self-report Morisky Medication Adherence Scale (MMAS-8)”, the “4-item self-report Morisky Medication Adherence Scale (MMAS-4)”); different outcomes (many evaluated non-adherence, but others evaluated adherence, discontinuation, persistence); and the way to quantify them (odds ratio, risk ratio, hazard ratio, or a simple qualitative assessment). This heterogeneity does not seem to be enough to condition our results, as more than the quantification of non-adherence, it was more important to identify their causes. Some factors that our review does not account for are the impact of media and more precisely social media platforms. In fact, in today’s society, these platforms of social interaction are of the utmost relevance as, in our view, they can contribute to changes in patients’ perceptions and preconceptions regarding medications. We think that this reality possibly negatively influences patients’ medical literacy and could, in most cases, create misconceptions and negatively affect patients’ motivation to adhere to medications, undermining the importance of adherence.

This review allows to draw the profile of the non-adherent patient, to whom we should pay special attention and a hard intervention to lead them to take the lipid-lowering medications appropriately, in their own benefit. Several strategies help to enhance the adherence to medications.63 First of all, it is really important that healthcare providers can detect non-adherence. This should be a systematic practice in every visit, in every patient, to quantify the problem and to register it in the clinical file. There are many ways to do it, both directly (counting pills or renewal of the prescriptions) or indirectly (by questionnaire or by comparison with the expected effect). Then, we must think about the best simplification of the regimen (using longer acting drugs or pills combing several drugs, adapting drug regimen to patients’ daily activities and preferences, and thoroughly introducing the therapeutic regimen, gradually increasing complexity). Literacy plays a major role.64 Patients need to be aware of their disease and the associated risks. We must be sure that they, and most of times also their families, understand reasonably the instructions and have enough time to express their expectations, their fears and doubts:65 using a simple and clear language; writing the instructions and letting the patient to repeat them; giving the patient several options, if possible. Communication skills are crucial, as it is a good doctor–patient relationship and the continuity of care, creating a trustful environment and empathy. The strengthened therapeutic alliance facilitates the sharing of responsibility toward health, where physicians and patients play their own specific role. Time is also a major factor to consider, as it is increasingly scarce in nowadays medical practice. Indeed, all the above items require time: time investment in the visit, time to listen to the patients and time to establish a real and meaningful doctor–patient relationship.

This review provides a synthesis of the determinants for non-adherence in patients with dyslipidemia in a practical perspective. Our results allow the physicians to identify the factors in each patient and adjust their intervention to better deal with the risk of non-adherence in a preventive approach. With this information, the healthcare providers may better estimate the probability of non-adherence of the patient since the initial evaluation. Patients with higher probability of non-adherence would benefit from earlier and more robust approach to minimize the risk of not taking the medications. Logically, being a qualitative assessment, this probability is subjective and is highly dependent on the physician expertise and experience to evaluate the patient in each visit. Perhaps more work to further assess the determinants of non-adherence and the posterior development of a score based on that information could allow a more objective approach to these situations, further facilitating the systematization of this practice of evaluation of non-adherence profiles.

In the future, more research is necessary to further understand the weight of the different aspects and the way they interact with each other, including the patients’ preferences, expectations and perceptions. More than a disease, dyslipidemia is a cardiovascular risk factor. Its treatment is mainly a preventive approach, and the efficacy is the non-occurrence of the event. This is very different from myocardial infarction, stroke or heart failure, where every measure presents a quick effect, measurable by the patient’s own perception. In fact, patient’s perception of the importance of adherence in dyslipidemia, along with the associated potential benefit, is a fundamental part of the problem that we ought to collectively improve.

Conclusion

There are important identifiable determinants of non-adherence in patients with dyslipidemia. Healthcare providers should identify such characteristics to approach patients in a personalized way, investing more time in improving adherence. These patients benefit from a specific approach to minimize the problem of non-adherence and improve the therapeutic success, improve health and save lives.