Beyond the prescription: a novel five-item risk score to predict medication non-adherence in older cardiac outpatients

Background Medication non-adherence is a major challenge in the management of cardiovascular disease in older adults, contributing to preventable hospitalisations, adverse events and mortality. In resource-limited public health systems, simple tools to identify outpatients at high risk of non-adherence are lacking. Purpose To identify determinants of medication non-adherence in older cardiac outpatients (≥60 years) within a public health system and to develop a simple tool to predict non-adherence during routine visits. Methods A multicentre cross-sectional study enrolled 559 cardiac outpatients from public clinics; after application of eligibility criteria, 502 self-administering patients were analysed. Medication non-adherence (primary outcome) was assessed using the validated Medication Adherence Measure (MAT); non-adherence was defined as a MAT score <5.429 (sample median). Cognitive status (Montreal Cognitive Assessment), functional status (Katz Index of Independence in Activities of Daily Living, chair stand test, handgrip strength), mental health (locally validated UCLA Loneliness Scale), and medication beliefs (adapted Brief Medication Questionnaire) were evaluated using validated instruments. Univariate and multivariable logistic regression analyses identified independent predictors, which were integrated into a risk score and calculator to estimate non-adherence risk. Results Among 502 older cardiac patients (mean age 71 ± 7 years; 56.6% women; 75.9% with ≥12 years of formal education), 41.2% were non-adherent. The most prevalent cardiovascular conditions were dyslipidaemia (81.9%), hypertension (76.3%), coronary artery disease (36.5%) and heart failure (27.9%); 79.9% had ≥2 concomitant cardiovascular diagnoses. Non-adherent patients had more complex dosing regimens, more negative medication beliefs, and more frequent barriers to remembering and obtaining medications. Independent predictors of non-adherence were: two daily dosing periods versus once daily (odds ratio [OR] 2.68, 95% confidence interval [CI] 1.02–7.03, p=0.045), three daily dosing periods (OR 3.19, 95% CI 1.23–8.29, p=0.017), negative medication beliefs (OR 2.55, 95% CI 1.56–4.17, p<0.001), difficulty remembering medications (OR 14.95, 95% CI 4.84–46.12, p<0.001), and difficulty obtaining medications (OR 3.49, 95% CI 1.46–8.35, p=0.005). The model demonstrated good discrimination (area under the curve 0.778, 95% CI 0.738–0.819, p<0.001). A five-item score was derived to generate individualised non-adherence risk estimates during ambulatory visits. Conclusion A simple five-item score enables identification of older cardiac outpatients at high risk of medication non-adherence in a public health setting. Its implementation and prospective validation may facilitate targeted adherence interventions, optimise therapeutic resource allocation, reduce preventable hospitalisations, and improve cardiovascular outcomes in resource-constrained health systems.