Frailty, comorbidity, and multimorbidity and their relation with medications adherence in primary care older adults.

Purpose: To assess and compare, through a retrospective cohort study, the relationships between frailty, comorbidity, multimorbidity, and levels of adherence to lipid-lowering drugs (LLDs), antihypertensives and antidepressants.

Methods: In a primary care database, we selected a cohort of patients aged 60 or older on December 31, 2022. The date of the first prescription of the aforementioned medications was the study index date. Patients with Variable Medication Possession Ratio (VMPR) > = 80% were classified as properly adherent. Frailty (i.e. Primary Care-Frailty Index), comorbidity (i.e. Charlson Index) and multimorbidity (i.e. disease counts) alternatively entered multivariate logistic regressions along with age and sex. Models' performances in prediction of medications adherence were compared in terms of information (AIC; BIC) and discrimination values (AUC).

Results: Incident users of LLDs, antihypertensives or antidepressants were 4310 (mean age: 67.9 (SD: 6.9); 56.0% females), 5969 (mean age: 69.1 (SD: 7.6); 58.0% females), and 3834 (mean age: 68.7 (SD: 6.9); 66.5% females), respectively. Among users of LLDs (46% adherent) and antidepressants (22% adherent), those who were moderately or severely frail showed a significant 30-32% decrease in adherence. In contrast, users of antihypertensives (46% adherent) showed a 41% increase in adherence when multimorbid. As a whole, the three multivariate models were equally effective in informing on medication adherence, as per AIC and BIC. They also displayed similar discriminatory ability, with AUC scores ranging from 53 to 58%. Regarding the workload of GPs, the number of elderly patients classified as moderately/high frail was less than those with co-morbidities or multimorbidities. For instance, there were approximately 35 users of antihypertensive medications per GP for the moderately frail group, compared to 46 and 66 for the co-morbid and multi-morbid groups, respectively.

Conclusions: These findings showed similar capacity for frailty, comorbidity, and multimorbidity in capturing medications adherence. Given the existence of a validated tool in primary care that aligns well with GPs' workload, frailty seems the most suitable measure for assessing the complexity of older adults in relation to their adherence to long-term medications.