Antidepressant adherence patterns in older patients: use of a clustering method on a prescription database.

According to the World Health Organization, depression will become the second most important cause of disability worldwide by 2020. Our objective was to identify patterns of adherence to antidepressant treatments in older patients using several indicators of adherence and to characterize these patterns in terms of medication exposure. We conducted a retrospective cohort study using the French National Health Insurance reimbursement database. Incident antidepressant users aged more than 65 were included from July 1, 2010, to June 30, 2011, and followed up for 18 months. Antidepressant and other psychotropic drugs (opioids, benzodiazepines, antipsychotics, anti-epileptics) were recorded. Adherence to antidepressant treatment was assessed by several measures including proportion of days covered, discontinuation periods, persistence of treatment, and doses dispensed. Patients were classified according to their adherence patterns using a mixed clustering method. We identified five groups according to antidepressant adherence. One group (n = 7505, 26.9%) was fully adherent with regard to guidelines on antidepressant use. Two patterns of nonadherent users were identified: irregular but persistent users (n = 5131, 18.4%) and regular but nonpersistent users (n = 9037, 32.4%). Serotonin reuptake inhibitors were the most frequently dispensed antidepressant class (70.6%), followed by other antidepressants (43.3%, mainly serotonin-norepinephrine reuptake inhibitors and tianeptine) and tricyclic antidepressants (TCAs) (13.4%). Nonadherent users more frequently had a dispensing of TCA, opioid, and anti-epileptic medication than adherent users. Health policies to improve adherence to antidepressant treatment may require better training of physicians and pharmacists, insisting on the important role of the continuation period of antidepressant treatment.