Introduction: Suboptimal adherence to antipsychotics leads to poorer outcomes and relapse. The adherence behaviour of people may be influenced by several factors including the number of antipsychotics used and their formulation. This study aimed to identify longitudinal adherence patterns to oral and long-acting injectable (LAI) antipsychotics in monotherapy or polypharmacy through group-based trajectory modelling (GBTM). Methods: This was a retrospective cohort study that linked prescription and dispensing data of adult patients with a new antipsychotic prescribed between 2015-2019 in Catalonia (Spain). GBTM was used to classify patients following a similar longitudinal pattern of adherence. The response variable was adherence, estimated through the continuous medication availability measure (CMA), in each 30-day period during 12 months of follow-up. Baseline and treatment characteristics were used to characterize the trajectories identified. Results: Among the 7,730 patients included in the study, we identified seven clinically distinct trajectory groups of adherence to antipsychotics: “non-adherent” (19%), “low adherent” (9%), “early-decline” (6%), “mid-decline” (5%), “late-decline” (5%), “high adherent” (21%), and “fully adherent” (35%). Trajectories with better adherence were more likely to receive the prescription from a psychiatrist, receive LAIs and have previous exposure to other antipsychotics. Intermittent medication use patterns and high levels of polypharmacy were characteristics of the “low” and “high adherent” groups. Conclusions: The trajectories reflect three adherence behaviours: stable over time; patients who discontinue treatment and; patients with an intermittent refill pattern. Patients on polypharmacy should have more regular adherence monitoring and LAIs should be considered, as they appear to be associated with better adherence.

Introduction: This study aimed to assess the frequency of dosing inconsistencies in prescription data and the effect of four dosing assumption strategies on adherence estimates for antipsychotic treatment. Methods: A retrospective cohort, which linked prescription and dispensing data of adult patients with ≥ 1 antipsychotic prescription between 2015-2016 and followed up until 2019, in Catalonia (Spain). Four strategies were proposed for selecting the recommended dosing in overlapping prescription periods for the same patient and antipsychotic drug: 1) the minimum dosing prescribed; 2) the dose corresponding to the latest prescription issued, 3) the highest dosing prescribed, and 4) all doses included in the overlapped period. For each strategy, one treatment episode per patient was selected and the Continuous Medication Availability measure was used to assess adherence. Descriptive statistics were used to describe results by strategy. Results: Of 277,324 prescriptions included, 76% overlapped with other prescriptions (40% with different recommended dosing instructions). The number and characteristics of patients and treatment episodes (18,292, 18,303, 18,339, and 18,536, respectively per strategy) were similar across strategies. Mean adherence was similar between strategies, ranging from 57-60%. However, the proportion of patients with adherence ≥ 90% was lower when selecting all doses (28%) compared to the other strategies (35%). Conclusions: Despite the high prevalence of overlapping prescriptions, the strategies proposed did not show a major effect on the adherence estimates for antipsychotic treatment. Taking into consideration the particularities of antipsychotic prescription practices, selecting the highest dose in the overlapped period provided a more accurate adherence estimate.