OBJECTIVES: Several strategies have been described assessing treatment persistence and the extent of adherence based on claims data. This study investigated the effect of different parameters on the calculation of persistence and adherence based on an example using oral novel hormonal therapies (NHT) in prostate cancer (PC).

METHODS: Using anonymized claims data from a statutory German sickness fund (AOK PLUS), male adult patients with ≥2 confirmed outpatient or ≥1 inpatient diagnosis of PC followed by an incident NHT prescription were identified in the period 01/01/2012-30/06/2021. The base case of the persistence/adherence assessment in the identified NHT starters considered the following definitions: supply approximated by the defined daily dose (DDD; WHO/WIdO), stockpiling allowed, drug coverage during hospitalizations assumed, supply gaps of >45 days indicate treatment discontinuation, extent of adherence within persistent periods calculated using the proportion of days covered (PDC). We altered one parameter at a time for a total of 16 scenarios. Non-persistence rates, time to discontinuation (TTD), PDC, and non‑adherence rates (PDC<80%) during persistent periods were analyzed for the first used NHT.

RESULTS: 3,438 PC patients (mean age: 76.3 years; mean Charlson Comorbidity: 9.2) who initiated an NHT (abiraterone: 2,045/apalutamide: 124/enzalutamide: 1,265/darolutamide:4) were observed for an average of 24.5 months (median: 20.4). Non-persistence/non-adherence rates varied widely depending on the altered-parameter scenario (30-66%/0-86%, respectively). Specifically, an increase in the coverage of one prescription (i.e., DDD multiplied by a factor of 2) and an extension of the permissible gap drastically reduced non-persistence rates and TTD, while the same scenarios exerted a differential effect on non‑adherence rates.

CONCLUSIONS: Treatment adherence and persistence analysis using claims data requires careful selection of analysis parameters depending on the research question. This study portrays the effect of specific parameters on outcomes and highlights the need for nuanced consideration of applied methodologies when interpreting results of adherence/persistence analyses.