IntroductionOptimal adherence thresholds can vary across medications and disease states. The objective of the study was to determine the optimal threshold of the proportion of days covered (PDC) for tumor necrosis factor (TNF) inhibitors in patients with rheumatoid arthritis (RA).MethodsPatients with RA initiating self-administered TNF inhibitors were identified using 2012–18 Medicare fee-for-service claims. Time-varying PDC was calculated every day for the preceding 90 days during follow-up. Oral and injected glucocorticoid use, hospitalizations, emergency room (ER) visits, serious infections, and a composite of these were measured as outcomes. Time to first occurrence of each outcome as a function of time-varying PDC for TNF inhibitors was evaluated using Cox regression. Incident/dynamic time-dependent receiver operating characteristic curves and Youden’s J index were used to obtain the optimal PDC threshold for outcomes at 365 days.ResultsOf the 1190 patients who met the study inclusion criteria, almost 75% (865 patients) experienced at least one of the outcomes. Increasing PDC by 10% was significantly associated with decreased risks of the composite outcome (HR 0.98, 95% CI 0.96–1.00), oral glucocorticoid use (HR 0.93, 95% CI 0.91–0.96), and hospitalization (HR 0.96, 95% CI 0.94–0.99) but an increased risk of ER visits (HR 1.04, 95% 1.01–1.07). Optimal PDC thresholds for the composite outcome, oral glucocorticoid use, and hospitalization were 0.64, 0.59, and 0.56, respectively.ConclusionsIncreased PDC was associated with a decreased risk of adverse outcomes, except ER visits. The optimal PDC for TNF inhibitors in Medicare patients with RA based on clinical outcomes was about 60%.Key Points• The optimal proportion of days covered threshold for tumor necrosis factor inhibitors at 365 days based on clinical outcomes was found to be about 60%, which is lower than the traditional 80% used to define adherence.• Increased adherence was associated with decreased risks of oral glucocorticoid use, hospitalization, and the composite outcome. However, it was also associated with an increased risk of emergency room visits.• The mean time-varying 90-day proportion of days covered decreased throughout the study starting 92% at day 1 of follow-up to 62% at day 365.