Predictors, mortality, and health outcomes of intensive phase non-adherence to a regimen in patients with drug-susceptible tuberculosis: a nationwide linkage database

Objectives: The clinical importance of adhering to the regimen in tuberculosis patients has been widely investigated, but most studies were conducted in controlled settings and in limited populations. We aimed to measure the level of real-world adherence during intensive phase and investigate the predictors and the risk of mortality and health outcomes of intensive phase non-adherence in tuberculosis patients. Study design: We conducted a nationwide cohort study by linking the Korean National Tuberculosis Surveillance System and the National Health Information Database. Methods: We included all incident drug-susceptible tuberculosis patients who initiated the regimens recommended by the World Health Organization from 2013 to 2018. Adherence was measured using the proportion of days covered (poor [<50%], moderate [50%–79%], and high [≥80%]). We used logistic regression model to assess predictors and the Cox proportional hazard model to evaluate the risk of mortality and health outcomes with intensive phase non-adherence. Results: Of 46,818 patients, there were 8% and 11% with poor and moderate adherent groups, respectively. Age ≥45 years, insulin use, and history of renal failure were predictors of non-adherence. Compared with high adherent group, poor and moderate adherent groups were associated with a substantial risk of mortality (poor: hazard ratio, 2.14 [95% confidence interval, 1.95–2.34]; moderate: 1.76 [1.62–1.92]). Similar trends were observed for health outcomes. Stratified analyses showed a higher risk of mortality in patients with medical aid, low income, and history of renal failure, systematic corticosteroids, and immunomodulators. Conclusions: Non-adherence during intensive phase increased mortality risk by twofold, underscoring targeted intervention for high-risk population, including advanced diabetes, and immunocompromised patients.