ABSTRACT
PURPOSE: Despite increasing literature on the association between treatment delay and outcomes, cut-off point (1 month or median) selection in almost all studies for treatment delay is too subjective. This study explored more scientific cut-off points of treatment delay for poor treatment outcomes and death at the clinical level. PATIENTS AND METHODS: A total of 18,100 newly confirmed pulmonary tuberculosis (TB) cases in Dalian, China were used in the final analysis. A 3-knotted restricted cubic spline (RCS) fitted for Cox proportional hazard regression models is used to analyse the effects of cut-off points of treatment delay on incident poor treatment outcomes. To explore the moderating effects of age, gender and diabetes, we added the interaction terms of these moderating variables and treatment delay to Cox proportional hazard regression models. RESULTS: The median time of treatment initiation was 30 days (IQR: 14-59 days). The risk of incident poor treatment outcomes increased when the time was greater than cut-off point 1 (53 days; adjusted HR: 1.26; 95% CI: 1.00-1.60) of treatment delay, and the risk of incident death events increased when the time was greater than cut-off point 2 (103 days; adjusted HR: 1.56; 95% CI: 1.00-2.44) of delay. In addition, treatment delay was associated with an increased risk of incident poor treatment outcomes and death, and older age, male sex, and diabetes may increase the risk of treatment delay for poor outcomes. CONCLUSION: This study is the first to identify scientific cut-off points of treatment delay for poor treatment outcomes and death, and this method of exploration should be popularized. In addition, the knowledge of tuberculosis must be spread to every adult. Moreover, the tuberculosis diagnosis level of community level health workers should be enhanced.