Factors associated with tuberculosis-attributed mortality in the Brazilian Amazon: a population-based retrospective cohort study with survival and competing-risk analyses

Abstract Background Tuberculosis mortality remains a major public health challenge in the Brazilian Amazon, where clinical vulnerability, social inequities, and heterogeneous access to healthcare may contribute to poor outcomes. This study aimed to identify factors associated with tuberculosis-attributed mortality in Pará, Brazilian Amazon, from 2012 to 2022. Methods We conducted a population-based retrospective cohort study using tuberculosis cases notified in the Brazilian Notifiable Diseases Information System between January 1, 2012, and December 31, 2022. After excluding cases without valid treatment closure date, post-mortem notifications, and inconsistent follow-up time, 52,066 cases were included in the main survival cohort. Tuberculosis-attributed death was the primary outcome. Cause-specific Cox proportional hazards regression was used to estimate adjusted hazard ratios. A Fine–Gray competing-risk model was fitted as a complementary analysis, considering cure and death from other causes as competing events. Results Among 52,066 tuberculosis cases, 1,032 tuberculosis-attributed deaths were recorded. Older age was strongly associated with mortality, with a 50% higher hazard per decade of life in the Cox model (aHR = 1.50; 95% CI: 1.45–1.55; p < 0.001), consistent with the Fine–Gray model (sHR = 1.49; 95% CI: 1.44–1.54; p < 0.001). Male sex was also associated with mortality (aHR = 1.16; 95% CI: 1.01–1.32; p = 0.035). TB–HIV/AIDS positivity increased the hazard of death (aHR = 1.64; 95% CI: 1.32–2.03; p < 0.001), as did unknown or unrecorded TB–HIV/AIDS status (aHR = 1.38; 95% CI: 1.20–1.58; p < 0.001). Alcohol use (aHR = 1.53; 95% CI: 1.26–1.86; p < 0.001), smoking (aHR = 1.30; 95% CI: 1.06–1.61; p = 0.014), and pulmonary plus extrapulmonary tuberculosis (aHR = 3.00; 95% CI: 2.29–3.92; p < 0.001) were independently associated with mortality. Incarceration showed an inverse association with mortality (aHR = 0.42; 95% CI: 0.27–0.65; p < 0.001), which should be interpreted cautiously as a possible marker of structured follow-up or differential surveillance. Conclusions Tuberculosis-attributed mortality in Pará was associated with older age, male sex, TB–HIV/AIDS, alcohol use, smoking, mixed pulmonary and extrapulmonary disease, and incomplete HIV-related information. The consistency between Cox and Fine–Gray models supports the robustness of the findings. Reducing tuberculosis mortality in the Brazilian Amazon requires integrated strategies focused on early diagnosis, systematic HIV testing, continuity of care, treatment supervision, and improved surveillance completeness among clinically and socially vulnerable groups.