A Retrospective Cohort Study of Mortality Associated with Tuberculosis in North Carolina, 1993-2003

Introduction Previously reported predictors of mortality have focused on mortality during TB treatment. We propose instead to examine TB deaths relative to TB treatment. We predict that different characteristics will predict death before TB treatment, early in TB treatment and later in TB treatment. Methods A retrospective cohort study was performed using data collected on all patients reported to the North Carolina TB Control Program in conjunction with data obtained from the U.S. National Death Index. Bivariate, multivariate and survival analyses were performed to examine relationships between patient characteristics and time of death relative to treatment. Results Age (adjusted OR=l.04 per year of age, 95%CI 1.03-1.05, HIV/TB coinfection (adjusted OR=3.44, 95%CI 2.04-5.79) and being an American Indian (adjusted OR=4.00, 95%CI 1.47-10.93) were predictors of death before initiation of TB treatment. Age (adjusted HR=1.06 each year of age, 95%CI 1.05-1.07), black race (adjusted HR=l.35, 95%CI 1.05-1.75), HIV/TB co-infection (adjusted HR=1.66, 95%CI 0.99-2.76), history of alcohol abuse (adjusted HR=1.68, 95%CI 1.17-2.41) and residence in a long-term care facility (adjusted HR=1.70, 95%CI 1.23-2.34) were predictors of death in the first 8 weeks of treatment. After the initial 8 weeks, only age (adjusted HR=1.05 for each year of age, 95%CI 1.04- 1.06) and HIV/TB co-infection (adjusted HR=5.53, 95%CI 3.65-8.44) continued to be predictors of mortality. Discussion Overall, increasing age and HIV/TB co-infection were predictors of death prior to starting TB therapy and during all stages of TB therapy. Additional TB patient characteristics also existed that predict death specifically before treatment or in the early weeks of treatment.