Association between diabetes mellitus and mortality among patients with tuberculosis in California, 2010-2014.

Abstract

Studies of US populations have produced conflicting findings about the impact of diabetes mellitus (DM) on tuberculosis (TB) treatment outcomes. To investigate the association between DM and all-cause mortality among patients on anti-tuberculosis treatment in California, USA. Using TB surveillance data, we conducted a retrospective analysis of California patients with culture-confirmed TB who started anti-tuberculosis treatment during 2010-2014. We used Cox proportional hazards models to estimate the association of DM with all-cause mortality and conducted a sensitivity analysis to estimate the attenuating effect of unmeasured confounding by body mass index. Among 8461 patients with TB, 2124 (25.1%) had DM and 713 (8.4%) died during anti-tuberculosis treatment. A higher proportion of TB-DM patients died (13.1% vs. 6.8% TB-no DM). After adjusting for confounders, DM was associated with mortality (adjusted hazards ratio [aHR] 1.35, 95%CI 1.15-1.57). There was effect modification by human immunodeficiency virus (HIV) status, with HIV-positive patients having an aHR of 5.33 (95%CI 1.76-16.12). TB patients with DM had a greater hazard of death during anti-tuberculosis treatment than those without DM. Further investigation into the impact of HIV on the relation of DM to death is necessary.