Tuberculosis and diabetes: bidirectional association in a UK primary care data set.

Abstract

BACKGROUND: Many studies have found an increased risk of pulmonary tuberculosis (PTB) among those with diabetes mellitus (DM). However, evidence on whether the association is bidirectional remains sparse. This study investigates DM rates among those with and without prior tuberculosis (TB) disease as well as the reverse. METHODS: Data on a UK general practice population, between 2003 and 2009, were obtained from The Health Improvement Network database. A series of retrospective cohort studies were completed. Individuals were successively classified as 'exposed' or 'unexposed' to TB, PTB, extrapulmonary TB (EPTB) or DM. Multivariate negative binomial regression was used to calculate incidence rate ratios (IRR) among each exposure group for outcomes of interest (TB, PTB, EPTB or DM in turn) adjusting for plausible confounding variables (age, sex, region, Townsend quintile and smoking status). Potential confounding due to ethnicity was adjusted for using McNamee's external method. RESULTS: DM risk was substantially raised among individuals with a history of TB disease (IRR 5.65 (95% CI 5.19 to 6.16)), PTB (IRR 5.74 (95% CI 5.08 to 6.50)) and EPTB (IRR 4.66 (95% CI 3.94 to 5.51)) compared with those without; results were attenuated after external adjustment for ethnicity (IRR 2.33 (95% CI 2.14 to 2.53)). TB risk was raised modestly among individuals with DM (IRR 1.50 (95% CI 1.27 to 1.76)) and was attenuated slightly after adjustment for ethnicity (IRR 1.26 (95% CI 1.07 to 1.48)). CONCLUSION: DM risk was raised among those with previous TB disease; this finding has implications for follow-up and screening of patients with TB, who may be at high risk of developing DM or related complications.