Diabetes mellitus and tuberculosis, a systematic review and meta-analysis with sensitivity analysis for studies comparable for confounders

Introduction
Meta-analyses conducted so far on the association between diabetes mellitus (DM) and the tuberculosis (TB) development risk did not sufficiently take confounders into account in their estimates. The objective of this systematic review was to determine whether DM is associated with an increased risk of developing TB with a sensitivity analyses incorporating a wider range of confounders including age, gender, alcohol consumption, smoke exposure, and other comorbidities.

Methods
Pubmed, Embase, Web of Science and Global Index Medicus were queried from inception until October 2020. Without any restriction to time of study, geographical location, and DM and TB diagnosis approaches, all observational studies that presented data for associations between DM and TB were included. Studies with no abstract or complete text, duplicates, and studies with wrong designs (review, case report, case series, comment on an article, and editorial) or populations were excluded. The odds ratios (OR) and their 95% confidence intervals were estimated by a random-effect model.

Results
The electronic and manual searches yielded 12,796 articles of which 47 were used in our study (23 case control, 14 cross-sectional and 10 cohort studies) involving 503,760 cases (DM or TB patients) and 3,596,845 controls. The size of the combined effect of TB risk in the presence of DM was OR = 2.3, 95% CI = [2.0–2.7], I2 = 94.2%. This statistically significant association was maintained in cohort (OR = 2.0, CI 95% = [1.5–2.4], I2 = 94.3%), case control (OR = 2.4, CI 95% = [2.0–2.9], I2 = 93.0%) and cross-sectional studies (OR = 2.5, CI 95% = [1.8–3.5], I2 = 95.2%). The association between DM and TB was also maintained in the sensitivity analysis including only studies with similar proportions of confounders between cases and controls. The substantial heterogeneity observed was mainly explained by the differences between geographic regions.

Conclusions
DM is associated with an increased risk of developing latent and active TB. To further explore the role of DM in the development of TB, more investigations of the biological mechanisms by which DM increases the risk of TB are needed.