Background It has previously been suggested that diabetes worsens tuberculosis treatment outcomes, but studies included in previous systematic reviews had substantial limitations in defining tuberculosis and diabetes, and made little adjustment for potential confounders. Many new studies have been published in this area since 2010. We performed an updated systematic review and meta-analysis.
Methods We searched bibliographic databases including MEDLINE/Pubmed, Cochrane CENTRAL, EMBASE, CINAHL, Web of Science, Scopus, and LILACS between January 1st 1980 and July 1st 2015, not restricted by language or region. We included cohort (prospective or retrospective) studies, case-control studies, and randomised controlled trials which investigated the difference in tuberculosis treatment outcomes between patients with both tuberculosis and diabetes and those with tuberculosis only. Study participants were not restricted by sex, race, drug resistance, or HIV status. Tuberculosis treatment outcomes were defined by WHO guidelines. Additional outcomes including relapse and recurrence were also considered. The Newcastle-Ottawa Scale and Cochrane review guidelines were used to assess the methodological quality and risk of bias of the included studies. Two reviewers independently reviewed all titles, abstracts, and extracted data. Random effect meta-analysis using generic inverse variance method was used for data synthesis using Stata 12, and main analyses were stratified by low and high income countries.
Results Of 15476 titles identified, 4156 articles were duplicates, and another 11028 studies were excluded after the abstract screening. Full-text screening was performed in 292 articles, of which 100 studies were eligible for inclusion for systematic review (including 70178 tuberculosis only patients and 11562 patients with both tuberculosis and diabetes). Patients with both tuberculosis and diabetes had double the risk of treatment failure or death compared with those with tuberculosis only (Odds ratio = 2.06, 95% CI: 1.68–2.53). Diabetes was also associated with increased risk of recurrence (1.57, 1.38–1.79), but not relapse (1.12, 0.50–2.49). Moreover, the odds ratio for death, and treatment failure were similar in low- (2.08, 1.54–2.82) and high-income countries (2.00, 1.52–2.52).
Conclusion Death or treatment failure was twice as common in tuberculosis patients with diabetes, compared with those without; longer-term outcomes such as recurrence were also adversely affected. Diabetes burden is rising globally, particularly in urban centres in low-income countries where tuberculosis is still endemic. Given the high risk of a poor outcome among those with co-morbid disease, heightened clinical attention to improve both diabetes control and tuberculosis outcomes.