RESUMO
Background: Patients with diabetes mellitus (DM) are at increased risk of developing TB, but the best screening algorithm for early detection and treatment of TB remains unknown. Our objective was to determine if combining routine chest X-ray screening could have a better yield compared with symptom-based screening alone. Methods: We conducted this cross-sectional study between September 2020 and September 2021 in 26 public health facilities in Addis Ababa, Ethiopia. All DM patients attending the clinics during the study period were offered chest X-ray and symptom screening simultaneously followed by confirmatory Xpert testing. We analyzed the number and proportion of patients with TB by the diagnostic algorithm category and performed binary logistic regression analysis to identify predictors of TB diagnosis. Results: Of 7394 patients screened, 54.6 % were female, and their median age was 53 years. Type-2 diabetes accounted for 89.6 % of all participants of the patients. Of 172 symptomatic patients, chest X-ray suggested TB in 19, and 11 of these were confirmed to have TB (8 bacterilogicially confirmed and 3 clinically diagnosed). Only 2 of the 152 asymptomatic patients without X-ray findings had TB (both bacteriologically confirmed). X-ray was not done for one patient. On the other hand, 28 of 7222 symptom-negative patients had X-ray findings suggestive of TB, and 7 of these were subsequently confirmed with TB (6 clinically diagnosed). When combined with 8 patients who were on treatment for TB at the time of the screening, the overall point prevalence of TB was 380 per 100,000. The direct cost associated with the X-ray-based screening was 42-times higher. Conclusion: Chest X-ray led to detection of about a third of TB patients which otherwise would have been missed but the algorithm is more expensive. Its full cost implication needs further economic evaluation.
RESUMO
Background: Tuberculosis (TB) and diabetes mellitus (DM) have negative synergistic impact on each other. Global guidelines recommend collaborative efforts to address this synergy, but implementation has been slow. Part of the reason is lack of adequate evidence on the operational feasibility of existing tools and mechanisms of collaboration. The objective of this study was to assess the yield of DM screening among TB patients using risk scoring tools combined with blood tests as a feasible strategy for early detection to improve TB/DM treatment outcomes. Methods: Between September 2020 and December 2021, we conducted a cross-sectional study among patients receiving TB treatment in public health facilities in Addis Ababa, Ethiopia. Trained health workers collected data on symptoms and risk scoring checklists before testing for random and fasting blood glucose levels. We used logistic regression analyses techniques to determine factors associated with increased DM detection. A receiver-operating characteristic curve was constructed to determine the performance of the risk scoring checklist. Results: Of 2381 TB patients screened, 197 (8.3%) had DM of which 48.7% were newly diagnosed. Having a family history of DM predicted DM with Odds Ratio (OR) of 5.36 (95% Confidence Interval, [3.67, 7.83]) followed by age ≥ 45 years (OR = 4.64, [3.18, 6.76]). Having one or more "symptoms" of DM was a significant predictor (OR 2.88, 95% CI, 2.06-4.01). Combining risk scores with symptoms predicted DM diagnosis with sensitivity of 94.7%, but specificity was low at 29.4%. In patients with known treatment outcome status, death rate was high. Conclusions: Almost a half of TB patients with DM did not know their status. A simple tool that combined risk factors with symptoms accurately predicted a subsequent diagnosis of DM. Such tools can help avoid high rates of death among TB patients suffering from DM through early detection.
