Abstract
Systematic screening for tuberculosis (TB) is recommended in people with diabetes; however, data on the accuracy of screening tools in this population are lacking. We assessed the accuracy of symptom and chest X-ray screening among people with diabetes. We consecutively enrolled adults with diabetes attending routine care in South Africa. All participants underwent symptom screening and chest X-ray. A single sputum specimen was collected from all participants and tested by Xpert Ultra. A positive Xpert Ultra result was used as the reference standard. We enrolled 673 participants. The median age was 54 years (interquartile range 47–60 years), and 63.8% were female. HIV prevalence was 17.2%. Prevalent TB was diagnosed in nine participants (1.33%). Any cough had a sensitivity of 22.2% (95% confidence interval [CI] 2.1-60.0%) and a specificity of 97.5% (95%CI 96.0-98.6%). Expanding the symptom definition to include any of cough, fever, weight loss, or night sweats did not improve sensitivity (22.2%, 95% CI 2.1–60.0) and slightly reduced specificity to 96.0% (95% CI 94.2–97.0). Chest X-ray abnormalities suggestive of TB demonstrated a sensitivity of 55.6% (95%CI 21.2-86.3%) and a specificity of 95.4% (95%CI 93.4-97.0%). The specificity of chest X-ray was significantly lower in participants with prior TB (87.6%, 95% CI: 79.8–90.6%), compared to 97.2% (95% CI: 95.2–98.5%) in those without (p < 0.01). Symptom-based TB screening has poor sensitivity in people with diabetes. Although chest X-ray improved the sensitivity, it remained suboptimal, with a reduced specificity in people with previous TB.