A clinical scoring system to predict tuberculous lymphadenitis in settings with high disease burden.

Limited diagnostic service availability challenges the timely diagnosis and treatment of tuberculous lymphadenitis, the most common extrapulmonary tuberculosis in endemic areas. To address this, our study aimed to develop a clinical scoring rule for estimating the likelihood of tuberculous lymphadenitis, designed as a simple and easy-to-apply classification tool to facilitate timely diagnosis. We used a cross-sectional study to collect data from 1364 patients with lymphadenopathy. A prediction model was developed by incorporating predictors from multivariable logistic regression analysis. Optimal cutoff point was determined using Youden's Index. Performance was evaluated using discrimination and calibration. Internal validation and clinical utility were assessed using bootstrapping and decision curve analysis, respectively. A total of 790 patients (57.9%, 95% CI 55.3%-60.5%) were diagnosed with tuberculous lymphadenitis. Sex, duration, size, location, and consistency of lymphadenopathy, presence of pain, sinus tract, systemic symptoms, and erythrocyte sedimentation rate were predictors incorporated to develop the scoring rule. With a discrimination power of 96.9% (95% CI 96.1%, 97.8%), the clinical scoring rule using these predictors showed a sensitivity, specificity, and accuracy of 93.7%, 89.7%, and 92%, respectively. It demonstrated internal validity with near-zero optimism coefficients for various performance metrics. Using easily accessible clinical parameters, we developed a scoring rule that reliably predicts tuberculous lymphadenitis. Besides its use in clinical settings for making timely clinical decisions, the scoring rule is a valuable candidate for community-level screening, facilitating early case detection.