Development and Validation of A Prediction Model for Tuberculous Pleural Effusion: A Large Cohort Study and External Validation

Abstract Background: Distinguishing tuberculous pleural effusion (TPE) from non-tuberculosis (TB) benign pleural effusion (BPE) remains to be a challenge in clinical practice. The aim of the present study was to develop and validate a novel nomogram for diagnosing TPE. Methods: In this retrospective analysis, a total of 744 consecutive patients with TPE and non-TB BPE from Ningbo First Hospital were divided into the training set and the internal validation set at a ratio of 7:3, respectively. The clinical and laboratory features were collected and analyzed by logistic regression analysis. A diagnostic model incorporating selected variables was developed and was externally validated in a cohort of 110 patients from another hospital. Results: Six variables including age, effusion lymphocyte, effusion adenosine deaminase (ADA), effusion lactatedehy drogenase (LDH), effusion LDH/effusion ADA, and serum white blood cell (WBC) were identified as valuable parameters used for developing a nomogram. The nomogram showed a good diagnostic performance in the training set. A novel scoring system was then established based on the nomogram to distinguish TPE from non-TB BPE. The scoring system showed good diagnostic performance in the training set (area under the curve (AUC), 0.932, sensitivity, 93.7%, and specificity, 85.4%), the internal validation set (AUC, 0.934, sensitivity, 93.5%, and specificity, 82.9%), and the external validation set (AUC, 0.938, sensitivity, 93.0%, and specificity, 83.1%), respectively. Conclusions: The study developed and validated a novel scoring system based on a nomogram originated from six clinical parameters. The novel scoring system showed a good diagnostic performance in distinguishing TPE from non-TB BPE and can be conveniently used in clinical settings.