DOI: https://doi.org/10.36347/sjpms.2026.v13i09.001

Pages: 277-291

Total adenosine deaminase activity in pleural fluid (P-ADA) has been proposed as a biomarker for differentiating the causes of pleural effusion syndrome (PES). This retrospective study included 157 patients with PES, classified as transudative (n = 33; 21%) or exudative (n = 124; 79%). A model was created to predict transudative pleural effusions (TPE) using age, sex, and P-ADA activity as predictors. According to the Akaike information criterion, P-ADA was included as a continuous variable (U/L). A univariable logistic regression model was employed for all independent variables, but only those with a P-value of ≤0.25 were used for the Enter multivariable binary logistic regression. The variance inflation factor (VIF = 1) indicated no multicollinearity. The model was internally validated using 1,000 bootstrap resamples. In the final multivariable model, P-ADA activity (U/L) was the only independent predictor of TPE, whereas age and sex were not retained in the model. The Youden index on the ROC curve established a P-ADA < 9.0 U/L for pleural transudates. The aOR for P-ADA was 0.89 (95% CI: 0.82–0.96; P = 0.001). The regression equation was logit (p)= −1.106 −0.12183 × P-ADA (U/L), that is, increasing P-ADA (U/L) reduced the estimated probability of TPE. The discriminative performance was good, with an AUC of 0.82 (95% CI, 0.76–0.87; P < 0.05). Calibration was satisfactory according to the Hosmer–Lemeshow test. Bootstrap validation demonstrated stable model performance, with an optimism of approximately 2%. The overall classification accuracy was 82%. The model demonstrated substantial explanatory power or predictive performance, with a Nagelkerke pseudo-R² of 0.33. The clinical utility was considered high because the AUC exceeded 70%. The final model was not excessively overfitted, despite the relatively small number of TPE. In conclusion, P-ADA is a useful independent biomarker that may enhance diagnostic decision-making for pleural transudates.