Nutritional scores predict the prognosis of patients with pulmonary tuberculosis

Abstract Background: Although malnutrition is associated with poor prognosis in Pulmonary Tuberculosis (PTB) patients, there is currently no nutrition-based prediction model for PTB. Herein, we explored the applicability of common nutrition scores on the prognosis of PTB patients, aiming to construct a nutrition-based prediction model related to their prognosis. Methods: We retrospectively collected clinical baseline data from 167 patients with secondary PTB who had not previously received anti-TB treatment. Subsequently, we determined the CONUT score, PNI index, and NPS score and evaluated the treatment efficacy using changes in lung lesions revealed by the chest CT scan. The Area Under the Receiver Operating Characteristic (AUROC) curve was used to quantify the predictive values of CONUT, PNI, and NPS scores for anti-TB efficacy in new-onset PTB patients, and the critical CONUT, PNI, and NPS values were determined using the Youden Index. We also performed univariate and multivariate analyses of prognostic factors in PTB patients to determine the nutrition scores and other clinical factors associated with the prognosis of patients with the new-onset PTB. Results: The Youden Index revealed that the critical value of the CONUT score in PTB patients was 4.5, at which the sensitivity (72.2%) and specificity (96.6%) were highest. On the other hand, the critical cut-off values of the PNI index and the NPS score were 39.825 and 3.5, respectively. Univariate analysis of the predictors of poor prognosis in PTB patients showed that patients with diabetes, COPD, pneumonia, and hypoproteinemia (which were risk factors) had a poor prognosis (p<0.05). Multivariate analysis revealed that the CONUT score, PNI, NPS, and NRS-2002 were independent predictors of unfavourable PTB outcomes, with adjusted ORs of 0.016 (95%CI: 0.936-0.993, p<0.0001), 23.667 (95%CI: 9.317-60.115, p<0.0001), 0.119 (95%CI: 0.054-0.262, p<0.0001), and 0.124 (95%CI: 0.055-0.279, p<0.0001), respectively. The Area Under the Curve (AUC) of the CONUT score in predicting poor prognosis of PTB patients was 0.885 (95% CI：0.830-0.940，p<0.0001), which is comparable to that of the PNI index (0.862, 95% CI: 0.805-0.920, p<0.0001), but higher than that of NPS (0.774, 95% CI: 0.702-0.846, p<0.0001), BMI (0.627, 95% CI: 0.541-0.717, p<0.0001), and NRS-2002 (0.763, 95% CI: 0.688-0.838, p<0.0001). We discovered that older patients (p<0.0001), male participants (p< 0.05), and patients with Diabetes Mellitus (DM) (p<0.0001) and Chronic Obstructive Pulmonary Disease (COPD) (p<0.05) were more likely to have a high CONUT score. Conclusion: The poor prognosis of PTB patients was related to a high CONUT score, low PNI index, and high NPS score, of which the specificity and sensitivity of the CONUT score were higher than those of the PNI index and the NPS score.