Wengao Zeng, Xiangzhi Xiao, Qing Liu, Huashan Zhou, Liu Chen, Yan Ouyang, Zhen Wang, Sufen Chen, Jue Hu, Yanhua Zhou, Qiong Wu
We found no clear evidence that PPD positivity was associated with in-hospital mortality in the full cohort. The pediatric association was large but imprecise and depended on nine deaths among PPD-positive children; it should be treated as an exploratory signal rather than an established age-dependent prognostic association. Multicenter studies using standardized PPD procedures, quantitative induration and testing-time data, and detailed baseline severity measures are needed to independently replicate the pediatric association.
BACKGROUND: Purified Protein Derivative (PPD) skin test reactivity reflects cell-mediated immune response to Mycobacterium tuberculosis. Evidence regarding the prognostic association of PPD reactivity in TBM, particularly its variation across age groups, remains limited.
METHODS: We conducted a single-center retrospective cohort study of consecutive non-HIV TBM patients. Firth's penalized logistic regression was used to estimate the association between PPD status and in-hospital mortality using fixed covariate sets selected on clinical grounds. Entropy balancing was used as a complementary covariate-balancing analysis. Age-related heterogeneity in the PPD-mortality association was evaluated on multiplicative and additive scales and explored using restricted cubic splines (RCS). Discrimination, calibration, and exploratory decision-curve net benefit were summarized using the area under the receiver operating characteristic curve (AUC), optimism-corrected AUC, Brier score, calibration intercept and slope, and decision curve analysis.
RESULTS: Among 1,574 patients, 187 (11.9%) died during hospitalization. In the fixed clinical model, there was no clear evidence that PPD positivity was associated with mortality in the full cohort (OR 1.30, 95% CI 0.74-2.19) or among adults (OR 0.68, 95% CI 0.31-1.34). Among 211 children, 9 of 25 PPD-positive children and 13 of 186 PPD-negative children died. The age- and diagnostic-category-adjusted pediatric estimate was OR 7.20, 95% CI 2.70-18.94; however, it was based on only 22 pediatric deaths and should be considered hypothesis-generating. The multiplicative interaction OR for adults versus children was 0.08 (95% CI 0.02-0.28), and the pediatric-minus-adult difference in standardized mortality-probability contrasts was 0.321 (95% CI 0.158-0.510). Full-cohort model performance was modest (AUC 0.673; optimism-corrected AUC 0.647; Brier score 0.0994). For the primary pediatric model, the E-value was 4.81 for the point estimate and 2.67 for the lower confidence-limit bound after approximate OR-to-RR conversion.
CONCLUSION: We found no clear evidence that PPD positivity was associated with in-hospital mortality in the full cohort. The pediatric association was large but imprecise and depended on nine deaths among PPD-positive children; it should be treated as an exploratory signal rather than an established age-dependent prognostic association. Multicenter studies using standardized PPD procedures, quantitative induration and testing-time data, and detailed baseline severity measures are needed to independently replicate the pediatric association.