Early multi-system physiological biomarkers of bronchopulmonary dysplasia in preterm infants: a matched case-control study based on bedside monitoring.

OBJECTIVE: To identify early multi-system physiological biomarkers that can be obtained at the bedside prior to the formal diagnosis of bronchopulmonary dysplasia (BPD). METHODS: We conducted a single-center retrospective 1:1 matched case-control study enrolling preterm infants of less than 32 weeks' gestational age admitted to the neonatal intensive care unit (NICU) of our institution between June 2022 and June 2025. Infants with confirmed BPD were assigned to the BPD group, and non-BPD controls were matched 1:1 by gestational age and birth weight. Baseline characteristics, bedside monitoring parameters, blood gas indices, and quantitative echocardiographic features were compared between the two groups. Candidate early variables associated with BPD were first screened using LASSO-penalized logistic regression, after which representative indicators from each physiological domain were selected to construct a multivariate logistic regression model. RESULTS: A total of 50 infants were included in each group, and no significant differences in baseline or maternal perinatal characteristics were found between the two groups (all P > 0.05). Multivariate analysis identified six independent predictors of BPD: invasive mechanical ventilation (IMV) [adjusted odds ratio (aOR): 21.312, 95% CI: 2.108-36.296, P = 0.023], peak FiO2 (aOR: 1.404, 95% CI: 1.022-1.931, P = 0.036), lower PI trend slope (aOR: 0.041, 95% CI: 0.002-0.974, P = 0.048), lower PaO2/FiO2 ratio (aOR: 0.026, 95% CI: 0.001-0.848, P = 0.040), lower LVEF (aOR: 0.008, 95% CI: 0.000-0.602, P = 0.028), and thinner IVSd (aOR: 0.043, 95% CI: 0.002-0.933, P = 0.045) as independent predictors of BPD. The combined model achieved excellent discrimination for BPD with an AUC of 0.982 (95% CI: 0.963-1.000), substantially exceeding the discriminative performance of each individual marker (individual AUC range: 0.625-0.864). At the optimal Youden index threshold, the model achieved a sensitivity of 93.47% and a specificity of 94.62. CONCLUSION: An early six-variable physiological profile integrating IMV, peak FiO2, PI trend slope, PaO2/FiO2 ratio, LVEF, and IVSd may facilitate early identification of preterm infants at increased risk of BPD.