Evaluation of the combined predictive value of multiple indicators based on diaphragmatic ultrasound using logistic regression and ROC curve in weaning from mechanical ventilation in pediatric patients.

Background: Conventional single indicators have low sensitivity and specificity for predicting weaning from mechanical ventilation in pediatric patients, necessitating the establishment of a combined prediction model for predicting weaning outcomes.

Objectives: To explore the combined predictive value of PaO/FiO Ratio (P/F ratio), diaphragm excursion-rapid shallow breathing index (DE-RSBI), diaphragm thickening fraction-rapid shallow breathing index (DTF-RSBI), and Pediatric Critical Illness Score (PCIS) in weaning from mechanical ventilation in pediatric patients.

Methods: Sixty critically ill pneumonia pediatric patients requiring mechanical ventilation treatment from July 2022 to June 2023 at the Second Affiliated Hospital of Jiaxing University were selected. They all underwent a spontaneous breathing trial (SBT) and were divided into the weaning success group (42 cases) and weaning failure group (18 cases) based on the weaning outcome. Parameters including total duration of illness, mechanical ventilation duration, heart rate (HR), P/F ratio, diaphragm excursion (DE), DE-RSBI, diaphragm thickening fraction (DTF), DTF-RSBI, and PCIS were included in univariate and multivariate logistic regression analyses to determine independent factors affecting pediatric weaning success. Receiver operating characteristic (ROC) curves were plotted to evaluate the predictive value of P/F ratio, DE-RSBI, DTF-RSBI, PCIS alone or in combination for weaning success.

Results: Comparing P/F ratio, DE, DE-RSBI, DTF, DTF-RSBI and PCIS, there were statistically significant differences (< 0.05). Through collinearity analysis and binary logistic regression analysis,P/F ratio [OR=0.777, 95% CI (0.641,0.941)], DE-RSBI [OR=1.694, 95% CI (1.172, 2.447)], DTF-RSBI [OR=1.057, 95% CI (1.002, 1.114)], and PCIS [OR=0.661, 95% CI (0.445, 0.982)] were identified as independent factors affecting successful weaning(< 0.05).The regression equation was: Logit = 73.299-0.253 P/F ratio + 0.525DE-RSBI + 0.055DTF-RSBI-0.43PCIS.The sensitivity of the combined indicator Logit() in predicting successful weaning from mechanical ventilation in pediatric patients was 88.9%, with a specificity of 95.2% (optimal cutoff value of 0.511), and the area under the ROC curve (AUC) was 0.960 [95% CI (0.915, 1.000)]. The AUC of the combined prediction model for predicting pediatric weaning was greater than that of P/F ratio, DE-RSBI, DTF-RSBI and PCIS alone ( values=9.129, 2.061, 2.075, 8.326, < 0.05).

Conclusions: In mechanically ventilated pediatric patients, the combined prediction model has better predictive value for weaning success compared to using P/F ratio, DE-RSBI, DTF-RSBI, or PCIS alone.