Abstract:

Objective To construct a prediction model of pediatric refractory mycoplasma pneumoniae pneumonia (RMPP) based on lung ultrasound score (LUS) and inflammation index. Methods The clinical data of children with mycoplasma pneumoniae pneumonia (MPP) hospitalized in the Respiratory Department of Shanghai Children's Hospital, Pediatrics Departmen of the First Affiliated Hospital of University of Science and Technology of China or Jieshou People's Hospital from June 2023 to January 2024 were retrospectively analyzed, and the children were assigned into RMPP group and non-RMPP group according to the diagnostic criteria. The independent risk factors of RMPP were screened by univariate and multivariate logistic regression analysis, and the predictive model was constructed, and the predictive efficiency of the model was verified by receiver operating characteristic curve (ROC). Results A total of 876 children with MPP were included, including 498 in RMPP group and 378 in non-RMPP group. Binary Logistic regression analysis showed that LUS (OR=1.276, 95%CI 1.196-1.362), C-reactive protein (CRP, OR=1.098, 95%CI 1.062-1.119), D-dimer (D-D, OR=2.683, 95%CI 1.923-3.742) and lactate dehydrogenase (LDH, OR=1.015, 95%CI 1.013-1.018) were independent risk factors for RMPP in children (P<0.05). The ROC showed good predictive value of LUS or each inflammatory index alone in predicting RMPP, while, the combined predictive efficacy was the best, with an area under the ROC curve (AUC) of 0.969 (95%CI 0.958-0.979), a sensitivity of 97.4% and a specificity of 84.7%. Conclusion The regression model based on LUS combined with inflammatory indicators has a good predictive efficacy and provides a good basis for clinical diagnosis of children with RMPP. It is convenient for pediatricians to identify and intervene early and improve the dangerous outcome of RMPP.

Key words: pneumonia, mycoplasma, refractory, children, lung ultrasound score, indicators of inflammation, prediction model