Background: Refractory Mycoplasma pneumoniae pneumonia (RMPP) has a serious, rapid progression that can easily cause a variety of extra-pulmonary complications. Therefore, the early identification of RMPP is crucial. This study aimed to construct and validate a risk prediction model based on clinical manifestations, laboratory blood indicators, and radiological findings to help clinicians identify patients who are at high risk of RMPP.
Methods: We retrospectively analyzed the medical records of 369 children with Mycoplasma pneumoniae pneumonia (MPP) admitted to Xi'an Children's Hospital, China. The demographics, clinical features, laboratory data, and radiological findings between the RMPP group and the general Mycoplasma pneumoniae pneumonia (GMPP) group were compared and subjected to univariate and multivariate logistic regression analyses.
Results: The fever peak and duration of the children in the RMPP group (n=86) were higher and longer compared with those in the GMPP group (n=283) (P<0.05). There was a significant difference in the incidence of lobar pneumonia and pleural effusion in pulmonary imaging between the two groups (P<0.05). Laboratory tests showed that the children with RMPP had lower serum uric acid (SUA) and albumin (ALB) as compared with the GMPP group (P<0.05). White blood cells (WBCs), neutrophil count (NEP), erythrocyte sedimentation rate (ESR), procalcitonin (PCT), C-reactive protein (CRP), and neutrophil-to-lymphocyte ratio (NLR) were higher in the RMPP group (P<0.05). Binary logistic regression analysis showed that the fever duration, pleural effusion, WBC, NEP, lactate dehydrogenase (LDH), CRP, NLR, and SUA levels were independent predictors of RMPP (P<0.05). The receiver operator characteristic (ROC) curve results showed fever duration, WBC, NEP, CRP, LDH, SUA, and NLR had good predictive value. The areas under the curve (AUCs) were 0.861, 0.730, 0.758, 0.837, 0.868, 0.744, and 0.713 and the best cutoff values were 10.50, 10.13, 6.43, 29.45, 370.50, 170.50, and 3.47, respectively. Finally, fever duration of more than 10.5 days, pleural effusion, WBC >10.13×109/L, NEP >6.43×109/L, CRP >29.45 mg/L, LDH >370.50 U/L, NLR >3.47, and SUA <170.5 µmol/mL constructed a prediction model of RMPP. According to internal validation, the mean AUC of the nomogram based on the development dataset was 0.956 [95% confidence interval (CI): 0.937-0.974] with good discrimination ability for predicting RMPP patients. The calibration plot and Hosmer-Lemeshow test (P=0.70) of the prediction model showed good consistency between the predicted probability and actual probability. Decision curve analysis (DCA) showed that the nomogram is clinically useful.
Conclusions: The simple and easy-to-use nomogram can help clinicians, especially primary doctors, to make early diagnoses of RMPP.