Frontiers in Pediatrics (Feb 2022)
Development of a Nomogram for Predicting Refractory Mycoplasma pneumoniae Pneumonia in Children
Abstract
BackgroundIn children, refractory Mycoplasma pneumoniae pneumonia (RMPP) may result in severe complications and high medical costs. There is research on a simple and easy-to-use nomogram for early prediction and timely treatment of RMPP.MethodsFrom December 2018 to June 2021, we retrospectively reviewed medical records of 299 children with Mycoplasma pneumoniae pneumonia (MPP) hospitalized in Tianjin Children's Hospital. According to their clinical manifestations, patients were divided into the RMPP group and the general Mycoplasma pneumoniae pneumonia (GMPP) group. The clinical manifestations, laboratory indicators, and radiological data of the two groups were obtained. Stepwise regression was employed for variable selection of RMPP. The predictive factors selected were used to construct a prediction model which presented with a nomogram. The performance of the prediction model was evaluated by C statistics, calibration curve, and receiver operating characteristic (ROC) curve.ResultsThe RMPP group significantly showed a higher proportion of females, longer fever duration, and longer hospital stay than the GMPP group (P < 0.05). Additionally, the RMPP group revealed severe clinical characteristics, including higher incidences of extrapulmonary complications, decreased breath sounds, unilateral pulmonary consolidation >2/3, and plastic bronchitis than the GMPP group (P < 0.05). The RMPP group had higher neutrophil ratio (N%), C-reactive protein (CRP), interleukin-6 (IL-6), lactic dehydrogenase (LDH), and D-dimer than the GMPP group (P < 0.05). Stepwise regression demonstrated that CRP [OR = 1.075 (95% CI: 1.020–1.133), P < 0.001], LDH [OR = 1.015 (95% CI: 1.010–1.020), P < 0.001], and D-dimer [OR = 70.94 (95% CI: 23.861–210.904), P < 0.001] were predictive factors for RMPP, and developed a prediction model of RMPP, which can be visualized and accurately quantified using a nomogram. The nomogram showed good discrimination and calibration. The area under the ROC curve of the nomogram was 0.881, 95% CI (0.843, 0.918) in training cohorts and 0.777, 95% CI (0.661, 0.893) in validation cohorts, respectively.ConclusionC-reactive protein, LDH, and D-dimer were predictive factors for RMPP. The simple and easy-to-use nomogram assisted us in quantifying the risk for predicting RMPP, and more accurately and conveniently guiding clinicians to recognize RMPP, and contribute to a rational therapeutic choice.
Keywords
