Construction and validation of a predictive model for adverse reactions to mite subcutaneous immunotherapy in children with allergic asthma

Objective To construct a nomogram prediction model for adverse reactions of mite subcutaneous immunotherapy（SCIT） in children with allergic asthma， and to evaluate and validate it. Methods Clinical data of children with allergic asthma who were managed by mite SCIT in Department of Pediatrics， the Second Hospital of Tianjin Medical University were retrospectively analyzed. According to the adverse reactions in the initial treatment stage of SCIT， they were divided into adverse reaction group and no adverse reaction group， which together formed the modeling group. Through univariate logistic regression analysis and multivariate logistic regression analysis， independent risk factors were screened out. Then construct the prediction model， draw a nomogram， eventually evaluate and validate the model. Results A total of 350 children were included， including 176 in the adverse reaction group and 174 in the no adverse reaction group. Univariate logistic regression analysis was performed on the baseline data of the children in the modeling group， and the results were as follows. Gender， age， number of asthma exacerbations in the past 1 year， percentage of eosinophils （EOS%）， forced expiratory volume in the first second in percent predicted values（FEV₁%pred）， total IgE（tIgE）， dermatophagoides pteronyssinus （Der p） sIgE， dermatophagoides farinae （Der p） sIgE， and the number of other inhaled allergenic species were the risk factors of SCIT-related adverse reactions （all <i>P &lt; </i>0.1）. These were included in the multivariate logistic regression analysis， and the results showed that： age， the number of asthma exacerbations in the past 1 year ， tIgE， Der p sIgE， Der f sIgE， and the number of other inhaled allergen species were independent risk factors （all <i>P &lt; </i>0.05）. Based on the results， nomogram was drawn in R language. Model evaluation by receiver operating characteristic curve showed an area under curve of 0.877， a predictive model sensitivity of 73.9%， and a specificity of 90.8%. The predicted probability was consistent with actual probability of occurrence. The model has good prediction performance. Conclusions Age， number of asthma exacerbations in the past 1 year， tIgE， Der p sIgE， Der f sIgE， and the number of other inhaled allergen species were independent risk factors for adverse reactions to mite SCIT in children. The nomogram drawn in this study has good predictive value.