Predicting Intravenous Immunoglobulin Unresponsiveness In Kawasaki Disease

Objective:This study retrospectively evaluated the predictive efficiency of six different prediction scoring systems and identified independent risk factors to generate a new predictive scoring system for intravenous immunoglobulin(IVIG)unresponsiveness in Kawasaki disease in China.Methods:The data of 1655 Chinese Kawasaki disease patients were retrospectively analyzed.Kobayashi score,Egami score,Sano score,Fu score,Formosa score and Tang score were performed for each selected patient to confirm their predictive capability.Independent risk factors of IVIG unresponsiveness were examined by univariable analyses,multiple logistic regression analyses and receiver operating characteristic(ROC).Ultimately,a new predictive scoring system for IVIG unresponsiveness was established according to the screened predictors.Results:Kobayashi's scoring system had relatively optimal predictive efficiency(Youden index was 0.43)in those six scoring systems,but its sensitivity was low(sensitivity=58.0%).The Youden indexes of other 5scoring systems were less than 0.4.Platelet count(PLT),neutrophil percentage(N%),total bilirubin(TB),and serum sodium were selected as independent risk factors of IVIG unresponsiveness in Kawasaki disease(P<0.05).Using the 4 predictors,a simple scoring system was established:PLT?278×10~9/L(1 point),N%?80%(1 point),TB?18.5?mol/l(1 point),and serum sodium?135mmol/L(1 point).The scores?2 points was defined as high risk for IVIG unresponsiveness in Kawasaki disease(70.4%sensitivity and 72.7%specificity),that gave an area under the ROC curve of 0.767.Conclusion:None of those six scoring systems was appropriate for detecting IVIG unresponsiveness in Chinese population in Kawasaki disease.PLT,N%,TB,and serum sodium were the best independent risk factors of IVIG unresponsiveness in Kawasaki disease.The sensitivity and specificity of our new scoring system was satisfied to be used,however,its predictive ability also should be validated in different races and populations.