Abstract O.45: Kawasaki Disease Refractory To Primary IVIG Treatment - Use Of Scoring Systems And Predictive Modelling Based On Data From Singapore Children
Objectives: Although first-line treatment for Kawasaki Disease (KD) is IVIG, some children may not respond, requiring repeat IVIG or another second-line therapy. It is important to predict non-responders for timely management decisions in this group of patients. The objective of our study was to assess the predictive value of the currently available scoring systems and propose an improved model for predicting refractory KD, based on data from KD patients in Singapore.
Method: We performed a retrospective review of the medical records (2001-2014) of 180 children with KD and identified 17 refractory cases. The 4 scoring systems; Egami, Kobayashi, Sano and Fukunishi; were applied to the data of refractory KD cohort to determine their predictive value based on sensitivities (SN) and specificities (SP). A logistic regression model was built using 5 predictors (age, AST, CRP, Hb, platelet and duration of illness) which were common among the existing models to propose a new scoring system based on our data -. The formula used was 1/(1+e^(-(-5.642839-0.192452*(zHb)-0.011597*(Age)+0.003756*(PLT)+0.009441*(CRP)+0.006015*(AST)+0.084408*(Days).
Results: The SN and SP obtained were; (1) Egami: SN 35.3%, SP 73.9% (original SN 78% and SP 76%) (2) Kobayashi: SN 23.5%, SP 67.5% (original SN 86% and SP 68%) (3) Sano: SN 18.8% and SP 88.2% (original SN 77% and SP 86%) (4) Fukunishi: SN 53.3% and SP 76.0% (original 84.6% and 87.0%). As the existing scoring systems proved less predictive in our population, we used logistic regression model to derive a formula incorporating the common variables to predict the likelihood of non-responders. This model attained an SN of 80% and SP of 80%.
Conclusion: The proposed model should be validated prospectively in our local population for predicting unresponsiveness, and the formula can be applied to different datasets from other KD populations to test its robustness.
Author Disclosures: S. Quek: None. Y. Leow: None. D.D. Rajgor: None. T. Lim: None. C. Heng: None. R. Grignani: None.
- © 2015 by American Heart Association, Inc.