Abstract O.41: Assessment of Kawasaki Disease Risk Scores for Predicting Coronary Artery Aneurysms at a North American Center

Abstract

Aim: To assess the performance of established Japanese risk scores (RS) to identify patients (pts) with Kawasaki disease (KD) at high-risk for developing coronary artery aneurysms (CAA). Methods: We reviewed clinical, laboratory and echocardiographic (echo) data for pts with KD treated with IVIG from 1/2006 to 5/2014. We defined CAA as z score ≥2.5 in the right coronary artery (RCA) or left anterior descending artery (LAD) at 4-8 weeks of illness. Relationships with Kobayashi, Sano, Egami and Harada RS and CAA were examined. The maximum z score of LAD or RCA (zMax) at baseline was a covariate in logistic regression. The discrimination of each model was assessed using the c statistic. Results: Of 268 pts with complete data, 173 (65%) were male and median age was 3.1 y (range 0.1-14.1 y). At diagnosis, 74 (28%) had ≤3 classical criteria for KD, and 70 pts (26%) received IVIG retreatment. On baseline echo, 75 pts (27%) had a zMax ≥ 2.0. CAA occurred in 15 pts (5.6%). The Harada RS predicted development of CAA (low risk = 0% (0/80), high risk=9% (12/140), p=0.005), but the Kobayashi, Sano, and Egami RS were not associated with CAA. CAA were associated with baseline zMax ≥2.0 vs. <2.0 (12 [16%] vs. 3 [2%], respectively, p<0.001) and as a covariate in logistic regression (Table 1). Conclusions: With the exception of the Harada score, established RS were ineffective at predicting the development of CAA at a cosmopolitan center. Baseline z scores were highly associated with CAA. However, adding baseline z scores to a logistic regression model did not improve discrimination of the RS.