Introduction: Thromboprophylaxis in patients with coronary artery aneurysms (CAAs) after Kawasaki Disease (KD) is managed according to their risk of thrombosis. The intensity of management for a given patient is directly related to the maximum CAA z-score, calculated based on body surface area using various equations. It remains unclear whether using different equations results in clinically meaningful differences in calculated risk.

Methods: CAA z-scores were calculated for all echocardiograms from patients (N=1,650) with CAAs enrolled in the International Kawasaki Disease Registry (IKDR) using 9 sets of CAA z-score equations derived from various cohorts of healthy and febrile children. Rank plots, correlation matrices, and disagreement matrices visualized the risk concordance between equations. Hazard ratios for thrombosis were used to assess the effect of using different equations on the risk of outcomes.

Results: Z-scores from multiple equations were generated for 10,147 RCA and 9,781 LAD measurements. While the statistical correlation between the equations was high (r2 > 0.85), there was substantial disagreement in the resulting AHA risk stratification when different equations were used. Disagreement was highest in the LAD (vs. RCA) and most pronounced between equations including exponential/logarithmic terms vs. those with only linear parameters. Higher LAD z-scores were equally predictive of coronary artery thrombosis (AUC 0.90-0.91, p<0.001) as were higher RCA z-score (AUC 0.85-0.89, p<0.001) regardless of the equation, however the z-score thresholds at which the statistical odds of thrombosis exceeded 1 was substantially different between the various equations: Olivieri z>5.2, Yu z>6.8, Kobayashi >6.8, Kurotobi >7.9, de Zorzi >10.8, Dallaire >12.0, Colan >13.7, McCrindle >13.9 and Lopez >15.4.

Discussion: While effective in indicating increased risk of thrombosis, z-score equations are generally not concordant with one another, nor do they provide a completely accurate prediction for a KD patient’s risk of thrombosis. The z-score threshold for intensive thromboprophylaxis in these patients should be indexed to specific z-score equations as opposed to being universally applied to all units of measurement.