103 - KAWASAKI DISEASE: A NEW APPROACH TO RISK STRATIFICATION IN THE POMERANIAN REGION IN POLAND

Background:

It is well-known that genetic, ethnic, and environmental factors significantly influence the course of Kawasaki disease (KD) and the risk of coronary complications. To date, studies have not provided evidence that tools such as the Kobayashi score, Sano score, or Egami score can be used for risk stratification of IVIG resistance and the development of coronary artery aneurysms in non-Asian populations. Our objective was to create a simple method for assessing the risk of coronary artery aneurysms.

Methods:

The study included 80 patients diagnosed with KD between 1997 and 2021. A retrospective analysis of the acute phase of the disease was conducted, focusing on fever duration, symptoms, and abnormalities observed in transthoracic echocardiography (TTE) and electrocardiography (ECG). After the acute phase, an assessment of coronary artery changes was conducted. All measurements were normalized using Z-scores. Coronary aneurysms were defined as Z-scores ≥ +2.5. Multivariable logistic regression analysis was used to identify independent risk factors for coronary artery lesions (CAL).

Results:

Seventeen (21.3%) out of 80 patients had CAL according to the diagnostic criteria. Multivariate logistic regression analysis identified that the duration of fever (OR = 1.24; 95% CI, 0.07-0.36; P = 0.005), age < 6 months (OR = 12.53; 95% CI, 0.16-4.9; P = 0.037), the number of additional atypical symptoms of KD (OR = 1.71; 95% CI, 0.04-1.04; P = 0.036), and ST-segment changes on ECG (OR = 17.97; 95% CI, 0.67-5.1; P = 0.011) were independent risk factors for CAL. These four variables were used to generate a simple scoring model that yielded an area under the receiver-operating-characteristics curve of 0.94. For a cut-off of 2 points or more, sensitivity and specificity were 94.1% and 81.0% in the designed model. Six points or more were associated with an 87.5% risk of CAL (95% CI, 0.58-1.2).

Conclusions:

Our simple clinical-based KD predictive model demonstrated high sensitivity and specificity in identifying patients at high risk of CAL.