Performance of Kawasaki Disease Risk Scores in a 30-Year Regional Spanish Cohort.

Several risk scores have been developed to predict intravenous immunoglobulin (IVIG) resistance and coronary artery involvement in Kawasaki disease (KD), but their performance outside Asian populations remains uncertain. To evaluate the diagnostic performance of major KD risk scores in predicting IVIG resistance and coronary artery involvement in a regional cohort from Northern Spain. Retrospective observational study including 81 KD patients diagnosed at a tertiary pediatric center (1994–2024). We applied the following IVIG-resistance scores: Kobayashi, Egami, Sano, Kawanet, Kawanet-echo, Kawa-Race, and the Spanish Pediatric Association (AEP) consensus. For coronary involvement prediction, the Son score and the Kawa-Race coronary score were used. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated, restricting IVIG-resistance analyses to patients who received IVIG as initial therapy. An exploratory stratified analysis according to therapeutic era (pre-2017 vs. ≥ 2017), reflecting the progressive introduction of adjunctive corticosteroids, was performed. IVIG resistance occurred in 7/81 (8.6%) and coronary artery involvement (dilatation or aneurysm) in 23/81 (28.4%). Most IVIG-resistance scores showed low sensitivity but high negative predictive value, while the AEP criteria achieved maximal sensitivity at the expense of very low specificity. Coronary involvement frequently occurred in IVIG-responsive patients (20/23, 87.0%). For coronary prediction, the Son score at the ≥ 2-point threshold showed the most balanced performance (sensitivity 34.8%, specificity 93.1%), whereas the Kawa-Race coronary score demonstrated high specificity (94.8%) but very low sensitivity (13.0%). Stratified analysis by therapeutic era showed persistently low positive predictive values and heterogeneous sensitivity across IVIG-resistance scores, with no clear improvement in discriminative performance after the introduction of adjunctive corticosteroid therapy. KD risk scores showed limited sensitivity for identifying children at risk of coronary artery involvement in this Western cohort and this limited performance remained consistent across therapeutic eras, including after the progressive introduction of adjunctive corticosteroid therapy. The marked discordance between IVIG resistance and coronary outcomes suggests that refractoriness-based prediction alone may be insufficient to guide early treatment intensification. A sequential strategy integrating early clinical assessment with coronary-focused prediction tools may support individualized management. Prospective multicenter validation is warranted.