In this single-center cohort of infants aged < 6 months with KD, the main findings were threefold. First, pre-treatment coronary involvement and higher inflammatory burden were the factors most strongly associated with additional IVIG requirement. Second, pre-treatment coronary disease severity and early refractory course were the factors most strongly associated with CAA at 1 month. Third, the small subgroup with residual CAA at ≥ 6 months clustered among infants with more severe coronary involvement at approximately 1 month after illness onset. These findings address the study aim more directly than does the diagnostic label of complete versus incomplete KD, supporting an echocardiography-anchored approach to early risk stratification.

Because this study did not evaluate missed cases or diagnostic performance, we focused our interpretation on risk stratification after clinical KD recognition rather than on diagnosis itself. Even in this cohort, in which recognition and initial treatment occurred relatively early, 22.4% required additional IVIG and 18.4% had CAA at approximately 1 month after illness onset, underscoring the intrinsic vulnerability of this age group.

Pre-treatment coronary involvement and higher NLR were independently associated with additional IVIG, whereas fewer illness days at initial IVIG showed an inverse association with retreatment. Prior studies have also shown that abnormal baseline echocardiographic findings, particularly coronary dilation, are associated with IVIG non-response across cohorts [17,18,19,20]. For example, a multicenter European study identified baseline CAA Z-max ≥ 2.0 as an independent predictor of second-line treatment; likewise, Korean studies reported increasing IVIG non-response rates with increasing coronary disease severity [17, 19]. Therefore, our findings do not identify an entirely novel associated factor; rather, they confirm and refine the clinical relevance of early coronary involvement in a cohort restricted to infants aged < 6 months.

Pre-treatment CAA Z-score ≥ 2.5 was associated with retreatment, reinforcing the value of early coronary Z-score assessment [1,2,3,4,5]. NLR was also associated with retreatment, suggesting that coronary severity and an age-aware inflammatory marker together may help stratify infants according to retreatment risk [21,22,23,24].

The inverse association between illness day at initial IVIG and retreatment is most consistent with confounding by indication rather than a harmful effect of earlier treatment itself. In our setting, some infants underwent early evaluation because of strong clinical suspicion, evolving classical clinical features, and early echocardiographic abnormalities suggestive of KD before the classical diagnostic window had fully elapsed [1,2,3,4,5]. Prior Japanese analyses and systematic reviews likewise suggest that very early IVIG does not reliably prevent CAA and may instead reflect greater baseline severity [25, 26]. Consistently, our IPTW analyses showed persistent imbalance in severity-linked covariates among those treated before illness day 3 [27]. Taken together, these findings support the interpretation of earlier IVIG primarily as a marker of baseline severity rather than a causal timing effect, underscoring the importance of severity-informed risk stratification rather than attribution of retreatment risk to treatment timing alone.

For CAA at 1 month (Z ≥ 2.5), both the baseline-only Model A (pre-treatment CAA Z-score, albumin, illness day at initial IVIG) and the course-aware Model B (pre-treatment CAA Z-score plus additional IVIG) achieved AUCs around 0.90 with acceptable calibration. The distinction between these models is clinically important. Model A uses information available at initial assessment and is intended for early bedside risk stratification. Model B incorporates additional IVIG not as a causal exposure but as a marker of an early refractory course. Pre-treatment CAA Z-score remained independently associated in both models, whereas lower albumin showed the expected inverse tendency, consistent with prior reports linking baseline dilation and hypoalbuminemia to later coronary outcomes [9, 28, 29].

Threshold analyses further supported guideline-based practice while suggesting potentially meaningful granularity within the “small” range. Risk increased at pre-treatment CAA Z-score ≥ 2.5 and increased further at ≥ 5.0, consistent with AHA and Japanese Z-score bands [1,2,3,4,5]. At approximately 1 month after illness onset, coronary dimensions generally regressed; however, all infants with residual CAA at age ≥ 6 months had 1-month Z-max ≥ 5.0. Although the residual event rate was low (3/76), persistence of CAA at 1 month with Z-max ≥ 5.0 appears to identify a subgroup warranting particularly careful longer-term surveillance. Coronary size at diagnosis also carried information about late persistence, but its clinical performance was more modest than that of CAA at 1 month. In particular, pre-treatment Z-score ≥ 5.0 showed high specificity but limited sensitivity for residual CAA at ≥ 6 months, whereas Z-score ≥ 5.0 at 1 month more strongly enriched late persistence.

Elevated pre-treatment CAA Z-score likely reflects a heavier coronary vasculitis burden at presentation and may identify infants with more advanced vascular remodeling [9, 28,29,30,31]. This interpretation is consistent with prior pathologic and clinical studies linking baseline coronary dilation with later aneurysm persistence or progression [28, 29].

Incomplete KD remains an important diagnostic challenge in very young infants. In our cohort, however, incomplete KD was not independently associated with CAA at 1 month after adjustment for baseline coronary severity and retreatment. In a setting with a low threshold for early echocardiography and repeated clinical assessment in febrile young infants, the excess coronary risk often reported in incomplete presentations may have been attenuated [1,2,3,4,5].

This study has limitations that must be acknowledged. First, the modest cohort size (n = 76) and rare residual CAA at age ≥ 6 months (3/76) limited precision and precluded multivariable modeling for the late outcome; we therefore emphasized exact methods and prespecified thresholds. Second, the retrospective, single-center design may introduce residual confounding and limit generalizability. Third, treatment practices evolved across the long study period. Infants aged < 6 months are recognized as a particularly high-risk subgroup in contemporary guidance, and intensified initial therapy may be considered in current practice; however, our cohort reflects real-world management from 2011 to 2024, a period during which treatment strategies were not uniform and primary adjunctive corticosteroids were not mandated for all infants. Accordingly, any descriptive associations involving adjunctive therapies should be cautiously interpreted.

In conclusion, in this cohort of infants aged < 6 months with KD, proactive echocardiography enabled relatively early diagnosis and treatment, yet pre-treatment coronary involvement and simple inflammatory indices showed the strongest associations with early outcomes. Pre-treatment CAA Z-score ≥ 2.5 and higher NLR were associated with additional IVIG, whereas very early IVIG administration primarily reflected greater baseline disease severity rather than a causal treatment effect. For CAA at 1 month, pre-treatment coronary disease severity, lower albumin, and an early refractory course were the factors most strongly associated with the outcome. Together, these findings support an echocardiography-anchored, Z-score-based strategy, supplemented by age-aware inflammatory markers, to optimize risk stratification and management in very young infants with KD.