Abstract
To examine changes in peripheral blood complete blood count (CBC) parameters during acute Kawasaki disease(KD), compare immunoglobulin(IVIG)-sensitive and IVIG-resistant groups, and develop an IVIG resistance model. A retrospective review of clinical and lab data from 282 KD patients (2014-2024) was conducted. CBC parameters were collected at initial, pre-IVIG, and post-IVIG stages. The rank-sum test assessed parameter differences over time. Patients were categorized into IVIG-resistant (n = 29) and IVIG-sensitive (n = 253) groups. Univariate and multivariate logistic regression analyses identified IVIG resistance risk factors, resulting in four predictive models (A, B, C, and D) based on blood changes and clinical experience. The models' effectiveness was evaluated using receiver operating characteristic (ROC) curves, the Hosmer-Lemeshow test, and decision curve analysis, with the bootstrap(BS) method confirming performance. Significant differences were found in post-IVIG blood parameters, including white blood cell count (WBC), neutrophils, lymphocytes, eosinophils, hemoglobin, platelets, neutrophil to lymphocyte ratio (NLR), platelet to lymphocyte ratio (PLR), and mean platelet volume to lymphocyte ratio (MPVLR), compared to pre-IVIG and initial CBC (P < 0.05). In IVIG-resistant patients, NLR, PLR, MPVLR, neutrophil percentage were higher, while lymphocyte percentage was lower than in IVIG-sensitive patients (P < 0.05). The resistant group also showed smaller changes in neutrophil percentages (△N) and lymphocyte percentages (△L). Area under the curve (AUC) values for BS-ROC curves were as follows: model A: 0.758 (95% CI: 0.636-0.878), model B: 0.917 (95% CI: 0.852-0.982), model C: 0.949 (95% CI: 0.909-0.978), and model D (NLR post-IVIG administration combined with △L): 0.910 (95% CI: 0.857-0.963). Hosmer-Lemeshow test P values for all four models were > 0.05. DCA indicated clinical value for all models, especially model C. Blood routine parameters in children with KD vary over time, and IVIG administration alters these parameters. We developed and validated four prediction models for IVIG resistance in KD patients using blood routine data. This indicates that ongoing monitoring of these parameters can predict IVIG resistance and enhance patient outcomes.