Construction of a risk prediction model for intravenous immunoglobulin non-reactive type Kawasaki disease based on peripheral blood systemic immune-inflammation index
Liu Junyan, Zhang Lihua, Xin Yi, et al
1.Department of Laboratory, Hengshui Hospital of Traditional Chinese Medicine, Hengshui, Hebei 053000, China; 2.Second Clinical College of Binzhou Medical University, Binzhou, Shandong 056603, China
Objective To analyze the risk factors of intravenous immunoglobulin (IVIG) non-reactive type Kawasaki disease (KD) based on peripheral blood systemic immune-inflammation index (SII), and construct a
nomogram prediction model. Methods A prospective study was conducted. Two hundred children with KD
admitted to Hengshui Hospital of Traditional Chinese Medicine from January 2022 to January 2024 were
selected. According to whether there was IVIG non-reactive type KD, 49 children with IVIG non-reactive
type KD were set as the observation group. and 151 children with non-IVIG non-reactive type KD were set
as the control group. The total sample was divided into modeling sets (n=147) and verification sets (n=53)
in a 7∶3 ratio. The SII, fever duration C-reactive protein (CRP), N-terminal pro-B-type natriuretic peptide
(NT-proBNP) and general data of children in the observation group and the control group were compared.
Multivariate Logistic regression was used to analyze the risk factors for the occurrence of IVIG non-reactive type KD. A nomogram prediction model was constructed through R 3.6.3 software. Internal validation was conducted using the Bootstrap method, and external validation was performed using the validation set. The receiver operating characteristic (ROC) curve and decision curve analysis (DCA) were used to analyze the predictive efficacy of SII for IVIG non-reactive type KD. Statistical analyses were performed using independent sample t-test and χ2test. Results The fever duration, CRP, NT-proBNP and SII in the observation group were all higher than those in the control group [(5.01±0.98) d vs (4.01±0.92) d, (28.66±6.91) mg/L vs (20.46±4.16) mg/L, (1 782.62±128.66) ng/L vs (998.62±82.64) ng/L, 619.52±52.66 vs 452.62±30.07], while the albumin (ALB) level was lower than that in the control group [(33.26±1.45) g/L vs (41.62±2.84) g/L], and the differences were statistically significant (t values were 6.506, 10.038, 49.752, 27.555, 19.763, respectively, all P<0.05). Multivariate Logistic regression analysis showed that SII (OR=2.137, 95% CI: 1.052-4.482), fever duration (OR=1.979, 95% CI: 1.130-3.433), CRP (OR=2.657, 95% CI: 1.288-3.581), and NT-proBNP (OR=3.957, 95% CI: 1.384-5.270) were risk factors for the occurrence of IVIG non-reactive type KD, ALB (OR=0.476, 95% CI: 0.257-0.912) was the protective factor (all P<0.05). The external validation C-index of the SII+fever duration+CRP+ALB+NT-proBNP prediction model was 0.854 (95% CI: 0.781-0.923), the predicted risk of occurrence was in good agreement with the actual situation, and the Brier score was 0.142. The C-index was 0.914, indicating that the SII+fever duration+CRP+ALB+NT-proBNP prediction model had a better discriminatory ability in predicting the occurrence of IVIG non-reactive type KD. The area under the curve (AUC) of the SII+fever duration +CRP+ALB+NT-proBNP prediction model was 0.914 (95% CI: 0.838-0.940), with high discriminative ability and predictive value. ROC curve analysis showed that the AUC of SII for predicting IVIG non-reactive type KD was 0.859 (95% CI: 0.813-0.947), with a sensitivity of 92.84%, and a specificity of 82.26%. Conclusions The abnormal increase of SII is closely related to the occurrence of IVIG non-reactive type KD, and SII has a high predictive efficacy for IVIG non-reactive type KD.
[J]. 发育医学电子杂志, 2026, 14(5): 420-427.
Liu Junyan, Zhang Lihua, Xin Yi, et al.
Construction of a risk prediction model for intravenous immunoglobulin non-reactive type Kawasaki disease based on peripheral blood systemic immune-inflammation index
. Journal of Developmental Medicine(Electronic Version), 2026, 14(5): 420-427.
2026, Vol. 14 
