Article Text

Download PDFPDF

230 Establishing a risk stratification scoring system for ivig resistance in kawasaki disease in Hong Kong
  1. Chi Tak Enoch So,
  2. Yan Wah Cheng,
  3. Man Ching Yam,
  4. Lai Ling Hui,
  5. Albert Martin Li
  1. Hong Kong


Background Kawasaki disease (KD) is an acute febrile multi-organ vasculitic syndrome primarily affecting infants and children with potentially devastating cardiovascular sequelae. The mainstay treatment for KD is intravenous immunoglobulin (IVIG).

Regrettably, IVIG resistance occurs in a sizeable portion of patients. International guidelines advocate primary intensification therapy with adjunctive steroids, in addition to IVIG, as initial treatment for this sub-group of KD.

In order to enable the aforementioned potential, however, there is a need to establish a validated methodology which segregates high and low risk IVIG-resistance patients in Hong Kong.

Objectives Our primary objective is to determine the applicability of the current international risk scores in predicting IVIG resistance in Hong Kong Chinese.

Our secondary objective is to design a risk stratification scoring system for use in our local population.

Methods A 5-year retrospective case control (IVIG resistance vs. responsive) study (2015–2019) of patients aged 0–18 years diagnosed with KD from two tertiary hospitals was performed. Patients in which IVIG was never administered, or administered at another hospital, were excluded. Resistance cases were defined as relapse or persistent fever with temperature ≥ 38.0°C at least 36 hours after the end of first IVIG infusion. The remaining subjects were defined as responsive and grouped as controls.

Applying established international risk scores (Kobayashi, Egami, Sano, Harada, Fu) in our sample generated sensitivity, specificity, positive predictive value (PPV) and negative predictive values (NPV) of each score for comparison. Modification of Kobayashi and Sano scores were necessary by substituting AST (aspartate transaminase) with ALT (alanine transaminase) due to lack of data in our population.

In composing a clinical risk scoring system, multi-variate logistic regression model was constructed using variables selected by univariate analysis. Demographic variables and variables with p < 0.1 in univariate analysis were included. Continuous variables were converted to dichotomous variable based on 25th or 75th centile value. Backward elimination based on likelihood ratio was performed. A p value of > 0.1 was required for elimination.

Results This study included 210 patients, of which 39 patients (18.6%) were IVIG resistant. The median age was 21 months old (IQR 10.8 – 40.3), with 61.0% male patients. Among all included risk scores, modified Kobayashi score performed best with sensitivity (53.8%), specificity (74.9%), PPV (32.8%) and NPV (87.7%).

Our preliminary analysis offered an alternative 4-variable model, code named HK-KD score v1, based on gender, neutrophil percentage, urea, and globulin. Using an aggregate score cutoff of ≥ 6 in defining high-risk IVIG resistant patients, HK-KD score v1 yielded sensitivity (66.7%), specificity (78.4%), PPV (41.3%), and NPV (91.2%). Higher aggregate score of HK-KD score v1 is suggestive of a higher chance of IVIG resistance.

Conclusions Modified Kobayashi score performed best among the established international risk scores for IVIG resistance with good specificity but fair sensitivity in our population. We have designed a preliminary novel clinical risk score for IVIG resistance, HK-KD score v1, with sufficient sensitivity and specificity for potential clinical use.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.