Data source
This retrospective cohort study used data from the Diagnosis Procedure Combination database.4 All 82 academic hospitals are obliged to participate in the database. Data of approximately 8 million hospitalized patients of all ages are collected annually, which is equivalent to approximately 50% of the total acute-care hospitalizations in Japan. The database includes the following information: unique identifiers of hospitals, patient characteristics, diagnosis and comorbidities at admission, and complications after admission recorded with text data in Japanese and International Classification of Diseases, Tenth Revision (ICD-10) codes. Additionally, the database contains information regarding medical treatments (including administration of drug, use of devices, and surgical and nonsurgical procedures) based on Japanese original codes, length of stay, discharge status, and medical costs during hospitalization. For the accuracy of the recorded diagnoses, the physicians in charge are obliged to record the diagnoses with reference to medical charts. Licensed medical information managers and trained medical clerks accurately record all major and minor procedures and drugs and devices used. Because the entry of accurate data is mandatory for reimbursement of medical costs, hospitals have a strong incentive to comply with these rules. The present study was approved by the Institutional Review Board of The University of Tokyo (approval number: 3501-(3); 25 December 2017). The requirement for informed consents was waived because of the anonymous nature of the data.
Participants
We used the database to identify patients who were diagnosed with KD (ICD-10 code: M30.3) between July 2010 and March 2020. We included patients with a history of receiving at least 1 g/kg IVIG and aspirin. We reviewed the Japanese text describing the detailed diagnoses for each patient to include atypical KD patients and exclude patients with a “suspected” diagnosis of KD. To exclude recurrence of Kawasaki disease, readmissions after 6 months from the initial hospitalization were excluded from the analysis. We also excluded patients aged > 6 years because the appropriate IVIG dose for older patients with KD remains controversial.5 Furthermore, we excluded patients who weighed < 3 kg or had missing data. We classified the study population according to the IVIG dose at the initial treatment: low dose group, IVIG < 1.9 g/kg; standard dose group, >=1.9 ~ < 2.1 g/kg; and high dose group, ≥ 2.1 g/kg.
Outcomes
The primary outcome was the occurrence of coronary artery abnormalities (CAAs) upon discharge. CAAs were identified with recorded diagnosis of CAAs (ICD-10 code: I25.4) and/or text data of CAAs in the Japanese language. The secondary outcomes were the proportion of IVIG resistance, duration of hospital stay, and medical costs. IVIG resistance was defined as the use of IVIG at a total dose of ≥ 4.0 g/kg and/or combination of any steroid, infliximab, cyclosporine, and/or plasma exchange that were not given during the initial IVIG treatment.
Covariates
The patient characteristics included sex, age, weight, height, hospital days of illness at the initial IVIG treatment, hospital type, complex chronic conditions,6 Japan Coma Scale at admission, transportation by ambulance, transportation from other hospitals, fiscal year, and hospital volume. In Japan, all medical expenses for patients in this age group are covered by public funds, so there is no possibility of changing the dosage of IVIG for financial reasons. For the hospital type, academic hospital was defined as university hospital and related hospitals. A previous study confirmed the correlation between the Japan Coma Scale and the Glasgow Coma Scale.7 Hospital volume was defined as the number of new KD patients at each hospital annually. The patients were divided into tertiles according to hospital volume to equalize the number of patients in each group.
Statistical analysis
Categorical variables are presented as numbers and percentages and were compared using the Fisher’s exact test. Continuous variables are presented as means and standard deviations (SD) or medians and interquartile ranges (IQR). Non-normally distributed variables (length of stay and medical costs) were compared among three groups using the Kruskal-Wallis test. In the sensitivity analysis for CAA, CAAs were defined as a diagnosis of CAAs plus the use of anticoagulants such as warfarin or clopidogrel, or a diagnosis of CAAs by cardiac catheterization.
Restricted cubic spline functions
Studies regarding the relationship between the IVIG dose and outcomes converted continuous measures of IVIG into categorical variables. However, this might result in loss of informations and decrease in statistical power.8 Therefore, we used multivariable regression models with restricted cubic spline functions to assess the potential nonlinear association of IVIG dose with CAA, IVIG resistance, length of stay, and medical cost. Multivariable logistic regression models were applied for CAA and IVIG resistance, and multiple regression models were applied for length of stay and medical cost with adjustment for age, sex, body weight, body height, hospital days of illness at the initial IVIG treatment, type of hospital, complex chronic conditions, transportation by ambulance, Japan Coma Scale at admission, transportation from other hospital, fiscal year, and hospital volume. We created restricted cubic splines with five knots at pre-specified locations according to the percentiles of the distribution of IVIG doses: 1.8, 2.0, 2.2, 2.4, and 2.6 g/kg.
Tests for overall and non-linear associations were performed using the χ2 test.9 A two-sided value of p < 0.05 was considered significant. Statistical analyses were performed using the Stata software version 17 (StataCorp LP, TX, USA).