Incidence Rate , Cumulative Incidence , and Cohort Effect of Kawasaki Disease in Japan

To get information about secular trend of age-specific incidence rate, cumulative incidence up to 5 years of age, and whether there is a cohort effect of incidence rate of Kawasaki disease, we analyzed the data of nationwide incidence surveys of the disease in Japan from 1970 through 1990. The age-specific incidence rate increased gradually except for three nationwide epidemic years. Cumulative incidence from birth through 4 years of age also increased, and was about 0.5% for males and 0.4% for females who were born in early and middle 1980's. Although some birth cohort showed a high relative risk of incidence rate with statistical significance, there was no meaningful cohort effect on Kawasaki disease. J Epidemiol, 1994; 4 : 13-16.

Twelve nationwide incidence surveys for Kawasaki disease have been done to date in Japan and approximately 5,000 patients with the disease per year were reported in recent 10 years except epidemic years (1982,1-3) Due to the decrease of the number of children, incidence rate increased gradually from 1987 through 19903).However, chronologic change of age-specific incidence rates has not been observed in detail so far.
Cumulative incidence of Kawasaki disease is another aspect of epidemiologic interest because of its cardiac sequelae.Many pediatricians need to know the prevalence of children with a history of the disease in school child health management.Some researchers observed the prevalence of such children among school children and it was estimated that approximately from 0.3 through 0.4 percent of pupils in elementary schools had the history4.5).However, both researches were questionnaire surveys and their validity was unknown.
It is expected that the prevalence of the history of Kawasaki disease among school children is high in specific age groups which consist of those who were 0 or 1 year of age when the epidemics of Kawasaki disease were observed.It is important to know whether specific birth cohorts have high incidence rate of Kawasaki disease after adjusting for period effects.In other words, our interest is whether the disease is likely to affect those who were born in a specific year or period excluding the effects of three epidemics.If such a cohort effect6 exists, intrauterine exposure to a agent, such as a viral, bacterial, or chemical one, which was prevalent in a specific period should be discussed as a causal agent.
To observe chronologic change of age-specific incidence rate and cumulative incidence of Kawasaki disease, and to observe whether there is a cohort effect of incidence rate of Kawasaki disease, we analyzed the results of the nationwide incidence surveys of Kawasaki disease in Japan.

MATERIALS AND METHODS
We used the yearly numbers of patients with Kawasaki disease reported to the nationwide surveys of the disease conducted by the Japanese Kawasaki Disease Research Committee.Observed period of time was the 21 year period from 1970 through 1990, and observed age was from 0 through 4 years, which was classified according to year.
To observe the cohort effect of incidence of Kawasaki disease adjusted for age, sex, and period effects, we used the Poisson regression model).As independent variables in the model, dummy variables were used for age, calendar year, and birth cohort.A relative risk (RR,) of a birth cohort for Kawasaki disease compared with other birth RRl =eXP(*T) where ,3 was the estimated coefficient for the corresponding independent dummy variable of the birth cohort 1.The program used was PROC LOGISTIC of the SAS Institute9).RESULTS Age-specific incidence rates increased gradually in the observed 21 year period except for the three epidemic period, 1979, 1982, and 1985-86 (Figs.I and 2).The incidence rate in 1990 was higher than that in 1979, the first epidemic year, due to this increasing, for all age groups in both sexes.
For males, incidence rate for infants (0 year old) and that for 1 year old boys were similar in every year .
However, incidence rate for 1 year old boys was higher than that for infants in epidemic years.The incidence rates were lower in older age groups according to the increase of age.
For females, incidence rate for every age group was smaller than that for corresponding age group of males in every year.Incidence rates for infants were lower than that for 1 year old girls till mid 70's, and after the mid 70;s the incidence rates were similar in both of the two age groups.
Like males, incidence rate for 1 year old girls was higher than that for infants in epidemic years.The incidence rates for girls was also lower in older age groups according to the increase of age after 1 year old.
Cumulative incidence also increased according to the period of time (Table 1).Among the observed 17 cohorts, the birth cohort 1981, who experienced the greatest epidemic of the disease in 1982 when they were 1 year of age , had the highest cumulative incidence in both sexes , which were 0.66% and 0.51%, respectively.This cumulative inci-Figure 1. Incidence rate of Kawasaki disease for male by year and age (<5 years), from 1970 through 1990, Japan.

Calendar year
Calendar year dence of 0.66% for the birth cohort 1981 for males, which comprised those who were 0 years old in 1981, means that 0.66% among this group were affected by Kawasaki disease by the time when they became 5 years old.Cumulative incidence for males was higher than that for females in every birth cohort.In addition, those who were born in early and middle 1980's had an approximately 10 times higher probability to have the disease than those born in early1970s After adjusting for age and period effect of the incidence of Kawasaki disease, no meaningful cohort effect was observed in spite of significantly higher or lower relative risks in some birth cohorts (Table 2).

DISCUSSION
This study in Japan demonstrates that incidence rate of Kawasaki disease increased gradually in the observed 21 years, that cumulative incidence up to 5 years old also increased and was about 0.5-0.7% for males and about 0. 4-0.5% for females who were born in early and middle 1980's , and that there was no meaningful cohort effect of incidence of Kawasaki disease adjusted for age and period effects.
Since the last nationwide epidemic of Kawasaki disease in Japan in 1985-86, no epidemic of the disease has been observed to date.However, incidence rate increased and that in 1990 was higher than that in 1979, the first nationwide epidemic year in Japan.In other words, yearly incidence rate of Kawasaki disease in recent years was as high as that in the first epidemic year though a manifest epidemic was not seen.Therefore, the disease becomes more important for pediatric practitioners.
Cumulative incidence observed in this study indicates that about one of 200 children born in early and middle 1980s has a history of Kawasaki disease when they became elementary school pupils.Former Japanese studies Figure 2. Incidence rate of Kawasaki disease for female by year and age (<5 years), from 1970 through 1990, Japan.showed that 0.3-0.4% of pupils in elementary schools had the history4,5), which was a little lower than the current result.However, we can explain the difference by the yearly increase of incidence.Thus, the current results are comparable to the former ones.this purpose.If a cohort effect of incidence rate of Kawasaki disease had been observed, we should have considered intrauterine exposure to specific agents as a cause of the disease like rubella",").
Although some birth cohorts had slightly high relative risk of the disease with statistical significance, the relative risk was at most 15% higher than unity.Therefore, we considered that no meaningful cohort effect of the incidence of Kawasaki disease was observed.The sample size was so large that some birth cohorts showed statistically significant cohort effects.Accordingly, researchers should not consider intrauterine infections as the etiology of Kawasaki disease because of this epidemiologic result.
In conclusion, we observed the increase of incidence rate and cumulative incidence, and no meaningful cohort effect of Kawasaki disease in Japan using nationwide incidence survey data.