Mortality Among Persons with a History of Kawasaki Disease in Japan : Existence of Cardiac Sequelae Elevated the Mortality

Objective. To clarify whether patients with Kawasaki disease have a higher death rate than the age-matched healthy population after the disease occurrence. Study design. Between July 1982 and December 1992, 52 collaborating hospitals collected data on all patients having a new definite diagnosis of Kawasaki disease. Patients were followed until December 31, 1997, or death. The expected number of deaths was calculated from Japanese vital statistics data and compared with the observed number. Results. Of 6576 patients enrolled, 25 (19 males and 6 females) died. The standardized mortality ratio (SMR) was 1.35. In spite of the high SMRs during acute phase, the mortality rate was not high after the acute phase for all patients. Although the SMR after the acute phase was 0.76 for those without cardiac sequelae, 6 males (no females) with cardiac sequelae died during this period and the SMR for the male group with cardiac sequelae was 2.77. Conclusion. The mortality rate among males with cardiac sequelae due to Kawasaki disease was 2.8 times as high as in general population, whereas mortality rates for females with the sequelae and both males and females without sequelae were not elevated. J Epidemiol, 2000 ; 10 : 372-375

Since 1991, the Kawasaki disease follow-up group have been following up a cohort consisted of persons with a history of Kawasaki disease 1-3).The study question is whether the mortality among those persons is higher than that in general population.
Because Kawasaki disease as vasculitis may progress atherosclerosis rapidly, the history may be a risk factor of cardiovascular disease when the patients become adults.
Differing from long-term follow-up data from hospitals, our data are unbiased because the cohort includes all patients who fulfill the inclusion criteria from the nationwide surveys' database.The latest follow-up ended on December 31, 1994 3), and the current study prolonged the observed period till the end of 1997.

MRTHODS
The items and methods to be followed up were almost same as the previous follow-up [1][2][3].We show them briefly.

Inclusion Criteria
In foreigners.The reasons about the inclusion and exclusion criteria have been shown in the previous paper 3).Finally, we enrolled 6576 patients.

Protocol
The cohort members had been followed from the time they first came to the hospitals until December 1997, or the time of death if it occurred before that date.For all the members excluding those whose deaths were known through the previous observations, we checked resident registration records or Koseki system (permanent resident registration system) 5) in municipal offices in 1998.The registration of at least one of the two systems told us that the patient lived on or after January 1, 1998.If a patient died before the last day of 1997, the systems also provided the information about the death.In case of death, copies of death certificates were obtained to determine the causes of deaths from local officers of the District Legal Affairs Bureaus, the Ministry of Justice; official approval was obtained to use these records.

Statistical analysis.
The duration of observation was calculated according to sex, age, and calendar year for each patients.The expected number of deaths was calculated by multiplying the observation time for each patient by the death rate calculated from Japanese vital statistics data for each group defined by sex, age, and calendar year.Standardized mortality ratios, or the ratio of the observed numbers of deaths to expected ones, were observed by sex, phase of illness (acute phase within 2 months after the onset, and after then), and whether or not to have cardiac sequelae after the acute phase.A 95% confidence interval was calculated for each ratio based on Poisson distribution 6).A standardized mortality ratio of which 95% confidence interval did not include 1.0 was considered to be statistically significant.

RESULTS
The sex and age distributions of the cohort members are shown in Table 1.The distributions at the first visit were similar to the epidemiologic features of Kawasaki disease in Japan 75.On December 31, 1997, the oldest cohort member was 27 years and the youngest was 5 years of age.Before that day, 25 died.We could not certify whether 28 persons were still alive after January 1, 1998, and they were follow-up loss in the current observation.Finally, the follow-up rate was (6576-28)/6576, or 99.6%.Of these members, 1003 (15.3%) were reported to nationwide surveys as cases with cardiac sequelae; the proportion was also reasonable in comparison with the whole Kawasaki disease patients in Japan 7).
The observed person-years were 70087.7 for all members, 39951.7 for males, and 30136.0 for females; therefore, the average observed periods were 10.7 years for both sexes, 10.6 years for males, and 10.7 years for females, respectively.
Table 2 shows the numbers of deaths and standardized mortality ratios for the 5 sub-groups by sex.Among the whole cohort members, 19 male deaths and 6 female ones were observed.The standardized mortality ratio for males elevated slightly, and that for females was quite similar to the unity; both were without significance.It has been reported in the previous papers that the mortality was high during the acute phase of Kawasaki disease, but not high after the acute phase; the same results were obtained in the current study.However, the elevated mortality rate after the acute phase was observed only among those with cardiac sequelae, especially for males.Six cases with cardiac sequelae, all of whom were male patients, died after the acute phase, and the standardized mortality ratio was 2.77 with statistical significance.
Detailed information including causes of deaths about the 6 cases are shown in

DISCUSSION
This is the fourth look of the cohort consisting of all patients with Kawasaki disease who were eligible for some conditions.
The current follow-up data show that persons with cardiac sequelae due to Kawasaki disease were more likely to die than general population.This tendency was only for males; the mortality rate for males with the sequelae was 2.8 times as Table 1.Sex and age distribution of the 6576 study patients.
Table 2. Observed and expected deaths according to disease phase , sex, and existence of cardiac sequelae.** The acute phase of the disease was defined as the first two months after the onset of symptoms .
CI denotes confidence interval.
Table 3. Six death cases after 2 months from the onset of Kawasaki disease with cardiac sequelae .
KD denotes Kawasaki disease, and MI denotes myocardial infarction.
The order is according to the cause of death.
high as in general population with statistical significance.
Many pediatricians may have such impression, but this is the first time to provide epidemiologic data of the elevated mortality.
The study question of the cohort observation is whether the mortality among those with a history of Kawasaki disease was higher than in general population.Several hypotheses should be considered about the study question.One is about the cardiac sequelae due to the disease, which appear on 10-15% of the patients.This issue is revealed by the current study.Six males with cardiac sequelae died after the acute phase whereas the expected number of death was 2.2 for the group (Table 2).This means that there were 3.8 (6 minus 2.2) excess deaths.Among the 6 deaths cases, 2 were caused by coronary artery lesions due to Kawasaki disease (Nos. 1 and 2 in Table 3), and another (No. 3) was acute myocardial infarction when he was 17 years old; it may not be problematic to consider that this infarction was due to Kawasaki disease because myocardial infarction among such age groups was very rare.Although this is only a speculation because of no autopsy data was available, it is possible that the drowning case (No. 4) was caused by heart attack due to coronary lesions by Kawasaki disease.Thus, the 3.8 excess deaths were reasonable.The rest 2.2 (expected number) includes the Nos. 5 (acute pneumonia) and 6 (suicide).
The mortality rate for females with cardiac sequelae did not elevate; only for males.Kawasaki disease is more prevalent among males than females, and the proportion of patients with cardiac sequelae is higher in males 7).Therefore, the expected number of death was small among female patients with cardiac sequelae, and the high mortality rate was observed only for males in this study.The longer observation is required, however, to reveal whether females with cardiac sequelae have the high mortality rate.
Another issue is ischemic heart disease and cerebrovascular disease due to atherosclerosis induced by the systemic vasculitis in childhood.Some pathologic observations tell us that atherosclerosis progressed in autopsy cases with a history of Kawasaki disease more than those with similar age but without the history a 9).Unfortunately, because the cohort was established in 1991, the oldest person in the cohort was 27 years old on the last day of observation, and the risk of cardio-and cerebro-vascular diseases was still low.On the other hand, a recent study has shown that coronary endotherial function was impaired after 1 to 12 years from Kawasaki disease onset, even in cases with regression of coronary artery aneurysms 10).If so, the mortality is expected to become close to those without the history of Kawasaki disease according to the passage of time.Further long-term observation is required to discuss the issue.
The final issue is intravenous gamma globulin therapy and immune system.Abnormal immune system response of the disease is well known 11, 12).In addition, many Kawasaki disease patients are treated with gamma globulin recently 13) Large amount of external immunoglobulin for children may affect the premature immune system, and the incidence of immunologic disease may increase after years.Fortunately, such diseases or abnormalities are not reported to date.Two persons have died of malignant neoplasms of the lymphatic or hematopoietic tissue, but the mortality was not significantly high 3).Further observation is necessary about this issue as well.
Cause-specific mortality rates among the cohort are of the interest.We have shown them in the third observation 3), but the expected numbers of deaths were so small because of the size of the cohort that the precision was low with wide 95% confidence intervals of the SMRs.In the current study, we refrained from observing them because the observed period of time prolonged was only 3 years.However, continuing the cohort will provide such data near future.
In conclusion, the current epidemiologic observation showed that mortality among persons with cardiac sequelae due to Kawasaki disease was significantly high among males.

Table 3 .
Three persons died of coronary disease; two coronary artery lesions due to Kawasaki disease,