The Incidence of Kawasaki Disease in the United States Did Not Increase Between 1988 and 1997

To the Editor.—In my study1 using the National Inpatient Sample (NIS) database to calculate the incidence of Kawasaki disease (KD) in the United States between 1988 and 1997, I did not use the recommended weighting technique to account for the differences in stratified sampling across different years. As a result, miscalculated KD incidences were presented in Fig 2 of the article,1 and an erroneous conclusion was made that KD incidence was rising over the study period. Dr Holman and colleagues2 used the same NIS data and repeated the calculations using weighting techniques. The results from Holman et al did not show a clear trend of increasing KD incidence between 1988 and 1997. Because of the concern that future researchers conducting a literature search may not be able to cross-link between my article and Dr Holman’s, I feel I must emphasize once more that a mistake occurred in my study. The incidence of KD between 1988 and 1997, calculated from the NIS data, did not increase.Although there is no KD active surveillance system in the United States, many researchers rely on hospital discharge data to study KD epidemiology. It is important to recognize the limitations and pitfalls of using administrative data to conduct epidemiology research. In another study investigating the KD epidemiology in California,3 I found that the KD incidence in California increased by 30% in 1997 when compared with 1995. Because the data used in that study were based on a statewide discharge database, there was no need to account for stratified sampling. In the study of California,3 the KD incidence rate decreased in 1999 after peaking in 1997. The KD incidence in 1997 from Dr Holman’s results also seemed to be higher than previous years.2 Whether the higher incidence in 1997 was within the year-to-year fluctuation or if it represents an outbreak of KD in 1997 requires additional investigation.I thank Dr Holman and colleagues for providing the results of weighted calculations and wish to alert the readers of Pediatrics that KD incidence between 1988 and 1997 in the NIS data did not increase.