Rising incidence of Kawasaki disease in England: analysis of hospital admission dataBMJ 2002; 324 doi: http://dx.doi.org/10.1136/bmj.324.7351.1424 (Published 15 June 2002) Cite this as: BMJ 2002;324:1424
- Anthony Harnden (), university lecturera,
- Bernadette Alves, health services researcherb,
- Aziz Sheikh, NHS research and development national primary care training fellowb
- a Department of Primary Health Care, Institute of Health Sciences, University of Oxford, Oxford OX3 7LF
- b Department of Primary Health Care, Imperial College of Science, Technology and Medicine, London W2 1PG
- Correspondence to: A Harnden
- Accepted 6 December 2001
Kawasaki disease is the leading cause of acquired heart disease in children in the developed world and may be a risk factor for adult ischaemic heart disease.1 A fifth of untreated children develop cardiac lesions during the acute phase of the disease. The cause remains uncertain. Epidemiological studies support an infectious agent inducing the disease in a genetically susceptible minority. Superantigen toxins have been implicated. Reported incidence rates differ considerably throughout the developed world with rates in Japan 10 times those in the United States and 30 times those in the United Kingdom and Australia.2–4 Hospital surveillance data suggest the incidence of Kawasaki disease in Japan has risen by over 50% between 1987 and 1998.2 To ascertain whether there had been a similar rise in England, we investigated trends in hospital admissions for Kawasaki disease using routinely collected statistics from 1991 to 2000.
Methods and results
The hospital episode statistics database provides information on every inpatient admission to English NHS hospitals. Coding accuracy and reproducibility are better for acute conditions than for chronic disorders.5
We examined all emergency inpatient admissions for children younger than 17 years primarily diagnosed as having Kawasaki disease between 1 April 1991 and 31 March 2000. We excluded 293 interhospital transfers to avoid duplication. Coding of diagnoses with ICD-10 (International Classification of Diseases, 10th revision) began in 1995. We analysed admissions coded with ICD-9 code 446.1 and ICD-10 code M30.3 separately and together. We derived overall admission rates specific to age from the 1999 estimates of the national midyear population. With monthly admissions as the dependent variable and the number of months from 1 January 1991 as the independent variable, we used linear regression to look at time trends over the study period.
During the nine years there were 2215 emergency admissions in children with a primary diagnosis of Kawasaki disease, 666 in the four year ICD-9 coded period and 1549 in the five year ICD-10 period. Median age at admission was 2 (interquartile range 1-4) years. Almost two thirds (61%) of children admitted were boys; this proportion was consistent in all ages and across the study period. Annual admissions increased from 143 in 1991-2 to 308 in 1999-2000.
The incidence per 100 000 children younger than 5 years doubled from 4.0 per 100 000 (95% confidence interval 3.4 to 4.8) in 1991-2 to 8.1 per 100 000 (7.1 to 9.2) in 1999-2000. The linear model (figure) provided evidence of a significant increase in monthly admissions: 0.15 (0.05 to 0.25; P<0.001) extra admissions per month during ICD-9, 0.20 (0.07 to 0.32; P<0.001) during ICD-10, and 0.21 (0.17 to 0.25, P<0.001) in the two periods combined.
Hospital admissions for Kawasaki disease increased in England between 1991 and 2000. It is possible that this reflects an increase in recognition rather than incidence. After the demonstration in 1991 that in the United Kingdom only 7% of children with Kawasaki disease received optimum immunoglobulin treatment, awareness of the importance of early diagnosis has been heightened.3
The incidence of Kawasaki disease among English children may have truly risen as it has in Japan. Explanations include a change in the infecting agent or a shift in susceptibility among young children. We recommend improved surveillance of Kawasaki disease to record further rises in incidence, to examine seasonal and geographical variations, and to report any changes in the pattern of clinical severity or the rate of complications.
We thank Tim Lancaster and David Mant for comments on earlier drafts of this paper and Adrian Cook for help with the data analysis.
Contributors: AH and AS had the idea for this paper. All three authors devised the study. BA analysed the data. AH drafted the paper and BA and AS commented on the text. AH is guarantor.
Funding AS is supported by a NHS research and development national primary care training fellowship.
Conflict of interests None declared.