Background: Kawasaki disease (KD) is an acute febrile vasculitis, which primarily affects children. The etiology of KD is unknown; while certain characteristics of the disease suggest an infectious origin, genetic or environmental factors may also be important. Seasonal patterns of KD incidence are well documented, but it is unclear whether these patterns are caused by changes in climate or by other unknown seasonal effects.
Methods: The relationship between KD incidence and deviations from expected temperature and precipitation were analyzed using KD incidence data from Japanese nationwide epidemiologic surveys (1991-2004) and climate data from 136 weather stations of the Japan Meteorological Agency. Seven separate Poisson-distributed generalized linear regression models were run to examine the effects of temperature and precipitation on KD incidence in the same month as KD onset and the previous 1, 2, 3, 4, 5 and 6 months, controlling for geography as well as seasonal and long-term trends in KD incidence.
Results: KD incidence was negatively associated with temperature in the previous 2, 3, 4 and 5 months and positively associated with precipitation in the previous 1 and 2 months. The model that best predicted variations in KD incidence used climate data from the previous 2 months. An increase in total monthly precipitation by 100 mm was associated with increased KD incidence (rate ratio [RR] 1.012, 95% confidence interval [CI]: 1.005-1.019), and an increase of monthly mean temperature by 1°C was associated with decreased KD incidence (RR 0.984, 95% CI: 0.978-0.990).
Conclusions: KD incidence was significantly affected by temperature and precipitation in previous months independent of other unknown seasonal factors. Climate data from the previous 2 months best predicted the variations in KD incidence. Although fairly minor, the effect of temperature and precipitation independent of season may provide additional clues to the etiology of KD.