Introduction: Previous wind pattern studies have suggested that the etiologic and/or triggering agent for Kawasaki disease (KD) originates from Central Asia and that westward tropospheric winds are associated with the incidence pattern in Japan. This hypothesis has not been formally tested elsewhere. Should it be correct, it follows that incidence patterns should be similarly associated with North/North-West winds in Canada.
Methods: We queried the Canadian hospital discharge database (Canadian Institute for Health Information) for hospital admissions associated with a discharge diagnosis of KD from 2004 through 2011; only acute admissions were included. Hourly weather data, including wind direction, was obtained from 9 weather stations across Canada situated closest to clusters of cases. The number of new KD admissions per day and per weather station was calculated. Wind direction was divided in 8 different quadrants covering 45° regions starting with 0°. A Poisson regression model for count data, adjusted for repeated measures (9 measures per day) was used to determine whether wind patterns were associated with incident cases. A number of separate models were tested using theoretical incubation periods of 0, 7, 14 or 21 days.
Results: A total of 3363 KD cases were assigned to one of 26,283 time/space periods for which wind direction was available. Wind direction was not associated with incidence patterns in any of the 4 incubation models: incubation time 0 day (global wind pattern effect chi-square: 4.28, p=0.75). 7 days (chi-square: 5.08, p=0.65), 14 days (chi-square: 7.32, p=0.29) or 21 days (chi-square: 4.74, p=0.69). Focusing on North/North-West wind patterns also failed to show a significant association in all models: incubation time 0 day (+0.0021 (0.0151) cases/day, p=0.89), 7 days (-0.0071 (0.0142) cases/day, p=0.62), 14 days (+0.0009 (0.0148) cases/day, p=0.95) or 21 days (+0.0073 (0.0156) cases/day, p=0.64). Results were similar when stratifying Eastern vs. Western Canada and when each weather station was considered separately.
Conclusions: Wind patterns were not found to be associated with incident KD cases in Canada. It is unclear if the reported association is specific to Japan or an ecologic fallacy.