51. Kawasaki Disease in Japan: A spatiotemporal evaluation of epidemiological features and linkage to wind-borne agents

Background: Kawasaki Disease (KD) is an acute systemic vasculitis that mainly affects children younger than 5 years old. Although KD has been diagnosed in over 60 countries across several continents, its incidence is the highest in East Asia, particularly in Japan. After more than five decades since its discovery and active research, the etiology of KD is yet to be elucidated. Recent studies have analyzed the association between KD and diverse environmental factors, with some advances pointing towards a relevant role of the atmospheric transport of a wind-borne agent triggering the disease. While some candidates have been proposed, the actual nature of this agent(s) is still unknown. We sought to characterize the seasonal dynamics in KD incidence for all of Japan and across its 47 prefectures, and to propose etiological candidates driving these dynamics.

Methods: Onset and admission dates of 393,376 KD patients in Japanese hospitals from 1970 to 2018 were collected from the biennial nationwide epidemiological surveys of KD in Japan. For cases during the 1979-2018 period (369,480), the prefecture of the hospital of admission was also collected. 92 weekly air samples were measured at the Noto Ground-based Research Observatory (NOTOGRO), at the tip of the Noto peninsula in the Ishikawa prefecture in Japan, from August 2014 to March 2016. These samples were subsequently analyzed for the concentrations of 57 different metals, general PM1, PM2.5 and PM10 content was quantified, and DNA was extracted and sent for metagenomic analysis to quantify abundances of bacterial and fungal clades. A combination of classical time-series methods such as modified ARIMA models and spectral analysis are used to decompose the epidemiological series data into its trend and seasonal components at different frequencies. To study association between the measured environmental variables and KD incidence, a combination of lagged correlation analyses and more complex methods such as Scale Dependent Correlation (SDC) are used.

Results: KD incidence in Japan data shows a strong increasing trend from 2000 onwards, with a marked yearly seasonal effect with maxima in January/February and minima in October/November. The amplitude of the yearly seasonal component has also been increasing together with the trend. The features observed at the national level are consistent for the data of most prefectures when stratifying the epidemiological records at the regional level. The temporal variability of several of the measured environmental factors were found to be moderately associated with the temporal changes in incidence of KD in the region, more so when considering transient couplings between the variables.

Conclusions: The coherence between the trend and seasonal patterns of KD incidence in Japan and among its individual prefectures points towards a common source potentially driving the etiology of the disease. The coherences between some of the environmental factors and the temporal changes of incidence in the disease warrant further and deeper study, ideally at higher temporal resolutions.

Disclosures. None