Abstract
Weather watch/warning systems have been established for human health outcomes. Our study aims to develop and demonstrate a weather watch/warning system for asthma and stroke within the whole of South Korea, using a stratified regression approach. We converted claim-based health insurance data covering almost all medical claims for the only health insurance system in Korea for asthma and stroke from 1996–2003 into personalized disease episode data, and combined them with meteorological data. We utilized a step-wise regression method using factors extracted from the meteorological data to develop stratified models for six (stroke) and nine (asthma) regional and day-of-week strata. Validation studies showed that the actual number of hospitalizations in 2003 increased according to the three-leveled predictions (levels I, II, and III) from the model based on the 1996–2002 data. This system is accessible via the internet (http://industry.kma.go.kr/APP/sub_APP15_H01.htm) at the Korean Meteorological Administration website.
Acknowledgements
This study was supported by the Korean Meteorological Administration through the ‘Development of the weather watch/warning system for health events’ project. Ho Kim was partially supported by the Korean Ministry of the Environment, Republic of Korea (Eco-technopia 2005, 091-052-036).