Title: Use Bayesian conditional logistic regression to build up a localized medical-meteorological index
Authors: Charlotte Wang - Tamkang University (Taiwan) [presenting]
Abstract: The environmental factors like meteorological factors and air pollutants have been recognized as important factors for human health, where mortality and morbidity of certain diseases may be related to the abrupt climate change or the air pollutant concentration. We use the National Health Insurance database and define the people aged 51-90 years who were free from cerebrovascular disease (ICD9: 430-438) or ischemic heart disease (ICD9: 410-414) in 1996-2002 as the susceptible group. We then adopted the case-crossover study design and used a Bayesian conditional logistic regression to predict personal risks for suffering cerebrovascular diseases or ischemic heart diseases via environmental factors. Based on the predicted odds ratios, we defined the interval of the alert for the disease risks and evaluate the performance of the interval of the alert for the disease risks. We also explored the association between meteorological factors and two diseases in six areas in Taiwan. The results show that people living in different areas of Taiwan have different risk levels of two diseases and the intervals of the alert for the disease risks vary in six areas. In addition, health risk index, provided by this personal risk prediction model, can be a reference for weather bureaus to issue health warnings in the future. With the early warnings, the susceptible group will be able to prevent from suffering the diseases when meteorological conditions change.