目標：本研究探討台灣七大空品區內，相對極端氣溫的天氣型態對民眾健康之影響。各地區有不同的相對極端氣溫門檻值，而健康評估目標則以需住院治療的腦血管疾病與缺血性心臟病新發生病例個數作為反應變數。方法：分別針對312個行政區，定義2003-2010年間每日最低氣溫的第5百分位及最高氣溫第95百分位，為各行政區的極端溫度門檻值，再將每日天氣型態分類為相對極端高溫、相對極端低溫、或正常氣溫之天氣類型。以各空品區中老年人每日新發生病例個數為目標，利用負二項分配描述發生個數的多變異（extra-Poisson variation）特質，考慮同一地區每日觀察值之間相關性，納入空氣污染因子及四季等變數，以廣義線性模式進行分析。結果：居住於不同空品區的居民有顯著不同的發病風險（p＜0.05）。在考慮各地區不同氣溫門檻值下，就腦血管疾病而言，相對極端高溫的天氣型態會增加各空品區的新發病風險；若是夏天的相對極端高溫的天氣型態，則中部、高屏、宜蘭、花東空品區增加的新發病風險將分別由5%、16%、16%、52%增至6%、17%、17%、54%；若是發生於冬天的相對極端高溫的天氣型態，高屏、宜蘭、花東空品區的新發病風險也會增加（以上p值皆＜ 0.05）。就缺血性心臟病而言，各空品區的居民雖仍有顯著不同的發病風險（p ＜ 0.05），但相對極端高溫的天氣型態只顯著影響雲嘉南及高屏空品區易感受族群的新發病風險（風險增加21%與11%，p ＜ 0.05）；相對極端低溫的天氣型態只顯著影響宜蘭空品區易感受族群的新發病風險.，尤其是冬天的低溫使得風險增加22%（p ＜ 0.05）。結論：極端高溫之預警已逐漸受到國際重視，而評估發病風險應考慮不同地區的氣溫異質性，才能針對不同地區的易感受族群，適時提出抗暑或禦寒等保護措施建議，以因應地球暖化影響下越來越多的高溫天氣型態。

Objectives: This research investigated the effects of area-specific extreme temperatures on health. The outcome variable was defined as the number of new cases of cerebrovascular disease and ischemic heart disease in each of seven regions comprising Taiwan. Methods: For each of the 312 administrative areas, the extreme temperature thresholds were defined as the 5th percentile of daily minimum temperatures and 95th percentile of daily maximum temperatures in 2003-2010, to account for the heterogeneity between areas. Each day was then categorized as experiencing unusually high temperature, unusually low temperature, or normal temperature weather. A generalized linear model was adopted to model the daily number of new cases of disease in each region, with a negative binomial link function for extra-Poisson variation, along with covariates including air pollution and season. Results: Residents in different areas experienced significantly different risk for both of the diseases considered (p ＜ 0.05). Under different threshold values of extreme temperature for the seven regions, extremely-high temperature days were associated with a higher risk of cerebrovascular disease. In summer, extremely high temperature increased risk even more. For instance, the increase of risk for the Central region was 5% overall but 6% in summer (from 16% to 17% in the Kao-Ping region, from 16% to 17% in the Yi-Lan region, and from 52% to 54% in the Hua-Dong region). In winter, the pattern of increased risk remained for extremely-high temperature days (all p-values above ＜ 0.05). For ischemic heart disease, the seven regions also exhibited significantly different risk (p ＜ 0.05). However, extremely-high temperature days only significantly influenced risk in the Yun-Chia-Nan and Kao-Ping regions by 21% and 11%, respectively; while days with extremely-low temperature only affected the susceptible residents in the Yi-Lan region, especially in the winter (risk increased by 22%, p ＜ 0.05). Conclusions: The evaluation of the effect of air temperature on health has attracted much attention in many countries including Taiwan. To evaluate such an effect, heterogeneity in the cutoff temperature values for temperatures among regions should be taken into account. In response to extreme temperatures, in particular to heat waves due to climate change, appropriate measures should be developed to alleviate or prevent negative health consequences for susceptible residents in different regions.