| 摘要: | 本研究結合三軍總醫院2013至2023年的癌症登記資料與2016至2024年間衛星遙測推估的PM2.5 資料，針對肺和氣管、乳房、大腸直腸三類癌症，探討病患於確診後至復發或死亡期間，暴露於不同空氣污染來源（人為污染AP、生質燃燒BB、沙塵DS）與生活習慣（吸菸、飲酒、嚼檳榔）對癌症結局（復發/無復發、死亡/存活）的影響。分析方法包含中位數檢定、log-rank test 與 Cox 比例風險模型。
結果顯示，在中位數檢定與 log-rank test 中，多數空污來源與癌症結局之間具顯著差異；但進入Cox模型後，顯著風險來源通常僅限於其中一、二項，顯示在控制個人條件與生活習慣後，比較不同癌症與結局發現，復發結局與死亡結局所受影響的污染來源都包含BB(生質燃燒來源)，對於死亡結局都額外包含了DS(沙塵來源)，且復發結局與死亡結局都不包含AP(人為來源)。反映不同污染機轉可能於病程不同階段發揮作用。此外，飲酒在三類癌症中皆與復發風險呈顯著正相關，而吸菸與嚼檳榔則在特定癌別中具風險意義。
本研究嘗試結合衛星遙測資料所區分的PM2.5來源與醫院癌症資料進行預後分析，作為初步探索，期望能為後續針對多源空污與生活習慣整合分析的研究提供經驗參考與方法上的試探，進而協助建立更完善的研究架構與實證基礎。;This study integrated cancer registry data from the Tri-Service General Hospital (2013–2023) with satellite-derived PM2.5 source estimation data (2016–2024) to investigate the effects of different air pollution sources and lifestyle behaviors on cancer outcomes among patients diagnosed with lung and trachea, breast, and colorectal cancers. Specifically, it examined post-diagnosis exposure to three PM2.5 sources—Anthropogenic (AP), Biomass Burning (BB), and Dust (DS)—as well as personal habits including smoking, alcohol consumption, and betel nut chewing, in relation to cancer recurrence and mortality. Statistical methods included median tests, log-rank tests, and Cox proportional hazards models.
The results indicate that in both median and log-rank tests, most air pollution sources showed significant differences in relation to cancer outcomes. However, when adjusted for personal and behavioral factors in the Cox model, only one or two sources typically remained statistically significant. Specifically, among the three cancer types and two outcome types, BB (biomass burning) was consistently associated with both recurrence and mortality, while DS (dust storms) was additionally associated with mortality. AP (anthropogenic pollution) was not significantly associated with either outcome. This suggests that different pollution sources may exert influence at distinct stages of disease progression. In terms of lifestyle factors, alcohol consumption was significantly associated with increased recurrence risk across all three cancer types. Smoking and betel nut chewing were associated with increased risk in specific cancer types.
This study represents a preliminary attempt to combine satellite-based PM2.5 source classification with hospital cancer data for prognosis analysis. It aims to provide methodological insights and serve as a reference for future research exploring the combined effects of multi-source air pollution and lifestyle behaviors, ultimately contributing to the development of more comprehensive research frameworks and evidence-based approaches. |
