台灣災難流行病學研究(2009 ~ 2016)

Abstract

研究背景 近年來台灣發生的災難(包含自然與人為技術性災難)伴隨著大量傷患與嚴重社會衝擊，引起社會高度重視系統性的防災準備與醫療應變。雖然現在台灣已有各式防災應變計畫，仍然有許多可以改進的空間。從過往的事件，仍有需多經驗需要發掘。過去有許多研究著墨於單一災難事件，但仍缺乏系統性地檢視災難事件的特性與醫療衝擊。在本研究，以文獻回顧與實證資料來釐清台灣當今的災難流行病學。 研究目的 本研究目的： (1) 對於災難事件發生率與帶來的醫療衝擊提供基本描述 (2) 闡明不同類型災難發生的時間趨勢 (3) 整合個人因子、災難特徵與地區因子，以建立風險預估模型，提升對於災難傷患產生的危險因子認識。 (4) 根據(3)的模型來預測災難傷患發生的數量，希冀提升台灣的災難準備。 研究方法 本研究為時間序列研究設計，資料來自2009年1月至2016年12月台灣衛生福利部建置的緊急醫療管理系統中的災害事件登錄系統內的災難通報事件資料。個人層級之特性如傷患年齡、性別、地域、傷病情，災難事件特性包含災難類別、日期、地點皆包含於內。依照災難類型區分為自然災害、人為/技術災害、生物災害以及群眾聚集相關災害。依照緊急醫療分區將台灣縣市分成六區。 本研究首先對台灣發生災害事件的時間趨勢進行分析亦對於災難事件對醫療系統之衝擊包含傷患數、傷患運送方式、傷病嚴重程度(死亡以及加護醫療需求度)進行描述與分析。本研究進而運用階層卜瓦松迴歸模式，整合個人層級因子與地域層級因子，對於災難發生之傷患數量之影響加以量化與評估。根據台灣災害登錄數據得出的估計結果，進一步提供了按地區和災害特徵預測的人員傷亡人數預測。 研究結果 研究時間自2009至2016年，涵蓋了902件災難事件，總傷患人數為34949人，總死亡人數為477人，總住院個案數為3610人。以生物災害、群眾聚集相關災害、自然災害，以及人為/技術性災害作為災難事件類別區分，其中308件(34.2%)屬於生物災害、82件(9.1%)屬於群眾聚集相關災害、52件(5.8%)為自然災害，406件(81.0%)屬於人為/技術性災害。相對應總傷患人數分別為6364人(18.2%)、2857人(8.2%)、17372人(49.7%)、8356人(23.9%)。 災難病患平均年齡為41歲(SD=21.6)，自然災害中之患者年紀最長，其平均值為47歲 (SD=20.9)；生物災害傷患其年齡最輕，平均值為28歲(SD=18.8)。 分析患者到院方式，可以發現生物災害以及自然災害患者以自行前往醫院為主(73.57%及70.11%)，但是群眾聚集以及人為/技術性災難患者主要以救護車送達醫院(54.98%及61.81%)。生物災害與群眾聚集事件患者需要加護病房住院比率相對其他兩類災難較低，人為技術災難有最高住加護病房比率(5.65%)。 運用階層卜瓦松迴歸模型評估個人層級與縣市層級之分析結果，顯示自然災害、生物災害以及群眾聚集相關災害相較於人為/技術性災害對於造成災難個案，其危險對比估計值分別為2.24 (95% CI: 2.17-2.31)、1.11 (95%CI: 1.07-1.15)，1.17(95%CI: 1.11-1.23)。都會區之居民成為災難個案之危險顯著低於非都會區者，其相對危險對比估計值為0.16 (95% CI: 0.07-0.36)。 災難事件個案人數預估值，以自然災害為86.4人(95% CI: 70.1-102.7人)，生物災害為68.8人(95% CI: 50.5-87.0人)，群眾聚集災害為43.8人 (95% CI: 30.5-57.0)，人為技術災害為37.6人 (95% CI: 21.7-53.5)，顯示台灣以自然災害帶來最為顯著衝擊。 結論 自然災害的颱風可能帶來許多傷患，造成醫療衝擊。對於災難的衝擊，在台灣有地域性的差異。南臺灣較常因自然災難罹災，而有更多傷患產生。非都會區有較高的災難衝擊。本研究提供未來防災與災難應變的實證依據。

Background： The occurrence of disasters including the natural disasters and the technological disasters associated with mass casualty and social impact in the recent years in Taiwan have drawn great attention on the systematic approach including the preparedness and medical response toward the prevention of these events. In spite of the established disaster prevention planning in Taiwan, there are still rooms for improvement and also experience to be learned from these events. Although previous studies provided detailed description and exploration on single event, a systematic approach for the elucidation of the characteristics and medical burden of disaster is lacking. In this thesis, we thus provided a study by using literature review and empirical data for a better understanding of disaster epidemiology in Taiwan. Objective: The aim of this study was (1) to provide basic description including incidence rate and medical burden brought by disaster events; (2) to elucidate the time trend of the occurrence of different types of disasters; (3) to establish risk prediction model incorporating the individual level factors, characteristics of disasters, and county level factors for a better understanding of the impact of these factors on the occurrence of disaster casualties; (4) to predict the occurrence of disaster casualties based on (3) toward a better preparedness of disaster in Taiwan . Methods: We applied a time sequence study design using data derived from the registry of emergency medical management system for disaster reporting provided by the Ministry of Health and Welfare. The study period spanned from January, 2009 to December, 2016. Information on patient-level characteristics including age, sex, and severity of injury, and the characteristics of disaster event including the type, date, and location of the event were collected. The disasters were categorized into the types of natural disaster, technical disaster, bio-disaster, and mass-gathering disaster. The areas in Taiwan was dived into six zones according to the emergency medical division. We first focused our analysis on the time trend of the occurrence of disaster events in Taiwan. The medical burden including the number of casualties, the method of transporting casualties, and the severity of casualties such as death and the requirement of intensive care was then depicted. We applied an multilevel Poisson regression to incorporate the effect of factors at individual level and county level on the occurrence of disaster casualty to elucidate their impact. Based on the estimated results derived from the empirical data on disaster in Taiwan, we further provided the prediction on the expected number of casualties by area and disaster characteristics. Results: During the study period between January, 2009 and December, 2016, the disaster registry included 902 events were recorded. A total of 34949 victims including 477 deaths and 3610 hospital admissions were enrolled. There are 308 (34.2%), 82(9.1%), 52(5.8%), and 406 (81.0%) events for the disaster type of Bio-disaster, mass-gathering associated, natural disaster, and technical disaster, respectively. The corresponding casualties associated with each type of disaster were 6364(18.2%), 2857(8.2%), 17372(49.7%), and 8356 (23.9%) respectively. The average age of disaster victims was 41 years old (SD: 21.6). The average age of natural disaster victims was the eldest ( 47 years, SD: 20.9) while the average age of bio-disaster victims was the youngest (28 years, SD: 18.8). As to the method of transportation of the casualties to hospital, patients of bio-disaster and natural disaster seek for medical care by themselves mostly. (73.57% and 70.11%). Half of the patient of mass-gathering and technical disaster were brought to hospital by the vehicle of emergency medical service (54.98% and 61.81%). The ICU admission rate were lower in bio-disaster and mass-gathering disaster (0.16% and 1.1%, respectively). The patient of technical disaster had the highest ICU admission rate (5.65%). Based on the results derived by multilevel Poisson regression model, the relative risks of natural disaster, bio-disaster and mass-gathering associated disaster for the disaster casualty occurrence compared with technical disaster was 2.24 (95% CI: 2.17-2.31), 1.11 (95%CI: 1.07-1.15), and 1.17(95%CI: 1.11-1.23) respectively. The relative risk of urban citizen as the victim of disaster casualty was lowest 0.16 (95% CI: 0.07-0.36), as comparing to the non-urban citizen for being the victims during disaster. The predicted casualties for natural disaster, bio-disaster, mass-gathering associated disaster, and technical disaster was 86.4 (95% CI: 70.1-102.7), 68.8 (95% CI: 50.6-87.0), 43.8 (95% CI: 30.5-57.0), and 37.6 (95% CI: 21.7-53.5), showing the significant impact of natural disaster in Taiwan. Conclusions: The analysis of the empirical data on disaster in Taiwan revealed that natural disaster was associated with massive casualties and will result in remarkable medical impact and burden. There are geographical variation considering the impact of type of disaster. Souther area of Taiwan was more vulnerable to natural disaster which results in increasing number of disaster casualties. The non-urban area also carried a higher burden of disease casualties. Our results provide an empirical evidence on guiding the preparedness and prevention of disaster.