以短天期歷史資料建構台灣地區流行性感冒監測模式，2005-2008

Short-term of Historical data to Construct Influenza Surveillance Model in Taiwan, 2005-2008

張育誠(Yu-Cheng Chang)

指導教授 ： 李子奇

高雄醫學大學/健康科學院/公共衛生學研究所/碩士(2010年)

https://doi.org/10.6832/KMU.2010.00101

摘要

研究背景：
根據WHO統計季節性流感對於死亡率具有很大的影響，而這些死亡患者有90%為身體虛弱的老人。幾次流行性感冒的大流行也造成很大的傷亡，像是1918年的西班牙流感、1957年的亞州流感及1968年的香港流感。建立精準且有效率的監測系統是很重要的。若能應用短天期資料建構台灣地區流行性感冒監測模式，並對其即時性、敏感度及特異度進行評估，將可提供建構台灣地區流行性感冒監測模式的參考。除此之外，很多研究也指出環境因子會影響流行性感冒，在研究此類議題時，必須控制相關風險因子。
研究方法：
為了控制與流行性感冒相關的風險因子，我們以自我迴歸模型找出與流行性感冒相關的風險因子，並以主成分分析將風險因子和流行性感冒通報率合併為一觀察指標。本篇研究使用的流行性感冒監測模式為simple linear regression來定義定點醫師通報資料和合併觀察值的警報發布頻率，將預測的警報發布頻率與黃金標準警報發布頻率作ROC分析，以評估此監測系統的敏感度和特異度。
結果：
控制其他相關風險因子及其自我相關下，與肺炎及流行性感冒標準化死亡率具有相關的風險因子為溫度和流行性感冒病毒活性，溫度越高，肺炎及流行性感冒標準化死亡率就越低，流行性感冒病毒活性越高，肺炎及流行性感冒標準化死亡率就越高。而與實驗室流行性感冒病毒檢出率具有相關的風險因子為溫度、臭氧及定點醫師流感通報率。溫度越高，實驗室流行性感冒病毒檢出率就越低，臭氧濃度越高，實驗室流行性感冒病毒檢出率就越低，定點醫師流感通報率越高，實驗室流行性感冒病毒檢出率就越高。以肺炎及流行性感冒死亡資料警報發布頻率作為黃金標準依據時，主成份資料作為監測模組所得到的結果都優於以定點醫師流感通報資料作為監測模組。但是當以實驗室流感病毒檢出資料警報發布頻率作為黃金標準依據時，其結果剛好與以肺炎及流行性感冒死亡資料警報發布頻率作為兩者比較的黃金標準依據時的結果相反，以定點醫師流感通報資料作為監測模組的結果都優於以主成份資料作為監測模組。
結論：
以定點醫師流感通報資料並配合控制溫度、臭氧及流行性感冒病毒活性等相關風險因子來建構台灣地區流行性感冒監測系統，經評估其敏感度、特異度和即時性良好。

關鍵字

流行性感冒監測系統 ； 主成分分析 ； ROC曲線下面積

並列摘要

Background: According to WHO statistics, seasonal influenza has a great impact on mortality, with 90% of frail elderly people. Several influenza pandemic, causing considerable impacts, such as the Spanish flu in 1918, Asian flu in 1957 and Hong Kong flu in 1968. To construct the accurate and efficient monitoring system is very important. By using the short-term history data to construct influenza surveillance model in Taiwan, if its timeliness, sensitivity and specificity are well provide. We could the suggestions on the construction monitoring system of influenza in Taiwan. In addition, many studies have pointed out that environmental factors will affect influenza, in the study also included such issues to control risk factors.
Methods: In order to control risk factors associated with influenza, we use auto-regression model to identify risk factors, and use principal components analysis to combine risk factors and sentinel surveillance data as one observed indicator. In this study, the influenza monitoring model as simple linear regression, the regression predicted values and the gold standard for ROC analysis to evaluate the monitoring system by the sensitivity and specificity.
Results: Controling of other risk factors and the autocorrelation, the standardized mortality of pneumonia and influenza is related to temperature and influenza virus activity. Controling of other risk factors and the autocorrelation, the influenza virus with the laboratory detection rate is related to temperature, ozone and communications rate of influenza sentinel physicians. To use pneumonia and influenza mortality data as the gold standard, principal component data obtained as a result of the monitoring model is better than influenza sentinel surveillance data obtained as a result of the monitoring model. But to use the influenza virus detection laboratory data as the gold standard, the results is different to pneumonia and influenza mortality data as the gold standard. Influenza sentinel surveillance data obtained as a result of the monitoring model is better than principal component data obtained as a result of the monitoring model.
Conclusion: The influenza surveillance system by using sentinel surveillance data, ozone, flu virus activity and temperature with simple linear regression model has good performance by the assessing of sensitivity, specificity and timeliness.

並列關鍵字

Influenza surveillance system ； Principal components analysis ； Area under ROC curve

參考文獻

參考文獻

1. WHO. Influenza (Seasonal). 2010-02-13.

2. Barker WH. Excess pneumonia and influenza associated hospitalization during influenza epidemics in the United States, 1970-78. American journal of public health. 1986;76(7):761-765.

3. Simonsen L, Clarke MJ, Schonberger LB, Arden NH, Cox NJ, Fukuda K. Pandemic versus epidemic influenza mortality: a pattern of changing age distribution. The Journal of infectious diseases. 1998;178(1):53-60.

4. Simonsen L, Reichert TA, Viboud C, Blackwelder WC, Taylor RJ, Miller MA. Impact of influenza vaccination on seasonal mortality in the US elderly population. Archives of internal medicine. 2005;165(3):265-272.