年齡、年代與世代模式之序位法的應用

The application of sequential method in age-period-cohort model

蘇士詠(Shih-Yung Su)

指導教授 ： 廖勇柏

中山醫學大學/健康管理學院/公共衛生學系碩士班/碩士(2012年)

https://doi.org/10.6834/CSMU.2012.00095

摘要

本碩士論文有別於一般碩論的作法，主要分為兩階段的過程，第一為方法學上的評估，主要目的在於提出一套具有可行性的方法學應用流程，內容在於評估並探討年齡、年代與世代效應之序位法的方法學，並提出合乎數理原則的統計方法輔助序位法的應用。第二為該方法學的實際應用，為了考驗應用的可行性，本碩士論文將評估後的結果體現在兩篇應用於流行病學的議題，最後將這兩篇投稿到國際期刊，藉以了解該方法學在實際應用上是否可行。
本碩士論文建議採用MAPE技術選擇適當的序位法適配模式，並且搭配AIC技術找出spline函數下最佳的節點參數量，如此便能解決在實際應用序位法時可能會面臨的統計方法學問題。然而在後續的實際應用上，選擇了結直腸癌與子宮頸癌的兩個流行病學議題進行分析與討論，隨後也發表在國際期刊上。
然而，序位法的方法僅是眾多年齡、年代與世代模式的其中之一，目前在國際上仍然陸續有新的年齡、年代與世代模式持續發展，這意味著這套技術仍然無法廣泛地應用在所有議題的疾病，這也突顯了未來在這套技術的發展上，有必要進行整合，或者是提出更合乎理論的新方法。

關鍵字

年齡、年代和世代模式 ； 結直腸癌 ； 子宮頸癌 ； 死亡率

並列摘要

This dissertation can separate two steps in integrant framework. The first step evidences the methodology of sequential method of age-period-cohort (APC) model. The aims try to establish an appropriate methodological procedure to support the application of sequential method. In consequence, this dissertation suggests using mean absolute of percentage error(MAPE) to choice the more accurate one of the two submodels(AP-C and AC-P) and using Akaike information criterion(AIC) to find the parameters in splines function of the best model-fitting.
Second, in order to test the practicability of first step, I apply this methodological procedure in two epidemiological topics, colorectal cancer and cervical cancer. However, the two epidemiological topics have submitted to international journal. Up to now, the manuscript of colorectal cancer has published in international journal of colorectal disease, and another manuscript has under review in BMC public health.
The sequential method is just one of the many methods of APC analysis. Furthermore, the new methods still continuously develop. It means that the technology of APC model still cannot practice in mostly epidemiological topics in existence. Therefore, that will be necessary to reorganize and integrate the presently methods of APC model in subsequent development, or address another new method in the future.

並列關鍵字

age-period-cohort model ； colorectal cancer ； cervical cancer ； mortality

參考文獻

68. Yeh CC, Hsieh LL, Tang R, et al. Risk factors for colorectal cancer in Taiwan: a hospital-based case-control study. Journal of the Formosan Medical Association = Taiwan yi zhi 2003;102(5):305-12.

1. Fienberg SE, Mason WM. Identification and estimate of age-period-cohort models in the analysis of discrete archival data. In : Schuessler KF (ed): Sociological Methodology, Jossey-Bass, San Francisco. 1979.

2. Osmond C, Gardner MJ. Age, period and cohort models applied to cancer mortality rates. Statistics in medicine 1982;1(3):245-59.

3. Boyle P, Robertson C. Statistical modelling of lung cancer and laryngeal cancer incidence in Scotland, 1960-1979. American journal of epidemiology 1987;125(4):731-44.

4. Clayton D, Schifflers E. Models for temporal variation in cancer rates. II: Age-period-cohort models. Statistics in medicine 1987;6(4):469-81.