病例雙親對照研究法中基因型相對危險性之估計：Mantel-Haenszel法之應用

Estimating Genotype Relative Risks in Case-parental Control Studies: A Mantel-Haenszel Approach

張晉豪(Chin-Hao Chang) ； 李文宗(Wen-Chung Lee)

《中華公共衛生雜誌》 19卷6期 (2000/12) Pp. 485-490

https://doi.org/10.6288/CJPH2000-19-06-09

摘要

目標：以病例對照研究法探討致病基因與疾病的相關強度時，研究族群中若有分層／融合的情況，基因型相對危險性的估計就可能有偏差。此時可使用「病例雙親對照研究法」之研究設計，以克服之。針對此種研究設計之資料，最近文獻上刊載三種非遞迴計算基因型相對危險性的方法。本研究應用Mantel-Haenszel（M-H）法，提出一個全新的非遞迴估計法。方法：上述四法之估計值公式頗為類似，其差異僅在加權常數的不同。而套用M-H法所推導出之加權數為1.5。結果：吾等進行電腦模擬試驗以比較此四法之統計性質，模擬結果顯示四法皆近似無偏差；而M-H法之變異數較小，表示其為較穩定之估計值。結論：以M-H法估算病例雙親研究法中之基因型相對危險性計算簡單且較前述三法穩定。

關鍵字

病例雙親對照研究法 ； 基因型相對危險性、Mantel-Haenszel法 ； 流行病學方法學 ； 遺傳流行病學

並列摘要

Objectives: Epidemiologists often study the relationship between a susceptibility gene and a disease using the case-control design. Due to stratification/admixture that may exist in the study population, the conventional case-control design may lead to biased estimation of ‘genotype relative risks’(GRRs). This problem can be circumvented by using the ‘case-parental control design’ instead. Under this design, there have been three non-iterative methods for estimating GRRs proposed in the literature. Here we propose a new non-iterative method using the Mantel-Haenszel(MH) concept. Methods: The above four methods have similar formula apar from the weighting constants. The weighting constants used in the M-H method are 1.5 Results: Monte-Carlo simulation shows that the above four methods all produce estimates that are approximately unbiased. However, the M-H method has the smallest variance and hence is the most stable among the four methods. Conclusions: The M-H method is simple to calculated and more stable than the other three non-iterative methods for the estimation of the GRRs in the case-parental control study.

並列關鍵字

case-parental control study ； genotype relative risks ； Mantel-Haenszel method ； epidemiologic methodology ； genetic epidemiology