以概似函數為基礎之鑑取雙胞對二元性狀資料相符度分析

Abstract

雙胞對研究常被用於尋找遺傳疾病的初期相關證據階段，倘若疾病確實由遺傳所控制，在MZ(同卵雙胞對)及DZ(異卵雙胞對)共享非遺傳效應因子(如，環境因子)之假設下，可由資料顯示出MZ雙胞對內兩個體的相似程度高於DZ雙胞對內的相似程度。完全隨機取樣設計下，衡量雙胞對組內兩個個體二元性狀資料相似程度的測度已有學者提出討論，但是當疾病盛行率過低或是考量成本效益，隨機取樣設計下的雙胞對資料在取得上有困難度，引入鑑取設計收集資料是一個可行的替代方式。因鑑取雙胞對二元性狀資料缺失正常個體易導致雙胞對兩個體的相似性不易被測得，惟病體狀態相符度以及成對狀態相符度能夠直接測量此鑑取資料之相似程度。在這個研究當中，我們沿用這個測度並討論不同鑑取設計下三種以概似函數為基礎的檢定統計量，包括Wald檢定、計分檢定及概度比檢定並計算不同鑑取研究設計下檢定量的所需樣本數。模擬結果中評估統計量在鑑取研究設計下參數組合的型一錯誤率和檢定力的優劣，最後建議研究者在特定的情況下可以採用的統計量以及進行鑑取雙胞對家族聚集性研究時的所需樣本數多寡。

The classical twin study has provided a useful tool for investigating the existence of genetic influences on a disease. A variety of methods have been proposed for analysis of twin data with dichotomous traits. Of these methods some are applied to randomly sampled twin data and some to ascertained twin data. Collection of twin pairs data by ascertainment is required when we study a rare disease or consider the cost-effectiveness. The common approach used in analysis of the ascertained dichotomous twin data is to test the equality of casewise or pairwise concordance between monozygotic (MZ) and dizygotic (DZ) twins. Significant test result indicates the potential role of genetic factors played in the development of a disease under the assumption that MZ twin pairs share nongentic effects to the same extent as DZ twin pairs. In this research, we aim to evaluate the applicability of likelihood-based concordance tests for ascertained dichotomous twin pair data and derive the formulas for calculation of the required sample sizes as well. The six tests which include the Wald, score and likelihood ratio tests are taken into account under the different ascertainment design. Simulations are conducted to compare the empirical type I errors and powers for these test statistics under medium or small sample size. Required sample sizes and the corresponding empirical powers are obtained and discussed for the above-mentioned tests.