檢測疾病是否存在家族聚集性(familial aggregation) 通常是遺傳流行病學的首要工作。在家族聚集性研究中，病例對照家庭設計 (case-control family design) 是典型的抽樣方式，此種抽樣方式是先取得病例和對照首被鑑取者 (case and control proband)，然後招募該家族其他成員進入研究中。當有些主要的干擾變數難以衡量時，則採用配對病例對照家庭設計。雖然關於二元性狀或是未配對之家庭資料的分析文獻上被廣泛討論，然而針對病例對照家庭設計下數量性狀的分析卻少有討論。本文討論應用Liang & Pulver (1996) 所提出的廣義估計方程式方法，以及Wang & Zhao (2008) 所提出的加權分量迴歸分析，探討疾病的家族聚集性。我們亦應用 Hoffman et al. (2001) 所提出的群內重抽樣方法，分別配適線性迴歸模型以及分量迴歸模型以評估家族聚集性。本文將透過模擬研究比較這些方法實際執行的性質，並進行實例分析。 The preliminary work in genetic epidemiology is to determine whether a given disease shows familial aggregation. A typical design used for determination of familial aggregation should be the case-control family design which a case or a control proband is ascertained first, with subsequent recruitment of other members in the family. When some major confounders are difficult to measure, the matched case-control family is adopted. Although methods for analyzing familial data with a binary trait or unmatched design are well discussed, methods proposed for analysis of a quantitative trait with matched case-control family design get less attention. In this study, we apply the generalized estimating equation method (Liang & Pulver, 1996) and the weighted quantile regression analysis for clustered data (Wang & Zhao, 2008) to detect familial aggregation. We also apply the within-cluster resampling method of Hoffman et al., (2001) to fit linear regression model or quantile regression model to assess the familial aggregation. We assess the performance of the proposed methods through simulation studies and analysis of one dataset.
