There is ample evidence that psychiatric disorders often run in families, and thus families represent an important source of information for determining the etiology of psychiatric disorders, with the study of familial aggregation being one of the central themes. The overall objective of this study is to evaluate existing epidemiologic approaches to the study of familial aggregation, and (where appropriate) to develop improved approaches to the design and analyses of such studies. The primary focus of epidemiologic approaches to family studies of psychiatric disorder is testing and estimating familial aggregation by comparing risk of disorder in relatives of case and control probands respectively. Implicit in current methods of analysis used to test and estimate these measures is the assumption that the mode of ascertainment of families of case an control probands is what geneticists refer to as """"""""single ascertainment"""""""" (where the probability of a case [control] family being ascertained is proportional to the number of affected [unaffected] members in the family). Researchers in the field rarely recognized that this assumption is implicit in all current methods of analyzing data from these studies. Yet, when this assumption is not met, biased estimates in all current methods of analyzing data from these studies. Yet, when this assumption is not met, biased estimates and tests of hypotheses could result, leading to incorrect inferences regarding the presence and degree of familial aggregation. Alternative modes of ascertainment have been studied intensively in the field of human genetics. Although these ascertainment scenarios were developed in the context of segregation analysis, they appear to be equally applicable to epidemiologic studies. We propose: a) to investigate the magnitude of the bias in testing and estimating familial aggregation when single ascertainment does not hold; and b) to develop methods of testing and estimating familial aggregation which allow for other, realistic modes of ascertainment. In addition, it has been shown that standard methods of survival analysis will also give biased estimates when used to estimate the lifetime risk of disorder in relatives of case probands in family studies and methods for eliminating this bias have been recently developed. We propose to c) extend these methods to estimate lifetime risks of disorder for relatives of control probands and develop tests of familial aggregation comparing lifetime rates in these two groups.
