Imputing diagnoses for genetic epidemiologic studies without face-to-face interviews

Abstract

Studies of diseases in families help clarify contributions of genetic and cultural factors in the development of psychiatric illnesses. Since not all family members can be evaluated due to inaccessibility, refusal, or death, information is obtained from relatives regarding these non-interviewed individuals, and diagnoses are assigned. When standardized interviews are used, family history reports are reliable and valid. Relatives are able to correctly report the absence of disorders in family members, and affected individuals are often identified as ill. Factors which influence the accuracy of family reports include diagnosis under study, severity of illness, relationship of informant and subject, sex of informant and subject, comorbid illnesses in subject, diagnosis of informant, and number of informants. These factors can be adjusted given a study population to maximize the predictive value of family history reports.