Differential validity of informant-based diagnoses of dementia and depression in index subjects and in their first-degree relatives.

There is no study indicating that informant-derived information on dementia and depression (i.e. family history information) is equivalently valid for first-degree relatives and for index subjects (i.e. patients and control subjects). However, this unproven assumption is the basis for the frequent, possibly inappropriate, use of instruments validated for patients and control subjects in family studies which focus on frequencies of psychiatric disorders in first-degree relatives. Consequently, there is a need to compare the validity of family history information for both disorders in index subjects and their first-degree relatives. Validity was assessed by comparison of family history information for dementia and depression with interview-derived diagnoses in 75 index subjects and 195 age-matched first-degree relatives. The validity of informant-derived information varied for different disorders, i.e. dementia and depression, and different samples, i.e. index subjects and first-degree relatives. In agreement with the study hypothesis, the sensitivity of surrogate information on dementia was significantly reduced in first-degree relatives in comparison with index subjects. In contrast, the sensitivity to detect depression was equivalent in subjects and in relatives. The results indicate the necessity to assess the validity of the psychiatric diagnoses of interest in the sample of interest, e.g. dementia or depression in first-degree relatives of patients and of control subjects. Observations in selected samples, i.e. subjects treated, hospitalised and/or autopsied, cannot be generalised to first-degree relatives in family studies.