TO THE EDITOR: Chochinov et al. (R1557291C) compared four brief screening measures for depression in terminally ill patients. They concluded that for diagnostic purposes, brief screening measures do not approach the validity of a single-item interview that asks, in effect, "Are you depressed?" We have a number of reservations with regard to this conclusion.

First, Chochinov and his colleagues compared the brief measures with the standard of a semistructured interview and the Research Diagnostic Criteria. Such standard criteria may not be valid for elderly (mean age of the study group was 71 years) depressed patients (R1557292C). For example, community surveys of geriatric depression using instruments designed and validated for the elderly report much higher rates of depression than do epidemiological studies using standard criteria (R1557293C).

Second, depressed patients over 65 years old are less likely than younger patients to complain of low mood (R1557294C). This observation suggests that the single-item interview could be expected to have low sensitivity for significant depressive illness in this population.

Finally, a comparison of liaison versus consultation models for geriatric inpatients (R1557295C) reported a much higher rate of diagnostic accuracy for depression by referring doctors in the liaison model. On the basis of data from this study (R1557295C), the approach suggested by Chochinov et al. would be expected to have a low specificity. A more appropriate approach to detection of depression in the physically ill would be an emphasis on liaison rather than encouragement of nonpsychiatric colleagues simply to ask their patients, "Are you depressed?"

In conclusion, we believe that the standard used by Chochinov et al. may be invalid for the population studied. A more appropriate standard would have been an instrument validated for an elderly population with physical illness or a positive response to psychiatric intervention.