Assessing Diagnostic Approaches to Depression in Medically Ill Older Adults: How Reliably Can Mental Health Professionals Make Judgments About the Cause of Symptoms?

OBJECTIVE: To evaluate the reliability of the DSM-IV approach and five other schemes for counting symptoms toward the diagnosis of depression in hospitalized medically ill older patients and to examine whether mental health professionals can reliably make judgments about the etiology (medical or psychological) of depressive symptoms. METHOD: A sample of 38 patients aged 60 years or older admitted to the general medicine, cardiology, or neurology services at Duke University Medical Center were evaluated for depression using a structured psychiatric interview and the Hamilton Depression Scale. Interrater reliability for the diagnostic schemes, for unstructured clinical diagnoses, and for determinations of the causes of individual depressive symptoms was assessed by three pairs of mental health professionals. RESULTS: Agreement between raters for structured diagnoses was high regardless of diagnostic strategy, with the DSM-IV approach being only slightly less reliable than the strict inclusive approach (Kappa 0.88 vs Kappa 1.0, respectively). For all diagnostic approaches, there was perfect agreement between raters for eight cases of major depression. Agreement for unstructured clinical diagnoses of depression (K = 0.50) was much lower than for the structured diagnoses. Agreement between raters on the etiology of individual depression criterion symptoms assessed by structured interview was greater than 80% for 14 of 19 symptoms. Correlation between raters' depression severity ratings on the Hamilton Scale using the DSM-IV etiologic approach was equivalent to that using the strict inclusive approach (0.98 vs 0.95, respectively). CONCLUSIONS: Mental health professionals can be trained to make judgments reliably about the cause (medical or psychological) of symptoms in hospitalized older medical patients. The "strict inclusive" and other diagnostic schemes for counting symptoms toward the diagnosis of depression have only marginal, if any, benefit compared with the current DSM-IV approach.