Background. Clinicians have questioned the validity of depression criteria in patients with chronic medical illness, but few empirical data address this question.
Method. Computerized records identified two samples of managed care patients beginning depression treatment: 235 with diabetes, ischemic heart disease, or chronic obstructive lung disease and 204 without those conditions. Telephone assessments at baseline, 2 months, and 6 months included the Structured Clinical Interview for DSM-IV and other measures of depression severity and functional status. Item Response Theory analyses compared patterns of depressive symptoms across groups and specifically evaluated somatic symptoms (fatigue, change in weight or appetite, psychomotor agitation/retardation, and sleep disturbance) as indicators of depression. Logistic regression examined how change in somatic symptoms during treatment varied between patient groups.
Results. Overall item response analysis indicated differential item functioning between groups (χ2=33·7, df=18, p=0·017). Two of eight item-level comparisons were statistically significant; one in the predicted direction (patients with co-morbidity reported more fatigue at low levels of depression: χ2=17·9, df=1, p<0·001) and one in the opposite direction from predicted (patients with co-morbidity reported less psychomotor agitation/retardation at low levels of depression: χ2=8·0, df=1, p=0·005). Observed differences were modest: at the midpoint of the depression severity scale, patients with medical co-morbidity had a 54% probability of reporting fatigue compared to 45% in those without co-morbidity. All four somatic symptoms showed robust improvement with treatment, and this improvement did not differ significantly between patients with and without medical co-morbidity.
Conclusions. We find only limited evidence that fatigue, changes in weight or appetite, psychomotor agitation/retardation, and sleep disturbance are less valid indicators of depression in patients with chronic medical illness. DSM-IV criteria for diagnosis of depression do not require significant modification for patients with medical co-morbidity.