A confirmatory factor analysis of the Beck Depression Inventory in chronic pain

The Beck Depression Inventory (BDI) is widely used to assess depression in chronic pain despite doubts about its structure and therefore its interpretation. This study used a large sample of 1947 patients entering chronic pain management to establish the structure of the BDI. The sample was randomly divided to conduct separate exploratory (EFA) and confirmatory factor analyses (CFA). EFA produced many satisfactory two-factor solutions. The series of CFA generated showed reasonable fit for ten of those solutions. All included a first factor identified as negative view of the self (items: failure, guilt, self-blame, self-dislike, punishment and body image change), and a second factor identified as somatic and physical function (items: work difficulty, loss of appetite, loss of libido, fatigability, insomnia and somatic preoccupation). The remaining items (suicidal ideation, social withdrawal, dissatisfaction, sadness, pessimism, crying, indecisiveness, weight loss, irritability) loaded infrequently or not at all in the CFA solutions. They did not form a coherent factor but comprised items associated with negative affect. When compared with published data from samples of depressed patients drawn from mental health settings the mean item scores for items reflecting the negative view of the self were consistently statistically lower that that observed in samples; there was no consistent difference between the samples on the items reflecting somatic and physical function; but the mean scores for the remaining affect items were significantly greater in the mental health samples. This version of depression is strikingly different from the psychiatric model of depression (e.g. DSM-IV or ICD-10), which is primarily defined by affective disturbance, and secondarily supported by cognitive and somatic symptoms. The finding is consistent with a reconsideration of what constitutes depression in the presence of chronic pain. It also has important clinical implications: it may provide a way to distinguish depressed patients with typical cognitive biases, who require specific treatment for depression alongside pain management.