A role for profiles of patient-specific depression characteristics and socioeconomic factors in the prediction of antidepressant treatment outcome.

To the Editor: In their recent article “Prognostic Subgroups for Citalopram Response in the STAR*D Trial,” Jakubovski and Bloch1 conclude that baseline socioeconomic variables “are likely to be more informative than routine clinical variables such as past medication response, duration and severity of illness, and comorbid psychiatric illnesses” in the prediction of citalopram treatment outcome. In particular, their assertion carries the curious implication that demographic factors are more strongly related to treatment response than patient-specific ones. However, their findings are not in agreement with an earlier study we published using the same data and analysis methods.2 In our study, we described prognostic subgroups for citalopram remission that included a combination of patient-specific depression characteristics and socioeconomic variables. For example, our analysis of baseline factors suggested that a person making at least $40,000 per year would have markedly different remission rates (12% vs 55%) depending on depression-specific characteristics such as depressed mood, interest in activities, and insomnia. The 12% rate of remission we identified for a subgroup characterized by higher socioeconomic status but worse patientspecific depression symptoms was actually lower than the rate of remission for any subgroup of lower socioeconomic status patients. Jakubovski and Bloch, on the other hand, suggested that depression-specific characteristics at baseline do not have stronger discriminative power than socioeconomic variables. We believe that Jakubovski and Bloch’s results may reflect variations in subject categorization that deviate from established findings. Notably, in Figure 1A, Jakubovski and Bloch identified 1,023 remitters by the Hamilton Depression Rating Scale and thus indicated that 41% of the subjects in the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) study remitted, whereas the initial STAR*D report3 identified only 790 remitters and reported a remission rate of 28%. Furthermore, Jakubovski and Bloch included only completers in their analysis (~2,500 subjects), as opposed to the full analyzable sample of 2,876 patients. Finally, they utilized a more limited set of predictors than we in our analysis: specifically, they did not utilize any individual depression symptoms, nor did they utilize the clinically important anxious depression construct.4 We believe that these variations markedly affected their findings. We feel that researchers and clinicians should focus attention on subgroups of patients with combinations of specific, severe depression characteristics and socioeconomic variables. Patientspecific depression characteristics, in addition to socioeconomic factors, can help psychiatrists guide their medication choices and provide additional accuracy over reliance on socioeconomic factors.