Interventions:
The selection of the comparators was appropriate because it included the two available delivery methods for CBT; individual or group therapy.
Effectiveness/benefits:
The clinical data were from a naturalistic study, which as the authors stated was not ideal. The allocation procedure was by therapist preference rather than random, which could have reduced any selection bias. At baseline, the group-therapy patients were older, less educated, and had a longer history of depression, but their depression scores were comparable to those of individual-therapy patients. Statistical analysis was conducted to control for baseline differences, but the authors acknowledged that confounding could not be ruled out. The outcome measures were specific to depression.
Costs:
The perspective was not reported, but the categories of costs appear to have reflected the viewpoint of the mental health institutions. The authors reported the categories that were considered, but they gave no unit costs and resource quantities. The resource use was from the clinical study, which should have ensured satisfactory data collection. The authors acknowledged that there was considerable variation in treatment duration, which might have affected the costs. No sensitivity analysis of the costs was conducted. The price year was reported, but the sources for the unit costs were not.
Analysis and results:
The results were clearly presented, with the total costs and outcomes for each strategy. No incremental ratios were calculated as a cost-consequences analysis was conducted. The uncertainty was not investigated; no sensitivity analysis was carried out. The authors acknowledged that the main limitation of their study was the naturalistic nature of the clinical analysis. The results appear to be specific to the authors’ setting and cannot be transferred to other settings.
Concluding remarks:
There were some methodological limitations that might affect the validity of the authors’ conclusions. As the authors stated, a large randomised controlled trial is needed.