Epidemiological studies indicate that depressive disorders are highly prevalent in the general population worldwide . Most cases are seen and managed in primary care, and only a small proportion of these are referred to specialty care . A number of studies suggest that primary care patients with depressive disorders are less severely depressed , experience a milder course of illness , have a distinct symptom profile with more complaints of fatigue and somatic symptoms , and are more likely to have accompanying physical complaints  than patients referred to specialty mental health care.
The cornerstones of antidepressant treatment are pharmacotherapy and psychological interventions . However, while the vast majority of patients with depression are dealt with in primary care, most of the research findings upon which decisions are made have involved secondary care patients. It is not fully clear whether the findings from trials in specialty settings can be generalized to primary care. Meta-analyses restricted to primary care patients have been performed for SSRIs and tricyclics compared to placebo [8, 9], SSRIs compared to tricyclics , and psychological interventions [11, 12]. They concluded that these treatments are effective in primary care settings. In some countries a relevant proportion of primary care patients with depressive symptoms is treated with hypericum extracts . The co-morbidity and symptom pattern of primary care patients described in recent studies [14, 15] fits well to the traditional indications of hypericum extracts (psycho-vegetative disorders, depressive disorders, anxiety and/or nervous agitation) . Systematic reviews of hypericum extracts include a considerable number of randomized trials in primary care patients [17, 18]. However, in these reviews the results of trials in primary and secondary care settings were pooled and not analyzed separately. Systematic reviews on music therapy, acupuncture, exercise, relaxation, and family therapy for treating depression published in the Cochrane Database of Systematic Reviews include only few or no trials conducted in primary care settings [19–23].
The systematic reviews and meta-analyses cited above [8–12, 17, 18] summarize the majority of the available randomized trials of depression treatments in primary care. However, it is not possible to answer the question how the available treatment options compare with each other (i.e., whether some treatments are superior to others in primary care). Traditional meta-analyses are restricted to the direct comparison of two interventions by pooling data only from trials with similar treatment arms. By consequence, they allow no decision about the relative effectiveness of two treatments, if they have not yet been directly compared in at least one randomized controlled trial (RCT). However, in case of insufficient or missing direct comparisons of available interventions the utility of indirect evidence may be considered. For example, RCTs of treatment A vs. placebo and treatment B vs. placebo would provide indirect estimates on the comparative effectiveness of A vs. B through the common reference placebo. The inclusion of more interventions would result in more complex networks and involve more complex indirect comparisons.
Network (or multiple/mixed treatment) meta-analyses are an enhancement of the traditional meta-analysis methodology to more than two interventions . They estimate the comparative effectiveness between two treatments based on all available direct and indirect evidence that is available in a network of treatments and comparisons. Besides augmenting validity of comparisons between available treatments through including indirect evidence, network meta-analyses allow for a formal assessment of evidence inconsistencies. Not least, they suggest a ranking of interventions according to their relative effectiveness, which may be of high relevance for clinical decision making. Network meta-analyses have been performed, for example, to compare newer antidepressant agents [25, 26]. However, most trials included in these research syntheses were not performed in the primary care setting. This is of importance as patients in primary care might differ from those in secondary care.
For results of network meta-analyses to be valid three important pre-conditions should be met : 1) the findings of each of the meta-analyses of the direct comparisons should be homogeneous (not suggesting that trials investigated slightly different questions); 2) for the indirect comparisons to be valid, patients included in the separate subgroups of trials need to be sufficiently similar; and 3) if both direct and indirect comparisons are available the pooled estimates for these need to be consistent. It might be that these pre-conditions will not be met in case of the primary care trials of various treatments for depression.
We will systematically review all randomized trials investigating treatments for depression performed in primary care settings. As we will use the same review methods across all treatments, a comparison of the evidence regarding effectiveness, feasibility, and safety will be possible. If adequate, we aim to perform a formal multiple treatment meta-analysis.