Study selection
The search which forms the basis of the present review was conducted in October 2005 in PubMed using "Delivery of health care" OR "Patient Health Care" as MeSH terms with "depression" and "trial", and in PsycInfo with "Primary Health Care or Health Care Delivery" and Major Depression as a subject heading. MeSH terms, which are used for indexing PubMed publications, provide a reliable means of capturing papers which use different terms for similar concepts. 1,691 papers were found of which 164 were identified from the abstract description as potential papers for inclusion. An additional 85 papers were located in the reference lists of previous reviews, and by undertaking a search of the Cochrane Library. Search terms "patient care management" OR "delivery of health care" and 'depression" were used to search the Cochrane Central Register of Controlled Trials [See Additional file 2]. The paper with the longest follow-up period was selected for inclusion for studies with more than one published outcome report. The original efficacy or effectiveness trial was sought to replace trials that reported cost effectiveness outcomes, and no publication date restrictions were applied. Figure 1 describes the flow chart for inclusion. 55 studies were identified. Because one study could contain more than one experimental arm, the review undertook an analysis of 70 comparisons. Included in this review were those comparisons where the active enhancement intervention was compared to a control group of either TAU or waitlist control. Studies involving quality and safety issues, side effects, comorbidity or adolescent populations, or published in languages other than English were excluded. To be included, the trial must have contained at least one outcome measure of depression. A summary of the papers included in the review [see Additional file 3] and a quorum statement checklist [see Additional file 4] are provided.
Extraction of data
Study characteristics
A standard coding sheet was developed, using previous formats with sections based on the Effective Practice and Organisation of Care (EPOC) [7] codes. Three coders (two per paper) coded the following variables: Country, type of control group (TAU, attention placebo, waitlist control, other model of treatment), comorbidity (social anxiety, panic or agoraphobia, generalised anxiety disorder], type of intervention (promotion, prevention, early intervention, treatment, recovery), study design (RCT, CCT), setting (family practice, university affiliated primary care clinic, community service provider, community mental health centre, www, pharmacy, work, other), recruitment procedures (screening in general practice, medical setting, electoral roll, advertisements, other, time period of the intervention (0–6 weeks, 7–11 weeks, 3–4 months, 5–6 months, 7–9 months, 10–12 months, 13–24 months, more than 24 months), treatment content (antidepressants, CBT, etc.), job description of the individual undertaking the intervention [general practitioner (GP), allied health], age of participants, gender, ethnicity, whether consumers were involved in the design, conduct or interpretation of the study. The length of time to the final post intervention follow-up was also recorded. The quality of research papers was rated according to the EPOC Group's criteria [7]. Ratings were made of masking of allocation, blinding, withdrawals, and performance bias. Discrepancies between coders were resolved through discussion, with a third rater acting as an arbiter if required.
Content of the intervention
Coding of components of care
We used the classification components outlined in Figure 2 which have previously been used in review of general practice, taking the terms and descriptors from the relevant research articles [3, 5, 6]. Full descriptions of these categories are available in the original publications.
Coding of treatment
Treatment coding was complex because multiple treatments were involved. We sought to determine the most important treatment, or whether combined treatment was used.
Coding of intentions
Eight types of intervention were investigated. Most studies focused on only one, but a number of interventions sought to determine the effects of more than one simultaneously. In particular, a number of studies sought to investigate both training and care management, or both enhanced care and care management. For these studies, both relevant categories were scored as positive, and each category of studies was the compared to the rest.
(1) Training and feedback directed at general practitioners. These enhancements were provided to general practitioners to improve their skills, and included the components of provider education, the provision of feedback to improve diagnosis, and the provision of clinical practice guidelines.
(2) Providing assistance within the general practice so that patients received 'care management'. This was achieved variously through the provision of tracking and monitoring by nurses, the provision (and implementation) of procedures designed to encourage adherence to medication or treatment, the use of practice managers to keep patients in care.
(3) Enhancements or extensions to general practice care. These involved the use of a specialist involved with the practice, the referral of the patient to health professionals attached to the practice, or the direct provision of enhanced therapy such as problem solving or CBT within the practice.
(4) The provision of self help materials or computer guided programs within the practice to improve efficacy.
(5) The provision of assistance from teams external to the practice such as mental health teams.
(6) The linking of the patient to community based health professionals other than doctors but including pharmacists.
(7) Interventions that occur in health maintenance organisations.
(8) Interventions that are initiated in the broad community or in smaller facilities within the community, such as educational groups or programs in residential facilities.
Detailed component checklist
The checklist included details about the following components: (a) guideline implementation, (b) provider training in depression care, other than by guidelines, (c) patient education including mental health literacy and self help training, (d) the inclusion of patient preferences in the type of care, (e) systematic tracking of patients (other than by the doctor), including details about the nature of the tracking, who does it and whether it is supervised, (f) monitoring of medication adherence, (g) the use of a team based approach, and the nature of the registry or record, (h) additions to usual care provided by the doctor, including enhanced care (CBT, for example), or assistance from another person, including a psychiatrist or social worker, (i) provision of initial patient diagnosis of depression to the GP, and (j) peer support. The full set of questions is provided [see Additional file 1].
Intervention outcomes
The primary outcome variable was the key depression variable used in the trial. One study which included a quality of life measure was also included [8]. A rating was made of whether there was significant change on that variable relative to the relevant control group (expressed as a dichotomous variable: improved/not improved compared to TAU).
Analysis
Descriptive data are presented as percentages. Dichotomous outcomes (improved above TAU) were analysed as odds ratios (ORs). Logistic regression was used to predict improvement where multiple components were involved. The Chi Square (χ
2) statistic was used to determine differences in the proportion of positive outcomes as a function of intervention type.