Fortney 2007.

Methods	Randomized controlled trial
Participants	The study location was Veterans Affairs (VA) community‐based outpatient clinics (CBOCs), South Central Veterans Healthcare Network, USA 177 patients (3 practices) were randomized to the intervention group and 218 patients (4 practices) were randomized to the control group The inclusion criteria were patients with depression that PCPs would be comfortable treating. Patients were screened with a Patient Health Questionnaire depression scale (PHQ9) The exclusion criteria were diagnosis of schizophrenia, current suicide ideation, recent bereavement, pregnancy, a court‐appointed guardian, substance dependence, bipolar disorder, cognitive impairment, or receiving specialty mental health treatment
Interventions	Intervention: TELEMEDICINE ENHANCED ANTIDEPRESSANT MANAGEMENT (TEAM) Patients at intervention sites received a stepped‐care model of depression treatment for up to 12 months. Treatment intensity was increased for patients failing to respond to lower levels of care by involving a greater number of intervention personnel with increasing mental health expertise. The intervention involved 5 types of providers: (1) PCPs located at CBOCs; 2) consult telepsychiatrists located at parent Veterans Affairs Medical Centers (VAMCs); (3) an off‐site depression nurse care manager (RN); (4) an off‐site clinical pharmacist (PharmD); and 5) an off‐site supervising psychiatrist. The consult‐telepsychiatrist accepted consultations or referrals from PCPs. The supervising psychiatrist provided clinical supervision to the care manager and clinical pharmacist via weekly face‐to‐face meetings. Patients and providers could choose either watchful waiting or antidepressant treatment (Step 1). Nurse care manager encounters were conducted via telephone and were scripted to enhance standardization and reproducibility. During the initial care management encounter, patients were: (1) administered the PHQ9 symptom monitoring tool; (2) educated and activated using a semi‐structured script4; and 3) assessed for treatment barriers using semi‐structured scripts for endorsed barriers. Follow‐up encounters to monitor symptoms, medication adherence, and side effects were scheduled every 2 weeks during acute treatment and every 4 weeks during watchful waiting or continuation treatment. Non‐adherent patients or those experiencing severe side effects were administered semi‐structured scripts. A trial was considered to have failed in the acute phase if the patient: (1) was non‐adherent to the medication, (2) experienced severe side effects, (3) experienced = 5‐point increase in their PHQ9 score, or (4) did not respond (50% decrease in PHQ9 score) after 8 weeks of antidepressant therapy. All feedback was provided to PCPs using the electronic medical record. Progress notes reporting failed trials requested an electronic co‐signature from the PCP. If the patient did not respond to the initial antidepressant, the pharmacist conducted a medication history and provided pharmacotherapy recommendations to PCPs via an electronic progress note (Step 2). The pharmacist also provided non‐scripted medication management over the phone to patients experiencing severe side effects or problems with non‐adherence. If the patient did not respond to 2 antidepressants trials, the protocol was to recommend a telepsychiatry consultation followed by additional treatment recommendations to the PCP (Step 3) Control: USUAL CARE Control groups received usual care. Usual care patients like the intervention group were provided with provider education and patient education. Also, the screening results were entered into the electronic medical record
Outcomes	The measures of adherence were assessed using an instrument specifically designed for the study. Prior to administering the follow‐up survey, a study psychiatrist examined the patients VA electronic medical records to determine what antidepressants had been prescribed (or refilled) during the previous 6 months. The most recently prescribed/refilled antidepressant medication was recorded prior to the follow‐up interview and the adherence questions referred to the antidepressant explicitly by name. Research assistants asked all patients with an active prescription if they were currently taking the medication. If the patient was currently taking the medication, the interviewers asked how frequently they took the medication in the previous month and if they took the dosage prescribed. Patients were classified as adherent if item responses indicated they took the full dosage > = 80% of the days in the previous month The patient outcomes were treatment response, remission, health status, and quality of life. Treatment response and remission were measured using the Hopkins Symptom Checklist. Response is measured dichotomously as a 50% improvement in depression severity between baseline and follow‐up. Remission is defined dichotomously as SCL‐20 < 0.5. Health status was measured by the change in the physical health and mental health component scores (PCS and MCS) of the Short Form (SF12V) between baseline and follow‐up. Improvement in health‐related quality of life was measured by the change in the Quality of Well Being (QWB) score. Measures were taken at 6 months and 12 months follow‐up.
Notes	―
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	The article gives no details on the randomization process. Protocol paper does not provide any additional information. (pg 1087) The 7 eligible CBOCs in the South Central Veterans Healthcare Network were matched by parent VAMC, and one CBOC within each pair was randomized to the intervention
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgment of 'Low risk' or 'High risk' of bias. (pg 1087) "The 7 eligible CBOCs in the South Central Veterans Healthcare Network were matched by parent VAMC, and one CBOC within each pair was randomized to the intervention. Five of the CBOCs had on‐site midlevel mental health specialists (e.g. social workers)."
Selective reporting (reporting bias)	Low risk	All relevant results from the methods paper seem to be reported
Other bias	Unclear risk	None noted but insufficient information provided to judge
Blinding of outcome assessment (detection bias) Adherence measure	Low risk	(PRIMARY) SELF REPORT ‐ INTERVIEW ‐ "Data were collected via blinded telephone interview." (pg 1087)
Blinding of outcome assessment (detection bias) Patient outcome	Low risk	(PRIMARY) SELF REPORT ‐ QUESTIONNAIRES ‐ "Data were collected via blinded telephone interview." (pg 1087)
Blinding of participants (performance bias) Adherence measure	High risk	(PRIMARY) SELF REPORT ‐ INTERVIEW ‐ Patients were not blinded to the intervention. This is a subjective measure
Blinding of participants (performance bias) Patient outcome	Unclear risk	(PRIMARY) SELF REPORT ‐ QUESTIONNAIRES ‐ Insufficient information about patient blinding to permit judgment of 'Low risk' or 'High risk'. Patients would have been aware if they received the intervention; might have affected their questionnaire responses
Blinding of personnel (performance bias) Adherence measure	Low risk	(PRIMARY) SELF REPORT ‐ INTERVIEW ‐ "Data were collected via blinded telephone interview." (pg 1087)
Blinding of personnel (performance bias) Patient outcome	Low risk	(PRIMARY) SELF REPORT ‐ QUESTIONNAIRES ‐ Author note: due to the nature of the intervention, it was not possible to blind primary care providers to randomization status. However, I do not believe that this introduces bias because the care management intervention was specifically designed to support primary care providers treating depression (e.g. alerting them to side effects and non‐adherence)
Incomplete outcome data (attrition bias) Adherence measure	Low risk	(PRIMARY) SELF REPORT ‐ INTERVIEW ‐ Author indicated that this is low risk of bias
Incomplete outcome data (attrition bias) Patient outcome	Low risk	(PRIMARY) SELF REPORT ‐ QUESTIONNAIRES ‐ Missing outcome data approximately balanced in numbers across intervention groups, with similar reasons for missing data across groups. Refer to Figure 1 in the article. Plus author note: missing data from completed follow‐up interviews were imputed using multiple imputation