3041 — Evaluating Implementation of Best Practices for Depression Care: 18 Month Results

Chaney EF (COE - Seattle), Rubenstein LV (COE - Sepulveda), Yano EM (COE - Sepulveda), Liu CF (COE - Seattle), Bolkan CR (COE - Seattle)

Objectives:
Over 37 randomized trials show collaborative care for depression in primary care is effective and cost-effective. We combined a quality improvement intervention (QI) to help practices design and implement evidence based depression care models with a randomized trial that referred a representative sample of depressed patients to the care model. We aimed to understand the 18-month outcomes of the randomized trial.

Methods:
In randomly allocated QI practices, clinical leadership adapted evidence-based depression care models. Primary care clinicians (PCPs) began referring patients to care managers at model start-up without researcher involvement. We followed the first 800 of these PCP-referred patients using de-identified care manager data through completion of six months of care management. We also screened 23,000 primary care patients prior to primary care visits for depression and enrolled 761 with probable major depression to the care manager (research-referred). We surveyed enrolled research patients by telephone at baseline and eighteen months (49% completion).

Results:
Comparing research-referred QI practice patients to usual care at 18 months, depression symptom severity improved for both similarly. On further analysis, IRB-related delays and success of the PCP referral process resulted in higher than planned care manager loads, with DCMs unable to accommodate all patients. Research-referred patients not receiving comprehensive DCM assessment improved less than all other enrolled patients (p < .02). Research-referred patients refused DCM assessment at higher rates than PCP-referred patients (41% versus 20%), and experienced significantly lower rates of guideline concordant care than PCP-referred patients (25% versus 80%).

Implications:
Our attempt to combine a naturalistic intervention process with a randomized trial was unsuccessful in that care managers were unable to accommodate the increased patient flow. In addition, without PCP involvement in the referral, patients were less likely to accept care management and less likely to complete it successfully.

Impacts:
Researcher referral of patients directly to a care manager, without patient selection and preparation by a PCP, may not accurately predict how a depression care intervention will work when PCPs are carrying out the referrals. Future implementation research should consider the role of PCP referral in assuring successful care management.