LV Rubenstein, VA Greater Los Angeles HSR&D Center of Excellence; LE Parker, RAND Health Program; LS Meredith, RAND Health Program; A Altschuler, RAND Health Program; UC Berkeley; E dePillis, RAND Health Program; University of Hawaii, Hilo; J Hernandez, PriceWaterhouseCoopers, Los Angeles, CA; N Gordon, Division of Research, Kaiser Permante of California

Objectives: We evaluated the attributes, implementation, and success of two forms of team-based quality improvement (QI) for depression in primary care (PC) practices. Both forms emphasize team use of evidence from the literature, including national guidelines.

Methods: We partnered with two non-profit managed care organizations--the VA Greater Los Angeles Healthcare System and Kaiser Permanente (KP) of Northern California--to initiate and evaluate five QI teams tasked with improving care for depression. We asked participating organizations to set priorities for depression QI and form two types of depression QI teams. One type, the (local team or LT), emphasized local leadership, planning, and implementation by each participating practice. The other type (central team or CT) emphasized central leadership and planning by content experts combined with local implementation by the participating practices. The 3 LTs involved 3 PC practices and the 2 CTs involved 3 PC practices. To evaluate QI team performance, we identified theoretically important positive features expected to support team success in carrying out quality improvement (e.g., team composition and organizational environment), and identified variations in team characteristics through observer process notes, national expert ratings of the quality of team plans, and interviews. We used qualitative predictor-outcome matrix analyses of team features to assess relationships between team characteristics and the quality, degree of implementation, and longevity of their depression improvement programs.

Results: Organizational priorities differed between KP and VA, with KP emphasizing depression follow-up in PC and VA emphasizing screening in PC and follow-up in mental health specialty. In general, experts rated VA team plans lower than KP plans because VA teams did not incorporate support for PC follow-up of depressed patients. As designed, the LT and CT approaches incorporated theoretically important positive features to an equivalent degree. As implemented, adherence to the design was high (e.g., access to content expertise among the CTs), but features not specified by design varied (e.g., some CTs had high local involvement). CT plans as a group were rated as higher quality, but the plan rated highest of all was from an LT. LTs, more than CTs, showed a strong relationship between the quality of their plans and the degree to which their local environments supported depression QI. All teams successfully implemented most elements of their plans. Longevity of implementation was equivalent between the two types of teams.

Conclusions: QI teams can successfully design and implement evidence-based models of depression care. LTs were characterized by both the potential to achieve the highest quality plans and by the greatest variation in plan quality.

Impact: 1. A health care organization's specific depression-related goals should be assessed and made available to QI teams.

2. QI teams can successfully base the key design elements of a new care model primarily on relevant evidence from the literature, when this is available, rather than on extensive data collection and analysis within their own organization. Implementation of the design, however, requires local testing and adaptation.

3. Local-only team-based QI should be reserved for environments strongly capable of supporting depression care improvement.