Comparing Long-Term Outcomes of Two Collaborative Care Approaches for People with Depression [Internet]

Background: Depression is a leading cause of morbidity with disparities in care. Depression collaborative care—a team-based approach supporting care management, patient activation, and evidence-based treatments—is effective relative to usual care but often unavailable in underresourced communities. Community Partners in Care (CPIC) randomized 95 programs in health care and community-based (eg, social services, faith-based) service sectors in 2 communities, into Resources for Services (RS) for individual program technical assistance vs Community Engagement and Planning (CEP) for multisector coalition support for depression collaborative care. For depressed clients, CEP relative to RS reduced having poor mental health-related quality of life (MHRQL) and behavioral health hospitalization over 6 to 12 months. Longer-term outcomes are unknown.

Aims: This extension study of CPIC aims to do the following:

1. Compare CEP and RS effects at 3-year follow-up 2 years after intervention support, on primary (depression and MHRQL), community-prioritized outcomes of physical health-related quality of life (PHRQL) and behavioral health hospitalization, and outpatient services use (secondary).

2. Compare effects of CEP and RS at 6, 12, and 36 months for CPIC participants enrolled in health care or community-based service sectors.

3. Describe outcome priorities for African American and Latino depressed clients, and as perceived by providers.

4. Describe views of stakeholders planning the Los Angeles County Health Neighborhood Initiative (HNI) informed by CPIC and national stakeholders briefed on CPIC and HNI.

We hypothesized that CEP relative to RS would improve long-term mental and physical health and reduce hospitalization nights, overall and within-sector. We expected stakeholders to prioritize social factors and mental health.

Methods: Extension Study (Aims 1-3): Between January and October 2014, we approached 1004 CPIC clients initially screened as depressed from 89 programs and eligible for 3-year surveys; 600 participated (60%; RS, 293; CEP, 307). Aim 3: For client qualitative interviews, we approached 163 consecutive 3-year survey participants stratified by depression, race/ethnicity, gender, and intervention; 104 participated (64%). We notified 289 staff completing 12-month surveys and used purposive follow-up stratified by community, service sector, and intervention, to complete 51 interviews. Aim 4: We interviewed 49 county and community leaders planning HNI and 14 of 34 leaders attending a national briefing.

Data Sources: Self-report surveys (Aims 1 and 2) and semistructured interviews (Aims 3 and 4).

Measures: Client outcomes are poor MHRQL (12-item Mental Composite Score ≤40) and depression (PHQ-8 ≥10) (primary); PHRQL (12-item physical composite score, mental wellness, and behavioral health hospitalization nights (community-prioritized); and use of outpatient services (secondary). Clients and providers were asked about priorities for care, and national stakeholders for responses to CPIC findings and related policy initiatives.

Analyses: We conducted 3-year end-status intent-to-treat analyses, with intervention status as the main independent variable, adjusted for baseline status and covariates, with response weights and multiple imputation (Aim 1); subanalyses of intervention effects within service sector at each follow-up (Aim 2); sensitivity analyses (raw data, longitudinal modeling, and sector-by-intervention interactions); and thematic analyses of qualitative data (Aims 3 and 4).

Results: Aim 1: CEP vs RS did not affect 3-year depression or MHRQL, modestly improved PHRQL, reduced behavioral health hospital nights, and increased having any social-community depression service and mood-stabilizer use. Aim 2: CEP vs RS improved some primary and secondary outcomes within health care and community sector clients at 6 or 12 months with no significant sector-by-intervention interactions. Significant 36-month interactions suggested greater increases under CEP than RS in primary care and self-help visits, antidepressant use, and probable appropriate depression treatment, for enrollees from community compared with health care sectors; and in community depression services for enrollees from health care compared with community sectors. Aim 3: Minority clients prioritized physical and mental health, housing, relationships, and employment. Providers prioritized housing, employment, and mental health. Aim 4: HNI leaders emphasized addressing social determinants and mental health; national stakeholders concurred and saw local initiatives informed by CPIC as innovative policy models.

Conclusions: For CEP relative to RS, there were no long-term effects on primary mental health outcomes, but modest improvements in PHRQL and reduced behavioral health hospital nights, and greater increases in depression services in health care settings for community sector enrollees and in community-based settings for health care sector enrollees. Clients, providers, and national stakeholders prioritized mental health and underlying social factors.

Limitations and Subpopulation Considerations: The study was limited to 2 communities, used self-report measures, and had moderate follow-up (60%). The significance of some findings was sensitive to modeling. The study included a largely minority (African American and Latino) subpopulation of adults screened for depression in underresourced communities.