"Aligning Resources to Care for Homeless Veterans" (ARCH) will study ways to best organize and deliver primary care for homeless Veterans. The investigators will assess 4 different adaptations of the PACT primary care model in a mixed methods study that includes multi-center, randomized-controlled trials of embedded peer-mentoring within different iterations of the PACT model, focus groups of study participants assessing satisfaction, treatment engagement and self-efficacy within the different care models and a cost-utility analysis to determine the most cost-efficient approach to organizing care for this population. Findings from this study will help determine optimal care approaches for reducing emergency department visits and acute hospitalizations, increasing patient satisfaction, and improving chronic disease management. Findings from this study will also substantively add to our understanding of health seeking behavior and the care of vulnerable/high-risk Veteran populations as well as clinical systems design. This project reflects a true "field-based study" to identify optimal and feasible approaches to patient care within our current VHA system. Finally, it will help inform pressing policy issues relevant to two identified T-21 priority areas: Ending Veteran Homelessness in 5 Years and Transforming to a Patient Centered Primary Care model.

Background:

Primary care, and specifically primary care directed to homeless Veterans represents an opportunity to engage individuals in care, address unmet health needs and facilitate receipt of services necessary to exit homelessness. However, it is unclear what the best and most cost-efficient approach is to providing this care. Past research suggests two alternative approaches to organizing and delivering primary care to homeless Veterans: (1) structurally realigned and organized care and (2) embedded peer mentoring. The overall purpose of our research is to compare and contrast outcomes from 4 different adaptations and combinations of primary care delivery to homeless Veterans within the construct of the Patient Aligned Care Team (PACT) model for primary care.

Objectives:

1. To test whether a peer mentor intervention embedded in the Patient Aligned Care Team (PACT) model will be more effective than usual-care PACT or, in a separate randomized controlled trial, within a homeless-oriented PACT (H-PACT) model, in reducing emergency department use and hospitalizations, improving chronic disease management, and increasing participation in homeless programming.
2. To compare clinical outcomes, service use, treatment engagement, self-efficacy, and patient satisfaction of participants in usual care-PACT with and without peer mentoring to H-PACT with and without peer mentoring.
3. To determine differential costs and cost offsets associated with each PACT model adaptation in relation to care outcomes for homeless Veterans.
4. To determine whether a structurally adapted health care delivery model for homeless Veterans (homeless PACT) affects treatment engagement, as measured by utilization of services over time, compared with assignment to a general population Patient Aligned Care Team or no primary care assignment.

Methods:

Substudy #1- Two multi-center Randomized Controlled Trials: The first comparing PACT to PACT+Peer Support (PACT+P); and the second comparing Homeless-oriented PACT (H-PACT) to H-PACT+Peer Support (H-PACT+P). Within each site we will conduct a 1:1 RCT of embedded peer support.

Substudy #2- A qualitative study using focus groups of study participants from each of the intervention arms to assess perceptions of care, treatment engagement, and satisfaction within each approach. These findings will be triangulated with survey data and conditional logistic regression modeling to address the question of how each model is perceived by those receiving care within it and what outcomes can be ascribed to each care approach. This submission will occur at the end of Year 2 of the project and be specific for the focus group activities.

Substudy #3- Cost-Utilization Analysis Study: We will conduct a cost-utilization analysis assessing cost offsets using CPRS, DSS, and PCMM labor mapping data to develop cost models for each care approach.

Substudy #4- VINCI Data Extraction & Natural Language Processing: Use VINCI to analyze for PACT and H-PACT emergency department visits, including diagnosis, whether substance abuse was a factor, whether it resulted in a hospital admission, and what type of aftercare occurred (primary care follow-up, case manager telephone call note, etc.); hospital admissions (diagnosis, length of stay, and aftercare follow-up), ambulatory care utilization (primary care, mental health, specialty clinics, outpatient substance abuse treatment, and homeless programming - VRRC), including both face-to-face and remote-based care (My HealtheVet, telehealth, telephone notes), medication compliance with continuous prescriptions (i.e. insulin, antihypertensives), and chronic disease monitoring and management (blood pressure, diabetes care, hyperlipidemia in heart disease and diabetic patients). Baseline utilization (prior 6 months) of emergency department, inpatient and primary care prior to cohort tracking will be conducted to allow for post-hoc stratification of patient subgroups based on predicted risk for high use patterning.

Inclusion Criteria:

• The study population will be homeless Veterans enrolled in primary care (including both new and established patients who are homeless at the time of enrollment).
• Currently homeless to include: unsheltered; staying in an emergency shelter; in transitional/Grant and Per Diem housing; or doubled-up with a family member or friend and not paying rent.

Exclusion Criteria:

• Currently enrolled in Mental Health Intensive Case Management (MHICM) or other VA-based case/care managed program;
• Stated plans to leave the area within 6 months of enrollment;
• Unable or unwilling to provide informed consent;
• Pregnant women will because excluded because we do not wish to detract from the amount of specialty care and services they receive and need.

Adverse event data were collected during active enrollment in the randomized controlled trial, May 2012 - March 2014 (22 months).

We submitted an amendment about adverse events. It outlined "expected AEs" commonly seen in the homeless population. As this study is considered "low risk" and unrelated to specific disease management, we defined the medical management issues/expected hospitalizations we anticipated among our homeless cohort. In accordance with all VA policies, we report all unexpected events occurring during study participation, events impacting research staff, and any events involving research privacy.

All AE and SAE monitoring was discussed with and approved by all regulatory authorities (IRB, DSMB, and research compliance officer).