Comparison of Patient Experience Between Primary Care Settings Tailored for Homeless Clientele and Mainstream Care Settings

Abstract

Background: More than 1 million Americans receive primary care from federal homeless health care programs yearly. Vulnerabilities that can make care challenging include pain, addiction, psychological distress, and a lack of shelter. Research on the effectiveness of tailoring services for this population is limited.

Objective: The aim was to examine whether homeless-tailored primary care programs offer a superior patient experience compared with nontailored ("mainstream") programs overall, and for highly vulnerable patients.

Research design: National patient survey comparing 26 US Department of Veterans Affairs (VA) Medical Centers' homeless-tailored primary care ("H-PACT"s) to mainstream primary care ("mainstream PACT"s) at the same locations.

Participants: A total of 5766 homeless-experienced veterans.

Measures: Primary care experience on 4 scales: Patient-Clinician Relationship, Cooperation, Accessibility/Coordination, and Homeless-Specific Needs. Mean scores (range: 1-4) were calculated and dichotomized as unfavorable versus not. We counted key vulnerabilities (chronic pain, unsheltered homelessness, severe psychological distress, and history of overdose, 0-4), and categorized homeless-experienced veterans as having fewer (≤1) and more (≥2) vulnerabilities.

Results: H-PACTs outscored mainstream PACTs on all scales (all P<0.001). Unfavorable care experiences were more common in mainstream PACTs compared with H-PACTs, with adjusted risk differences of 11.9% (95% CI=6.3-17.4), 12.6% (6.2-19.1), 11.7% (6.0-17.3), and 12.6% (6.2-19.1) for Relationship, Cooperation, Access/Coordination, and Homeless-Specific Needs, respectively. For the Relationship and Cooperation scales, H-PACTs were associated with a greater reduction in unfavorable experience for patients with ≥2 vulnerabilities versus ≤1 (interaction P<0.0001).

Conclusions: Organizations that offer primary care for persons experiencing homelessness can improve the primary care experience by tailoring the design and delivery of services.

Figure 1:. CONSORT Diagram of Inclusion/Exclusion Criteria and Survey Respondents in the PCQ-HoST study.

* Ineligible PC locations are clinics in locations without any Homeless-Patient Aligned Care Teams † No address or telephone contact information in either VA administrative records or in commercially available address verification software. No attempt was made to reach these HEV. ‡ HEV who could not be reached by either mail (address missing or non-deliverable) or phone (incorrect number or disconnected) § Percentage is calculated based on a denominator (n=14340) which excludes the 316 HEV that were unable to receive survey

Figure 2:. Unfavorable Experience and Vulnerability Count for Homeless-Experienced Veterans

Figure shows the percentage of homeless-experienced veterans (HEVs) qualifying for unfavorable experience on each of 4 patient-reported Primary Care Quality-Homeless (PCQ-H) scales, among all respondents regardless of clinic type: Relationship (≥3 unfavorable responses), Cooperation (≥2 unfavorable responses), Access/Coordination (≥3 unfavorable responses), and Homeless-Specific Needs (≥1 unfavorable responses). Vulnerabilities (minimum=0, maximum=4) were counted based on survey responses as follows: severe chronic pain (pain score ≥7 and bodily pain lasting longer than 3 months), unsheltered homelessness (self-report of ≥1 nights unsheltered in prior 6 months), severe psychological distress (score on a modified Colorado Mental Health Symptom Index score of ≥10 of 24), and self-report of drug or alcohol overdose in the preceding 3 years (self-report of having “an overdose where you needed to go to the emergency room or get medical care right away”).

Figure 3:. Interaction of Clinic Type and Vulnerability Status for Unfavorable Experience

Figure shows the percentage of respondents qualifying for unfavorable experience on each of the 4 patient-reported Primary Care Quality-Homeless (PCQ-H) scales: Relationship (≥3 unfavorable responses), Cooperation (≥2 unfavorable responses), Access/Coordination (≥3 unfavorable responses), and Homeless-Specific Needs (≥1 unfavorable responses). Low and high vulnerability were designated based on having ≤1 or ≥2 of the following: severe chronic pain (pain score ≥7 and bodily pain lasting longer than 3 months), unsheltered homelessness (self-report of ≥1 nights unsheltered in prior 6 months), severe psychological distress (based on combining the 4-item PHQ-4 and 2 items related to psychotic symptoms from the Colorado Mental Health Index, with a score of ≥10), and self-report of drug or alcohol overdose in the preceding 3 years (self-report of having “an overdose where you needed to go to the emergency room or get medical care right away”). Each multivariable-adjusted model is adjusted for age, gender, race, Hispanic/Latino ethnicity, chronic homelessness, low income (<$1000 per month), social support, number of medical conditions, the presence of an alcohol or drug use problem on the Two-Item Conjoint Screener, and receipt of psychiatric medication. For all 4 scales, p<0.0001 for the clinic type (H-PACT versus Mainstream) and vulnerability level (low versus high). The interaction of clinic type by vulnerability was significant at p<0.05 only for the Relationship and Cooperation scales.