HSR&D Home » Research » IAA 07-069 – HSR&D Study
Primary Care Quality and Service Customization for the Homeless
Stefan G Kertesz, MD MSc
Birmingham VA Medical Center, Birmingham, AL
Funding Period: September 2008 - August 2012
While the VA has in many areas surpassed mainstream health systems in improving quality or reducing disparities, some reports from the 1990s suggested VA had not met with similar success in assuring primary care access to homeless veterans. Conversely some studies from organizations that tailor primary care services for the homeless suggest that patient-level outcomes are improved in regard to appointment attendance, reductions in some hospital admissions, and improvements in some diseases. The question of how best to align primary care services for the homeless has drawn attention as the Veterans Health Administration recently initiated a 32-site program in tailored primary care. This team sought to assess whether the patient experience was superior in primary care service settings tailored to homeless primary care (tailoring includes aspects of mission, staffing, service location, training and data systems) compared to settings where such tailoring is less prominent or absent. Although numerous patient-reported measures of care exist, none were developed for the unique concerns of persons who have been homeless. Thus, this study first developed a psychometrically valid measure of the primary care experience appropriate to the population, the Primary Care Quality-Homeless (PCQ-H) survey. We then assessed whether ratings of care were superior in two settings that tailor primary care for the homeless, compared to three where such tailoring was less present or absent.
Objective 1: To develop an understanding of the major aspirations for the primary care of homeless patients, as voiced by the patients themselves and experts in their care.
Objective 2: To develop a psychometrically valid survey to assess primary care quality from the perspective of homeless-experienced patients.
Objective 3. To compare primary care quality across 5 locations differing substantively in regard to the organization and delivery of primary care services for persons who have been homeless. We tested this hypothesis: homeless-experienced patients rate primary care more highly when it is obtained in a setting where services are explicitly tailored to the homeless population.
Qualitative interviews were necessary to properly draft a new, psychometrically valid patient survey, the Primary Care Quality-Homeless (PCQ-H) tool. To narrow interview focus to constructs of greatest relevance, 26 homeless-experienced patients and 12 expert providers were asked to prioritize aspects of quality drawn from 2 major reports from the Institute of Medicine, using a cardsort exercise. The resulting 8 constructs were then explored through interviews with 36 homeless-experienced patients (in Birmingham and Boston) and 22 experts in their care. Based on coding of interviews, the team developed a preliminary 78-item survey (PCQ-H version 0), administered to 601 randomly selected patients across 5 organizational settings. These included three VA facilities that did not tailor primary care for the homeless, one tailored VA program, and a 26-year old non-VA primary care program for homeless individuals recognized as a national exemplar. A random sample of homeless-experienced primary care patients was sought through random selection from medical record systems. The 78-item PCQ-H version 0 was administered as part of a battery including measures of other patient characteristics important either to instrument validation analysis or for purposes of casemix control required for statistical comparison of the 5 sites under study.
Analyses occurred in two phases. Phase 1 applied elements of classical psychometrics, confirmatory factor analysis and Item Response Theory to identify a subset of items having high informational value relative to the scales of interest. Phase 2 involved comparison of mean scale scores across sites to test the hypothesis that patients would rate care more highly when obtained from sites that tailor care for homeless clients. This analysis compared continuous scores and a categorical indicator of an unfavorable experience, defined as falling into the lowest tertile of scores across all sites.
Qualitative analyses under Objective 1 yielded a large number of themes of special concern to homeless-experienced primary care patients. As one example, patient participants offered thoughtful reservations concerning the notion promulgated by the Institute of Medicine that patients should have "control" in their care. A voice in care, while sharing control, was a favored motif. Cardinal concerns included trust and respect, and responsiveness of caregivers to certain needs unique to the homeless condition.
Under Objective 2, we arrived at a 33-item instrument with a 4-scale solution comprised of Patient-Clinician Relationship (13 items), Cooperation among clinicians (3 items), Accessibility/Coordination (11 items) and Homeless-Specific Needs (4 items). Collectively these are termed Primary Care Quality-Homeless (PCQ-H) scales. These 4 scales met study criteria for convergent validity, divergent validity and internal reliability.
Site comparisons (Objective 3) are framed in terms of a 4-point scale (higher is better) where the options in response to an item (e.g. "If my doctor and I were to disagree about something related to my care, we could work it out") were Strongly Disagree, Disagree, Agree and Strongly Agree. Across all participants, mean scores fell between 2.5 and 3.5 (generally, participants offered moderate agreement with favorably-worded items), with standard deviations of 0.4-0.5. Scores were lowest for perceptions of cooperation among caregivers. Comparing sites,the highly-tailored non-VA site scored higher than all others, significantly so for Patient-Clinician Relationship (p<0.001), Cooperation (p=0.004), and nearly significantly so for Access/Coordination (p=0.055), but not so for Homeless Specific Needs (0.22). When analyses were restricted to persons reporting Poor or Fair Health (n=256), there was a far greater magnitude in the difference between the highly-tailored non-VA site and the non-tailored VA sites (differences ranged between 1/4 and 1 full standard deviation, all p-values significant <0.05). In general the tailored VA site performed similarly or slightly better than the non-tailored VA sites. When analyses focused on the percentage of individuals with lowest-tertile scores, the differences were large. In absolute terms, at the tailored non-VA site, no more than 1/3 and typically just 20-25% of patients had scores in the lowest tertile. By contrast, at the non-tailored sites, the percentage with lowest-tertile scores was routinely 10%-30% (in absolute terms) higher, a difference that was significant for all 4 measured scales (all p<0.01).
These results support the hypothesis of this study. Homeless-experienced patients rate their primary care more highly in settings that have instituted homeless-tailored primary care through service modifications including outreach/shelter care, specially recruited and trained staff, a homeless-specific identity, and inclusion of formerly homeless persons in governance roles. Additionally, we found that patients were more likely to give unfavorable ratings in regard to cooperation among their caregivers than in regard to other aspects of care, and that the problems with "Cooperation" among caregivers stood out most prominently at all sites we studied.
The Secretary has declared ending veteran homeless a cardinal priority at this time. VA's recent 18-month, 32-site deployment of homeless-tailored Patient Aligned Care Teams is a potential asset to this effort. The current Homeless PACT initiative will continue only if outcomes appear to suggest it results in superior experiences for patients. Our results support the validity of the PCQ-H instrument developed as part of this study. Because it is deeply grounded in homeless patient experiences we believe it should be advanced for approval for use in clinical settings. Such advancement will likely require collaboration with VA Central Office, and other agencies with authority or stakeholder interest in patient surveys including the Office of Management and Budget and the Bureau of Primary Health Care within the US Department of Health and Human Services. The results of our analysis do highlight that homeless patients' primary care patient experiences are measurable and amenable to enhancement through proper service design.
External Links for this Project
Dimensions for VADimensions for VA is a web-based tool available to VA staff that enables detailed searches of published research and research projects.
If you have VA-Intranet access, click here for more information vaww.hsrd.research.va.gov/dimensions/
VA staff not currently on the VA network can access Dimensions by registering for an account using their VA email address. Search Dimensions for this project
DRA: Health Systems
DRE: Treatment - Observational, Technology Development and Assessment
Keywords: Homeless, Organizational issues, Primary care
MeSH Terms: none