3036 — Systematic Methodology for Developing a Patient-Centered Primary Care Measure for Homeless Veterans
Kertesz SG, and Johnson-Roe NK, Birmingham VA Medical Center; Steward J, University of Alabama at Birmingham; Stringfellow E, Boston Healthcare for the Homeless; Austin EL, Birmingham VA Medical Center; Holt C, University of Maryland; Gordon AJ, Pittsburgh VA Medical Center; Davis L, Tuscaloosa VA; Kim T, Boston VA; Pollio DE, University of Alabama
Patient-centered primary care is a central goal of the Patient-Centered Medical Home (PCMH; in VA, Patient-Aligned Care Teams). Existing patient-reported surveys evaluate patient perceptions. However, they are not designed to reflect the unique challenges and concerns of homeless persons. This abstract’s objective is to describe the method used to develop a patient-centered primary care measure specific to the needs of persons who have been homeless.
We undertook a phased, systematic development of a primary care survey specific to the experiences of homeless persons. In Phase 1, constructs relevant to primary care quality were drawn from Institute of Medicine (IoM) reports. In Phase 2, a cardsort with patients and homeless-expert providers prioritized constructs for exploration in interviews. In Phase 3, semi-structured interviews and focus groups were conducted among patients (n = 24) and “leaders” (n = 34 homeless-expert clinicians, administrators, researchers), within and outside VA. In Phase 4, a multidisciplinary team crafted survey items, followed by voting. Cognitive response interviews for item refinement followed, resulting in the Primary Care Quality-Homeless survey (v1.0). Results of administration to 329 participants are reported separately.
12 providers (6 non-VA, 6 VA) and 26 patients (5 non-VA, 21 VA) completed the cardsort, narrowing the IoM constructs from 16 to 8 (access, coordination, cooperation, evidence-based care, accountability, patient control, continuity, shared knowledge). Patients prioritized “shared knowledge and free flow of information” more highly than providers, while giving “patient control” a lower rating. Thematic coding of the interviews identified quotations illustrative of the 8 IoM-derived constructs as well as 3 emergent constructs: trust/respect, homeless-specific needs, and substance abuse/mental illness. Based on these quotes, 877 candidate items were proposed, and reduced to 78 through review and voting by a multidisciplinary team. Cognitive response interviews supported item refinement. Challenges to survey design included identification of a putative source of primary care, as well as tooling item language to avoid presuming mental illness or addiction while still addressing concerns important to patients with these problems.
The development of a quality survey to assess care for a particularly vulnerable VA-priority population can be accomplished through systematic, conceptually-grounded inclusion of patient and provider perspectives at every step of survey design.
Patient-centered measures of homeless persons’ primary care will help to guide VA’s new and substantive investment in novel service models for this population.