Lead/Presenter: Aerin deRussy,
Birmingham VA Medical Center
All Authors: deRussy AJ (Birmingham VA Medical Center), Jones AL (VA Salt Lake City Health Care System; University of Utah School of Medicine, Salt Lake City, UT), Gordon AJ (VA Salt Lake City Health Care System; University of Utah School of Medicine, Salt Lake City, UT) Hoge AE (Birmingham VA Medical Center, Birmingham, AL) Austin EL (Birmingham VA Medical Center; University of Alabama at Birmingham School of Public Health, Birmingham, AL) Montgomery AE (Birmingham VA Medical Center; University of Alabama at Birmingham School of Public Health, Birmingham, AL) Holmes SK (Birmingham VA Medical Center, Birmingham, AL) Kim YI (Birmingham VA Medical Center; University of Alabama at Birmingham School of Medicine, Birmingham, AL) Kertesz SG (Birmingham VA Medical Center; University of Alabama at Birmingham School of Medicine, Birmingham, AL)
Health systems seeking patient assessments of care face challenges due to nonresponse from clients who are economically vulnerable or residentially unstable. For example, VA's Survey of Healthcare Experiences of Patients (SHEP) obtains a 23% response rate from homeless-experienced Veterans (HEVs). We report on recruitment in the PCQ-HoST study to examine practices with potential to optimize survey response among HEVs.
We developed an 11-page survey to profile HEVs' perceptions of primary care. We identified HEVs with evidence of homelessness and primary care use from 26 VA facilities. An independent survey contractor used a commercial database, MelissaData (MD) to verify and update addresses and recruit HEVs through 4 mailings and up to 5 phone calls. We describe survey response overall and by address source. A mixed effect logistic regression model examined associations of survey response with predisposing, enabling, and need characteristics from VA administrative records.
Of the initial VA addresses, > 50% required reformatting/alteration to meet US Postal Service standards. Among 14,340 candidates, 12,635 (88.1%) surveys were mailed based on VA-supplied addresses; 1,557 (10.9%) were sent to new MD-identified addresses; 148 (1.0%) only had telephone contact information. During recruitment, 10,879 (75.9% of 14,340) VA addresses and 953 (6.6%) MD addresses required no further updates. 1,566 (10.9%) addresses were updated by outside sources (Veteran, US Postal Service, online search), and 941 (6.6%) were returned by mail with no address alternatives. Of the 14,340 HEVs with contact information, 5,766 (40.2%) responded: 5,337 by mail and 429 by phone. Survey response was higher for HEVs with VA addresses (46.3%), versus HEVs with MD-updated addresses (31.3%) or outside sources (27.1%). Respondents were more likely than non-respondents to be female (Adjusted odds ratio [AOR] = 1.18, p < .01), and > = 50 years (AOR = 2.0, p < .001); respondents were less likely to have psychosis or drug use disorder diagnosis (AORs = 0.72 and 0.84, respectively; p < .001), or to receive VA sheltering services (AOR = 0.86, p = .001).
A 40.2% response rate was obtained when VA address records were supplemented by address verification, tracking/updating procedures and telephone follow-up, for an 11-page survey.
Modifications to VA's standard patient experience survey methodology could foster greater representation of hard-to-reach Veteran groups. [Funding: IIR15-095]