JMIR Publications

ABSTRACT

Background:

People experiencing homelessness are at risk for gaps in care after an emergency department (ED) or hospital visit, which leads to increased utilization, poor health outcomes, and high health care costs. The majority of homeless individuals have a cell phone of some type, which makes mobile health interventions a feasible way to connect a person experiencing homelessness with providers.

Objective:

To investigate the accuracy, acceptability, and preliminary outcomes of a global positioning system-enabled mobile health (GPS-mHealth) intervention designed to alert community health paramedics when people experiencing homelessness were in the ED or hospital.

Methods:

This was a pre-post design with baseline and 4-month post-enrollment assessments. A person experiencing homelessness taking at least two medications for chronic conditions who scored at least 10 on the Patient Health Questionnaire-9 (PHQ-9) and had at least two ED or hospital visits in the prior 6 months was eligible. Participants were issued a study smartphone with a GPS app programmed to alert a community health paramedic when a participant entered an ED or hospital. For each alert, community health paramedics followed up via telephone to assess care coordination needs. Participants also received a daily e-mail to assess medication adherence. GPS alerts were compared to ED and hospital data from the local health information exchange (HIE) to assess accuracy. Paired t-tests compared scores on the PHQ-9, Medical Outcomes Study Social Support Survey, and ASK-12 adherence survey at baseline and exit. Semi-structured exit interviews examined perceptions and benefits of the intervention.

Results:

Thirty participants enrolled; the mean age was 44.1 years (SD 9.7). Most were male (67%; n = 20), White (57%; n = 17), and not working (63%; n = 19). The GPS app showed limited accuracy in ED or hospital visit alerts. Only 18.8% of the alerts aligned with HIE data (3/16), mainly due to patients not having the phone with them during the visit, phone being off, and gaps in GPS technology. There was a significant difference in depressive symptoms between baseline (M=16.9, SD=5.8) and exit (M=12.7, SD=8.2); t(19)=2.9, p=.009 and a significant difference in adherence barriers between baseline (M=2.4, SD=1.4) and exit (M=1.5, SD =1.5); t(17)=2.47, p = .025). Participants agreed that the app was easy to use (M=4.4/5 with 5 = strongly agree (SA)) and indicated the e-mail helped them remember to take their medications (M=4.6/5). Qualitative data indicated that unlimited phone access allowed participants to meet social needs and maintain reliable contact with case managers, healthcare providers, family, and friends.

Conclusions:

mHealth interventions are feasible for and acceptable to people experiencing homelessness. Objective data from the HIE provided more accurate ED and hospital visit information, but unlimited access to reliable communication provided benefits to participants beyond the study purpose of improving care coordination. Clinical Trial: Not applicable