Rideshare-Based Medical Transportation for Medicaid Patients and Primary Care Show Rates: A Difference-in-Difference Analysis of a Pilot Program. Chaiyachati, K., H., Hubbard, R., A., Yeager, A., Mugo, B., Shea, J., A., Rosin, R., & Grande, D. Journal of General Internal Medicine, 33(6):863-868, Springer US, 1, 2018.
Rideshare-Based Medical Transportation for Medicaid Patients and Primary Care Show Rates: A Difference-in-Difference Analysis of a Pilot Program [pdf]Paper  Rideshare-Based Medical Transportation for Medicaid Patients and Primary Care Show Rates: A Difference-in-Difference Analysis of a Pilot Program [link]Website  abstract   bibtex   
BACKGROUND: Transportation to primary care is a major barrier for Medicaid patients despite access to non-emergency medical transportation (NEMT) insurance benefits. Rideshare services provided by Uber and Lyft have been proposed as potential NEMT alternatives because they are convenient and low cost. As part of a pilot-program, we designed a rideshare-based transportation intervention for Medicaid patients and tested whether primary care show rates improved. METHODS: Our study population included adults living in West Philadelphia, and were Medicaid patients at one of two Penn internal medicine practices-an intervention and control clinic-within the same building. Intervention clinic patients with a scheduled appointment received a telephone reminder 2 days before and were offered free transportation to-and-from clinic using Lyft. Up to 3 attempts were made to contact the patient. For those who accepted, staff pre-scheduled a Lyft ride using a web-based dispatch tool, circumventing the need for patients to have a smartphone or app. To return home, patients called staff to dispatch a Lyft ride after their clinic visits. Control clinic patients received an appointment reminder 2 days before appointments as part of usual care. For both clinics, we collected appointment attendance data froma period before (Aug 4 - Sep 16, 2016) and during (Oct 12-21, 2016) the intervention. The primary outcome was appointment show rate-the proportion who attended among those called. We employed a difference-indifference analytic approach using logistic regression with robust standard errors to compare show rate changes at the two clinics between the two data collection periods. We adjusted for patient demographics, zip code, and provider types (resident, attending, or mid-level). Our approach accounts for the influence of secular trends in the intervention clinic and time-invariant differences between clinics. RESULTS: The 60 individuals called in the intervention group had a median age of 52 years (IQR 41-58), and were female (70%), non-Hispanic (98%), and black (92%). Out of 45 who answered, 18 (40%) had a ride scheduled, among those 13 (72%) used the service. At the control clinic, the show rate declined from 60% (143/240) to 51% (34/67). At the intervention clinic, the show rate improved from55%(72/132) to 68% (41/60). In the adjustedmodel, those called to receive the intervention had a 2.51 (95% CI: 1.08-5.81, p = 0.032) higher odds ratio of attending clinic. The average cost of per ride was $8.10. CONCLUSIONS: Results of this pilot program suggest that a ride-share service can increase show rates to primary care appointments for Medicaid patients. These findings are not generalizable to non-urban environments and pre-appointment reminders with multiple follow-up attempts may explain these results, but patients received at least one reminder in both arms. Our findings have implications for payers and providers hoping to improve primary care show rates for low-income populations.

Downloads: 0