Community Paramedicine Program Cuts Hospital Admissions for Low-Income Seniors, Trial Shows

When two specially trained paramedics set up a weekly drop-in clinic in the lobby of a subsidized apartment building, the goal is simple: catch health problems before they become emergencies. A new cluster-randomized controlled trial published in Frontiers in Health Services suggests that this modest intervention, known as the Community Paramedicine at Clinic program, or CP@clinic, does more than improve individual health behaviors. Drawing on linked administrative health records for 3,751 older adults across 30 Ontario apartment buildings, the study found that buildings receiving the program had significantly lower hospital admission rates than matched control buildings, and that this effect persisted for a full year after the formal trial ended, even as the program shifted to a more pragmatic delivery model.

The research team, led by Gina Agarwal of McMaster University’s Department of Family Medicine, designed the study to answer a question that haunts community health interventions: do the benefits survive once the research spotlight fades? The original trial had already shown that CP@clinic reduced emergency medical services calls by 0.90 calls per month per 100 apartment units and improved health-related quality of life among residents. But the researchers hypothesized that the program’s impact on broader health service utilization would also be sustained beyond the 12-month randomized trial period, when paramedic services were no longer following the trial protocol and instead continued a usual community wellness clinic care model at frequencies they themselves determined.

The study employed a longitudinal, repeated cross-sectional design nested within an open-label cluster-randomized controlled trial. Thirty apartment buildings across five Ontario communities, home to low-income older adults whose rent was capped at 30 percent of household income, were paired based on similar size, demographic characteristics, and geographic location. One building in each pair was randomly assigned by a computer-generated tool to receive the CP@clinic intervention, while the other continued with usual care. Because the intervention was a voluntary drop-in program, masking residents was not feasible, though control residents were not explicitly told they were in the control group and intervention residents were not informed of the trial’s outcome measures.

The intervention itself was structured around health promotion and disease prevention. Each week, two specially trained paramedics offered a half-day drop-in session in a common area of the building, where they conducted health risk assessments using validated tools, including cardiometabolic risk screening and social determinants of health questionnaires. Based on the results, paramedics worked with residents to address key risk factors through tailored education, goal setting, and referrals to community resources such as homecare services. Residents were encouraged to return to monitor progress, and for those with a family physician, assessment results were shared with the physician after obtaining consent. All interventions ran between July 1, 2015, and November 30, 2016, with staggered start dates across the sites.

To measure the program’s effects, the researchers turned to ICES, an independent nonprofit research institute authorized under Ontario’s health privacy law to collect and analyze health data for research. Postal codes of the participating buildings were shared with ICES, which retrieved and securely analyzed demographic and health care records linked at the person level using unique, encoded identifiers. The sample included all residents aged 55 and older in the study postal codes, drawn from the Ontario Health Insurance Plan Registered Persons Database, with exclusions for people aged 120 or older, those ineligible for OHIP, and those with no health care encounters in the preceding seven years, a proxy for unrecorded emigration. The analysis followed an intention-to-treat approach at the building level.

The primary utilization outcome was emergency department visits by ambulance, with additional outcomes including overall emergency department visits, non-elective hospital admissions, the number of days admitted, and primary care visits. Emergency department data came from the Canadian Institute for Health Information’s National Ambulatory Care Reporting System, while urgent and emergent hospital admissions were identified in the Discharge Abstract Database using an episode-of-care method that grouped inter-hospital transfers occurring within 24 hours into a single admission. Primary care use was captured through OHIP physician billing records, counting multiple billings by the same physician for the same patient on the same date as one visit. Statistical comparisons used Generalized Estimating Equations adjusted for building pairing and each building’s baseline outcome value, with a first-order autoregressive correlation structure.

The results were striking in one domain and nuanced in others. After adjustment, CP@clinic buildings showed hospital admission rates 25 percent lower than usual care buildings during the 12-month trial period, with an adjusted incidence rate ratio of 0.75 (95 percent confidence interval 0.61 to 0.91, p equals 0.005). Crucially, the effect persisted in the 12 months after the trial, when the intervention was delivered pragmatically and participation rates remained stable: the post-trial ratio was 0.82 (95 percent confidence interval 0.70 to 0.96, p equals 0.017). Primary care visits told a two-phase story. During the trial, CP@clinic buildings had slightly fewer primary care visits (aIRR 0.93, confidence interval 0.88 to 0.99), but in the post-trial period visits rose above control levels (aIRR 1.05, confidence interval 1.00 to 1.10). No significant differences emerged in emergency department visits, ambulance-transported visits, or days admitted in either period.

The descriptive data reinforced the pattern. Monthly emergency department visits per 100 residents in CP@clinic buildings drifted down from 9.49 before the trial to 9.25 after it, while usual care buildings climbed from 7.49 to 8.74 over the same span. Non-elective hospital admissions fell from 3.62 to 1.82 per 100 residents per month in intervention buildings, compared with a rise from 1.63 to 2.25 in control buildings. A post-hoc sensitivity analysis excluding two intervention building pairs deemed ineligible partway through the trial, one because it shared a postal code with assisted-living units and another because a neighborhood crime surge following a police station closure made it incomparable to its match, strengthened the findings further, showing significantly reduced ambulance-transported emergency visits during the trial (aIRR 0.81) and even lower hospital admission ratios of 0.66 during the trial and 0.78 afterward.

The researchers interpret the sustained admission reduction and post-trial primary care increase as evidence of a shift from reactive to preventive care among social housing residents. Fewer hospitalizations, they note, can indicate better chronic disease management and earlier community intervention, consistent with the original trial’s finding that CP@clinic facilitated antihypertensive medication initiation. The rise in primary care visits is particularly promising, since greater use of primary care is associated with improved outcomes for ambulatory care sensitive conditions and reduced reliance on emergency services. As for the null finding on ambulance-transported emergency visits, the authors point out that many 911 calls do not result in transport: a Canadian population-based study found 19 percent of ambulance calls did not end in hospital transport, and Ontario’s Auditor General reported that over 25 percent of dispatched ambulances did not transport a patient. Among those transported, acuity may have been lower given the reduced admissions, and 63.7 percent of patients brought to Ontario emergency rooms by ambulance in 2019 were discharged without admission.

The study’s strengths include its combination of a pragmatic cluster-randomized design with routinely collected administrative data, eliminating the need for primary data collection, and its generalizability to other jurisdictions with publicly funded health care and social programs, such as the United Kingdom, the United States, and Australia. Limitations include the potential for misclassification of residents due to out-of-date addresses in the provincial registry, though matched timing and geography likely mitigated bias, and the fact that post-trial delivery frequency varied by paramedic service, a difference the trial was not powered to compare. The authors conclude that the lasting impact beyond the trial period suggests community paramedicine programs can improve health system efficiency, primary care access, and reduce health inequities in vulnerable populations, with future research needed to determine how session frequency can be optimized or reduced over time without losing benefit.

Subject of Research: Effect of the CP@clinic program (Community Paramedicine at Clinic program) on health care utilization from administrative data during and after the intervention: a cluster-randomized controlled trial

Article Title: Effect of the CP@clinic program (Community Paramedicine at Clinic program) on health care utilization from administrative data during and after the intervention: a cluster-randomized controlled trial

Article References: Agarwal, G., Pirrie, M., Vanama, M. S., Hu, Z. J., Paterson, J. M., Thabane, L., Angeles, R., & Marzanek, F. (2026). Effect of the CP@clinic program (Community Paramedicine at Clinic program) on health care utilization from administrative data during and after the intervention: a cluster-randomized controlled trial. Frontiers in Health Services, 6, Article 1939050. https://doi.org/10.3389/frhs.2026.1939050

Image Credits: AI Generated

DOI: 10.3389/frhs.2026.1939050

Keywords: community paramedicine, CP@clinic, hospital admissions, primary care, social housing, older adults, cluster-randomized controlled trial, preventive care, health services utilization, Ontario, Effect, clinic