A Study of 12.7 Million Ontarians Measured the Outcomes Associated with Primary Care Unattachment. Mortality Odds Were 85 Percent Higher Among Those Unattached Two to Five Years.

Analysis by DPC Insider — Research Desk
#research#primary care access#continuity of care#patient outcomes#attachment#mortality
Sources: Primary care unattachment: impact on mortality, hospitalizations and costs | Health Affairs Scholar · Direct Primary Care 2024 Data Brief | AAFP · Primary Care Unattachment; Impact on Mortality, Hospitalizations and Costs | ICES

Primary care researchers have long argued that continuity of care matters. A February 2026 study published in Health Affairs Scholar tried to measure how much.

The research team used health administrative data from 12,726,325 residents of Ontario, Canada, to examine how the duration of primary care attachment and unattachment influences mortality, hospitalizations, and costs. The central finding: people who were unattached from primary care for two to five years had 85 percent higher odds of all-cause mortality compared with those who had been continuously attached for fifteen years or more.

The Study

Fitzsimon, St-Amant, Green, Glazier, Gayowsky, Premji, Frymire, and Bjerre used data from ICES, an independent nonprofit research institute that maintains linked population health data for Ontario. The cohort of 12.7 million makes it a population-scale retrospective study in the primary care attachment literature.

“Attachment” in the Ontario context encompasses both formal enrollment with a specific family physician and informal affiliation with a primary care clinician or team. Ontario’s healthcare system has structured patient enrollment models, and the province maintains tracked affiliation status — but the study’s construct is broader than formal enrollment alone. The research classified individuals along a spectrum from long-term attached (fifteen or more years) through multiple durations of unattachment.

The unattached population in Ontario is not a small residual. The province has faced a worsening primary care shortage for several years, and a substantial share of residents cannot find a family physician. Some fall into unattachment by circumstance rather than choice — the study mentions clinician retirement, relocation, and career changes as illustrative factors.

The Findings

The headline number is the 85 percent higher odds of all-cause mortality for people unattached two to five years relative to those attached fifteen or more years.

That gap widened considerably for people managing multiple chronic conditions. Among multimorbid patients who were long-term attached, all-cause mortality odds were already roughly five times higher than for attached patients without comorbidities. For multimorbid patients who were recently unattached, the odds reached approximately twelve times higher.

The hospitalization findings followed a similar directional pattern to mortality. The cost findings are more complicated. For people unattached less than two years, the study found lower total healthcare costs than the long-term attached reference group — a cost ratio of 0.76 that the authors themselves flag as paradoxical, likely reflecting deferred or forgone care rather than a genuine reduction in healthcare need. Costs increased with longer duration of unattachment, reaching higher levels for those unattached fifteen or more years. Among highly multimorbid patients in the two-to-five-year unattachment band, cost ratios reached 22.83. The cost story is a time-dependent curve, not a flat rejection of any single assumption.

What the Study Can and Cannot Say

Three limits on generalizability are worth stating directly.

Ontario operates under universal coverage. Every person in the study was OHIP-eligible, which means the measured effect is not attributable to insurance access barriers. In the United States, primary care avoidance involves both physician availability and financial access. The Ontario data removes insurance access as a variable — what happens to a fully covered population that still cannot connect with a primary care clinician. The two mechanisms may compound in the United States, but the study cannot address that.

The design is observational and retrospective. The study notes that administrative morbidity scores may underestimate illness burden in unattached patients — fewer clinical encounters mean fewer coded diagnoses — so the risk adjustment may not fully account for unmeasured illness in the unattached group. That is a limit on how much of the mortality gap can be attributed to unattachment itself versus unmeasured health differences between groups.

Documented affiliation status is what the study measures, not the depth or quality of the relationship. Two patients both classified as “attached for fifteen years” may have meaningfully different relationships with their clinician. Whether the mortality benefit lies in the affiliation status itself or in what that status tends to produce — longer visits, better chronic disease monitoring, a clinician who knows the patient’s history — is a question the administrative data cannot answer.

What the study establishes, within those limits, is the scale of the association. A retrospective population cohort of 12.7 million, with a mortality odds gap large enough to take seriously even accounting for residual confounding. The study does not prove that attachment causes lower mortality. It establishes that the two are associated at a scale that warrants treating the relationship as real.

Where DPC Fits

This study was designed to quantify what happens when the underlying construct — documented affiliation with a primary care clinician — is absent. The research was not designed to evaluate any particular care delivery model.

That construct is what DPC is structurally designed to build. AAFP data puts the average DPC practice panel at about 400 patients, well below typical primary care panel sizes. Same-day appointments and direct phone and text access are structural features that keep the relationship active between acute episodes. The panel size makes sustained affiliation more achievable — it is harder to maintain the kind of connection the study’s long-term attached group represents when a physician carries a substantially larger patient list.

None of that guarantees the fifteen-year attachment the study’s best-outcome comparison group represents. A DPC practice that opened three years ago cannot manufacture that longitudinal history. What it can do is remove structural barriers — crowded panels, insurance-gated access, brief visits — that make sustained affiliation harder to build.

The Ontario study doesn’t validate DPC. The research design doesn’t allow it. What it does is put a number on the thing DPC practices are trying to build, and the number is 85 percent.