Does quality of primary care vary across delivery models? IJPDS (2017) Issue 1, Vol 1:055, Proceedings of the IPDLN Conference (August 2016)

Main Article Content

Alan Katz Dan Chateau Carole Taylor Jeff Valdivia
Published online: Apr 13, 2017


To determine the relationships between five models of primary care service delivery and quality of care indicators in an urban population. Two fee-for-service (FFS) and three alternative-funded models of primary care service delivery were studied

We allocated all Manitoba residents who had at least three visits to any primary care provider (PCP) at any Winnipeg clinic between 2010-2013 to the most responsible PCP (N = 626,264). We then allocated each PCP to a model of primary care service delivery. We created general linear mixed models to describe the relationship between each model of primary care and the dominant, traditional fee-for-service model for health services use, while controlling for a variety of PCP and patient factors, including patient social complexity.

Patient social complexity was associated with poorer crude rates for many of the indicators. There were no differences among the models for hospital readmission within 30 days or specialist referral by the assigned PCP. Hospitalizations for ACSC were higher for one alternative funded model (1.98 OR, 1.38-2.83 95% CI), while non-indicated low back X-rays were lower for a different alternative funded model (0.14 OR, 0.03-0.59 95% CI). Ambulatory care visits to any PCP were lower for all three alternative funded models than the two FFS models. The family medicine academic teaching sites had lower rates of continuity of care (p< 0.5)

Overall, no model of primary care consistently outperformed the others. FFS models had higher rates of visits, but appeared to satisfy patient needs better because they had less use of telehealth services following visits. Teaching sites appeared to sacrifice continuity of care potentially to support other academic activities. Controlling for social complexity was associated with a reduction in the differences between models in indicator outcomes.

Article Details

Most read articles by the same author(s)