Dissertation (Doctoral) Paying for Primary Care: The Relationship between Payment Change and Primary Care Physician Behaviour


David Rudoler


The Ontario government introduced several alternatives to fee-for-service (FFS) payment for primary care physicians (PCPs), including age-sex adjusted capitation (CAP). There is concern that this reform encouraged PCPs to avoid sicker, more complex patients due to incentives inherent in the new payment schemes. This dissertation analyzed the extent to which this occurred.

We conducted a series of studies using patient and PCP level administrative data (1999/00 – 2010/11) from the Institute for Clinical Evaluative Sciences to analyze changes in patient cost and case-mix across payment models in Ontario. Our data captured all Ontarians and PCPs in FFS, enhanced- FFS and CAP payment models during the study period. To ensure our findings were robust we analyzed PCP characteristics associated with selection into different payment models, and controlled for this selection effect in our analysis of the impact of the payment incentives. In our studies, we used both non-parametric (relative distribution) and parametric methods (multinomial selection models).

Our results demonstrate that there are differences in cost and case-mix across payment models and that these differences were both pre-existing and a result of responses to payment incentives. PCPs who self-selected into CAP models were more likely to have healthier and wealthier patients than PCPs in FFS models. Furthermore, we found that PCPs who treated healthier patients, benefited financially from payment reform. We did find evidence that PCPs altered the composition of their rosters because of payment incentives; however, the effects were small and largely accounted for by PCP self-selection effects. The mixed CAP payment scheme allowed PCPs to treat sicker patients off roster, which may have ameliorated some of the financial incentives to risk select.

Our findings demonstrate that PCPs did respond to financial incentives. In particular, PCPs self-selected into payment models, based on pre-existing characteristics that allowed them to maximize their preferences for consumption and leisure. We did not find strong evidence in support of risk-selection, but did find evidence that age-sex adjusted capitation disproportionately benefited PCPs with pre-existing patient rosters that were less complex. Future reforms to PCP payment should take into account how PCPs sort themselves into payment models, and better account for the heterogeneity of PCP and patient populations.


Raisa Deber


Audrey Laporte