Setting and context
Primary care in BC, Canada, is publicly funded and privately delivered. Primary care physicians, who deliver the vast majority of care, have traditionally worked in solo or small-group physician-owned and physician-operated practices. They are paid by the provincial health insurance plan on an FFS basis. Physicians pay overhead and staffing costs from gross FFS billings. There is a limited role for interdisciplinary teams or non-physician providers such as registered nurses and nurse practitioners. This is beginning to shift as the Ministry of Health implements some structural reforms (though these occurred after the data for this study were collected) [23,24,25].
BC is divided into five health regions. The largest region by population, Vancouver Coastal Health (VCH), provides services to 1.25 million British Columbians (25% of the provincial population) across 12 municipalities and four regional districts. It operates a network of hospitals and acute care facilities, specialized community health centers, residential care facilities, and home support services.
Data for this cross-sectional study were drawn from information that is routinely collected as part of VCH’s annual privileging process, which physicians are required to complete to confirm their eligibility to provide any services at a VCH facility. The research team designed a survey to address the annual privileging and workforce planning needs within VCH and to elicit physician perspectives on their models of practice. The survey was reviewed and approved by the Regional Medical Director of Primary Care for VCH.
The survey captures information on how physicians are currently structuring their practices, and asks additional questions that support health authority-wide human resource planning such as planned retirements. The data were linked with publicly available demographic and training information from the College of Physicians and Surgeons of BC, including gender, year of graduation, and training location. While the privileging process is required, the survey component is optional.
Physicians who had clinical privileges with VCH in 2018 were invited (via email) to participate in the survey. Participation reminders were sent 1 and 5 weeks following the initial invitation. Responses were collected between January 30 and April 15, 2018. Six weeks after the initial invitation, physicians were also invited to have their data removed if they had reconsidered since first answering the survey. The survey was delivered online using REDCap. Researchers were presented with de-identified data for analysis.
Survey content and variables
We asked respondents to select between four options that describe their main model of primary care practice: all or some community-based primary care (CBPC), hospital or inpatient facility only, locum only, or non-clinical (provide no patient care). We grouped those respondents who indicated that they provide at least some CBPC based on the number of self-reported hours per week spent working in locations where CBPC is provided (solo, two to four, or five + physician clinics in the community that a patient can access without a referral). The three groups used were full-time CBPC (> 37.5 h per week), mostly CBPC plus some other work (20–37.5 h per week CBPC), and mostly other work plus some CBPC (< 20 h per week CBPC). We also asked that they indicate whether their main model of practice was focused, general, or mixed (general but also with a specific clinical focus), and asked about the provision of specific services and care to special populations.
The survey also included a core set of demographic and work questions, including hours worked per “typical” week (broken into specific tasks areas such as patient care, paperwork, and business activities), number of practice locations, retirement intentions, and responsibilities for call coverage.
Statistical analyses
We excluded respondents who reported that they do not provide any patient care or who did not respond to core questions about model of practice. To determine the representativeness of our sample, we compared demographic (year of graduation and gender) and training (location) characteristics for survey respondents with the total population of primary care physicians practicing within BC area, using χ2 tests. These data are presented in Supplement 2. Data for this comparison were drawn from the College of Physicians and Surgeons of BC public listing.
For descriptive analysis, we used one-way ANOVA and χ2 tests to compare demographic and practice characteristics across core models of practice (some/all CBPC, hospital/facility only, or locum) and among physicians who provide at least some CBPC (less than half time, more than half time, or full time). We used the same test to compare demographic and practice characteristics between new graduates (physicians in practice less than 10 years) and more established physicians.
To assess the adjusted effect of demographic and practice characteristics on model of practice, we used a two-stage modeling approach. We used a multivariable logistic model to assess which variables were associated with providing at least some CBPC. Then among the subsample of physicians who did report that they provide at least some CBPC, we constructed a second logistic model to assess the effect of the same set of demographic and practice characteristics on whether or not they provide CBPC care full time.
Finally, we conducted a series of χ2 tests looking for differences in provision of specific clinical services across full time, greater than half time, and less than half time spent providing CBPC. Services included maternity care, substance abuse treatments, non-office-based care, serving special populations, and the use of technology to support accessible patient care. All analyses were conducted in Stata IC/15.1.