Abstract
Background: Variations in primary care practices may explain some differences in health outcomes during the COVID-19 pandemic. We sought to evaluate the characteristics of primary care practices by the proportion of patients unvaccinated against SARS-CoV-2.
Methods: We conducted a population-based, cross-sectional cohort study using linked administrative data sets in Ontario, Canada. We calculated the percentage of patients unvaccinated against SARS-CoV-2 enrolled with each comprehensive-care family physician, ranked physicians according to the proportion of patients unvaccinated, and identified physicians in the top 10% (v. the other 90%). We compared characteristics of family physicians and their patients in these 2 groups using standardized differences.
Results: We analyzed 9060 family physicians with 10 837 909 enrolled patients. Family physicians with the largest proportion (top 10%) of unvaccinated patients (n = 906) were more likely to be male, to have trained outside of Canada, to be older, and to work in an enhanced fee-for-service model than those in the remaining 90%. Vaccine coverage (≥ 2 doses of SARS-CoV-2 vaccine) was 74% among patients of physicians with the largest proportion of unvaccinated patients, compared with 87% in the remaining patient population. Patients in the top 10% group tended to be younger and live in areas with higher levels of ethnic diversity and immigration and lower incomes.
Interpretation: Primary care practices with the largest proportion of patients unvaccinated against SARS-CoV-2 served marginalized communities and were less likely to use team-based care models. These findings can guide resource planning and help tailor interventions to integrate public health priorities within primary care practices.
Among people in Canada, concerns about risks and adverse effects are the top reasons for non-intent to receive a SARS-CoV-2 vaccine.1 These trends are most commonly observed among communities that have low levels of trust and confidence in government because of deep histories of marginalization by institutions and harm, including being subjected to unethical medical procedures and experimentation.2
Family physicians can play an important role in increasing vaccine uptake through building vaccine confidence and debunking misinformation among their patients,3,4 as many people in Canada consider family physicians as their most trusted source of vaccine information.3,4 This requires that physicians first proactively identify vaccine-hesitant patients, and then have the necessary communication skills and capacity to engage in conversations that take time and may require several encounters.5 Some family physicians may prioritize such work more than others or may have more capacity to undertake it. Successful partnerships between public health and primary care requires understanding of which practices are in most need of support.
We sought to evaluate characteristics of family physicians with the largest proportion of patients unvaccinated for SARS-CoV-2. We sought to describe these physicians, their practices, and their patients, and to explore characteristics associated with vaccination, with the goal of informing tailored supports that can leverage primary care to support vaccination or similar public health efforts.
Methods
Study design
We conducted a cross-sectional, population-based analysis using linked health administrative data in Ontario, Canada, to assess the characteristics of family physicians with the largest proportion of patients (aged ≥ 12 yr) unvaccinated for SARS-CoV-2, namely patients who had not received any doses of SARS-CoV-2 vaccine as of Nov. 1, 2021. We a priori defined this group as the 10% of physicians who had the highest proportion of unvaccinated patients in their practices, knowing that future efforts to support primary care practices in achieving public health priorities would need to focus on those with the greatest opportunity for improvement.
We compared the group of comprehensive-care family physicians who cared for the largest proportion of unvaccinated patients to the remaining 90% of comprehensive-care family physicians in the province. We used data as of Nov. 1, 2021, just before the Omicron wave of the COVID-19 pandemic. At that time, we were launching an intervention to support primary care physicians in encouraging uptake of SARS-CoV-2 vaccinations among their patients. We designed this study to provide baseline insights for a randomized trial of practice supports to improve vaccination rates (clinical trial no. NCT05099497).
Setting
In Ontario, vaccination against SARS-CoV-2 began in December 2020 for priority groups; by May 2021, vaccines were available to everyone aged 12 years and older. In July 2021, people were eligible for their second dose using the shortened 28-day interval. Therefore by Nov. 1, 2021, everyone in the study cohort had the opportunity to have 2 doses. At the time of this study, SARS-CoV-2 vaccinations were freely available in Ontario in pharmacies, large public health immunization centres, and pop-up immunization clinics, as well as in a small number of primary care clinics. Furthermore, the vaccination of essential workers, their families, and other residents living in COVID-19 hotspots was accelerated and prioritized.6
Ontario has publicly funded health care for medically necessary physician and hospital services for permanent residents, without deductibles, and does not limit patients’ choice of physician. Almost all primary care is delivered by family physicians; 85% of the population is enrolled with a family physician.7
Study population
We included family physicians practising in a patient enrolment model, as opposed to a strictly fee-for-service model. In patient enrolment models, between 15% and 70% of payment for physicians is based on age-and sex-adjusted capitation.8 The enrolment model is meant to reinforce a mutual commitment between patient and physician. Around 80% of family physicians in Ontario work in a patient enrolment model.
Data sources
We used routinely collected administrative data from the Ontario Health Insurance Plan (OHIP) database for physician claims; the Registered Persons Database, which is Ontario’s health care registry for OHIP-eligible patients; the Client Agency Provider Enrolment tables for patients in primary care enrolment models; the Corporate Provider Database for physicians in patient enrolment models; and the ICES Physician Database for physician characteristics. We used the COVaxON database, Ontario’s central point-of care SARS-CoV-2 vaccine management system and database for the entire province. We also used sociodemographic data from the 2016 Canadian Census. Finally, we used the Immigration Refugee and Citizenship Canada (IRCC) Database, which includes people with landed immigrant or permanent resident status at any time from 1985 to 2014,9 to identify recent immigrants, defined as those identified who had immigrated to Canada within the 10 years before Nov. 1, 2021.
These data sets were linked using unique encoded identifiers and analyzed at ICES, an independent, nonprofit research institute whose legal status under Ontario’s health information privacy law allows it to collect and analyze deidentified health care and demographic data, without the need for explicit consent, for health system evaluation and improvement.
A review of Canadian studies of data quality in administrative databases revealed that demographic and clinical data have high levels of completeness and are reliable.10 Data used in this study relied on administrative fee codes for physician payments and, despite small differences in completeness of billing between capitation and fee-for-service practices, the data were sufficiently complete. A small proportion (< 5%) of health services would not have been captured in our data (e.g., interactions with salaried professionals such as nurse practitioners).10 Nevertheless, we are confident that the data are sufficiently comprehensive and valid for this study.
Physician and patient characteristics
For primary care physicians, we collected data on sex, years practising medicine, country of graduation, and total OHIP billings. Practice characteristics included primary care enrolment model (enhanced fee for service, blended capitation, blended capitation with an interprofessional family health team), roster size and payment via fee for service.11 An explanation of funding models and Ontario’s patient enrolment model can be found in Appendix 1, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.230816/tab-related-content. We calculated SARS-CoV-2 vaccination coverage among enrolled patients. We also looked at quality-of-care indicators including screening for colorectal and cervical cancer, and diabetic care (e.g., at least 1 test of low-density lipoprotein cholesterol in the previous 2 years).
Patient demographics included age, sex, public health unit, distance to the rostered physician’s location, and recent immigration status. We also included existing comorbidities (i.e., hypertension, congestive heart failure, diabetes, asthma, and chronic obstructive pulmonary disease) and SARS-CoV-2 vaccination.
We used postal codes at the neighbourhood level, linked to census data to assign income quintiles, marginalization quintiles, and rurality scores. We used Matheson’s Canadian Marginalization Index12,13 to assign marginalization quintiles for 4 components of marginalization — dependency, residential instability, material deprivation, and ethnic concentration — and presented these as a summary score. We assigned rurality categorically into urban areas (score 0–9), small towns (score 10–44), and rural areas (score ≥ 45) according to the Rurality Index of Ontario.14 Variables on health care use included continuity of care (i.e., percentage of primary care visits to the rostered physician, using a 2-year look-back) and any virtual visits in the previous 6 months. We assessed overall health care use using the Johns Hopkins Adjusted Clinical Group, determining resource utilization bands over the previous 2 years, with 0 being no health care use (no comorbidity) and 10 being the highest expected use (high comorbidity).15
Statistical analysis
We calculated the proportion of unvaccinated patients enrolled to physicians and then analyzed patient and physician characteristics, stratified by physicians in the top 10% of unvaccinated patients in their practices and the remaining 90%. We calculated standardized differences and considered differences of 0.1 or greater noteworthy.16
We used random-effects logistic regression models to evaluate variables associated with patient vaccination status. We entered covariates into the model based on what has been shown in the literature to affect preventive care, namely patients’ age, sex, neighbourhood income, neighbourhood ethnic diversity, comorbidity, rurality, and recent immigration status, and physicians’ age, sex, and enrolment model.17–20 We used the clustering of patients within physicians as a random effect. We ran separate models including patients of all physicians, those enrolled with physicians in the top 10% of unvaccinated patients, and those enrolled with physicians in the remaining 90%. The model for all patients included physician group (top 10% or remaining 90%) as a covariate.
We conducted 2 sensitivity analyses. The first defined the top decile as those with the largest number — rather than the largest proportion — of unvaccinated patients, compared with the remaining 90%. The second analysis compared physicians in the top 20% of unvaccinated patients enrolled with the remaining 80%. We used the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) and the Reporting of Studies Conducted Using Observational Routinely Collected Health Data (RECORD) checklists to ensure completeness of reporting of study background, methodology, and results.
Ethics approval
The use of data in this project was authorized under section 45 of Ontario’s Personal Health Information Protection Act, which does not require review by a research ethics board.
Results
We analyzed 9060 family physicians who enrolled 10 837 909 patients. The physicians with the highest proportion of unvaccinated patients (n = 906) cared for 259 130 unvaccinated patients as of Nov. 1, 2021 (Table 1). The proportion of rostered patients who received 2 or more doses of SARS-CoV-2 vaccine as of Nov. 1, 2021, was 74.2% (interquartile range [IQR] 69.2%–76.7%) in this group, compared with 87.0% (IQR 84.1%–89.8%) among patients of the remaining 90% of physicians. Compared with the remaining physicians, physicians with the largest proportion of unvaccinated patients were more likely to be male (64.6% v. 48.1%), to have trained outside of Canada (46.9% v. 29.3%), to be older (mean age 56 yr v. 49 yr), and to work in an enhanced fee-for-service model (49% v. 28%) (Table 2).
Physicians practising in a patient enrolment model, ranked by proportion of enrolled patients unvaccinated against SARS-CoV-2
Characteristics of physicians practising in a patient enrolment model with the highest proportion of patients unvaccinated against SARS-CoV-2 (top 10%), compared with the remaining 90% of physicians
Patients enrolled with physicians in the most unvaccinated group tended to live in places with more ethnic diversity, higher material deprivation, and lower incomes. More patients in this group were recent immigrants. The 2 groups had similar comorbidity indices, rates of in-person and virtual visits in the previous year, and diabetes quality-of-care indicators, but cancer screening indicators such as colorectal cancer screening and Papanicolaou smears for cervical cancer screening were slightly lower among patients enrolled with physicians in the most unvaccinated patient group (Table 3).
Characteristics of patients (aged ≥ 12 yr) enrolled with physicians practising in a patient enrolment model by proportion of patients unvaccinated against SARS-CoV-2
Table 4 summarizes the model-adjusted associations of patient and physician characteristics with vaccination. The patient characteristics most strongly associated with being unvaccinated were younger age and less comorbidity, as well as living in rural locations and living in lower-income neighbourhoods. Physician characteristics that were most strongly associated with unvaccinated patients were older age, being male, and working in an enhanced fee-for-service payment model. Unique models for the top 10% and remaining 90% groups can be found in Appendix 2, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.230816/tab-related-content. The output from these 3 models suggests that the characteristics associated with unvaccinated patients were mostly similar. Two patient characteristics were not consistent across the models. Recent immigration was associated with being vaccinated in the top 10% group, but not in the 90% group; male patients were less likely to be vaccinated in the 90% group but not in the top 10% group.
Random-effect logistics model of characteristics of patients and physicians associated with unvaccinated status*
The sensitivity analysis grouping physicians according to the number of unvaccinated patients led to similar results (Appendix 3, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.230816/tab-related-content). The most noteworthy differences between the primary and sensitivity analyses were observed for the variables of practice size and proportion of vaccinated patients, and are a function of differences in how the deciles were defined. Small differences in variables measuring various aspects of structural marginalization likely also represent differences in the approach to analysis (percentage of unvaccinated patients v. number of unvaccinated patients). Our sensitivity analysis of the top 20th percentile produced similar results; however, differences regarding ethnic diversity were not observed (Appendix 3).
Interpretation
The family physicians in Ontario who had the highest proportion of unvaccinated patients in their practices during the COVID-19 pandemic were more likely to be male, to have trained outside of Canada, to be older, and to be working in an enhanced fee-for-service model. These physicians cared for patients living in areas with more ethnic diversity, more material deprivation, and lower incomes, with a higher proportion of immigrants. The family physicians with the most unvaccinated patients were also more likely to practise in enhanced fee-for-service models and less likely to practise in team-based models, meaning they may have had fewer support staff in their clinics. This illustrates the ongoing inverse relationship between the need for care, and its accessibility and utilization.21 In other words, the practices in highest need receive the fewest resources.22
Patient characteristics associated with not being vaccinated included younger age, having less comorbidity, being male, living in low-income neighbourhoods, and living in neighbourhoods with high ethnic diversity. Studies in the United States23 and the United Kingdom24 have reported similar findings. In the US, SARS-CoV-2 vaccine coverage was observed to be lower among non-Hispanic Black and Hispanic people.23 Similarly in the UK and Europe, lower rates of vaccine coverage were found among those who were Black, lived in the most deprived areas, and those with less comorbidity.24,25
In August 2020, only 4% of the first and second doses of SARS-CoV-2 vaccines were administered in primary care offices in Ontario.26 However, family physicians can influence patients’ decisions through opportunistic discussions during patient encounters. Studies have found that guidance from health care providers plays a crucial role in influencing both general27–29 and SARS-CoV-2-specific vaccine-related decision-making.30,31
Overall, the median percentage of vaccinated patients was high (86%). However, practices with a high proportion of unvaccinated patients may be a viable target group for efforts to coordinate public health and primary care. This approach would be similar to hotspotting, whereby databases are mined to identify patients who have the highest rates of health care system use and who are then prioritized to receive tailored services.32 The hotspotting approach typically includes team-based care with a focus on patient engagement and social determinants of health. Adapting the hotspotting model to support quality improvement initiatives in primary care is described in the practice facilitation literature.33 Practice facilitators use techniques to address gaps in care delivery, such as connecting physicians to outside resources, optimizing the use of electronic health records, implementing evidence-based practices, and addressing barriers to improve processes.33,34 When designing health system supports, including practice facilitation, the characteristics of the family physicians and the patients they serve should be considered. We found differences in these characteristics, indicating that a tailored approach may be beneficial when developing public health interventions for those in greatest need.
Many of the physicians who cared for the largest proportion of unvaccinated patients served patients living in marginalized neighbourhoods. When considering supports for primary care, cultural differences in perceptions toward vaccines and heath interventions should be considered.35 Many marginalized communities have a history of neglect from government (municipal, provincial, federal) and health care, and this may lead to mistrust in public health initiatives.2 Interventions to support these communities should include meaningful community engagement and consideration for age-, language-and culturally appropriate communication tools to assist primary care in boosting vaccine uptake.36–41
We did not see large differences between groups with respect to other quality metrics related to chronic disease management or cancer screening. We postulate that the high continuity of care seen across groups is protective for these metrics, while potentially unmeasurable influences known to be related to vaccine uptake (such as news sources or political affiliations) may have been different.24,42,43
Overall, our findings suggest that primary care practices serving communities with greater need may benefit from additional supports. During the COVID-19 pandemic, Ontario implemented a hotspot strategy whereby public health efforts targeted communities that were disproportionately affected.6 The practices identified in our analysis had opportunities for improvement in SARS-CoV-2 vaccine coverage, but the hotspot strategy was not implemented at the practice level. Further research is needed to understand local best practices for integration between primary care and public health when addressing public health issues.
Limitations
Our analysis of patient and physician characteristics was limited to administrative databases. Aside from the inability to measure many social determinants of health, psychological factors, and beliefs, which are all known to be associated with vaccine uptake, the most notable limitation of this study is that our analyses were limited to patients that were attached to a family physician. Patients with no attachment to primary care likely represented the greatest public health priority, during the pandemic and beyond.26,44–46 Likewise, the generalizability of our findings to jurisdictions — where the availability of primary care may differ or the links between primary care and public health are more (or less) formalized — is limited. Further, our analysis was cross-sectional in nature and did not account for temporal trends.
Conclusion
We found that family physicians in Ontario who cared for the largest proportion of unvaccinated patients had distinct patterns that may represent opportunities for targeted interventions. Overall, these physicians tended to serve patients living in marginalized neighbourhoods and were less likely to work in team-based models of care. More equitable resource allocation, specifically expanding primary care teams in equity-seeking neighbourhoods, should be considered when supporting primary care practices with public health efforts.
Footnotes
Competing interests: Michael Green reports research support from the Ontario Ministry of Health and the Canadian Institutes of Health Research (CIHR). He is president and board chair of the College of Family Physicians of Canada and sits on the board of AMS Healthcare. Tara Kiran reports honoraria from the Ontario College of Family Physicians, the Ontario Medical Association, the Canadian Medical Association, the Canadian College of Family Physicians, and the Association of Family Health Teams of Ontario. Mina Tadrous reports consulting fees from the Canadian Agency for Drugs and Technologies in Health and Health Canada. Dominik Nowak reports consulting fees from the Alliance for Healthier Communities, the Centre for Effective Practice, the Ontario College of Family Physicians, the Ontario Medical Association, TELUS, and Women’s College Hospital. He is president-elect of the Ontario Medical Association. Holly Witteman is supported by a Canada Research Chair and reports an honorarium from Stanford University. Isaac Bogoch reports research funding from CIHR and consulting fees from the Weapons Threat Reduction Program. David Kaplan reports travel support from the Public Health Agency of Canada; he is chair of the Post–COVID-19 Condition guideline advisory committee and sits on the board of directors with the Musical Stage Company. No other competing interests were declared.
This article has been peer reviewed.
Contributors: Jennifer Shuldiner and Noah Ivers conceived and designed the study. Shahriar Khan and Rahim Moineddin conducted the data analysis. All of the authors contributed to data interpretation. Jennifer Shuldiner drafted the manuscript. All of the authors revised it critically for important intellectual content, gave final approval of the version to be published, and agreed to be accountable for all aspects of the work.
Funding: Funding for this study was from a Canadian Institutes of Health Research project grant.
Disclaimer: This study was supported by ICES, which is funded by an annual grant from the Ontario Ministry of Health (MOH) and Ministry of Long-Term Care (MLTC). The opinions, results, and conclusions reported in this paper are those of the authors and are independent from the funding sources. No endorsement by ICES or the Ontario MOH and MLTC is intended or should be inferred. This document used data adapted from the Statistics Canada Postal CodeOM Conversion File, which is based on data licensed from Canada Post Corporation, and/or data adapted from the Ontario Ministry of Health Postal Code Conversion File, which contains data copied under license from Canada Post Corporation and Statistics Canada. Parts of this material are based on data and/or information compiled and provided by the Ontario MOH and the Canadian Institute for Health Information. The analyses, conclusions, opinions, and statements expressed herein are solely those of the authors and do not reflect those of the funding or data sources; no endorsement is intended or should be inferred.
Data sharing: The data set from this study is held securely in coded form at ICES. While legal data-sharing agreements between ICES and data providers (e.g., health care organizations and government) prohibit ICES from making the data set publicly available, access may be granted to those who meet prespecified criteria for confidential access, available at https://www.ices.on.ca/DAS (email: das{at}ices.on.ca). The full data set creation plan and underlying analytic code are available from the authors upon request, understanding that the computer programs may rely upon coding templates or macros that are unique to ICES and are therefore either inaccessible or may require modification.
- Accepted February 26, 2024.
This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) licence, which permits use, distribution and reproduction in any medium, provided that the original publication is properly cited, the use is noncommercial (i.e., research or educational use), and no modifications or adaptations are made. See: https://creativecommons.org/licenses/by-nc-nd/4.0/
References
In this issue
Article tools
Jump to section
Related Articles
Cited By...
- No citing articles found.
More in this TOC Section
Similar Articles
Collections
