Article Text

Original research
Describing primary care patterns before and during the COVID-19 pandemic across Canada: a quasi-experimental pre–post design cohort study using national practice-based research network data
  1. Shuaib Hafid1,
  2. Karla Freeman1,
  3. Kris Aubrey-Bassler2,
  4. John Queenan3,
  5. Neil Drummond4,
  6. Jennifer Lawson1,
  7. Meredith Vanstone1,
  8. Kathryn Nicholson5,
  9. Marie-Thérèse Lussier6,
  10. Dee Mangin1,
  11. Michelle Howard1
  1. 1Family Medicine, Faculty of Health Sciences, McMaster University, Hamilton, Ontario, Canada
  2. 2Primary Healthcare Research Unit, Memorial University of Newfoundland, St. John's, Newfoundland, Canada
  3. 3Family Medicine, Queen's University, Kingston, Ontario, Canada
  4. 4Family Medicine, University of Alberta, Edmonton, Alberta, Canada
  5. 5Epidemiology and Biostatistics, Western University, London, Ontario, Canada
  6. 6Médecine de famille et de médecine d’urgence, Université de Montréal, Montréal, Québec, Canada
  1. Correspondence to Dr Michelle Howard; mhoward{at}mcmaster.ca

Abstract

Objective The objective was to analyse how the pandemic affected primary care access and comprehensiveness in chronic disease management by comparing primary care patterns before and during the early COVID-19 pandemic.

Design We conducted a quasi-experimental pre–post design cohort study and reported indicators for the 21 months before and after the onset of the COVID-19 pandemic.

Setting We used electronic medical record data from primary care clinics enrolled in the Canadian Primary Care Sentinel Surveillance Network from 1 January 2018 to 31 December 2021.

Population The study population included patients (n=919 928) aged 18 years or older with at least one primary care contact from 12 March 2018 to 12 March 2020, in Canada.

Outcome measures The study indicators included three indicators measuring access to primary care (encounters, blood pressure measurements and lab tests) and three for comprehensiveness (diagnoses, non-COVID-19 vaccines administered and referrals).

Results 67.3% of the cohort was aged ≥40 years, 56.4% were female and 53.5% were from Ontario, Canada. Fewer patients received an encounter during the pandemic (91.5% to 81.5%), while the median (IQR) number of encounters remained the same (5 (2–1)) for those with access. Fewer patients received a blood pressure measurement (47.9% to 31.8%), and patients received fewer measurements during the pandemic (2 (1–4) to 1 (0–2)).

Conclusions Encounters with primary care remained consistent during the pandemic, but in-person care, such as lab tests and blood pressure measurements, decreased. In-person care indicators followed temporally to national COVID-19 case counts during the pandemic.

  • primary care
  • COVID-19
  • electronic health records
  • health services accessibility

Data availability statement

Date are available upon reasonable request to CPCSSN. The data set from this study is held securely by CPCSSN, and data sharing agreements prohibit the authors from making the dataset publicly available. Researchers can request access to CPCSSN data on their website.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

STRENGTHS AND LIMITATIONS OF THIS STUDY

  • Used a pre–post design allowing for temporal comparisons of care patterns, crucial for understanding changes during the COVID-19 pandemic.

  • Leveraged the large and longitudinal Canadian Primary Care Sentinel Surveillance Network electronic medical record data from 13 primary care research networks across 7 Canadian provinces, which included recent data with validated case definitions for chronic diseases.

  • Lack of randomisation and control groups in pre–post design studies prevents establishing causal relationships. Therefore, the findings are descriptive.

  • Analyses restricted to patients attached to a primary care physician, limiting the scope of healthcare use captured. Therefore, findings do not reflect the experiences of patients unattached to primary care physicians or the care provided in other settings.

Background

Strong primary care systems are associated with improved patient health outcomes.1–5 Evidence shows primary care’s positive effect on access to care, continuity and comprehensiveness of care, person focus (patient-centred care over time) and health equity.6 7 Millions of Canadians have life-limiting chronic diseases like chronic obstructive pulmonary disease, heart failure and chronic kidney disease,8–10 as well as risk factors for life-limiting illness, such as diabetes and hypertension.11–13 With Canada’s ageing population, the prevalence of multiple co-occurring conditions is rising.13 14 Systems that prioritise accessible, comprehensive primary care allow for the prevention and management of chronic conditions—an increasingly important component of improving population-level health outcomes.15 Thus, strong primary care is crucial in the management of chronic disease in Canada,16–19 and other jurisdictions with similar ageing populations.

In March 2020, the WHO declared COVID-19 a global pandemic. Governments around the world implemented public health measures to mitigate transmission. These initiatives had a global impact on non-COVID-19 healthcare provision.20–22 For example, in-person care was limited in many jurisdictions. While it continued to be recommended for complex care scenarios requiring physical examination, virtual appointments (ie, telephone, video) were offered for non-urgent care (eg, chronic disease management, screening and diagnostic testing, specialist referrals). At certain points during the pandemic, some aspects of care (eg, asymptomatic, or population-based cancer screening) were paused entirely.23 The ebb and flow of these changes throughout the pandemic exacerbated many pre-existing barriers to care across the healthcare system, including primary care. This study aims to estimate the impact of COVID-19 on primary care access and comprehensiveness by comparing the rates of several indicators before and during the pandemic, using a national database of electronic medical record (EMR) extracts.

Methods

Study design and data source

We conducted a quasi-experimental, pre–post design cohort study using data from Canadian Primary Care Sentinel Surveillance Network (CPCSSN) data extracts. CPCSSN consists of 13 primary care practice-based research networks from across Canada, resulting in a deidentified database containing EMR data from approximately 2 million patients and 1500 family physicians nationwide. CPCSSN updates its datasets every 6 months (January–June, July–December) to include recent encounters among existing and new patients. CPCSSN datasets are described in greater detail in our previously published study protocol.24

Study population

We applied a documented approach to create a 2-year contact group (2-YCG) of patients who attended primary care to form our population cohort.25 We obtained a national dataset from the 2019-Calendar Quarter 4 and 2022-Calendar Quarter 2 CPCSSN data holdings. Data were cleaned and prepared in a stepwise progression. Patients were excluded if they had a missing birth year or sex status, if their patient status was ‘duplicate’ if they were <18 years of age in 2018 or if they were >105 years of age at the start of 2018 (to remove paediatric patients and patients with potential birth year documentation errors). We also excluded patients without any records of encounters, encounter diagnoses, billings and medication prescriptions from 12 March 2018 to 12 March 2020, as most Canadians with one or more chronic conditions typically experience at least one primary care appointment within 2 years.26

Variables of interest

Primary exposure variable

The primary exposure is the start of the COVID-19 pandemic. 13 March 2020 was selected to represent this date as Canadian provincial governments began officially imposing public health measures in response to the official declaration of the pandemic during this week.27 The selected observation periods were 630 days prepandemic (22 June 2018 to 12 March 2020) and the first 630 days of the pandemic (13 March 2020 to 3 December 2021), with 21 30-day intervals in each period.

Outcome definitions

We conducted a literature search to identify common or previously used indicators and methods of measuring primary care access and comprehensiveness.28–37 Final selection of indicators was guided by the availability and consistency of EMR data in the CPCSSN data tables. However, for several indicators of interest, it was not possible to include data from each network due to lack of data availability. Similarly, we excluded indicators relying on information contained in clinician notes or attached documents within the EMR as these are not captured. Our selected access indicators included physician encounters, blood pressure (BP) measurements and lab tests conducted. We were unable to identify virtual encounters as this information was not consistently populated for all CPCSSN encounter records. Indicators measuring comprehensiveness included unique diagnoses addressed during encounters, vaccine administration (excluding COVID-19 vaccines) and specialist referral patterns. All indicators were measured across both study periods at population level, to measure patterns across the entire patient population, and patient level to measure the system-use patterns of individuals who accessed care.

Patient characteristics

We reported patients’ age in years in 2018 (categories: 18–39, 40–49, 50–59, 60–69, 70+), sex (male, female), rurality status (rural, urban) and province of the primary care network. Patient gender and ethnicity were unavailable in the CPCSSN data, and therefore not reported. The prevalence and number of chronic conditions with a validated CPCSSN case definition38 are reported per patient at the start of the pre-COVID-19 period (before 22 June 2018) to measure baseline comorbidity status.

Analyses

Descriptive results are presented as categorical variables and as mean and standard deviations (SD) for normally distributed variables or median with the interquartile range (IQR) for skewed variables. We produced medians for numeric indicators, and percentages for categorical indicators over both study periods. We calculated differences in indicators between the pre-COVID-19 and during-COVID-19 periods and conducted Kolmogorov-Smirnov Tests and measured the kurtosis of differences to assess the distribution of the differences. Subsequently, we conducted Wilcoxon signed rank tests for non-normally distributed numeric indicators, and McNemar’s Test for binary categorical indicators. We report effect sizes measuring the differences between both periods using phi coefficients for dichotomous categorical indicators, and rank-biserial correlations for non-normally distributed numeric indicators.39 40 We also produced time-series plots for each indicator, using 30-day intervals for the 2-YCG cohort. The time-series plots also present monthly national COVID-19 case counts to visualise temporal relationships between primary care patterns and COVID-19 case patterns. We conducted a sensitivity analysis by excluding the first 90 days before and after the start of the COVID-19 pandemic to measure the impact of periods affected by initial public health measures. Statistical significance was set at p value <0.05 (two-tailed). We used SAS Enterprise Guide V.8.2 for all statistical analyses and R-Studio V.2022.12.0 to produce time-series plots.

Patient and public involvement

The research study was originally designed without patient involvement and received funding. However, a patient advisor on the CPCSSN data access committee provided input in May 2022. The analysis plans were refined to consider the feedback provided by the patient advisor.

Results

Cohort characteristics

We identified 919 928 eligible patients after applying data quality exclusions, age exclusion criteria (<18 years or >105 years of age) and the 2-YCG exclusion criteria (figure 1). The median (IQR) age in 2018 was 50 years (35–64), with 32.7% of the cohort aged 18–39 years. 56.4% of patients were female, 16.0% resided in rural regions and 53.5% of patients were from Ontario (table 1). The median (IQR) number of CPCSSN chronic condition case definitions before the start of the pre-COVID-19 period was 1 (0–2). Including both CPCSSN risk factor and chronic disease case definitions, the top five prevalent cases were dyslipidaemia (32.6%), hypertension (21.4%), depression (18.5%), diabetes mellitus (10.9%) and osteoarthritis (10.6%).

Figure 1

Cohort creation flow chart. CPCSSN, Canadian Primary Care Sentinel Surveillance Network; 2-YCG, 2-year contact group.

Table 1

Profile of Canadian Primary Care Sentinel Surveillance Network (CPCSSN) patients aged 18 years or older with a primary care encounter from 12 March 2018 to 12 March 2020, in Canada

Longitudinal patterns of care

Decreases were found for almost all indicators of primary care access during the COVID-19 period relative to the prepandemic period. Fewer patients had encounters during the pandemic (91.5% to 81.5%; effect size: 0.265) (table 2); however, patients with one or more encounters had the same number of encounters during the pandemic (5 (2–10) to 5 (2–10); effect size: −0.620). Fewer patients received BP measurements (47.9% to 31.8%; effect size: 0.496), and patients had fewer BP measurements (2 (1–4) to 1 (0–2); effect size: −0.937) if they had a measurement. Similarly, fewer patients received a lab test during the pandemic (63.9% to 53.6%; effect size: 0.210), and patients with one or more tests had fewer lab tests (17 (7–13) to 15 (0–27); effect size: −0.715).

Table 2

Comparing provision of routine chronic disease management indicators measuring access and comprehensiveness, across pre-COVID-19 and during-COVID-19 periods, among CPCSSN patients aged 18 years or older with a primary care encounter from 12 March 2018 to 12 March 2020, in Canada

All indicators measuring comprehensiveness decreased during the pandemic, except for vaccines. Fewer patients received a non-COVID-19 vaccine during the pandemic (26.3% to 21.3%; effect size: 0.192), while patients with one or more vaccines had the same number of vaccines administered (1 (0–2) to 1 (0–2); effect size: −0.701). Fewer patients received specialist referrals during the pandemic (26.5% to 24.3%; effect size: 0.103), and patients with one or more referrals received fewer referrals (2 (1–3) to 1 (0–3); effect size: −0.642). Further, patients had fewer unique medical conditions addressed at encounters during the pandemic (4 (2–6) to 3 (1–5); effect size: −0.763). All outcome differences were statistically significant (p <0.001). Our sensitivity analysis produced comparable results despite excluding the 90 days before and after the pandemic onset (online supplemental file 1).

Our time series plots demonstrate large decreases in all indicators between January 2020 to April 2020 (figure 2). However, the number of encounters and diagnoses addressed per encounter returned to prepandemic frequency much quicker than other indicators. BP measurements, for example, did not return to similar prepandemic frequency.

Figure 2

Comparing the provision of population-level routine chronic disease management activities and national COVID-19 case counts across pre-COVID-19 and during-COVID-19 periods, among CPCSSN patients aged 18 years or older with a primary care encounter from 12 March 2018 to 12 March 2020, in Canada.

Discussion

Main findings

In this quasi-experimental, pre–post design cohort study, we used CPCSSN EMR data to compare indicators of primary care access and comprehensiveness in the 21 months before and after the onset of the COVID-19 pandemic. Patients experienced decreased access to primary care, as the percentage of patients who received any encounter decreased. However, care intensity remained unchanged for those who did access primary care. Patients also experienced a decrease in services accessed through primary care (eg, laboratory tests, and referrals), suggesting decreased primary care comprehensiveness. This likely reflects a combination of factors, including constraints of government pandemic measures on the accessibility of both primary care and primary care-related services, and patients’ personal choice to defer these services. For example, access to acute and elective laboratory testing was also restricted during the pandemic, resulting in longer wait times and screening programmes being paused.41 42 Care comprehensiveness was reduced as the number of unique encounter diagnoses slightly decreased during the pandemic, suggesting that fewer unique conditions were treated and diagnosed.

Most indicators experienced marginal decreases during the pandemic, apart from the median number of encounters and unique encounter diagnoses. Reductions were also smaller than anticipated, except for indicators requiring in-person interactions, such as BP measurements. Further, the percentage of patients who experienced any encounters reduced by 10%, with 80% of patients having at least one encounter during the pandemic.43 The continued access to encounters aligns with the adoption of virtual appointments during the early pandemic, as virtual appointment billing codes were implemented between 12 March and 17 March 2020, across all jurisdictions included in the study data.44 The transition towards virtual care may have maintained access and enabled physicians to continue providing comprehensive care despite limited in-person encounters.

The pivot towards telemedicine may explain the reduction of indicators requiring in-person interactions, such as BP measurements, lab tests and administration of non-COVID-19 vaccines, which decreased by 16%, 10% and 5%, respectively. While physicians may record self-reported measurements from patients’ home BP monitors in the EMR, the number of measurements still decreased at the beginning of the pandemic and had not returned to baseline levels as of December 2021. Lab tests ordered by primary care physicians are fulfilled by external facilities, which also decreased operational capabilities during the pandemic while simultaneously prioritising COVID-19 testing. Testing facility operations likely increased as restrictions were lifted during the summer of 2020, resulting in increasing lab test rates. Restrictions appeared to affect referral patterns less. Most specialists also adopted virtual consultations, resulting in increasing referrals. The changes experienced were not constant over the two pandemic years observed, but rather represented waves: overall, patterns of indicators requiring in-person interactions followed an inverse temporal relationship to national COVID-19 case counts. This may be attributed to several factors including, but not limited to, provincial governments augmenting public health measures to mitigate the resurgence of cases due to the Delta and Omicron variants. The resurgence in cases and restrictions may have also influenced clinicians’ decisions about prioritising in-person care for medically complex patients, and patients’ decisions to not pursue in-person care.

We found statistically significant differences in indicators and meaningful differences in the distribution of all indicators between the pre-COVID-19 and during-COVID-19 periods. The largest effect size observed was for reductions in the number of BP measurements experienced, which aligns with the large decrease in BP measurements in our time series plots. Some indicators, such as the number of encounters and number of vaccines administered, experienced strong effect sizes despite reporting no differences in the medians, which may be attributed to shifting distributions between the two periods.

Implications for clinical practice, policy and research

These patterns have several clinical practice, policy and research implications. First, some care provision patterns never returned to prepandemic levels. This may be attributed to a reduction in the frequency of some investigations, treatments or tests that may be considered lower priority, overused or unnecessary. It may also represent a backlog in care provision—a phenomenon also observed in other care settings.45 46 Missing or outdated BP measurements, as well as backlogs in laboratory testing and specialist referrals, may result in delays in diagnosing illnesses or adjusting treatment. Similarly, delays in non-COVID-19 vaccine administration may increase patients’ exposure to other infectious diseases, potentially resulting in additional illness, morbidity and complexity.

Since the pandemic, attrition of primary care physicians has increased, both in Canada and globally,47–51 leading to physician shortages and reduced primary care access.52–54 Backlogs in care caused in part by surging COVID-19 cases may exacerbate pre-existing human resource challenges, including clinician burnout and high expectations for patient care despite workforce and resource constraints.48 51 55 56 As such, increased investment in primary care may help address the challenges faced by the workforce and the downstream health effects on patients.

Finally, future research exploring the appropriateness of primary care provision during the pandemic is warranted. The observed reductions in care likely represent reductions in both effective and ineffective care. This would support healthcare-governing bodies in identifying a clear and systematic approach for deprioritising less appropriate or unnecessary care when clinical needs exceed clinical capacities.

Strengths and limitations

This study has several strengths. First, pre–post design studies allow for temporal comparisons, critical for measuring differences in care patterns that occurred during the COVID-19 pandemic. Next, we used the large and longitudinal CPCSSN EMR data, containing data from 13 primary care research networks across 7 Canadian provinces. Thus, these findings may be relevant to similar jurisdictions relying on strong primary care systems. CPCSSN has developed several data-cleaning algorithms to prevent data quality issues; we report the application of these without major modifications.57 The CPCSSN dataset is routinely updated to include data from new patients across research networks and to add newly developed and validated chronic disease case definitions.58 59 Hence, this study includes recent data with validated case definitions for chronic disease.

This study also has several limitations. First, pre–post design studies lack randomisation and control groups. Therefore, causal relationships between exposures/interventions and outcomes cannot be established. Our findings can only describe changes in healthcare use and cannot confirm causation. Next, this study only captured the healthcare use of patients attached to a primary care physician. With EMR data as our data source, we were able to assess some aspects of access and comprehensiveness as core mechanisms by which primary care improves patient outcomes. Two other evidence-supported mechanisms cannot be measured using CPCSSN data: continuity of primary care and person-focused care. Patient-level medication dosage and dispensing data were also unavailable at the time of analysis, and other nuances such as time to access appointments or patient perspectives on care patterns were similarly unable to be captured. There are also limitations with our selected indicators: namely, we cannot determine the modality of encounters (ie, virtually or in-person) because of how EMR encounter data are translated within the CPCSSN dataset. We aimed to provide an overview of aspects of primary care that changed within Canada, but this required prioritising patterns using data readily available within CPCSSN. Moreover, the unique diagnoses indicator was sourced from 6 of 13 CPCSSN research networks due to limited data availability from the remaining indicators, while all other indicators reported are sourced from all CPCSSN networks. Regardless, the six networks represent patients from five different provinces across Canada. Lastly, over half of the patients included in the cohort were from Ontario as 5 of the 14 CPCSSN primary care research networks are from Ontario. Therefore, this may limit the generalisability of our findings as healthcare delivery and primary care patterns may vary across provinces in Canada.

Conclusions

While the percentage of patients with any access to primary care reduced slightly, the intensity and provision of comprehensive remained stable during the pandemic for those who did access primary care. Lab tests and BP measurements were most affected as they require in-person interactions. The time series plots of indicators requiring in-person interactions followed fluctuating COVID-19 cases temporally. Overall, we assessed the impact of the pandemic on two core pillars of primary care: access and comprehensiveness. The results highlight the importance of identifying high-quality evidence regarding the core functions of primary care to inform appropriate health system responses and the vital need for increased investment in primary care and investment in longitudinal data that captures the impact of primary care in Canada.

Data availability statement

Date are available upon reasonable request to CPCSSN. The data set from this study is held securely by CPCSSN, and data sharing agreements prohibit the authors from making the dataset publicly available. Researchers can request access to CPCSSN data on their website.

Ethics statements

Patient consent for publication

Ethics approval

This study was approved by the Hamilton Integrated Research Ethics Board (#14782-C) to access and analyse deidentified data from CPCSSN. The CPCSSN contributing networks have waivers of consent in place to contribute deidentified EMR data.

Acknowledgments

The authors would like to thank Rachael Morkem, Stephanie Garies, Mike Cummings, Chad Herman, and the entire CPCSSN team for their generous help, guidance, and support with this study.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • X @@shuaib_hafid, @MGVanstone, @DeeMangin, @mhoward101

  • Presented at Canadian Association for Health Services and Policy Research (CAHSPR) Annual Conference (2023); 29-31 May 2023; Montreal, Canada

  • Contributors MH, KF, DM and SH conceived the study. All authors designed the study and interpreted the results. SH analysed the data. SH and KF wrote the manuscript. All authors revised the manuscript 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. MH is the guarantor for this study.

  • Funding This study was supported by CIHR (grant # W11-179885). 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 by CIHR is intended or should be inferred.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.