The Canadian Primary Care Sentinel Surveillance Network (CPCSSN [40]) is a multidisease electronic medical record (EMR) research and surveillance data repository that securely collects deidentified health information from EMRs of participating primary care clinicians (eg, family physicians and nurse practitioners) who are referred to as “sentinels.” CPCSSN currently brings together 13 primary care research networks for the purposes of research, disease surveillance, and quality improvement, with more networks due to join in the near future. Every 6 months, CPCSSN acquires deidentified patient-level clinical data from primary care EMR systems held by 1583 primary care providers across Canada, with approximately 200 patient encounters per month per physician [20]. CPCSSN acquires most of the data fields in Canadian primary care EMRs with the exceptions of portable document format such as imaging data and the narrative “Subjective, Objective, Assessment, and Planning” notes. The acquired data are processed into a standard format and are made available for surveillance and research [41]. The relative frequency, longitudinal format, and comprehensiveness of CPCSSN data are ideal for the study of changes in practice, through time, in response to significant environmental imperatives, such as public health or research initiatives [41]. Numerous case definition validation studies have been completed [42-44], and the data have also been used to evaluate patterns and quality of health care [19,45].

This protocol was designed using the RECORD (REporting of studies Conducted using Observational Routinely collected Data) statement [46]. We will conduct a retrospective closed cohort study with a single group using a pre-post design. We will examine prevention and management activities for exemplar conditions on the basis of having a validated CPCSSN case definition [47,48] (ie, diabetes and heart failure), having substantial prevalence in primary care, and either having high morbidity or being risk factors for morbidity if not well managed. Patient-level data extracted from the CPCSSN repository between 2018 and 2022 will be used to investigate changes in rates of presentation and clinical management of various chronic diseases before, and through the phases of, the pandemic.

The study population is all patients who were 18 years and older in 2018, and some analyses will focus on subgroups of patients with specific chronic conditions. We will use the 2-year contact population as recommended previously for defining practice populations for primary care [49]. This method uses contacts within encounter, medication, and billing-related data tables, and has been used previously with CPCSSN data to produce national chronic disease estimates [50]. This method will be used in the period from March 13, 2018, to March 13 2020. We will use this population rather than the full CPCSSN patient population data (dating back more than 2 decades for some networks) to limit the size of the data set, while maintaining the patient population that is likely to have a relationship with a primary care practice during the study period. Although this approach may exclude patients who are younger, healthier, and less likely to seek primary care compared with patients with chronic conditions, we do not expect this to have a major impact on findings because this study focuses on the age group where chronic condition prevalence begins to increase and where screening guidelines apply (eg, diabetes). Therefore, this age group is more likely to see a primary care physician regularly.

People who died during the study period will be included up to the point of death using the death date if captured or the presence of a deceased status with date estimated from the most recent contact in CPCSSN data. Patients who were older than 105 years at the start of the COVID-19 pandemic will be excluded as a possible data error.

As of 2018, there were 1.13 million adult patients in the CPCSSN database, with 65% (734,500/1,130,000) having at least 1 chronic condition [55]. A sample size of 1276 paired (pre-post) individuals would be required to detect a change from 50% to 55% of patients experiencing an outcome, with a correlation between paired observations of 0.2 [56]. With a correlation between pairs of 0.1, the required sample size would be 1432. We expect to have adequate sample size to examine changes within specific conditions given the approximate prevalence of hypertension (22%), diabetes (10%), chronic obstructive pulmonary disease (4%), and heart failure (4%) among adults in CPCSSN data or other Canadian data [13,55].

This study was approved by the Hamilton Integrated Research Ethics Board on April 11, 2022 (project number 14782-C). Because we will be using deidentified data, no consent is required. This study is also registered with ClinicalTrials.gov (NCT05813652).

The start date of this study was October 5, 2022, and the prospective end date is January 31, 2024. At the time of writing (May 2023), we have finished data cleaning and begun exploratory analyses. As there are known consistencies with CPCSSN data, having a longer exploratory analysis phase has allowed us to liaise with central CPCSSN data managers in order to prepare data for the main analyses. We have now also defined our study cohort: a total of 875,934 individuals between the ages of 18 and 105 years as of 2018 were included.

We will attempt to extend our analyses to include the first half of 2022 when data become available if our timelines and resources allow. At the time of writing this protocol (May 2023), data were available up to December 2021. The additional 2022 data will allow us to investigate changes in primary care in the later portion of the COVID-19 pandemic, capturing the Omicron wave, and measure whether any changes identified in the earlier portion of the pandemic were sustained or returned to prepandemic patterns of care.

We plan to share knowledge products for patients and the public with the patient advisory committee being created as part of the David Braley Primary Care Research Collaborative at McMaster University, with whom several authors are affiliated. For clinicians, we will create evidence briefs to be shared with the national and provincial colleges of family physicians through social media and our CPCSSN practice-based research network linkages. For researchers and policy and practice decision makers (eg, College of Family Physicians of Canada), we will produce peer-reviewed publications and evidence briefs where we will contextualize the new knowledge against recent research relating to pandemic recovery in health care. The 1- to 2-page evidence briefs will be shared with knowledge users involved in the development and implementation of policies and resources related to pandemic recovery. For patients, we will create infographics and a short video that will raise awareness of the issue of chronic condition care gaps and describe patients’ perspectives. These will be circulated via social media and clinical organizations.

The aim of this study is to help identify signals of change across several domains of primary care, including access, comprehensiveness, and appropriateness, that may have occurred in response to the COVID-19 pandemic. This will be done by investigating the overall number of primary care encounters during the pandemic compared with prepandemic, as well as potential changes in the number of diagnoses addressed during encounters, number of blood pressure measurements, number of vaccines provided, number of blood tests carried out, and so on. Potential changes in indicators specific to the management of chronic conditions will also be investigated, for example, frequency of HbA_1c testing for those with diabetes, number and frequency of visits for monitoring of heart failure, and number of prescriptions for diabetes or heart failure. Potential changes in any of these indicators may indicate that gaps in care are present that may benefit from targeted recovery measures.

However, these potential changes in primary care are not inherently negative. For example, many blood tests, examinations, and procedures are considered overused in certain populations and may be conducted more frequently than necessary [53,57,58]. If decreases in number or frequency of these overused tests are identified during the pandemic, this may be considered more appropriate use when compared to prepandemic levels. Furthermore, decreases in frequency of certain primary care interventions may not be clinically significant or necessarily lead to negative health outcomes. Therefore, it is important that each identified change be carefully considered in a broad clinical context before firm conclusions can be drawn.

The strengths of this study include the use of a large, multiprovince primary care EMR data source, developed to understand the epidemiology of chronic diseases and care in Canadian primary care. The data set is regularly updated and includes coded information on most aspects of primary care practice. The relative currency, longitudinal format, and comprehensiveness of CPCSSN data are ideal for the study of practice changes, through time, in response to significant environmental imperatives, such as the sweeping public health measures put in place during the pandemic [41]. Numerous case definition validation studies have been completed [42,43], and the data have also been used to evaluate patterns and quality of health care [19,45]. CPCSSN has undertaken extensive data cleaning algorithms to address data quality issues [41].

There are limitations in this study. EMR data are collected for clinical and administrative purposes and are not primarily collected for research, which may result in incomplete capture of information. Furthermore, participating sentinel clinics are overrepresented by academic practices [59], and by extension, we cannot be certain to what extent the patients in the CPCSSN data differ from other Canadian primary health care patients. There may be issues of nonparticipation bias and variation in data quality. This study will not provide information on people who are not attached to a primary care provider. As with primary care patients generally, older adults and female individuals are overrepresented in the CPCSSN database as compared with the national population; however, as this study is about changes in care patterns, this data set is representative of the population of interest [59]. Clinical context for some research based on secondary use of EMR data can be difficult to infer. Detailed information on family and living circumstances, ethnicity, language spoken, education, and socioeconomic status is rarely captured in EMRs, so it is more difficult to conduct equity analyses, relying instead on neighbor-level indicators.

This research was initially designed and received funding without patient involvement. When the research team presented the data request to the CPCSSN data access committee (May 2022), a patient advisor on the committee provided input. The advisor noted that the proposal was broad in terms of the scope of chronic diseases to be investigated. In response, we refined our analysis plans to examine diabetes and heart failure in greater detail as exemplar diseases. The advisor also suggested that it would be important to consider the changes in the health care system and behaviors of the public over different phases of the pandemic, especially since the Omicron variant.

We anticipate that this study will provide evidence that will inform whether and in what aspects of primary care the COVID-19 pandemic is associated with interruptions that may require attention. It will also highlight gaps or provide reassurance with respect to chronic condition care for populations who were expected to be disproportionately affected by pandemic health care changes such as those who are older or of lower socioeconomic status.