Primary care insights on the management of diabetes: results from a mixed method study of care changes and impacts during the COVID-19 pandemic

Abstract

Background

In Canada, primary care manages most diabetes care. The COVID-19 pandemic disrupted primary care, reducing access to care and monitoring.

Objectives

We aim to describe changes in diabetes monitoring during the first 21 months of the pandemic, patients’ experiences with these changes, and impact on HbA1c and blood pressure (BP) control.

Methods

We conducted a two-phase mixed methods study: (i) A retrospective pre-post cohort study using de-identified electronic medical record data to compare HbA1c and BP measurement frequency and results in diabetic patients prepandemic (22 June 2018 to 12 March 2020) and during the pandemic (13 March 2020 to 3 December 2021); (ii) A qualitative descriptive analysis using semistructured interviews to understand patient experiences navigating diabetes care during the pandemic.

Results

The cohort included 84 617 patients with validated diabetes case definition. Proportion of patients with <1 HbA1c test increased by 10% during the pandemic. For those with ≥1 test, mean HbA1c remained unchanged [mean (SD) HbA1c_Pre: 7.2 (1.3); HbA1c_During: 7.2 (1.3); P = .51]. Proportion of patients with <1 BP measurement increased by 23%. For those with ≥1 measurement, mean BP remained clinically similar [mean (SD) sBP_Pre: 131.8 (13.7); sBP_During: 132.9 (15.2); P < .01. Mean dBP_Pre: 74.9 (8.6); dBP_During: 75.1 (9.1); P = .63]. Nineteen participants were interviewed, discussing virtual care, challenges with self-monitoring, and self-management strategies.

Conclusions

Mean HbA1c and BP values remained clinically stable during the pandemic despite reductions in monitoring frequency, likely due to continuity of care and patient self-management. Future research should evaluate a de-intensified frequency of diabetes monitoring and address care gaps.

Key messages.

• Frequency of HbA1c tests and BP measurements were reduced during the pandemic.

• Results of diabetes monitoring indicators changed little.

• Patients described increased self-monitoring and self-management.

• Patients preferred in-person appointments to virtual for diabetes monitoring.

• Less-intensive diabetes monitoring may be suitable for some patients and warrants future research.

• Need for more research into strategies to alleviate strains on primary care.

Introduction

Background

Approximately 3.7 million Canadians currently live with diabetes [1]. Between 2020 and 2021, the prevalence of diabetes in Canada was 9.2%, and rates are rising [1, 2]. In Canada, most diabetes management occurs within primary care [3–5]. Primary care is an ideal setting for diabetes management due to the equitable, patient-centered approach allowing for relational continuity, comprehensiveness, shared decision-making, and interprofessional care coordination [4, 6–8]. Patient health outcomes improve when primary care diabetes management is supported by a continuous, multidisciplinary team of allied health professionals, specialists, and diabetes educators [4, 9]. There is a strong association between higher continuity of primary care and reduced mortality rate, disease-related complications, hospitalizations, health service utilization, and healthcare costs among people with diabetes or hypertension, independent of disease-specific markers such as Hemoglobin A1c (HbA1C) and blood pressure (BP) [9].

Primary care faced enormous obstacles during the COVID-19 pandemic—public health restrictions led to increases in virtual care and reduced in-person services such as lab tests, vaccine administration, and BP measurements [10]. Many community-based health services relevant to diabetes care (e.g. patient education sessions) were canceled or suspended. In Canada and around the world, studies have shown that the pivot towards virtual care temporarily reduced in-person primary care appointments and services, including those related to diabetes management (e.g. BP monitoring and laboratory testing) early in the pandemic, with a gradual return to in-person activities [10–14]. The short-term interruption may have had little impact on health outcomes for people who were receiving recommended diabetes care prior to the pandemic. For some patients, the interruption may have led to longer-term delays in monitoring beyond the early pandemic period for reasons related to patients’ care-seeking decisions. This study investigated the extent to which patients in primary care continued to receive timely diabetes monitoring and remained controlled over the first nearly two years of the pandemic and how pandemic healthcare changes were experienced by patients in relation to their diabetes.

Objectives

The objectives of this parallel convergent mixed methods study were: (i) to compare the frequency and results of two intermediate diabetes health indicators (HbA1c tests and BP measurements) in the first 21 months of the pandemic (13 March 2020 to 3 December 2021) to the 21 months preceding the pandemic (22 June 2018 to 12 March 2020) among patients 50 years or older with diabetes, using a national database of de-identified electronic medical records (EMRs) ; and (ii) to explore how patients experienced and navigated changes in their diabetes care during the pandemic, and how these changes may have impacted their health and quality of life through a qualitative descriptive study.

Materials and methods

Quantitative: study design and population

We conducted a pre-post design retrospective cohort study using de-identified EMR data from the Canadian Primary Care Sentinel Surveillance Network (CPCSSN). CPCSSN consists of 14 practice-based research networks across Canada (13 available for analysis in this study), including data from approximately two million patients nationwide. CPCSSN data holdings are described in our previously published protocol [15]. We analysed active primary care patients using the 2019-Calendar Quarter 4 and 2022-Calendar Quarter 2 data extracts to identify a cohort of patients with diabetes. We included all patients aged ≥50 in 2018 with a validated CPCSSN disease case definition of diabetes mellitus (Supplementary File 1) before the prepandemic period (i.e. prior to 22 June 2018) [16].

Qualitative: study design and population

We conducted a qualitative descriptive study to explore how patients’ lived experiences with diabetes care during the COVID-19 pandemic were similar or different from prepandemic experiences. To complement and explain the quantitative findings, we inquired as to how the pandemic impacted diabetes-related healthcare delivery and the experience of navigating virtual care, continuity of care, and disruptions to self-monitoring and management routines. Eligible participants were community-dwelling Ontario residents aged ≥50 years with ongoing medical management of type 2 diabetes since 2018 or earlier, to ensure that participants had ample experience with diabetes management prepandemic and to increase the likelihood of recruiting participants with comorbidities, as these individuals may experience additional self-management burden. Participants were recruited through advertisements posted in family medicine clinics, on social media, and through personal and professional networks of the research team. Maximum variation sampling was used to recruit an initial sample of participants with differing contexts and experiences (e.g. different ages, comorbidities, rurality status); purposive sampling was also used to select participants offering experiential or demographic characteristics not yet present in the sample (e.g. males). Recruitment proceeded until data saturation was achieved.

Quantitative: variables of interest

Canadian provinces began to roll out public safety measures in response to the pandemic during the week of March 9-15, 2020, including introduction of telemedicine billing codes in many provinces [17]; thus, the primary exposure, the start of the COVID-19 pandemic, was defined as 13 March 2020. Due to data availability, the observation periods were the 21 months preceding the pandemic (22 June 2018 to 12 March 2020) and the first 21 months of the pandemic (13 March 2020 to 3 December 2021). The indicators selected for diabetes management were glycemic control monitoring (i.e. frequency and results of recorded HbA1c tests), and BP monitoring (i.e. frequency and results of recorded systolic and diastolic BP measurements, sBP and dBP), as these markers are important for assessing diabetes progression and risk of cardiovascular disease. Indicator frequency was reported at patient-level for both observation periods. Categorical indicator variables for both glycemic control and BP monitoring results during each period, informed by the 2018 Diabetes Canada clinical guidelines, were produced [18]. Variable categories for glycemic control include: (i) optimal (mean HbA1c <7.0%); (ii) suboptimal (mean HbA1c 7.0% to <8.5%); (iii) inadequate control (mean HbA1c >=8.5%). Variable categories for BP control include: (i) normal (mean sBP <130 mmHg and mean dBP <80 mmHg); (ii) elevated (mean sBP 130 mmHg—150 mmHg and mean dBP >=80 mmHg); (iii) very elevated (mean sBP >=150 mmHg). Inadequate HbA1c and BP monitoring frequency were defined as <1 HbA1c test or <1 BP measure in a given period, a conservative estimate of adequate monitoring frequency.

Quantitative: patient characteristics

Patients’ age in 2018 was reported numerically and categorically (50–59, 60–69, 70+). Sex (male, female), rurality (rural, urban), and province of residence were reported categorically. We reported the prevalence and number of validated CPCSSN case definitions before the prepandemic period (< 22 June 2018). Gender and ethnicity were not reported as these were unavailable in the CPCSSN data.

Qualitative: data collection and analysis

Individuals with type 2 diabetes participated in a single semistructured interview between April and December 2023. Interviews were conducted by a trained qualitative researcher (AC), unknown to participants. The interview guide was developed in consultation with the larger research team, which included physicians and researchers. Interviews were conducted in English via telephone or Zoom teleconference, audio-recorded, and transcribed verbatim. Analysis was inductive and iterative with data collection, moving from initial coding, which summarized and condensed findings into categories. Subsequent rounds of focused coding grouped categories and expanded on ideas of interest. Regular consultation and reflexive discussion with the larger team ensured that qualitative coding, category identification and development was responsive to the quantitative findings of significance, as well as enhancing rigour and ensuring a balanced synthesis between qualitative and quantitative findings.

Quantitative: statistical analyses

Categorical variables were reported using frequencies and percentages. Normally distributed numeric variables were reported as means with standard deviations (SD). Non-normally distributed variables were reported as medians with interquartile range (IQR). Differences in indicators across both time periods were assessed using paired-sample t-tests for normally distributed numeric variables, Wilcoxon signed rank tests for non-normally distributed variables, and marginal homogeneity tests for paired multicategorical variables. Effect sizes were calculated to measure the magnitude of differences between each period. Standardized mean differences (a.k.a. Cohen’s D) were produced for normal paired numeric variables, rank-biserial correlations for non-normal paired numeric variables, and Cramer’s V for paired multicategorical variables. A sensitivity analysis excluding the three months before and after 13 March 2020 was performed to assess the impact of initial public health measures. We set statistical significance at P-value < .05 (two-tailed). Given the exploratory nature of this study, we did not control for multiple comparisons, therefore all P-values reported are unadjusted. We used R-Studio v.2022.12.0 for all statistical analyses.

Data integration

Thematic data integration occurred twice. Data integration during study design guided the selection of qualitative participants, interview questions, and quantitative indicators. Data were integrated again after undergoing a separate analysis, with quantitative findings leading the assembly of relevant qualitative data. Qualitative data were organized into categories offering explanatory, contradictory, or refining information to the quantitative findings.

Results

In the quantitative arm, 84 617 patients aged 50 or older with a case definition of diabetes since at least June 2018 were identified (Fig. 1). Mean (SD) age in 2018 was 68.2 (10.4) years, 47.3% were female, 19.2% resided in rural regions, and mean (SD) number of prevalent CPCSSN conditions was 2.1 (1.1) (Table 1). Nineteen participants were enrolled in the qualitative arm. Mean (SD) age was 68.6 (10.3), 68.4% were women, and mean (SD) number of years since diagnosis was 20.8 (8.8). Few participants resided in rural regions (15.7%). Most had some post-secondary education (78.9%), lived with others (68.4%), and were retired (57.9%).

Figure 1.

Cohort creation flowchart.

Table 1.

Profile of active CPCSSN patients aged 50 years or older with a CPCSSN Diabetes Meletus disease case definition before the pre-COVID-19 pandemic period (< 22 June 2018) in Canada.

Variable	Total cohort (N = 84 617)
Age in 2018, n (%)
50–59	19 717 (23.3%)
60–69	27 686 (32.7%)
70+	37 214 (44.0%)
Mean (SDa)	68.2 (10.4)
Median (IQRb)	68 (60, 75)
Female, n (%)	39 989 (47.3%)
Rural residencec, n (%)	15 391 (19.2%)
Province, n (%)
Ontario	49 768 (58.8%)
Alberta	14 135 (16.7%)
Manitoba	10 531 (12.4%)
British Columbia	3763 (4.4%)
Nova Scotia	4262 (5.0%)
Quebec	1660 (2.0%)
Newfoundland & Labrador	498 (0.6%)
Prevalence of CPCSSN disease case definitions at the start of the pre-COVID-19 period (< 22 June 2018), n (%)
Osteoarthritis	22 135 (26.2%)
Depression	20 407 (24.1%)
Chronic Kidney Disease	19 910 (23.5%)
Congestive Heart Failure	13 473 (15.9%)
Chronic Obstructive Pulmonary Disease	8762 (10.4%)
Dementia	3827 (4.5%)
Non-Valvular atrial fibrillation	3140 (3.7%)
Epilepsy	947 (1.1%)
Parkinson’s Disease	633 (0.7%)
Number of prevalent CPCSSN disease case definitions at the start of the pre-COVID-19 period (< 22 June 2018)
Mean (SD)	2.1 (1.1)
Median (IQR)	2 (1, 3)

^aSD, standard deviation.

^bIQR, interquartile range.

^cRurality determined by participant’s Forward Sorting Area (FSA): An FSA = 0 indicates a rural location, approximately 10,000 people or less. FSA > 0 indicates an urban area of approximately 10,000 people or more.

Reductions in clinically recorded HbA1c and the roles of access and continuity of care

The percentage of patients with inadequate HbA1C monitoring frequency increased from 14% prepandemic to 23.9% during the pandemic (Effect Size:0.45; P < .01). Of those who received >=1 tests prepandemic (n = 72 672), 13.1% did not receive any HbA1c tests during the pandemic. The median number of HbA1c tests completed among patients with >=1 test (n = 64 376) also decreased during the pandemic (Median_Pre: 3 (1,5), Median_During: 2 (1,4); Effect Size: 0.49; P < .01). (Table 2). Results of the sensitivity analysis were similar (Supplementary File 2).

Table 2.

Comparing provision of routine chronic disease management indicators for diabetes, across pre- and during COVID-19 periods, among active CPCSSN patients aged 50 years or older with a CPCSSN Diabetes Mellitus disease case definition before the pre-COVID-19 pandemic period (< 22 June 2018) in Canada.

Indicators	Level	Total cohort (N = 84 617)		Effect size (95% CI)	P-value
		Pre-COVID-19	During COVID-19
Blood Glucose Monitoring:
Number of HbA1c tests per patient.	Median (IQR)	3 (1, 5)	2 (1, 4)	0.49 (0.48, 0.50)	<.01
Mean HbA1c test results per patient.	Mean (SD)	7.2 (1.3)	7.2 (1.3)	−0.003 (−0.011, 0.005)	.51
Blood Pressure Monitoring:
Number of blood pressure measurements per patient.	Median (IQR)	3 (0, 6)	1 (0, 2)	0.86 (0.86, 0.86)	<.01
Mean systolic blood pressure measurements per patient.	Mean (SD)	131.8 (13.7)	132.9 (15.2)	0.07 (0.06, 0.08)	<.01
Mean diastolic blood pressure measurements per patient.	Mean (SD)	74.9 (8.6)	75.1 (9.1)	0.002 (−0.007, 0.012)	.63

Notes: Rank-biserial correlations were produced for non-normal numeric indicators, interpretation: small or weak = 0 to ±0.30; moderate = ± 0.31 to ±0.50; strong = >= ±0.51. Values can vary from −1 to 1, where the sign of the correlation indicates the direction of the association.

Standardized mean differences, Cohen’s D, were produced for normal numeric indicators, interpretation: 0.2–0.5 (small), 0.5–0.8 (medium), >=0.8 (large). Values can vary from −1 to 1, where the sign of the effect size indicates the direction of the association.

P-values are unadjusted.

Qualitative interviews found that most participants continued to access regular bloodwork testing (including HbA1c) during the pandemic:

“I was in conversation with my doctor by phone and by virtual during the pandemic…and yes, they would request that I go for my bloodwork…they were continuing to monitor and remind me. And they would send the [requisitions] to a place that I could get it done. Okay, so yes, it did continue. They continued to monitor during the pandemic.” (P19)

Participants with continued access to bloodwork testing during the pandemic described requisitions being sent electronically by clinicians, who would follow up with results over the phone. Most did not experience any (or very minor) changes in completing regular bloodwork.

Conversely, several participants noted that they did not receive bloodwork during the early pandemic: “During the pandemic I wasn’t going….They weren’t doing any lab testing then…” (P07). Lapses in continuity of care, virtual-only appointments, closure or suspension of specialty diabetes clinics and programs, and reduced lab access were described as barriers to completing bloodwork: “They had a hard time finding a medical driver to take us – take me to my bloodwork and then – and then when they found it, it was like you got to sanitize your hands and wear our mask and everything else going into the van.” (P10). Some described their preferred labs closing during the pandemic, requiring them to go elsewhere. Interviewees discussed the strict public health restrictions at medical laboratories that occasionally contributed to increased wait times. Many participants felt online appointment booking helped streamline the experience, while others found this burdensome and preferred walking in at their own convenience.

No changes in glycemic control for those with continued monitoring supported by patient efforts with blood glucose self-monitoring, diet, and exercise

The Sankey plot reporting the changes in categorical HbA1c results shows a shift away from monitoring during the pandemic (Fig. 2a). The percentage of patients with optimal control decreased to 39.4% during the pandemic (prepandemic: 45.1%), patients with suboptimal control decreased to 26.7% (prepandemic: 29.6%), and patients with inadequate control decreased to 9.9% (prepandemic: 11.3%). Across all result categories, these shifts were explained by increases in the number of people with inadequate monitoring frequency (i.e. no testing during the pandemic), and to a lesser degree by bi-directional shifts across categories of control (Supplementary File 3). Among patients who had tests during both periods (i.e. continued to receive tests during the pandemic), mean HbA1c test results did not significantly change during the pandemic (Mean_Pre: 7.2 (1.3), Mean_During: 7.2 (1.3); Effect Size: −0.003; P = .51).

Figure 2.

Changes in blood glucose (a) and BP (b) monitoring across pre- and during COVID-19 periods among active CPCSSN patients aged 50 years or older with a CPCSSN Diabetes Mellitus disease case definition before the pre-COVID-19 pandemic period (< 22 June 2018) in Canada (N = 84 617).

Many participants monitored blood glucose at home. Finger-prick testing using lancets was frequently described as challenging: “I was doing the finger pricks, and my fingers were always bleeding, and they were getting bruised, and I just got tired of it. I should be doing it. I’m not.” (P19). Several participants described advantages to implanted devices (e.g. Freestyle Libre System), especially within the context of the pandemic, as it allowed clinicians remote access to patients’ blood glucose readings, meaning they could track results over time: “Because I use the LibreLink app…he has access to it, so he can see what my readings are. He doesn’t usually look at them unless he’s got a call to me, but he’ll look at it while he’s talking to me.” (P07)

Many participants discussed the success of at-home treatment and management tools facilitated by pandemic circumstances. For some, this included their ability to self-monitor blood glucose: “Again, maybe because of working from home, I probably was more routine at monitoring during the pandemic.” (P09). When discussing other factors that facilitated successful blood glucose control (e.g. exercise, healthy diet), participants mentioned features such as lessened commute times and more availability for self-care: “Well, my main exercise when I do it is walking. And if I walk a lot, which I have done, I can control it pretty well.” (P14).

Some participants attended virtual appointments with a dietician or nutritionist to discuss healthy eating practices. These provided patients with reassurance that their efforts were working, and helped identify areas for improvement. Combined with exercise and medication, many participants were able to keep blood glucose levels within range: “Basically, I watch a lot of what I eat, and I exercise every day.” (P02)

Reductions in clinically recorded BP measurements and challenges with at-home BP monitoring

The proportion of patients with inadequate BP monitoring (i.e. no BP recorded in the EMR) increased from 22.7% to 46.1% during the pandemic. Of those with any BP measurements prepandemic (n = 65 412), 26.3% did not have any BP measurements recorded during the pandemic period. The median number of BP measurements completed by patients with >= 1 measurement prepandemic decreased during the pandemic (Median_Pre: 3 (0,6), Median_During: 1 (0,2); Effect Size: 0.86; P < .01). Results from the sensitivity analysis were also consistent (Supplementary File 2).

Qualitative results showed that the pandemic shifted the onus for BP monitoring from the healthcare system onto patients. Several interviewees found this challenging and disruptive. For example, several participants used BP machines located within community pharmacies, which were often restricted or removed entirely during the early pandemic:

“I do have a blood pressure monitor but I don’t do it. I mean during COVID, I got one because my doctor wanted me to…you know, to check the blood pressure which I can’t do during you know, obviously when I wasn’t going to the appointments. But other than that, I’m not doing it for any other, you know. I just don’t monitor it, I guess.” (P12)

Some participants highlighted a lack of education and instruction on how to monitor BP at home, which impacted the frequency and accuracy of results: “Yes. I’m supposed to be doing that. I’m not probably doing it nearly enough.” (P18)

Most participants found at-home BP monitoring less desirable. They perceived in-clinic monitoring as giving accountability for their vital signs and body measurements; it also allows for review of trends over time with a clinician:

“I prefer in-person…I’m going in for my bloodwork, but if I have any other ailments, it’s easier just to talk to him, and/or show him where the ailment is…I can get my results on my phone and it just shows me the numbers – but when I go with him, he pulls up the graphs and it’s really easy...I like the in-person. I like seeing the graphs, I like having a conversation. I like having the conversation in-person more than on the phone. The phone sufficed, it served its purpose, but I was happy when it came back to being in-person.” (P02)

No changes in BP measurement readings for those with continued monitoring, supported by patient self-management

The Sankey plot shows movement of patients across all BP measurement categories into the category of inadequate BP monitoring, as well as bi-directional shifts across categories of control in the retrospective cohort study (Fig. 2b). The percentage of patients with normal BP measurements decreased to 19.7% during the pandemic (prepandemic: 30.3%), those with elevated measurements decreased to 31.2% (prepandemic: 43.8%), and those with very elevated measurements decreased to 2.9% (prepandemic: 3.2%). (Supplementary File 3). Among patients with recorded measurements during both periods, mean sBP and dBP increased marginally during the pandemic [sBP Mean_Pre: 131.8 (13.7), sBP Mean_During: 132.9 (15.2), Effect Size: 0.07, P < .01; dBP Mean_Pre: 74.9 (8.6), dBP Mean_During: 75.1 (9.1), Effect Size: 0.002; P = .63].

Interviewees discussed lifestyle actions they took to help improve both BP and blood glucose. Some of these were augmented by the circumstances of the pandemic itself (e.g. working from home):

“I was actually probably eating better because I was at home. And so it was just so much easier instead of like these, you know, commutes back and forth…I was also going out more for walks when I was working at home, just because, again, I wasn’t sitting in a car, going to and from work” (P19)

Pharmacists facilitated continuity of care for many patients by providing in-person advice and ensuring ongoing access to prescriptions and medical supplies. This was particularly important when physician’s offices were less accessible: “[Pharmacist] is pretty good, yeah he’s always making sure my BP’s down, my weight is like at--he’s always asking about my wellbeing” (P15).

Discussion

This mixed methods study investigated how the COVID-19 pandemic impacted patients’ experiences with diabetes monitoring and management, including potential downstream effects of changes on BP and glycemic control. We integrated results from qualitative participant interviews with those from a retrospective analysis of CPCSSN EMR data comparing diabetes care indicators in the 21 months before and after the start of the pandemic.

Reductions in primary care encounters early in the pandemic have been widely documented [10–14]. Our study found reductions in the intensity of both HbA1c and BP monitoring specifically. Fewer patients had any tests or measurements during the first 21 months of pandemic compared to the previous period. Amongst those who continued to be monitored during the pandemic, the mean number of measures also decreased. However, actual measures themselves changed very little for these patients.

Our results provide insight into patients’ experiences of factors influencing their diabetes care. Participants recognized the importance of virtual monitoring but preferred in-person encounters. Many discussed challenges with using at-home monitoring equipment, barriers, and facilitators to continuity of care, and restricted access to community-based programs and services. However, some described opportunities to improve lifestyle behaviors (e.g. diet and exercise) during the pandemic.

Implications for clinical practice, policy and research

Our study, alongside others, underlines important benefits of in-person appointments, such as opportunities for education, visual examination, and continuity of monitoring [19, 20]. The challenges described by participants reflect what is known about the burden of self-care for chronic disease, particularly in people with multimorbidity, as was common in our cohort. Among diabetic patients, the burden of self-management increases in a nonlinear way with additional comorbidities, potentially leading to decreased adherence to management recommendations and complicated management regimes that can exceed the patients’ ability to cope [21]. These results warrant consideration in the context of current recommendations supporting home BP monitoring as the gold standard for guiding treatment [22].

Despite many patients having fewer (or no) HbA1c monitoring tests and recorded BP measurements, we found that for those who continued to be monitored during the pandemic, on average there were no clinically relevant changes to these indicators. We were unable to determine changes in indicators of those who did not have BP or HbA1c monitoring during the pandemic. However, our results suggest that some diabetic patients may be able to maintain adequate glycemic and BP control with reduced monitoring frequency if continuity of care is maintained. Continuity of care may be more important for improved patient outcomes than the frequency of monitoring [9]; our results add to previous studies showing that reduced HbA1c and BP monitoring frequency may be appropriate for some patients while reducing overtesting [23–26].

Reduced monitoring may also reduce workload and ameliorate provider burnout, and in turn, improve healthcare access for emergent issues. Qualitative results suggest that this approach should focus on a de-intensification of care frequency rather than a switch to self-monitoring, which participants considered less desirable and more burdensome. This may be especially important in older adults, as treatment burden increases with age and multimorbidity [21, 27, 28]. Quantitative results showing a decrease in patients with any monitoring also reinforce the importance of maintaining access to care—the evidence linking access to primary care with positive health outcomes is as strong as for continuity of care [6, 29]. Further piloting and evaluation of less frequent but adequately intensive (based on individual patient context) care models on patients’ long-term health could inform guidelines that reduce both burden of care for patients and clinician workload. Decreasing unnecessary monitoring may help improve capacity for timely care of acute and subacute issues.

The COVID-19 pandemic created a natural experiment in decreased contact with health care. The pandemic exacerbated many preexisting challenges within primary care, such as lack of capacity, staffing shortages, provider burnout, and workforce attrition [11–16]. Our findings highlight the necessity of defining strategies for primary care to prioritize and de-prioritize aspects of care provision during situations of healthcare system duress—strategies that would also prove relevant in the face of the ongoing human resources challenges currently affecting primary care in Canada and globally [30–35]. Further research into patients who were not monitored during the pandemic is also necessary to ensure that future strategies for de-prioritization of care are both equitable and clinically appropriate.

Strengths and limitations

Using a mixed method design enabled us to comprehensively examine the primary care experiences of people with diabetes by illuminating the contextual factors shared through qualitative interviews. The use of longitudinal EMR data from primary care patients across multiple Canadian provinces resulted in a very large cohort size and increases the external validity of findings to other jurisdictions with robust primary care systems. Limitations within the quantitative arm may include the potential for data to be affected by an “inverse care law” bias; specifically, negligible changes in test results may be due to healthier patients seeking care. CPCSSN data does not reliably capture ethnicity, gender, or the modality of encounters; therefore, we could not assess differences between virtual and in-person care, or report on these characteristics. The CPCSSN-validated case definition used to identify patients with diabetes does not differentiate between type 1 and type 2 diabetes (however, this applies equally across both time periods). Finally, CPCSSN data only includes in-office BP measurements or those entered manually into the EMR during virtual visits. While clinicians would generally record patient-reported BP measurements in the EMR, it is possible that some were not adequately captured or transcribed [36]. Having patient-reported quality of life and outcome reporting data linked to EMR data within research network databases like CPCSSN would strengthen the quality of future analyses, as the interview participants were not recruited through CPCSSN and thus would not be included in the quantitative cohort. Qualitative data were limited by the distance of data collection from pandemic circumstances (approximately 3–3.5 years after pandemic onset), which may have introduced recall bias among participants.

Conclusion

This study found that despite significant disruptions to primary care during the early pandemic, including reductions in HbA1c and BP monitoring frequency, clinical measures remained consistent among patients who continued to received care. Triangulation with qualitative results indicated that many patients were able to maintain adequate glycemic and BP control, especially where continuity and access to diabetes care were maintained. Control of these measures may have been augmented by patient self-monitoring and lifestyle management despite these proving challenging to some. Our results demonstrate the potential for less intensive monitoring frequency without necessarily compromising diabetes control. Future research to explore the long-term feasibility and efficacy of less-intensive diabetes monitoring strategies in certain patient groups is warranted and could help inform development of guidelines for prioritization of care during times of healthcare system duress. Importantly, future research must ensure that overall access to care is not compromised and that the burden of care is not shifted onto the patient.

Trial Registration: The quantitative phase of this study is registered at www.clinicaltrials.gov (NCT 05813652). Registered 12 April 2023.

Conflicts of interest

None declared.

Funding

This study was supported by the Canadian Institutes of Health Research (CIHR, grant # WI1-179885). 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. M.V. is supported by a Canada Research Chair (Tier 2) in Ethical Complexity in Primary Care.

Data availability

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: https://cpcssn.ca/. Qualitative participants did not consent to their complete transcripts being shared beyond the research team, to maintain confidentiality. Accordingly, we do not have research ethics permission to make the qualitative data set publicly available.

Ethics approval

This pre-post design retrospective cohort study was approved by the Hamilton Integrated Research Ethics Board (#14782-C) on 11 April 2022. The CPCSSN practice-based research learning networks have waivers of patient consent to contribute de-identified EMR data to CPCSSN. Ethics approval for the qualitative descriptive study was obtained from the Hamilton Integrated Research Ethics Board (#14838) on 16 March 2023.