Impact of the COVID-19 Pandemic on Antihyperglycemic Prescriptions for Adults With Type 2 Diabetes in Canada: A Cross-sectional Study

Objectives Diabetes is a major public health problem in Canada and requires multifactorial, consistent clinical management. The COVID-19 pandemic has increased challenges in the management of many chronic ailments, including diabetes. Diabetes was associated with a higher risk of severe illness in the context of COVID-19. Pandemic restrictions also impacted diabetes care continuity, which may have contributed to an increased risk of diabetes-related complications and mortality. Methods This was a retrospective cross-sectional study of prescription patterns of antihyperglycemic medications claimed by individuals with type 2 diabetes (T2D) before and during the COVID-19 pandemic using the IQVIA Canada Longitudinal Prescription Claims database. The study period was from March 1, 2018, to February 28, 2021. The study outcomes are described on a monthly, quarterly, and yearly basis and overall, and by medication, medication class, and insurance coverage type. “New-to-molecule” patients were defined as those claiming a medication during the analysis period that they had no history of claiming in the database. Adults with at least 1 year of prescription history available and claiming their first prescription for an antihyperglycemic drug during the analysis period were classified as newly diagnosed with T2D. Results A similar number of people had at least 1 non-insulin antihyperglycemic prescription during the baseline, prepandemic, and pandemic periods in Canada (1,778,155, 1,822,403, and 1,797,272, respectively). However, the number of people initiating newer antihyperglycemic medications decreased at the beginning of the pandemic, in contrast to older medications, which remained consistent across the pandemic period. The number of people diagnosed with T2D decreased in the early months of the pandemic but recovered by October 2020. Conclusion The COVID-19 epidemic in Canada impacted clinical care for at-risk Canadians, with fewer being prescribed newer antihyperglycemic drugs and a reduction in the number of diagnoses of T2D.


Introduction
Diabetes is a major public health problem in Canada. The condition is associated with an increased risk of micro-and macrovascular complications, but comprehensive, multifactorial management of diabetes has been shown to reduce this risk [1,2].
The severe acute respiratory syndromeecoronavirus-2 (SARS-CoV-2, or COVID-19) pandemic was a devastating global health threat that increased challenges in the management of many chronic ailments, including diabetes [3]. In comparison to other comorbidities, diabetes was associated with a higher risk of severe illness in the context of COVID- 19 [3e6]. Furthermore, COVID-19 restrictions impacted diabetes care continuity, which may have contributed to an increased risk of diabetes-related complications and mortality [6e10].
In response to the COVID-19 pandemic in March 2020, the provincial governments implemented lockdowns across Canada, with a similar overall timeline in each of the provinces. However, compared with other provinces, Ontario had the most stringent inperson restrictions [11]. As a result of these restrictions, access to medicines may have been impacted, and in-person medical visits were limited to the most urgent cases, causing interruptions in comprehensive diabetes care and management [1,12]. A recent study conducted in Canada reported a 19% drop in glycated hemoglobin (A1C) tests for individuals with diabetes during the pandemic [3].
Retrospective studies from other countries have also reported the impact of the COVID-19 pandemic on diabetes management in their primary care settings. A United Kingdom (UK) study reported a 70% drop in diabetes diagnoses and a 30% drop in A1C testing at the onset of COVID-19, which gradually returned to normal over the fourth quarter of the pandemic [13]. Another UK study reported an 82% to 88% drop in A1C testing in April 2020, which failed to recover to expected levels even by September 2020 [14]. A study conducted in the United States (US) also reported a reduction in A1C testing (76.5% vs 81.8%) in 2020 as compared with 2019 [15]. Similarly, a study in the US showed a decrease in outpatient visits and laboratory testing at the beginning of the pandemic, thus raising concerns regarding gaps in diabetes management and glycemic control [16e18].
The COVID-19 pandemic has also been shown to have an impact on prescription medications for diabetes care. A cross-sectional study of a random sample of individuals with diabetes (N¼285,343) from the US IQVIA longitudinal prescription claims (LRx) database demonstrated a considerable reduction in the average number of weekly insulin prescription fills during the pandemic. Before the pandemic, the baseline average count of all existing insulin prescriptions in the first week of 2019 was 17,037.5 (95% confidence interval [CI], 16,728.7 to 17,346.4) and this increased weekly by 11.0 (95% CI, 2.8 to 19.3). During the first week of the pandemic, however, the mean number of prescriptions decreased by 395.6 (95% CI, 933.5 to 142.4), followed by a weekly decrease of 55.3 (95% CI, 78.6 to 32.0) during the pandemic period [19].
There is a lack of real-world evidence on the impact of the COVID-19 pandemic on prescription patterns of antihyperglycemic medications across Canada. The overall aim of this study was to describe prescription patterns of antihyperglycemic medications and type 2 diabetes (T2D) diagnoses in Canada before and during the COVID-19 pandemic, using IQVIA Canada's LRx database. The specific study objectives were to evaluate the following on a monthly and annual basis: 1) the number of individuals with T2D with 1 antihyperglycemic prescription during the prepandemic and pandemic periods; 2) the number of individuals with newly diagnosed T2D in the prepandemic and pandemic periods; and 3) the number of individuals with T2D with "new-to-molecule" claims in the prepandemic and pandemic periods.

Study design and data source
This was a retrospective cross-sectional study of prescription patterns of antihyperglycemic medications claimed by individuals with T2D before and during the COVID-19 pandemic from the Canadian LRx data set. The analysis period was March 1, 2018, to February 28, 2021, and was classified into 3 subperiods: baseline (March 1, 2018, to February 28, 2019), the prepandemic (March 1, 2019, to February 29, 2020), and the COVID-19 pandemic (March 1, 2020, to February 28, 2021).
The LRx database comprises de-identified pharmacy claim records collected from approximately 6,000 retail pharmacies, covering >70% of retail prescriptions dispensed across Canada. The database provides basic demographic and prescription-level information, including age, sex, province, payer, drug identification number, quantity, and days' supply. There are several antihyperglycemic drugs available for the management of T2D in Canada, including metformin, sodium-glucose cotransporter-2 inhibitors (SGLT2is), glucagon-like peptide-1 receptor agonists (GLP-1RAs), insulin secretagogues (meglitinides, sulfonylureas), dipeptidyl peptidase-4 inhibitors, thiazolidinediones, as well as fixed-dose combinations of some of the aforementioned agents [20,21].

Study population
The study population consisted of individuals with T2D identified as 18 years old at the start of the analysis period (March 1, 2018) and with 1 prescription for a non-insulin antihyperglycemic medication in their complete prescription history in the LRx database. All prescriptions for antihyperglycemic medications and basal insulin after the first prescription (index) were captured for adults with T2D during the analysis period.

Study outcomes
The study outcomes were described on a monthly, quarterly, and yearly basis and overall, and by medication, medication class, and insurance coverage type. "New-to-molecule" individuals were defined as those claiming a medication during the analysis period that they had no history of claiming in the database. Adults with a first prescription for an antihyperglycemic drug (index) during the analysis period were classified as newly diagnosed with T2D. Misclassification of newly diagnosed T2D due to individuals switching pharmacies was mitigated by applying this analysis only to people who were active in the database for 365 days before the prescription of the first non-insulin antihyperglycemic medication.

Data analysis
This study was a descriptive analysis of the LRx database. Data are based on the number of claims in the database and no projections were made. Estimates are assumed to be representative of Canada as a whole. Data were analyzed nationally and by province. Categorical variables are presented as number and proportion (%) of unique individuals. Change between time periods was calculated as the relative change between data points. Data extraction from the database, data manipulation, and data analysis were carried out using SAS version 9.4 (SAS Institute, Cary, North Carolina, United States).

Number of people with antihyperglycemic prescriptions
Similar numbers of people had 1 non-insulin antihyperglycemic prescription during the baseline, prepandemic, and pandemic periods in Canada (1,778,155, 1,822,403, and 1,797,272, respectively) (Supplementary Figure 1A). There was a small decrease of 4.7% of people with at least 1 non-insulin antihyperglycemic prescription in Québec from the prepandemic to the pandemic period (565,587 vs 538,786) (Supplementary Figure 1B). However, monthly trends of the number of people with antihyperglycemic medications displayed a transient increase of 21% (951,617 vs 1,151,207) at the start of the implementation of COVID-19 restrictions (April to June 2020) across Canada (Figure 1). This monthly rise in antihyperglycemic prescriptions from April to June 2020 in Canada was primarily attributed to the 46% (290,832 vs 423,387) increase seen in Ontario (Figure 2). The number of people with antihyperglycemic medications stratified by insurance coverage type was also similar across the 3 time periods for private (684,260, 707,103, and 703,156, respectively) and public (1,003,840, 1,026,499, and 1,008,772, respectively) insurance (Supplementary Figure 2). Similar patterns were observed with stratification by province.

Number of "new-to-molecule" people
The number of people who newly initiated GLP-1RAs, SGLT2is, or second-generation basal insulin analogues for the first few months of the pandemic (March to May 2020) decreased by 33% (9,706 vs 6,524), 31% (14,859 vs 10,226), and 35% (3,473 vs 2,258), respectively, but began recovery to prepandemic levels by November 2020 (Figure 3AeC). Similar patterns of decline in the number of people initiating GLP-1RA (Supplementary Figure 3A, B) and SGLT2i treatment in the first few months of the pandemic were seen in both Ontario and Québec. In contrast, the number of people who newly initiated older antihyperglycemic medications, including metformin, dipeptidyl peptidase-4, sulfonylurea, and first-generation basal insulin analogues, remained relatively stable during the COVID-19 period.

Number of people with newly diagnosed T2D
The number of people newly diagnosed with T2D decreased by 7% during the COVID-19 period as compared with the preeCOVID-19 period (175,444 vs 163,824). When broken down by month, a decrease of 33% (16,145 vs 10,860) was observed in the number of newly diagnosed people in the early months of the pandemic (March to May 2020), which recovered to prepandemic levels by October 2020 (Figure 4). Similarly, decreases of 7% (63,191 vs 58,596) and 9% (48,151 vs 44,004) were observed in people newly diagnosed with T2D in both Ontario and Québec during the pandemic period.

Discussion
In this study based in Canada, we have compared prescription patterns of antihyperglycemic medications among the T2D population during the COVID-19 pandemic and the prepandemic period, using IQVIA's LRx real-world prescription database. LRx covers regular pharmacies with in-person and prescription deliveries while leaving out exclusively online pharmacies, which account for <0.5% of monthly prescriptions in Canada. Previously, the LRx database was used to analyze insulin prescription trends in individuals with diabetes during the COVID-19 pandemic in the US [19].
Overall, little difference was exhibited over time in the number of people with antihyperglycemic prescriptions during the COVID-19 period compared with the preeCOVID-19 period. This included both individuals with public insurance and those with private insurance, despite the impact of the pandemic on unemployment nationwide. Nevertheless, at the beginning of the pandemic, there was a spike in antihyperglycemic prescriptions, mainly driven by prescribing patterns in Ontario and Alberta. This increase was timelimited, and the number of people with antihyperglycemic prescriptions returned to prepandemic levels by June 2020. The observed spike was most likely due to government restrictions imposed on the drug supply that a person could receive per pharmacy transaction (reduced to w30 days from the more common 90-day supply) [22]. This was implemented to counter potential drug shortages due to disrupted medicine supply and stockpiling in response to COVID-related restrictions [11,22].
Our analysis has also demonstrated that, in the first wave of the COVID-19 pandemic, there was a drop in the initiation of newer classes of antihyperglycemic medications, including specifically GLP-1RAs, second-generation basal insulin analogues, and SGLT2is, which recovered to preeCOVID-19 levels by the end of the study. The decreased numbers could be attributed to reduced visits of individuals to health-care touchpoints due to COVID-19 restrictions and concerns over exposure to the virus [2,23]. This may have resulted in greater practitioner discomfort or concerns regarding initiating "newer" medications over more established medications during a time when access to health care was uncertain. In support of this hypothesis, a decrease in new insulin prescriptions was not seen for first-generation basal insulin analogues but was seen with second-generation basal insulin analogues. Further exploration is needed to understand this pattern. This hypothesis is further supported by a UK study that showed a reduction of 22% in new antihypertensive prescriptions during the pandemic (between March and December 2020). This reduction was attributed to a decrease in both monitoring and clinical visits as well as an increase in risk of mortality and long-term complications [24]. In addition, there may have been reduced access to specialized diabetes clinics during this time, but data are not available to confirm this possibility.
The number of people newly diagnosed with T2D declined during the COVID-19 pandemic in comparison to the previous 2 years. This was likely due to disruptions in accessing primary care because of pandemic restrictions, including access to A1C testing. Although there was a considerable increase in diagnoses in August 2020, after the decline, the impact of these diagnostic delays on patient outcomes is not clear. The rate of T2D diagnoses stabilized to prepandemic levels by February 2021.
The finding of fewer people newly diagnosed with T2D in this study is consistent with that reported in UK studies [13,14]. In 2021, Carr et al reported that the rate reduction of T2D diagnoses was 0.32 (95% CI, 0.28 to 0.35) between May and December 2020, likely due to a concomitant decrease in A1C testing. The authors also reported an increase in the mortality rate of 0.19 (95% CI, 0.14 to 0.23) in England and of 0.13 (95% CI, 0.08 to 0.16) in all other UK nations between March 1 and December 10, 2020, due to a decrease in T2D diagnoses [13]. Further exploration of Canadian data is needed to understand the impact of diagnostic delays on patient outcomes.

Strengths and limitations of study
One of the major strengths of the study is the large size of the LRx patient-level database, which encompasses 70% of prescriptions dispensed across Canada, thus offering real-world insights into the impact of the pandemic on diabetes treatment. The study design allows a description of trends across multiple jurisdictions within Canada to identify how different policy responses to the COVID-19 pandemic may have impacted diabetes care. Another strength is that it allowed for stratification by payer type, which is an important consideration, given the known impact of the pandemic on employment.
Our study also has a few limitations. First, people who get their prescriptions from pharmacies outside of those captured by the LRx database would not be recorded in the database. This may have impacted the representativeness of the study across Canada, due to differential capture rates by province. However, our findings indicate that the prescription data captured by province for this study were roughly proportional to the population, and therefore we expect the risk of this to be low. In addition, the LRx database can only track people if they continue to get their prescriptions from the same pharmacies. If an individual switches from one pharmacy to another, then the individual appears as a new patient in the database, and any prescriptions filled at the second pharmacy are not linked to the prescriptions filled at the first pharmacy.
Second, the database does not include patient clinical information, such as diagnosis information, and therefore a diagnosis of T2D must be inferred from the medication history. A history of 1 antihyperglycemic medication is used to infer T2D because it is a standard treatment for individuals newly diagnosed with T2D. Because people who start on insulin therapy may have type 1 diabetes, we have conservatively excluded them from the analysis.
When a person has a non-insulin antihyperglycemic prescription for the first time in their whole recorded prescription history, the person could either have newly diagnosed T2D, have previously established diabetes that was lifestyle-managed only, or be "new to system." Those "new to system" do not have an adequate prescription history (<365 days) before the first non-insulin antihyperglycemic prescription and, therefore, may have received their antihyperglycemic prescriptions at a different pharmacy. The "newto-system" people were not counted as people with newly diagnosed T2D. This approach may underestimate the number of people with newly diagnosed T2D because some may not have an adequate prescription history when they receive their first antihyperglycemic prescription. To assess the potential impact of this misclassification, the number of people who were newly diagnosed with T2D and had a history at their current pharmacy was compared with the number who were new to the system, because the reduction in T2D diagnoses may have been related to people changing pharmacies due to COVID-19. However, the number of "new-to-system" T2D diagnoses did not increase over the pandemic period, which suggests that the decrease in T2D diagnoses reported in this study was a true decrease associated with pandemic-related restrictions and not related to individuals being misclassified between groups.
Finally, our findings of change between the time periods were calculated as the relative change. No measures of uncertainty were used in the calculation, so the percentages provided should be interpreted accordingly.
In conclusion, overall, this study has highlighted that the number of people taking antihyperglycemic medications was similar in the first year of the COVID-19 pandemic compared with the 2 previous years. However, fewer people with T2D initiated newer classes of antihyperglycemic medications in the first phase of the pandemic. In addition, there were fewer people newly diagnosed with T2D in the first few months of the pandemic, which may reflect the reduced access to health care during that time. Future research should explore how changes in prescription patterns may impact long-term clinical outcomes.

Supplementary Material
To access the supplementary material accompanying this article, visit the online version of the Canadian Journal of Diabetes at www. canadianjournalofdiabetes.com.