Background
Pharmacists’ role in the health care systems of different countries has expanded over the last three decades, for several reasons. First, demand for medication dispensing and counselling – core pharmacy services – has grown in lockstep with the variety of prescription medications that are available. Second, pharmacists’ scope of practice has expanded. In Canada and in various other jurisdictions, pharmacists are administering vaccines, screening for hypertension and other diseases, counselling on smoking cessation, prescribing drugs for minor ailments, and providing a variety of other primary care services [
1,
2].
For pharmacists to effectively deliver healthcare, they need to be accessible to patients. Patient access to pharmacist services has been the focus of a growing literature. These include studies of pharmacy service availability: across urban and rural areas of New Zealand [
3] and the United States; [
4] in areas populated predominantly by lower income households in England [
5] and in various cities in the United States; [
5‐
8] and among the elderly living in Scotland [
9] and Portugal [
10]. Waterson developed an economic framework to determine the optimal number of pharmacies and compared the density of pharmacies in Melbourne, Australia to the optimum [
11]. Jagadeesan and Wirtz review studies of local pharmacist service availability in other countries [
12]. The Organisation for Economic Co-operation and Development (OECD) has compiled data on the pharmacy density of member countries [
2].
Several studies have investigated patient access to pharmacy services in Canada. Two focused on access in the provinces of Ontario [
13] in 2010 and Nova Scotia [
14] in 2011. The studies found that most urban dwellers in these provinces live close to a pharmacy; access is somewhat lower in rural areas. Another study focussed on pharmacy access across neighbourhoods in the Greater Toronto Area in 2014; Toronto is the capital of Ontario [
15]. Neighbourhoods were characterized by level of healthcare need, i.e., the capacity to benefit from healthcare. Need was measured using average household income, and the elderly (65 +) and child (0–9) shares of the population. The study found that about 14% of the population had low access relative to healthcare need; about 13% of the population was over-serviced relative to need [
15].
Little is known about pharmacy access outside of Ontario and Nova Scotia. This paper, therefore, examined pharmacist services availability across Canada and also examined how pharmacist services availability varied with two indicators of healthcare need that are routinely used in the aforementioned literature. These characteristics are older age (the share of the population 65 +) and median household income.
Methods
This study used data on pharmacist services availability and population characteristics for each of roughly 1,300 geographic regions, called Forward Sortation Areas (FSAs), that span Canada’s 10 provinces. These data were used to estimate, via ordinary least squares (OLS), the parameters of linear regression models of FSA-level pharmacist services availability as a function of, inter alia, the share of the FSA population 65 + and median household income.
Measuring FSA-level pharmacist services availability
Pharmacist services availability was measured using the number, and usual hours of operation each day of the week, of community pharmacies located in residential FSAs in Canada’s 10 provinces. Pharmacy location data were obtained from an insurance claims adjudicator, Express Scripts International (ESI), which provided addresses of the community pharmacies in each province that billed private drug plans in the period April-June 2019. These lists were verified by comparing with lists obtained by the pharmacy regulatory bodies in each province, except Quebec. I did not have any master pharmacy lists for Quebec so I was unable to verify the ESI data as I did for the other provinces. The Ontario and British Columbia pharmacy regulatory authorities provided pharmacy-specific operating hours. For the remaining provinces, the study team manually determined the usual hours of operation each day of the week of each pharmacy by consulting the pharmacy website. If the website did not provide the information, a member of the study team called the pharmacy. Pharmacies were assigned to FSAs using the pharmacy postal code; the FSA identifier is the first three characters of the postal code [
16].
The ESI data exclude “central fill” facilities – pharmacies that prepare large number of prescriptions for distribution to residents of long-term care facilities and other institutions. However, not all of the pharmacies in the ESI data are traditional community pharmacies; some of these pharmacies specialize in mail-order (for both domestic and international patients), and some provide exclusively biologics and other specialty medicines. I removed these types of pharmacies.
Measuring FSA-level population characteristics
I obtained data on the residential population, by age group, and median total household income for 2015 of each FSA from the 2016 Census of Canada, [
16] which Statistics Canada conducted in July 2016. The population data were used to express the number of community pharmacies per FSA and the total number of hours that they are open into per capita terms.
One defect of the Census-based household income variable was that it was measured for 2015, four years prior to the 2019 measurement of pharmacy density. I thus used an alternative measure of income, collected in 2019. This was average total personal income for 2019 declared by tax filers in the FSA; these data were reported by the Canada Revenue Agency [
17]. I assessed whether results were robust to this alternative measure of income.
Selection of FSAs
There were a total of 1,632 FSAs that span the 10 provinces. Some of these FSAs consist mostly of commercial or government buildings and have very small or no residential communities. I removed these by requiring all FSAs to have an area of at least 8 square kilometers and a residential population of at least 100. This left 1,304 FSAs. Of these FSAs, census data were available for 1,282 FSAs [
18] and Canada Revenue Agency taxfiler data was available for 1,280 FSAs. The FSA of median size in the analysis sample encompasses an area of 92 km
2.
Descriptive statistics
I graphed province-level data on pharmacy availability, including the 10th, 50th and 90th percentiles of normal operating hours, for both the entire week and just weekends, of pharmacies in each province. I also graphed, by province, the fraction of pharmacies open on Saturdays and on Sundays. Finally, I graphed the number of community pharmacies per 10,000 population. I used Statistics Canada estimates of provincial populations for the second quarter of 2019 [
17]. (These population estimates were different than the FSA-level population data, which were available for 2016 only.)
I further plotted data on pharmacy availability at the FSA level (the number of pharmacies and total weekly pharmacy operating hours, both per 10,000 population) against median total household income for residential FSA regions across the 10 provinces of the FSA. A lowess curve (a local average of the y-axis variable evaluated at each income value) was superimposed in each of the plots.
Modelling FSA pharmacy availability
I specified linear regression models of the number of pharmacies, the total weekly pharmacy operating hours, and total weekend operating hours per FSA, all expressed per 10,000 population. These outcome variables were posited to depend inter alia on the fraction of the residential population that is 65 years or older and median household income. To allow for non-linearities in the relationship between income and pharmacy availability, I used indicators for 9 of the 10 deciles of median household income. The first decile (the 10% of FSAs with the lowest median household income) formed the reference group. To allow for non-linearities in the relationship between the elderly share of the FSA population and pharmacy availability, I used three indicators of the quartiles of the elderly share of population; the first quartile formed the reference group.
The regression model also includes a variable that indicates whether the FSA is rural or non-rural. (Rural regions are identified by a “0” in the second character of the FSA identifier.) Finally, the model included indicators for each of the provinces. Ontario is the reference group [
17]. The Stata program “coefplot” [
19] was used to graph regression parameter estimates. To account for possible heteroskedastic errors, the heteroskedasticity-consistent covariance matrix estimator was used.
One defect with OLS regression is that estimates can be sensitive to outlier values of the pharmacy accessibility outcome variables (either the number of pharmacies per capita or the total operating hours per capita). To assess the robustness of the OLS estimates, the models were re-estimated using quantile regression. Specifically, I estimated the conditional median of the outcome variables.
Discussion
Pharmacists’ expanded scope of practice raises questions about individuals’ access to pharmacy services. Previous studies for Canada have focussed on the geographic proximity of residential households to pharmacies in Nova Scotia and Ontario. In this paper, I focus on access to pharmacist services across the 10 Canadian provinces using more recent data. I find that pharmacy availability – both the number of pharmacies per capita and the total operating hours of these pharmacies, both overall and on weekends – was higher, the greater the elderly share of the population and the lower is median income.
Given that lower regional income is associated with higher levels of unmet healthcare need, this finding suggests that pharmacies may be well placed to provide care in areas that need it. As an example, earlier research has found that rates of lung cancer and cardiovascular-related mortality are higher in lower income Canadian neighbourhoods [
20]. Smoking cessation counselling and hypertension screening – services routinely provided in pharmacies – may be effective in addressing these health problems.
The density of pharmacies in Canada appears to be higher than in the UK, the US and other countries examined in the literature. For instance, Norris and colleagues note that in New Zealand in 2010 there were about 4,800 individuals served per pharmacy; [
3] in Canada in 2019 this number was 3,440. There are likely two reasons for the surfeit of pharmacies in Canada. The first reason lies with Canada’s approach to financing prescription drugs, which includes not only drugs themselves, but also payments to pharmacies and wholesalers. The public sector covers less than one half (about 45%) of prescription drug costs, typically for those 65 + , those with very low income and others who face high drug costs relative to income. Voluntary, private insurers cover about 35%; this is mostly employer-sponsored coverage extended to employees and dependents. The remaining 20% is covered directly by patients, typically in the form of copayments and deductibles [
21]. The voluntary, private insurance sector is non-existent, or at least much smaller, in other OECD countries [
2]. Instead, most OECD countries rely on public drug coverage, or compulsory social insurance coverage, although such coverage usually requires patient cost sharing. This gives the state some control over spending on pharmacist services. The New Zealand government, for instance, sets an annual budget for pharmacist services [
22]. Expenditure control is lower in mixed public–private insurance systems, especially systems such as Canada’s where private insurers tend to impose weaker reimbursement constraints than do public plans [
23].
While, to the best of my knowledge, there is no literature that compares pharmacy remuneration across countries, it seems to be the case that total pharmacy remuneration is higher in Canada than in other countries examined in the literature. This level of remuneration allows more pharmacies to be commercially viable than otherwise, including pharmacies with lower prescription volumes.
The second possible reason that Canada has a relatively large number of pharmacies is due to its permissive rules on pharmacy ownership and location. All provincial pharmacist regulatory authorities have rules on the books requiring that pharmacies be majority owned by pharmacists. But in most provinces, corporations can work around these rules; [
24] pharmacists are still required, however, to manage a corporate-owned pharmacy. Pharmacies in Canada also face no special location restrictions. In many jurisdictions in Europe and elsewhere, new pharmacies must locate a certain distance away from existing pharmacies, or satisfy a local pharmacy needs assessment [
25,
26].
The combination of relatively unconstrained pharmacy remuneration in Canada and permissive rules on pharmacy ownership and location help explain Canada’s density of community pharmacies. It is less clear why pharmacies tend to locate in regions with lower household income or higher elderly shares of the population. One possible explanation is that prescription demand (and thus pharmacy sales revenues) is particularly high in these areas. The prevalence of chronic health problems increases with age and those 65 and older are eligible for relatively generous provincial government drug plan coverage. Similarly, there is a well-established association between income and health. But lower income individuals under 65 in Canada typically would not have superior coverage to those with higher income given that they are less likely to qualify for generous employer-sponsored drug coverage. There is public drug coverage for the indigent in each province, and it is possible that lower income individuals who do not qualify for this public coverage are willing to pay out of pocket for pharmacy services. But it is unclear if these are the factors responsible for the income-pharmacy access gradient found in this paper. Further investigation seems warranted.
Our models of regional pharmacy density have limitations. As this analysis relied on observational data, it is possible that there are unmeasured confounders – variables associated with health need factors that independently determine pharmacy availability. However, I was unable to identify any such confounders. Also, individuals who live near a regional border may have access to a pharmacy in a neighbouring FSA, yet this would not be captured. The study is also silent on the distribution of pharmacies within an FSA. The previous Canadian studies, however, found that most households in Ontario and Nova Scotia live near a pharmacy. This study focussed on associations between indicators of health needs and availability of pharmacy services. The study, however, does not directly measure health need; different indicators could lead to different results. Finally, this study was silent on the health impacts of regional variations in access to pharmacist services. This remains an area for future research.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.