Mapping veterinary care in Canada: An index of care accessibility

Abstract

Background

Access to veterinary care has been identified as the largest animal welfare issue in contemporary society. Access to veterinary care is complicated by several factors, including the cost of care, potential language differences between providers and clients, the number of care providers, and distance to a care provider. Each of these factors alone can impact an individual’s ability to seek adequate veterinary medical care for their companion animal, with an additional burden when multiple factors are present.

Procedure

A veterinary care accessibility score (VCAS) was created, consisting of key variables for Canada, that measured these factors and scored them in relation to the rest of the country at the census division level.

Results

In this study, nearly 2 million households in Quebec and 700 000 in Ontario were in the lowest VCAS ranking. Further, nearly 75% of households in New Brunswick were in low-access census divisions. The ratios of care providers to the estimated numbers of pet-owning households and households were also derived. An estimated veterinary clinic employee shortage was calculated at a minimum of 6803 to simply bring every census division up to a weighted mean, although the actual shortage is likely higher.

Conclusion

This research could be used by policymakers, funders, and the animal welfare community to prioritize investment and design targeted solutions.

Introduction

Pet ownership in Canada includes ~60% of households and has grown by 2.6% for dogs and 4.9% for cats from 2020 to 2022, resulting in ~7.9 million pet dogs and 8.5 million pet cats in Canada (1). The companion animal population has a direct impact on demand for veterinary services. In recent years, there has been an increased demand for good-quality and accessible veterinary care across Canada, due in part to the growing pet population (1,2). With the increased demand for veterinary services, a 5% cost increase is predicted for veterinary practices (3). Compensation is forecasted to increase at rates above general inflation for veterinary workers (3), which may put continued upward pressure on costs for services. Additionally, Health Canada has enacted new service fees for veterinary licenses and the sale of medications; these fees are expected to increase from 39 to 500% by 2027 (4), increasing concern for veterinary affordability and accessibility in Canada (5). Although the Canadian Animal Health Institute states that more cats are receiving veterinary care than before, with a 3% increase from 2020 to 2022, only 61% of cats had been taken to a veterinarian within the past 12 mo in 2022, compared to 86% of dogs (1). Further, due to owner financial limitations and other factors, simply having visited a veterinarian does mean that an animal is receiving all care needed.

The cost of veterinary care is creating a situation where low-income pet owners may not be able to afford needed care for their animals; this has been identified as one of the top challenges facing Canadian veterinarians (6,7). Households that have low income or include a person with a disability (8) experience limitations to veterinary accessibility related to the affordability and convenience of veterinary services (9). Accessibility limitations in general are due to the overall cost of veterinary care, limited client education on animal healthcare, location of care, availability of transportation, veterinarian-client communication issues, and language spoken (6,9,10).

Indigenous communities located in northern regions of Canada face particularly challenging accessibility limitations (11). The communities are rural and isolated, with limited or no veterinary services available. According to a study conducted in Manitoba by Boissonneault and Epp (11), these regions have pet dogs and cats that may not receive any veterinary or preventative care within their lifetimes. It is important to note that financial limitations and the remoteness of these communities are primary reasons for limited access, though cultural differences and viewpoints on medicine in these communities can exacerbate the issue (11).

There are several public health benefits resulting from increased advocacy and accessibility for animal healthcare. In rural communities, for example, free-roaming dogs are prevalent and identified as pivotal for rabies control (11). The Centers for Disease Control define “social vulnerability” as economic and social barriers that may negatively impact communities (12). By increasing the level of accessibility for socially vulnerable groups and addressing the overall health of companion animals, we can treat sick animals and reduce numbers of feral strays through spaying and neutering, decreasing the risk of spreading zoonotic diseases (13,14). To increase access to veterinary care, it is important to identify areas most in need of intervention. Geographic information systems (GIS) are a valuable tool for ingesting, analyzing, and mapping complex data. This tool was used in the United States by Neal and Greenberg (15) to identify veterinary care deserts based on social vulnerability and the distribution of veterinary clinic employees. By using GIS to locate veterinary resources, the number of veterinary employees, and overall demographic variables that may contribute to inaccessible veterinary care, a geographic index was created for Canada to identify the relative level of veterinary accessibility. This index considers factors such as the ratio of veterinary clinic staff by pet and human populations, measures of income, and potential language barriers at the census division level. The purpose of this research was to create a geographic-based index across Canada to evaluate the relative accessibility of veterinary care for companion animals, based on knowledge about factors that contribute to care accessibility.

Materials and methods

Study area

Census divisions in Canada were the primary geography used for this study. According to Statistics Canada, census divisions represent a mid-sized geography between the larger, province-level unit and the finer-scale municipal level (16). Provinces and territories are on a geographic scale that is too large to produce a meaningful spatial index, as they each consist of a variety of communities with wide variations in population densities and demographics. Since this research was modeled after similar research completed in the United States, census divisions represented the closest geographic equivalent to United States counties. The 3 Canadian territories were not included in this study due to their extremely rural nature and likelihood that their mode of accessing veterinary care may be very different than that in the provinces; for example, with reliance on mobile sources of veterinary care. In addition, the survey used as the source for the pet-owning household calculations in the compilation of the spatial index did not include the territories. Ultimately, 302 of the 314 total census divisions of Statistics Canada within the provinces were included in this study.

Socioeconomic variables

Selected socioeconomic variables that were applied to the veterinary accessibility score included low-income measure, after-tax (LIM-AT) (17), median household income (18), and percentage of households whose residents do not speak or have knowledge of either official language (English or French). The 2021 count of total households by census division [sourced from Earth Science Research Institute’s (ESRI’s) ArcGIS Online] was included for deriving weighted statistics for veterinary employee density. The LIM-AT measures low income at a fixed percentage of 50% of median adjusted after-tax income of family households (8). These data were sourced from Statistics Canada. The adjustment recognizes that large households have greater needs than smaller households, but at a decreasing rate for each additional household member. The number of households that did not speak either official language (English or French) in 2021 (sourced from ESRI’s ArcGIS Online) was used to calculate the percentage of these households. This was calculated by dividing the number of households that did not speak English or French by the total number of households; this percentage was included in the final aggregated percentile rank.

Pet-owning households

There are limited data available on the total number of pets and total number of pet-owning households at the census division level in Canada. According to the Canadian Animal Health Institute, the number of pet-owning households increased during the COVID-19 pandemic, resulting in ~60% of households with at least 1 companion animal (1). A survey by Abacus Data in 2020 provided an estimate of the percentage of households per province that owned pets (19) and was used to estimate the number of pet-owning households. The total number of family households in each census division was multiplied by the percentage of pet-owning households reported by Abacus Data in the province in which each census division was located. As Prince Edward Island, Newfoundland and Labrador, New Brunswick, and Nova Scotia were combined under the Atlantic region category by Abacus Data, the same percentage was used for each of these provinces. Using estimates derived at the provincial level and forecasting those down to the census division level assumed that pet ownership rates were consistent across space within each province. Although this was unlikely to be true, it was a requisite assumption to make when data were not available at a finer scale.

Veterinary employees

Veterinary clinic locations and numbers of employees at each clinic location were acquired through ESRI ArcGIS applications (20–22). Veterinary clinic locations were identified using the North American Industry Classification System. Under this system, the 541940 code represents veterinary healthcare locations (15). This code was used to identify clinic locations throughout the study area. The number of individuals employed at each clinic location was also extracted and used as an indicator of veterinary care capacity, after being aggregated in each census division. Total number of employees was used, and not number of veterinarians, as clinics with more support staff likely have higher patient capacities.

Ratios

Simple ratios of the number of veterinary clinic employees to the total number of households in each census division of the study area were calculated, using the data described above, by aggregating number of clinic employees at the census division level. A ratio of the number of veterinary clinic employees to the estimated number of pet-owning households in each census division was also generated. These ratios were expressed per 1000 households, as well as per 1000 pet-owning households, to facilitate interpretation. Both ratios can be used to consider the distribution of veterinary employees across space. The ratio that only used total household count did not account for variation in pet ownership across space; however, it was not corrupted by potential errors in estimating number of pet-owning households.

Descriptive statistics were calculated to examine differences in ratios across the various census divisions in the study area. A weighted mean was calculated for each ratio of veterinary clinic employees to 1000 households and per 1000 predicted pet-owning households. Ratios were weighted by the number of households in the census division and a weighted mean was calculated across all census divisions. Weighted means were obtained with the following calculation:

$$\overline{x}=\frac{{\mathrm{\Sigma }}_{i=1}^n(x_i*W_i)}{{\mathrm{\Sigma }}_{i=1}^n{W}_i}$$

Here, x = the ratio of veterinary employees to 1000 households (or 1000 pet-owning households) in each division (i), and W = the number of households in each census division.

Weighted mean was used because it returns a mean that gives additional weight to census divisions that have more households; it is therefore a more accurate way to express the most common household experience of the ratios. The weighted mean provided a lower-bound estimate of the number of veterinary employees needed at the census division level. In addition, the number of employees needed to reach the existing 90th percentile of employees to households and pet-owning households was also calculated, to provide an upper-bound comparison.

Calculations were then completed to determine the total number of additional employees needed for all census divisions to achieve at least the national weighted mean of employees per 1000 households. This was accomplished by taking the variation of each census division ratio distance from the weighted mean. Values for census divisions below the weighted national mean were identified and numbers of employees needed to bring each geography to a minimum of the weighted mean were calculated. Census divisions that were above the national weighted mean were not included in the analysis under an assumption that there was no oversupply in those regions. Values were weighted for number of pet-owning households in the census division and a weighted standard deviation was calculated for ratio values.

To provide an additional metric for comparison, the unweighted 90th percentiles of the numbers of clinic employees to both the estimated number of pet-owning households (expressed per 1000, for consistency) and the number of households (also expressed per 1000) were also calculated. The number of employees needed to bring each census division to this level was calculated, providing a more upper-bound estimate of the employee shortage.

Ranking scores

Ranking scores were designed to account for available variables that may affect veterinary care accessibility in a comparative context. Each of the socioeconomic variables and the ratio of veterinary clinic employees to 1000 estimated pet-owning households were used to compile a composite statistic commonly referred to as an index. In this case, each variable contributed equally and was not weighted, though this could be considered in future work, if specific variables are expected to disproportionately affect care access.

Percentile rankings for the number of employees per 1000 pet-owning households were calculated using the rank.avg function to accommodate the large numbers of zero values, whereby the average percentile ranking is returned for all equal-value variables. Note that more veterinary clinic employees per 1000 pet-owning households returned a higher percentile ranking. Percentile rankings were generated using the following equation (with ranking modifications as mentioned):

$$\frac{rank-1}{N-1}$$

Percentile rankings were also derived for the low-income after-tax variable, the 2021 median household income variable, and the percent of households not speaking English or French variable. Because of the unique value and variation in these data, a rank.eq function was used. A zero-value reference point (higher levels of low income were equated with a lower percentile rank value) was used for median household income, whereas a reference point of 1 was used for the low-income measure and the percentage of households not speaking English or French percent variable. An aggregate ranking was then calculated by summing, at equal weight, each of the previously calculated rankings and returning a new value to provide an overall comparative metric.

Rank percentiles were calculated for each variable entered into the index. “Rank percentile” was defined as the proportion of scores in a distribution that an individual score met or exceeded. For the veterinary employee to pet population ratio, values were ranked in order from lowest to highest, as lower numbers of employees are equated with higher risk of inaccessible veterinary care. For census division-level aggregated index ratings, scores were ranked from highest to lowest. For the veterinary clinic employee to pet population variable, all sequences of ties were assigned the average of the corresponding ranks, so as not to underweight the more frequent zero value. For census division level percentile ranks, the smallest of the corresponding ranks was used for any sequences of ties.

Once percentile rank calculations were completed for veterinary coverage, all percentile ranking values were summed at the census division level and a composite percentile ranking was calculated for a final index value, or veterinary care accessibility score (VCAS), which was visualized. Final values were multiplied by 100 to facilitate interpretation. In the resulting index, values closer to 100 had the highest access to veterinary care, whereas those closer to zero had the lowest access to veterinary care. Thus, this index is a relative measure that evaluated accessibility to veterinary care in the study area of Canada.

Results

Overall, 302 census divisions were identified and entered into the analysis. Statistics were calculated first for the number of veterinary employees per 1000 estimated pet-owning households at the census division level. The weighted mean of these values was 3.82, the weighted standard deviation was 2.26, and the range was 0 to 23.47. There were 26 census divisions that had no clinic staff, according to available data. The distribution of the number of veterinary employees per 1000 estimated pet-owning households by census division level is reported as a histogram (Figure 1).

Figure 1.

Distribution of number of veterinary employees per 1000 pet-owning households at the census division level.

These same data were mapped as a choropleth at the census division level (Figure 2). Note that a portion of the country with high population density has been enlarged to promote interpretation. An interactive version of this map is available from: https://www.accesstovetcare.org/

Figure 2.

Map of the ratio of veterinary clinic employees per 1000 pet-owning households at the census division level. International abbreviations alpha codes from Statistics Canada are used to identify the provinces: https://www12.statcan.gc.ca/census-recensement/2021/ref/dict/tab/index-eng.cfm?ID=t1_8

Due to the described issues with calculating the number of pet-owning households with confidence, the same statistics were derived for the ratio of the number of veterinary clinic employees per 1000 households in each of the 283 census divisions. The weighted mean was 2.16, the weighted standard deviation was 1.21, and values ranged from 0 to 11.03. The number of veterinary clinic employees per 1000 households by census division is reported as a histogram (Figure 3).

Figure 3.

Distribution of number of veterinary employees per 1000 households at the census division level.

These same data were mapped as a choropleth at the census division level (Figure 4). Note that a portion of the country with high population density has been enlarged to promote interpretation. An interactive version of this map is available from: https://www.accesstovetcare.org/

Figure 4.

Map of the ratio of veterinary clinic employees per 1000 households at the census division level. International abbreviations alpha codes from Statistics Canada are used to identify the provinces: https://www12.statcan.gc.ca/census-recensement/2021/ref/dict/tab/index-eng.cfm?ID=t1_8

To determine the number of clinic staff that would be needed across the country, the weighted mean was used. This number represented a reasonable veterinary clinic employee goal, either as a function of pet-owning households or of households. To do this, we obtained the number of additional veterinary clinic employees that would be needed to bring any census division that was below the weighted mean up to the same ratio level as the weighted mean. Census divisions that exceeded the weighted mean were not entered into the calculations, as we do not suggest “robbing Peter to pay Paul.” When completed, the calculations yielded an aggregate clinic employee shortage across the country of 7395, using the clinic employees to estimated pet-owning households ratio. Furthermore, when calculations were completed using the clinic employees to households ratio, there was an estimated shortage of 6803. Results are described by province (Table 1).

Table 1.

Employee shortage distance to weighted mean and 90th percentile, estimated by pet-owning households (POHH) and households (HH).

Province	Weighted mean shortage by POHH	Weighted mean shortage by HH	90th percentile shortage by POHH	90th percentile shortage by HH
Alberta	17	36	2251	2779
British Columbia	1470	1902	5121	5957
Manitoba	50	78	657	933
New Brunswick	162	107	871	640
Newfoundland	144	115	722	551
Nova Scotia	72	48	975	675
Ontario	1489	1540	11 404	10 732
Prince Edward Island	0	0	51	36
Quebec	3956	2918	13 757	10 605
Saskatchewan	35	59	380	570
Total	7395	6803	36 138	33 380

The final results of the VCAS scoring were visualized as a choropleth map (Figure 5) and summarized by province in Table 2. Scores were grouped by high, moderate, and low accessibility, and the sums of the number and percentage of census divisions and households in each province that were within each of these levels are summarized (Table 2).

Figure 5.

Map of veterinary care accessibility index score at the census division level. Note that a portion of the country with high population density has been enlarged for ease of interpretation. International abbreviations alpha codes from Statistics Canada are used to identify the provinces: https://www12.statcan.gc.ca/census-recensement/2021/ref/dict/tab/index-eng.cfm?ID=t1_8

An interactive version of this map is available from: https://www.accesstovetcare.org/

Table 2.

Summary of veterinary care accessibility score (VCAS) at each census division level by province, including number of census divisions at each level, percentage of census divisions at each level, total number of households at each level (sum of the census divisions household count), and percentage of households at each level.

Province	Higher accessibility		Moderate accessibility		Lower accessibility
	Households n (%)	Divisions n (%)	Households n (%)	Divisions n (%)	Households n (%)	Divisions n (%)
Alberta	1 580 784 (96.3)	15 (79)	55 755 (3.4)	3 (16)	4682 (0.3)	1 (5)
British Columbia	1 379 326 (68.4)	7 (24)	315 156 (15.6)	10 (34)	322 124 (16.0)	12 (41)
Manitoba	364 002 (70.3)	8 (35)	77 961 (15.1)	8 (35)	75 736 (14.6)	7 (30)
New Brunswick	34 040 (10.2)	3 (21)	51 140 (15.3)	3 (20)	248 400 (74.5)	9 (60)
Newfoundland	9592 (4.4)	2 (18)	132 634 (60.7)	2 (18)	76 195 (34.9)	7 (64)
Nova Scotia	7104 (1.7)	1 (6)	153 417 (36.4)	7 (39)	261 063 (61.9)	10 (56)
Ontario	2 669 139 (47.5)	23 (47)	2 264 982 (40.3)	19 (39)	685 103 (12.2)	7 (14)
Prince Edward Island	7523 (11.6)	1 (33)	57 465 (88.4)	2 (67)	0 (0)	0 (0)
Quebec	950 304 (25.7)	26 (27)	750 874 (20.3)	36 (37)	1 994 723 (54.0)	36 (37)
Saskatchewan	235 530 (52.2)	12 (67)	57 927 (12.8)	3 (17)	157 619 (34.9)	3 (17)
Total	7 237 344 (48.3)	98 (34)	3 917 311 (26.1)	93 (33)	3 825 645 (25.5)	92 (33)

Discussion

This analysis provides the first index of veterinary care accessibility for Canada. Although it may not be surprising that care is less available in rural areas, there were also some centers of population in Canada that were in the lowest VCAS ranking. Two provinces stand out as examples: Quebec, with nearly 2 million households in census divisions scoring at the lowest level of accessibility, and Ontario, where nearly 700 000 households were located in areas of lowest access. Also notable was New Brunswick, where, despite a smaller population, nearly 75% of homes were in census divisions that scored in the lowest third of care accessibility relative to the rest of the country. That Quebec has significant potential issues with access to veterinary care was reinforced by the CVMA Workforce Study, reporting a year-over-year loss in the number of veterinarians despite population increases (5). The VCAS scoring highlighted overall vulnerability to lack of access to veterinary care, and these results point to some clear geographic regions that have heightened challenges of care access.

These analyses also enabled identification of census divisions that had lower-than-anticipated levels of veterinary clinic staffing based on 2 ratios, employees to 1000 estimated pet-owning households and employees to 1000 households. A similar analysis conducted for density of veterinary employees in the United States yielded different results than this analysis for Canada. The weighted mean of veterinary employees per 1000 United States pet-owning households across all counties was reported to be 5.53 (23), compared to 3.8 for the same metric across Canada’s census divisions. There are a few possible explanations for these differences. First, different geographies were being compared (census divisions versus counties). Second, determining the actual number of pet-owning households was challenging in both countries due to a lack of data and the need to extrapolate reported state- or provincial-level data down to a smaller geography. Reported values for veterinary employees per 1000 total households were 2.8 for the United States (23) and 2.2 for Canada. This measure, though less sensitive to potential errors introduced by pet-owning household estimates, may be affected by potential differences in overall percentages of petowning households in Canada versus the United States, as well as internal variation in percentages of pet-owning households at the sub-province level within Canada.

The numbers of additional staff needed to bring all census divisions up to a minimum level of the weighted mean in Canada were similar to those in the United States (23) when proportionally adjusted for total population. It is important to note that this analysis only indicated the number of clinic staff that would be needed to reach this weighted mean — not to reach some level of optimal accessibility. There is no published number of clinic employees that would provide an accessibility threshold, though this would be an area for potential future research. The 2 methods of calculating an employee shortage produced relatively similar results (7395 versus 6803), providing additional confidence. It is likely that the actual need is higher than this, considering the limitations of using a weighted mean versus some type of target or optimal number of employees. The idea that the weighted mean provides a lower-end threshold measurement is confirmed by findings from the CVMA that all provinces noted a need for additional veterinary capacity and that most existing veterinarians are working at or above capacity (5). For an upper-bound comparison, the 90th percentile statistics may be helpful to policymakers and advocates.

Several census divisions were noteworthy for having low relative access to veterinary care staff: in particular, a cluster of census divisions in southern Quebec and several remote, northern census divisions across Newfoundland and Labrador, Manitoba, Saskatchewan, and British Columbia. In most cases, low scores in these areas were attributed to a lack of veterinary hospitals. In turn, any solutions attempting to address challenges in these areas will need to account for a lack of existing capacity and consider that capacity will need to be created and not simply augmented. Perhaps some of these areas are served by mobile or temporary veterinary services not captured in the location-based database that was used.

Regarding the overall VCAS, there were also geographic clusters of areas with lack of access in a few regions, including the west coast and southern portions of Ontario. Notable about these locations was that the combined score accounted for other factors leading to issues with accessing veterinary care, such as income and language. Further, some of these areas were regions with substantial populations; they represent opportunities for strategic investment to supplement existing services, or provision of resources to support pet owners unable to afford care.

It is important to recognize that the index created here was a relative measure that considered a limited number of variables that were equally weighted. Areas indicated as having high access simply have access that is higher in relation to other census divisions in Canada. There is currently no “optimal” level of care access with an identified ideal number of clinic staff as a function of population (animal or human). There are also additional barriers to accessing care that were not included in this study due to a lack of available data, and some factors (e.g., limited income) may arguably affect access to veterinary care more strongly than others (e.g., language barriers). There is, similarly, no threshold at which some barriers (e.g., income) are no longer barriers to accessing care. Future research that adds data to the index, weights the variables, or analyzes those data at other levels of scale is desirable.

Future research and study limitations

Limited studies have been conducted on pet ownership across Canada. This is an opportunity for additional research on the total number of companion animals and how this varies by both demographics (income, housing density, housing type) and space (rural, urban, northern, southern). This type of information would add tremendous value to the understanding of access to veterinary care and many other animal welfare-related issues in Canada. The relative nature of the index helps to control for this limitation but does not eliminate it.

An additional limitation of this study was the use of census divisions to aggregate data. The modifiable areal unit problem elucidates the limitation that any given choice of spatial boundary for analysis necessarily has limitations and potential biases (24). Clinic staff and households are not evenly distributed over space. Similarly, the possibility of ecological fallacy exists any time data are aggregated in such a way. Ecological fallacies are logic errors that occur if one assumes that what is true for the group is true for any individual in the group. To avoid these errors when interpreting this work, one must remember that, just because a census division is reported as having higher access, this does not mean that everyone in that census division has high access. Although census divisions are a logical starting point, additional research on a smaller scale is needed.

Note that full-color, interactive versions of the maps presented in this manuscript are available on the Veterinary Care Accessibility Project website (www.accesstovetcare.org).