Introduction

Medicines play an integral part of health system [1]. Target 3.8 of SDG 3 is: ‘to achieve universal health coverage, including financial risk protection, access to quality essential healthcare services and access to safe, effective, quality and affordable essential medicines and vaccines for all’ [2]. Access to medicines is essential to improve the health outcome and achieving universal health care coverage [1]. Lack of access to medicines can result in increased preventable morbidity and mortality, loss of economic income and increased poverty. One of the main access points of medicines are pharmacies and other private sector outlets [3]. Pharmacies not only provide medicines but often offer primary care services, advice and consultation regarding common ailments which helps to improve the overall health of the population [4].

Penchansky’s and Thomas’ [5] concept of access states availability, accessibility, accommodation, affordability and acceptability as the different dimensions of access. Measuring access has often been limited to two of these five dimensions, namely availability and affordability. For instance, Target 3.b of the SDGs is measured as the proportion of health facilities that have a core set of relevant essential medicines available and affordable on a sustainable basis [2]. However, equally important for achieving access is geographical accessibility, commonly measured as geographical distribution of pharmacies within a region and density of pharmacies per population [6]. This is also reflected by the fact that the geographical location of residency is a core dimension when measuring health equities [7].

With developing technologies, we can identify geographical distribution of pharmacies within a region, by mapping pharmacies using spatial methods to determine the areas of pharmacy deserts and the areas of high pharmacy density [6]. However, the literature on geographical distribution of pharmacies is sparse. An exception is the study of pharmacy density in Chicago and the relation between their distribution and socio-economic characteristics of the neighborhood [8]. Our study aims to provide a systematic review of literature on the evidences available for geographical distribution of pharmacies across countries and their relationship between pharmacy density and other sociodemographic factors. This study contributes to the existing body of knowledge of measuring geographical accessibility of medicines (Additional file 1).

Methods

This study follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [9].

Data collection

Original studies describing the geographical distribution of pharmacies were included. We included articles from all countries regardless of the funding source of the study. The review timeline was restricted to 18 years (2000–2018) and only studies published in English.

A search strategy that combines key words “geographic distribution”,” spatial analysis”,” maps” and “pharmacy” was applied to PubMed and Web of Science (see Appendix 1 for the search terms). Additionally, we did open Google Scholar searches using the combination of terms outlined in Appendix 1. Duplicated articles were removed. We excluded articles that discussed the access to medicines but had no mention regarding the geographical distribution of pharmacies. Geographical distribution of pharmacies represented in any format (e.g.—maps and graphical representation) were included. Both qualitative and quantitative studies were included. The full text papers were assessed against the inclusion criteria by one researcher and those identified as relevant was checked again by another researcher.

Data extraction

Data from the included studies were then extracted into an excel database under the following headings: author, title, year and location of the study, objective of the study, type of pharmacy described- whether public or private, sampling and study design, pharmacy data collection source, pharmacy census validation, pharmacy density per 10,000 population, distance to pharmacies on relation to population, description of pharmacies based on urban/rural regions, ethnicity, socio-economic status and description of pharmacy characteristics. We classified study countries based on income according to World Bank Country classification [10]. If a study did not provide sufficient information to assess bias (e.g. no data on sampling and pharmacy data sources) it was excluded. Type of pharmacy was described as public if it was run by the government and private if it was run by an independent or chain pharmacy. Studies were classified as ‘national’ and ‘local’ depending on the region of data presented. If the data was presented for the entire country, we considered them as ‘national’ and if the data was presented for a state/city in a country, we considered it to be ‘local’. Data sources for pharmacies and population were identified and listed. We considered the pharmacy census to be validated if the data obtained were cross-checked by either visiting pharmacies in person or making phone call to certain pharmacy to check the accuracy of the data obtained from the sources. Ethnicity was described in some studies based on the population of pharmacy’s neighborhood. Socio-economic status was accounted for as described by each study using the countries deprivation indices and registries. Pharmacy characteristics such as hours of operation, presence of pharmacists, prescription medication delivery services, in-store medication availability and medicines price described were extracted and synthesized.

Data analysis

While the units of pharmacy density varied across studies, we retrieved and converted them to per 10,000 population to compare results across studies. Distance to pharmacies were described as mentioned in original studies, due to different units in which they were reported aggregation and comparison between studies was not feasible. While some studies measured distance to pharmacies based on population, some studies measured distance between two pharmacies. Studies that further described differences in distribution of pharmacies according to urban–rural regions, ethnicity and socio-economic status was identified. Key emergent themes recurring across the data were analyzed and a narrative review was conducted.

Results

Pubmed database search resulted in 3528 papers and Web of Science resulted in 2806 papers, which was narrowed to 4676 after removing duplicates. 176 articles were selected on the basis of title screen, 20 articles were identified based on the inclusion criteria. Only 3 of those studies are from lower middle- and upper middle-income countries (Brazil, India and South Africa) [6, 11, 12], the rest are from high income countries. Data from high income economies are available from countries namely New Zealand [13], England[14, 15], Scotland [16], Italy [17], Portugal[18], Canada [19, 20] and United States [8, 21,22,23,24,25,26,27,28]. Studies reported the geographical distribution of pharmacies, in terms of population density, rural and urban areas, characteristics and policy guidelines in the region.

The characteristics of studies included in the review are presented in Table 1. While the majority (n = 18) of the studies were observation, using census pharmacy data with geographical mapping, two studies [12, 16] did a cross-sectional survey to map the distribution of pharmacies in Scottish Highlands and India respectively. Except two studies from United States before 2010, most studies (n = 17) were conducted after 2010 making use of more developed geographical information systems. The term community pharmacies was employed in most of the papers. Table 1 stratifies the community pharmacies into public and private pharmacies.

Table 1 Study characteristics
Full size table

Table 2 outlines the methodology and results described in papers. Data reported were available both at national (n = 5) and local (n = 15) levels. While some studies (n = 11) obtained pharmacy data from the country’s department of health, studies from England used Fuse Geo-Health Care Access Database, studies from Canada used college of pharmacist’s data, two studies [12, 21] conducted cross-sectional surveys to identify pharmacies, two studies [16, 22] conducted surveys after obtaining data from a data-base. Ikram et al. [23] did not report the data source of pharmacy in their paper. Six studies conducted validation of pharmacy data and one study [24] used an internally validated database.

Table 2 Methodology and Results
Full size table

With respect to population, all studies (n = 20) utilized the latest available country’s population census. Pharmacy density data was reported in 15 studies. While Qato et al. [25] reported the density of pharmacies per square mile and Qato et al. [8] reported the density of pharmacies per census tract, other studies (n = 13) reported data based on population. Comparatively high pharmacy density per 10,000 population in Lisbon [18], Nova Scotia [19] and Ontario [20] respectively.

Geographical Information System (e.g. Arc GIS Software) was used for any necessary geo-processing in all the studies [29]. However, distance to pharmacies was reported only in 14 studies. The rest of the studies [6, 11, 14, 15], [26], [27] did not report the pharmacy distance even while using GIS software. While majority studies reported distance to pharmacies based on population accessibility, the study from India [12] reported data based on pharmacy distance from health care providers, the study from the UK [8, 14] reported pharmacy distance between two pharmacies. Studies that measured the distance of pharmacies in relation to population represented the data in Euclidean distance (straight line distance) from the center of the measured radius containing the population. Hot spot analysis, which is a spatial analysis technique used to identify clusters of high and low value, was used by one study [28] to measure accurate distance of pharmacies using their address and population address.

Studies (n = 11) differentiated the pharmacy density based on urban and rural areas according to country census classification. All studies concluded urban population have better access to pharmacies and shorter distances to travel. The study conducted in Lisbon [18], Portugal, found pharmacy deserts even in parts of urbanized corridors of Lisbon.

Some studies [21, 23, 24, 28] (n = 4) differentiated pharmacy density based on ethnicity. Interestingly the reports from these studies contrast each other. While Ikram et al. [23] and Pednekar et al. [28] found pharmacy deserts in areas with higher white population, Qato et al. [25] and Chisholm-Burns et al. [21] reported pharmacy deserts in areas of segregated black and Hispanic community.

Studies (n = 8) also differentiated pharmacy densities based on their socio-economic status. The majority of studies identified pharmacy deserts among population of lower socio-economic status. Conversely, two studies [13, 28] identified higher income communities within pharmacy deserts.

Several studies [12, 14,15,, 21, 22, 25, 27, 28] described pharmacy characteristics along with accessibility. Interestingly, two of them [21, 27] also surveyed the pricing information of medications and reported the data. Pharmacy characteristics described are hours of operation, pharmacists available, prescription medication delivery services, in-store medication availability and affordability. Appendix 3 describes the details of pharmacy characteristics.

Four studies [21, 22, 27, 28] described hours of operation in the pharmacies. They reported that the least populated area typically reported fewer hours per week. Although the study from India [12] and one study conducted in the US [21] described the availability of pharmacists in a pharmacy, the findings related to socioeconomic level of the population were not significant to arrive at a conclusion. Three studies from the US [21, 22, 28] reported data on prescription delivery services. It is important to note that all articles were representing different regions in United States and such services were not described in any middle-income economies. Two of these studies [12, 28] also described details on medication availability. While Sabde et al. [12] identified no difference in availability in urban and rural pharmacies, Pednekar et al. [28] reported 15% of pharmacies had at least one item out of stock. Two studies [21, 27] in the US collected data on affordability. However, it was hard to synthesize the data across studies due to different methods of measuring prices.

Discussion

This systematic review fills an important gap in our knowledge on geographical accessibility of pharmacies and contributes to develop measures for access to medicine. While the majority of the studies are from high income economies, pharmacy accessibility is particularly relevant in low- and middle- income countries since transportation can be expensive and difficult to access. The absence of studies from low- and middle-income countries could possibly be attributed to the challenges of maintaining an up-to-date pharmacy census with limited resources in these nations. The study team in India [12] addressed the challenge of the lack of a census by walking through a geographically defined area and identifying pharmacies, whichis a resource intensive method.

The majority of the studies measured the distance based on the ‘centroid’ approach, which considers the center of the geospatial unit (e.g., zip code, census block) and measures the straight line (also known as Euclidean distance) distance to the nearest pharmacy. As outlined by Pednekar et al. [28], this may lead to errors in measurement as the actual distance required to travel to pharmacy might be less or more depending on the geography of the surrounding land. Although Ikram et al. [23] explains that distance measured in ArcGIS software is an underestimate compared to Google Maps and the data can be correlated, with improving access to Google Maps technology and hot spot analysis, we recommend further studies to determine the actual distance instead of using the centroid approach. Some studies measure pharmacy distance to assess density in terms of population and other studies measure distance between two pharmacies. Standardized measure of representation of pharmacy density such as pharmacy per 10,000 people would allow for uniformity and comparability.

Perception is another important factor to assess accessibility. However, only one study analyzed perception of distance. Rushworth et al. [16] found that those in the most remote areas and those who prescribed five or more regular medicines were more likely to report inconvenience of access to prescribed medicines. While only 63.7% found traveling to pharmacy to be easy, 84.3% found access to pharmacies convenient. This shows that people’s perception of easy access to pharmacies also depends on other factors like number of prescribed medicines, mobility and reliance on others.

Finally, it would be important to correlate the different access dimensions including geographical distribution to measure access to medicines more comprehensively. As mentioned by Penchansky and Thomas, access is multi-dimensional. Higher density of pharmacy is found in urban areas and increased pharmacy density is associated with increased access to medicines. High income areas were noted to have increased access to pharmacies except for two studies [13, 28], which noted pharmacy deserts in higher median income regions. Amitslavski et al. [27] found significant difference and very limited stock and hours of operation of pharmacies in poor communities compared to those of wealthier communities. They also explained that higher odds of common medications being out of stock in poorer communities might be due to high poverty and low rates of prescription insurance coverage. Home medication delivery services, as mentioned by Chisholm-Burns et al. [21], would be a feasible option to explore in areas where access is limited. Overall, access to pharmacy appears sparse in low income and rural regions and measures to expand pharmacy access based on actual distance in those populations would be beneficial.

Limitations

Literature searches were restricted to PubMed, Web of Science and Google Scholar and did not use other search engines. Although comprehensive search terminology was used, there might be some articles that were missed. Literature was searched only in English language. While attempts to draw conclusion was made with regards to the distance and geographical distribution of pharmacies in various countries, heterogeneity in reporting on this measure reduced the ability to summarize trends and establish patterns. Studies were conducted in different countries with differences in cultures and health care system which affects consumer perception.

Conclusion

Geographical accessibility of pharmacies is one of the key dimensions of access to medicines. Disparities between rural and urban populations is an important challenge. The literature is scarce on studies assessing accessibility of medicines in particular in low- and middle- income countries. Expanding our knowledge on geographical access of pharmacies will enable us to provide better access to medicines, moving a step closer towards providing universal health coverage.