Population characteristics and geographic coverage of primary care facilities

Population characteristics and geographic coverage of primary care facilities Background: The location of General Practitioner (GP) facilities is an important aspect in the design of healthcare systems to ensure they are accessible by populations with healthcare needs. A key consideration in the facility location decision involves matching the population need for the services with the supply of healthcare resources. The literature points to several factors which may be important in the decision making process, such as deprivation, transportation, rurality, and population age. Methods: This study uses two approaches to examine the factors associated with GP accessibility in Northern Ireland. The first uses multinomial regression to examine the factors associated with GP coverage, measured as the proportion of people who live within 1.5 km road network distance from the nearest GP practice. The second focuses on the factors associated with the average travel distance to the nearest GP practice, again measured using network distance. The empirical research is carried out using population and geospatial data from Northern Ireland, across 890 Super Output Areas and 343 GP practices. Results: In 19% of Super Output Areas, all of the population live within 1.5 km of a GP practice, whilst in 24% none of the population live within 1.5 km. The regression results show that there are higher levels of population coverage in more deprived areas, smaller areas, and areas that have more elderly populations. Similarly, the average travel distance is related to deprivation, population age, and area size. Conclusions: The results indicate that GP practices are located in areas with higher levels of service need, but also that care needs to be taken to ensure rural populations have sufficient access to services, whether delivered through GP practices or through alternative services where GP practices are less accessible. The methodology and results should be considered by policy makers and healthcare managers when making decisions about GP facility location and service provision. Keywords: Accessibility, Facility location, Primary care location, Coverage, General practice, Healthcare need Background Facility location is also important in maximising cost and Primary care plays an important role in improving the utilisation efficiencies in the healthcare system [9], which health of surrounding populations [1–3]. However, GP is crucial in a climate of public funding constraints and practices must be accessible to facilitate utilisation and for the business efficiency of GP practices. Ensuring equity better health outcomes [4–7], thus highlighting the im- of access to facilities across the population is also import- portance of considering accessibility in the design and ant, and can be particularly challenging in areas with a management of healthcare systems. An important com- rural geography, deprivation inequalities, and other social, ponent of accessibility is related to geography, and in religious and political divides. particular to the distance that must be travelled to reach This study focuses specifically on the location of GP the facility, which can impact on service utilisation for practices, which are crucial components of the health- both preventative and curative interventions [3, 4, 8]. care system in providing both primary care and func- tioning as an access point to other parts of the healthcare system. In addition, GP’s provide a range of Correspondence: byron.graham@qub.ac.uk other services such as vaccinations and smoking Queen’s Management School, Queen’s University Belfast, Riddel Hall, 185 Stranmillis Road, Belfast BT9 5EE, UK © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Graham BMC Health Services Research (2018) 18:398 Page 2 of 10 cessation [10]. Although the importance of primary care such as cost effectiveness, the location of existing facil- accessibility is recognised, the inverse care law proposes ities, planned developments, and government and ad- that areas which are most in need of healthcare tend to ministrative policies. Despite this, much of the literature have poorer levels of access, particularly where market on the location of healthcare facility location only in- forces are at play [11], although the empirical evidence cludes a small subset of the possible factors influencing for this proposition is mixed [12–14]. the location decision, or focuses only on achieving max- Deciding on the location of primary care facilities is com- imal coverage regardless of the population characteris- plex, with accessibility and utilisation influenced by factors tics influencing the service need. in addition to geographical proximity, such as wait time, af- This study draws on the existing literature on health- fordability, healthcare knowledge, and the availability of care need, accessibility, and utilisation to examine the transportation [4, 15]. When studying access to services it factors associated with geographical location of GP prac- is important to take into consideration the factors associ- tices. Specifically, this study will seek to answer the fol- ated with population need for the service, such as age, gen- lowing research question: der, deprivation, and measures of health [15]. These are What factors are associated with the location of pri- considered in both the academic literature [10, 13], and by mary healthcare facilities? healthcare policy and planning authorities, through for ex- To answer this question, multivariable regression ample, capitation formulas [16, 17], which are used in the models are developed to examine the factors related to allocation of funding based on the population need. The the proportion of each Super Output Area (SOA) residing ability to access the service should also be considered, along within 1.5 km of the nearest GP practice, and the average with the need for the service and any predisposing risk fac- travel distance for each SOA. In contrast to some previous tors [15]. Moreover, these factors may also be important de- studies which have used Euclidian distance [10, 23], this terminants of the service mix required by the population. study uses the more accurate network distance. These Healthcare decision makers and practice managers should measures are combined with other sources of area level therefore consider these wider factors when making loca- data which may be related to the location of the GP prac- tional decisions and decisions around the services provided, tice, such as multiple deprivation, health deprivation, such as opening hours and specialist services. population age, area, and religion. This facilitates a de- The accessibility of healthcare facilities can take into tailed examination of the relationships between accessibil- account both spatial and non-spatial factors, and can be ity and population characteristics. measured and interpreted in a range of ways. These can This study makes several theoretical and practical con- include the distance between the population and facility, tributions to the literature. The first is the use of a novel facilities per population of geographical region, the num- dataset to study the factors related to the location of ber of staff and waiting times, satisfaction, utilisation, healthcare facilities. The paper adds to our understand- cost of access, transportation links, catchment areas or ing of the range of factors that could shape the location coverage, and gravity models [3, 15, 18, 19]. Studies such of GP practices, providing policy makers with insight as Hawthorne [20] have added to these accessibility about current levels of provision and the factors related measures by creating a measure of access based on dis- to accessibility. In particular, this study will contribute to tance and satisfaction with the service, to take account our knowledge of the population characteristics in which of the potential for people to travel further to access GP practices are located in Northern Ireland. This re- higher quality care. mains an important topic, with previous studies report- Despite the breadth and depth of research into the ac- ing conflicting findings around the existence of an cessibility of healthcare facilities, challenges remain in inverse care law, with some studies finding evidence in both the scientific literature and in healthcare location support of an inverse care law [12, 13, 24], and others policy and decision-making. One challenge is that acces- not [10, 25]. However, in general distance based studies sibility is a multifaceted concept, with no consistent def- tend to show that GP practices are more accessible in inition of ‘poor access’ to healthcare services [21]. more deprived areas [10]. Moreover, the relationship be- Moreover, the literature highlights a wide range of pre- tween deprivation and accessibility is made less clear disposing, enabling and need factors important in deter- due to differences in the age profile of areas, with more mining healthcare utilisation [15, 22]. This points deprived areas consisting of more younger people, com- towards the complexity of the facility location problem, pared with less deprived areas consisting of more older and makes it challenging to objectively evaluate and plan people and consequently more age related healthcare service provision. needs [26]. This study makes a further contribution to The multifaceted nature of accessibility makes the fa- this debate, by considering both deprivation and age. cility location decision particularly challenging due to From a practical perspective, this study will contribute the need to take into consideration a range of factors, to planning and policy decisions on the location of Graham BMC Health Services Research (2018) 18:398 Page 3 of 10 primary healthcare facilities through the measurement variety of measures such as straight line distance [10], of facility coverage, and its associated factors. The results drive time [13], availability of bus services [13], floating presented in this study will also help policy makers to catchment areas [18, 34, 35], and population to GP ra- examine the extent to which GP practice location is tios. For example, Todd et al. [10] measure the Euclidian matched to population characteristics. distance between population centroids and the nearest Identification of areas where accessibility is low is also GP practice. Rosero-Bixby [23] also uses a distance important to healthcare decision makers in planning the measure, but also creates an additional weighted index provision of services. This is of current policy relevance in measure taking account other relevant factors. primary care as pharmacies take on some of the roles his- As the focus of this study is on the factors related to torically reserved for GP’s such as minor ailment schemes distance accessibility, two measures of geographic acces- [27], and GP practices expand to take on additional roles sibility were developed for use as dependent variables. [28]. Incorporation of the insights gained in this study The first measures practice coverage as the proportion could therefore add to the planning and policy discussions of people in each SOA that live within 1.5 km of a GP around the provision of wider healthcare services. As this practice. The second measure of accessibility is the aver- study is based on the analysis of publicly available data, it age distance between the population of each SOA and also demonstrates the usefulness of open data in studying the nearest GP practice. healthcare geography and highlights the importance of re- Both measures were calculated based on network dis- cent government open data initiatives. tance using road network data from Ordinance Survey Northern Ireland [36]. Network distance and subsequent Methods mapping was carried out using ArcGIS (version 10.5). The Study setting 2017 GP list [37] and the 2017 central postcode directory This studyfocuses on thelocation of GP practices in were used to identify the postcodes and latitude and longi- Northern Ireland. In Northern Ireland there are 343 GP tude of the facilities and postcodes in Northern Ireland. practices and 1710 GP’s serving a population of over 1.9 The postcode location of the GP practice was identified by million registered patients [29, 30]. GP practices are run as combining the GP postcode with the central postcode small businesses, with practices responsible for day to day lookup, which contains the latitude and longitude of every management and recruitment of staff [30]. Practices are postcode in the United Kingdom. Postcodes have been funded by the Northern Ireland Health and Social Care widely used to calculate distance in previous GIS studies Board under the current GP contract, but remain as separ- e.g. [10, 38]. On average there are 31 people living at each ate businesses [30]. The initial location decision is therefore postcode in Northern Ireland, and there are over 58,000 likely to be crucial, as it is likely to be easier to expand an postcodes. Distance was measured by calculating the net- existing practice by employing additional staff than it would work distance between the GP postcode and every other be to open an entirely new facility. There may also be sce- postcode in Northern Ireland. narios where alternative services are more appropriate. The proportion of people within the coverage area was The locational decision of GP practices in Northern calculated by dividing the sum of the usually resident Ireland is based on criteria that aims to ensure accessi- population at each postcode location within 1.5 km of a bility, taking into consideration factors such as demand, GP by the total population of the SOA, resulting in a travel distance, new housing developments and difficul- value between 0 and 1. This measure acts as a proxy for ties registering on existing GP lists. Equality of access to individual level travel distance to the nearest GP prac- public services is a particularly important issue in North- tice. Due to a high proportion of areas having either full ern Ireland due to community divides on political and or no coverage, a new variable was computed for use in religious grounds. Ensuring equality of provision of pub- the subsequent modelling, consisting of four categories: lic services is also written into Northern Ireland legisla- no coverage, low coverage, high coverage, and full cover- tion [31], with Section 75 of the Northern Ireland Act age. This variable was based on percentage of the popu- 1998 requiring public services to consider equality of lation within 1.5 km of a GP practice, with no coverage provision across the areas of religion, political opinion, representing 0%, low coverage greater than 0% and less race, age, gender, marital status, sexual orientation, dis- than 50%, high coverage between 50% and less than ability, and dependants [32]. Equality of access to health- 100%, and full coverage indicating 100% of the area is care services in general has been widely studied in the within 1.5 km of a GP practice. literature across various geographical areas [33]. Similar to Todd et al. [10] who used a distance of 1.6 km, a distance of 1.5 km coverage was chosen as Dependent variables representing around a 20 min walk from the persons There is no universally agreed measure of coverage or house to the GP practice. Sensitivity analysis was also accessibility in the literature, with studies adopting a carried out based on other distances, as discussed Graham BMC Health Services Research (2018) 18:398 Page 4 of 10 further in the analysis section. The average travel dis- The final set of variables included in the analysis was de- tance to the nearest GP practice was also based on the termined by the literature reviewed, the data available, and network distance calculated for the coverage proportion significance in the models. All data included in the study measures. This was calculated by deriving the mean net- was measured at the SOA level in Northern Ireland, which work travel distance for the population of each SOA, is the lowest level at which most of the statistical measures again based on the usually resident postcode population of interest are made available. Northern Ireland is broken and the distance to their nearest GP practice. down into 890 SOAs, with an average of 2000 people or 700 households in each area [44]. Independent variables Data analysis In addition to geographic measures of accessibility, The initial analysis of the dataset involved producing de- previous studies have considered a range of other fac- scriptive statistics and visualisations, and investigating tors which may be important in the location of GP the correlation between variables. Due to the ordered practices. These include access to transportation, nature of the coverage dependent variable ordinal logis- healthcare needs, rurality, deprivation, and demog- tic regression would be the best choice as it utilises the raphy [10, 13, 14, 26, 38, 39]. Drawing on the wider information contained in the order of the variables. literature, the independent variables included in this However, the test of parallel lines assumption was vio- study were the Northern Ireland Measure of Multiple lated. Multinomial logistic regression was therefore used Deprivation (MDM), health deprivation, the SOA level to examine the relationships between coverage and the population aged 0–15 and over 65, the proportion of population characteristics. Ordinary least squares regres- the area who report being protestant religion, and the sion was used to examine the relationships between size of the area. In total six publicly available datasets population characteristics and the average travel distance were combined to form the final dataset. The 2017 within each SOA. Separate models were built to include MDM was used in the final models, which is a com- multiple deprivation and health deprivation due to high posite measure made up of 8 sub domains, which levels of correlation between these two variables. provides a deprivation ranking to each area [40]. The Sensitivity analysis was carried out by rerunning the health deprivation domain was also included individu- multinomial regression models based on different net- ally due to the particular relevance of this domain in work distances of 750 m, 1 km, 1.25 km, 1.75 km and the location of GP practices. 2 km. Each model was examined for differences in the The 2016 mid year population age groups [41] were significance of the variables compared to the focal used as the measure of age. Earlier models included four 1.5 km model. All data pre-processing, variable calcula- age bands: 0–15, 16–39, 40–64 and 65+, but the inclu- tion, visualisations and initial statistical analyses were sion of all four bands introduced high levels of multicoli- undertaken using the R statistical programming language nearity into the model, probably because of children [45], with multinomial regression analysis undertaken in aged 0–15 cohabiting with their parents or guardians SPSS Version 22. aged 15–64. The final models therefore only include the age bands of 0–15 and 65+. The SOA area was mea- sured in square kilometres as presented in the most re- Results cent Northern Ireland SOA’s[42]. Table 1 summarises the dependent variable and the Religion was captured in the models using the 2011 Northern Ireland SOA level measures that are included census measurement of the percentage of people report- in the study. Across all SOA’s the mean network distance ing their religion or religion brought up in was Protest- between the population and the nearest GP practice is ant and other Christian [43]. In the 2011 census, 93.5% 2684.51 m. This is displayed visually in Fig. 1, which of the population selected either this option, or Catholic, shows a map of Northern Ireland with the mean travel with other religions accounting for less than 1% of the distance within each SOA, along with the location of population, and no religion accounting for 5.6% of the each GP practice. In 18.7% of SOA’s all of the population population. Because of the very small proportions out- live within 1.5 km of a GP practice, and in 24.4% of side of Protestant and Catholic, the models focus on the SOA’s none of the population live within 1.5 km of a GP proportion of the population reporting as Protestant and practice. This highlights the differences in GP coverage Other Christian. Car ownership and rurality were in- levels across the region. SOA’s have a mean size of cluded in earlier models, but are not included in the 15.882 km . Across all SOA’s, a mean of 436 people are final models as these measures are likely to be captured aged 0–15, 654 are aged 16–39, 668 are aged 40–64, and by the multiple deprivation measure, and SOA area mea- 335 are aged 65 and over. The mean MDM and health sures respectively. deprivation rank is 445.5. Graham BMC Health Services Research (2018) 18:398 Page 5 of 10 Table 1 Descriptive statistics Statistic N Mean / (%) SD Min Max Average Distance to GP 890 2684.51 2561.73 132.93 17,455.07 MDM Rank 890 445.50 257.065 1 890 Health Deprivation Rank 890 445.50 257.065 1 890 Area (km ) 890 15.882 29.54 0.11 205.11 Persons Aged 0–15 890 435.95 176.96 48 1663 Persons Aged 16–39 890 654.12 250.942 94 2362 Persons Aged 40–64 890 667.68 200.58 110 1575 Persons Aged 65+ 890 334.55 123.57 6 1066 Protestant or Other Christian (%) 890 48.96 29.14 1.5 91.26 Full Coverage (%) 167 18.7 High Coverage (%) 260 29.2 Low Coverage (%) 246 27.7 No Coverage (%) 217 24.4 For subsequent multinomial models the proportion coverage. The results show that the most deprived of people in each SOA living within 1.5 km of a GP areas tend to have the highest level of coverage, practice was converted to four levels, ranging from across both MDM and health deprivation ranks. no coverage to full coverage. Table 2 presents the de- Areas with lower coverage also tend to be much lar- scriptive statistics broken down by the four levels of ger in terms of geographical area. No consistent Fig. 1 Mean travel distance to the nearest GP practice, and the location of GP practices in Northern Ireland (source: Authors’ own work) Graham BMC Health Services Research (2018) 18:398 Page 6 of 10 Table 2 Descriptive Statistics by Coverage Level. Mean with standard deviation in parentheses Full High Low None Average Distance 615.24 (252.14) 1495.02 (1782.43) 3205.39 (2076.71) 5111.68 (2649.21) MDM Rank 355.64 (295.17) 431.36 (259.25) 494.38 (244.02) 476.18 (216.19) Health Deprivation Rank 306.57 (256.47) 401.12 (245.67) 529.39 (244.28) 510.48 (227.65) Area (km ) 0.59 (0.788) 5.27 (12.19) 27.84 (39.15) 26.81 (32.79) Persons Aged 0–15 354.71 (127.08) 421.34 (167.73) 477.07 (201.51) 469.38 (168.12) Persons Aged 16–39 619.93 (255.23) 659.02 (267.42) 679.09 (262.81) 646.26 (208.13) Persons Aged 40–64 558.87 (139.07) 644.75 (192.15) 728.76 (210.96) 709.65 (200.41 Persons Aged 65+ 303.83 (104.89) 337.75 (124.20) 359.30 (133.89) 326.30 (118.56) Protestant or Other Christian 46.63 (30.68) 47.57 (28.87) 49.30 (28.83) 52.05 (28.52) pattern seems to emerge from the descriptive statis- because of children and their parents cohabiting. The re- tics around the population age and religion. sults also show a significant negative relationship be- tween the proportion of people who report being Regression results protestant, and the level of GP coverage. Smaller areas The results of the regression analyses are presented in also have higher levels of GP coverage. Tables 3 and 4. Model 1 shows the results of the multi- Model 2 (Table 3) focuses on the relationships with nomial regression, focusing on factors related to the pro- the average network distance that must be travelled by portion of the area level population living within 1.5 km the population to reach their nearest GP practice. This of a GP practice – grouped into full coverage, high model shows that areas with higher levels of health coverage, low coverage, and no coverage. This model deprivation have lower travel distances to their nearest shows that, compared to areas with no coverage, areas GP practice. There is a significant positive relationship with high and full coverage are more health deprived. between the number of people aged 0–15 and the aver- The results also show that areas with full coverage have age distance to the nearest GP practice. However, the a lower number of people aged 0–15 when compared number of people aged over 65 is not significant, nor is with areas with no coverage. Conversely, areas with a religion. A significant positive relationship is found be- higher number of people aged over 65 have higher levels tween the size of the area and the distance that must be of coverage. In earlier models, other age bands were in- travelled to reach the nearest GP practice. cluded, but were removed in the final models due to Model 3 focuses on the proportion of the population high levels of multicolinearity, probably occurring of each area within 1.5 km of a GP practice, but unlike Table 3 Model 1 shows the multinomial regression including Age, area, religion, and health deprivation influences on coverage. Reference category is no coverage. Model two shows the relationships with average travel distance Model 1 Model 2 Low coverage High coverage Full coverage Average distance Health Deprivation and Disability (reverse coded) .000 .001** .002*** −1.143*** (.000) (.000) (.001) (.306) Population aged 0–15 .000 −.001 −.002*** 1.387** (.001) (.001) (.001) (.437) Population aged 65+ .003*** .004*** .006*** −.855 (.001) (.001) (.001) (.630) Protestant −.010** −.012*** −.012** −.346 (.004) (.004) (.005) (2.756) Area (km ) −.004 −.051*** −1.272*** 45.741*** (2.653) (.003) (.007) (.224) Intercept −.079 .002*** .448*** 2165.628*** (337.347) (.472) (.476) (.567) N 890 Log likelihood 2010.943*** 2 2 Nagelkerke R : .412 R :.352 *p < 0.10; **p < 0.05; ***p < 0.01. Unstandardised coefficients with standard errors in parentheses Graham BMC Health Services Research (2018) 18:398 Page 7 of 10 Table 4 Model 3 shows the multinomial regression including Age, area, religion, and multiple deprivation influences on coverage. Reference category is no coverage. Model 4 shows the relationships with average travel distance Model 3 Model 4 Low Coverage High Coverage Full Coverage Average Distance Multiple Deprivation (reverse coded) .000 .001** .002*** −0.616** (.000) (.000) (.000) (0.296) Population aged 0–15 .000 −.001 −.002** 1.329*** (.001) (.001) (.001) (0.440) Population aged 65+ .003 .004*** .006*** −0.747 (.001) (.001) (.001) (0.637) Protestant −.009** −.013*** −.012** 1.351 (.004) (.004) (.005) (2.736) Area (km ) −.003 −.055*** −1.312*** 48.574*** (2.567) (.003) (.007) (.225) Intercept −.127 .053** .528 1791.466*** (328.601) (.465) (.466) (.550) N 890 Log likelihood 2011.011*** 2 2 Nagelkerke R :.412 R :.345 *p < 0.10; **p < 0.05; ***p < 0.01. Unstandardised coefficients with standard errors in parentheses model 1, it focuses on multiple deprivation rather than when different distances are used. This section reviews specifically focusing on health deprivation. Compared to the patterns in the differences that were found. For the low coverage areas, high and full coverage areas tend to purposes of this section, significance is considered at the have higher levels of deprivation. The number of people 5% level. Although, the population aged 0–15 was only aged over 65 is significantly positively related to higher significant in the full coverage models presented previ- levels of coverage, whereas the number of people aged ously, significant positive relationships were also found under 15 is significantly negatively related to full cover- in all sensitivity models at the high coverage level, and at age. There is a significant negative relationship between the low coverage level in the two models based on a dis- the proportion of the area level population reporting as tance of 1.25 km. Health deprivation and MDM were Protestant and Other Christian and higher levels of found to be significant in the 0.75 km and 1 km models, coverage. The results also show that smaller areas have but MDM was not significant in the model based on higher levels of coverage. 1 km distance when considering full coverage. The SOA Model 4 focuses on the relationships with average area was also found to be significant in the health travel distance, including multiple deprivation rather deprivation models when considering distances of than health deprivation. The results of this model show 0.75 km and 1 km. Some differences also emerged when a significant negative relationship between deprivation focusing on religion. In contrast to models 1 and 3, the and travel distance. The proportion of the population number of people reporting protestant religion was aged under 15 is significantly positively related to the found not to be significant in the models for full cover- average travel distance, but no evidence is found for a age based on 0.75 km and 1 km, nor in the low coverage relationship between the proportion of the population models based on distances of 1.75 km and 2 km. More- aged over 65 and travel distance. The proportion of over, the overall variance explained by the models in- people who report as Protestant and Other Christian is creased as the distance used to calculate the dependent not found to be significant. Evidence is also found for a variable was increased. Across all models the lowest positive relationship between the size of the area and the Nagelkerke R-squared was 0.266 in the model based on average travel distance. a distance of 0.75 km, including health deprivation. The highest R-squared was 0.513, which resulted in both Results of the sensitivity analysis models based on the 2 km distance. Sensitivity analysis was carried out by constructing a series of dependent variables based on different network Discussion distances of 0.75 km, 1 km, 1.25 km, 1.75 km and 2 km, This study focuses on the relationships between primary and comparing the results of these models with the focal care accessibility and population characteristics in 1.5 km model. Although the overall patterns were simi- Northern Ireland. The results provide insight into the lar, this analysis revealed some differences in the models relationships between the population characteristics and Graham BMC Health Services Research (2018) 18:398 Page 8 of 10 the location of GP practices, adding to our understand- across models. These differences highlight the importance ing of the populations in which GP practices are located. of considering different levels of coverage when examining The healthcare planning assumption underpinning the accessibility. The population aged 0–15 may bemoreim- study is that GP practices should be located closest to portant for other distances than the focal distance pre- areas with the greatest need, and therefore relationships sented here. Religion may be less important for other are expected between population need and accessibility. distances. One potential explanation for the differences in The results of the regression analysis support the the levels of significance could be the movement of people proposition that the most deprived areas should have from one category to another as the distance changes, the highest levels of GP coverage. This is the case for therefore changing the number of people in each category. both overall multiple deprivation and for health Healthcare planners may therefore want to consider mul- deprivation, even when controlling for area size, popula- tiple distances when examining GP practice coverage. tion age, and religion. This relationship exists when ac- Moreover, the larger the distance considered in the cessibility is measured using coverage and average dependent variable the more variance the model explains. distance. The only models where deprivation is not sig- Although this study contributes to the current body of nificant are those comparing no coverage with low levels literature on the geography of healthcare facilities, it is of coverage. This finding suggests that GP practices are not without limitations. This study focuses specifically located in areas with highest levels of deprivation related on coverage as one measure of accessibility, but does need. The relationship between deprivation and facility not consider which facilities people choose to attend. Al- location observed in this study is consistent with find- though there is some evidence from other contexts to ings from the wider literature [10, 25]. This importance suggest people tend to choose one of the closest facilities of this finding is further highlighted by the negative rela- [54], future studies could consider whether people tionship between deprivation and population health dis- choose to attend the facility that is closest to home. This cussed in the literature [46–50]. study also considers two specific measures of accessibil- The population age is an important consideration in the ity, and there are other potential measures that future location of healthcare facilities as older people tend to studies could incorporate to build on this research. Fu- have increased mobility problems [51], and also suffer ture research could consider the specific population from increased health problems [52]. The findings pre- needs and service profiles of the GP practices, as well as sented here show that areas with older populations also the impact of need and accessibility on utilisation. This have higher levels of coverage even when controlling for research also focuses specifically on factors that may in- other factors such as deprivation, area size, and religion. fluence the need for primary care services, rather than However, no evidence is found for a relationship between focusing on the factors related to the utilisation of ser- the population aged 65 and over and average travel dis- vices. Future work could consider which factors drive tance. Areas with more people aged under 15 are associ- need, accessibility, and utilisation of services, and ated with higher average travel distance. Moreover, areas whether these factors are the same. Consideration of with full coverage have lower numbers of people in this other healthcare services could also provide additional age group. These findings suggests that there is not neces- insight for healthcare decision makers. Moreover, the sarily a trade-off between coverage in deprived areas and measures considered here do not consider the ability of coverage in areas with an older population. the healthcare provider to deliver the care. Future stud- Smaller areas tend to have higher levels of coverage, ies could therefore incorporate factors such as the num- which shows that GP’s are located closer to population ber of GP’s and other staff working at the practice, the centres, where they can serve more dense populations. practice list size, specialist services, and opening hours Although this make sense conceptually, consideration matched to population need. also needs to be given to how the primary healthcare needs of more rural areas are met, whether this is Conclusions through GP practices or alternative services. These find- This study focuses on the relationship between popula- ings are consistent with previous research showing sub- tion characteristics and GP practice accessibility. The stantially better access to GP practices in urban areas in findings of the research show that GP practices are lo- England [10] and Ireland [53]. The results also show cated in areas with higher levels of need, as measured by some evidence of differing access to GP practices by reli- older populations and deprivation. This research con- gion. Historically, religious divides have been prominent tributes to the existing methodological and empirical lit- in Northern Ireland, with certain areas continuing to re- erature through the examination of multiple variables main predominantly Protestant or Catholic. which should influence the location decision making Although the overall general trend is similar across the process. From a policy perspective, the findings indicate models, the sensitivity analysis does show some differences some of the factors that are associated with GP location. Graham BMC Health Services Research (2018) 18:398 Page 9 of 10 However, a significant proportion of the population live Received: 8 November 2017 Accepted: 22 May 2018 outside of the 1.5 km coverage area, which is consistent with the rural geography of Northern Ireland. Planners should consider how best to serve this population, for References 1. Starfield B, Shi L, Macinko J, Starfield B, Shi L. Contribution of primary care example through other means of care, such as pharma- to health systems and health. Milbank Q. 2005;83:457–502. cies and district nursing services. Long travel distances 2. Thorlby R. Reclaiming a population health perspective. Future challenges for in rural areas also highlight the need for either private primary care. Nuff Trust; 2013. p. 1–24. https://www.nuffieldtrust.org.uk/ research/reclaiming-a-population-health-perspective. Accessed 27 May 2018. or public transportation to reach primary care services. 3. Guagliardo MF. Spatial accessibility of primary care: concepts, methods and From a methodological perspective, this study also high- challenges. Int J Health Geogr. 2004;3:3. https://doi.org/10.1186/1476-072X-3-3. lights the importance of considering multiple measures 4. Arcury TA, Gesler WM, Preisser JS, Sherman J, Spencer J, Perin J. The effects of coverage and accessibility when considering the loca- of geography and spatial behavior on health care utilization among the residents of a rual region. Health Serv Res. 2005;40:135–55. tion of GP practices. 5. Nemet GF, Bailey AJ. Distance and health care utilization among the rural The incorporation of the data and analysis presented elderly. Soc Sci Med. 2000;50:1197–208. here into a planning tool could assist with the location 6. Tanser F, Gijsbertsen B, Herbst K. Modelling and understanding primary health care accessibility and utilization in rural South Africa: an exploration decision-making process, as well as the wider provision using a geographical information system. Soc Sci Med. 2006;63:691–705. of healthcare services. This is particularly important in 7. Perry B, Gesler W. Physical access to primary health care in Andean Bolivia. the current Northern Ireland healthcare system, where Soc Sci Med. 2000;50:1177–88. 8. Bright T, Felix L, Kuper H, Polack S. A systematic review of strategies to the role of GP’s and pharmacies are evolving to meet increase access to health services among children in low and middle healthcare needs. For example, current GP practice de- income countries. BMC Health Serv Res. 2017;17:252. https://doi.org/10. velopments include practice based pharmacists, advance 1186/s12913-017-2180-9. 9. Afshari H, Peng Q. Challenges and solutions for location of healthcare nurse practitioners, telephone triage, and online services facilities. Ind Eng Manag. 2014;3(1):12. [28]. The implementation of these innovations should be 10. Todd A, Copeland A, Husband A, Kasim A, Bambra C. Access all areas? An driven by evidence, and informed by the local population area-level analysis of accessibility to general practice and community pharmacy services in England by urbanity and social deprivation. BMJ Open. characteristics. Moreover, this evidence should be con- 2015;5:e007328. https://doi.org/10.1136/bmjopen-2014-007328. sidered alongside alternative services, such as the phar- 11. Tudor Hart J. The inverse care law. Lancet. 1971;297:405–12. macy minor ailments scheme [27]. GP’s could also use 12. Mercer SW, Watt GCM. The inverse care law: clinical primary care encounters in deprived and affluent areas of Scotland. Ann Fam Med. 2007; this information to plan their own staffing requirements 5:503–10. based on population need. This is likely to become a 13. Lovett A, Haynes R, Sunnerberg G, Gate S. Car ravel time and accessibility more prominent concern in Northern Ireland which has by bus to GP services: a study using patient registers and GIS. Soc Sci Med. 2002;55:97–111. an ageing GP workforce [55]. In addition, the analysis of 14. Pearce J, Witten K, Hiscock R, Blakely T. Regional and urban-rural variations accessibility and population characteristics could be used in the association of neighbourhood deprivation with community resource retrospectively to evaluate the performance of past loca- access : a national study. Environ Plan. 2008;40:2469–89. 15. Aday LA, Andersen R. A Framework for the study of access to medical care. tional decisions. If locational decision making processes Health Serv Res. 1974;9:208–20. https://doi.org/10.3205/psm000089. are functioning effectively high levels of correlation 16. National Health Service England. Technical Guide to Allocation Formulae would be expected between the factors which are driving and Pace of Change. 2016. https://www.england.nhs.uk/wp-content/ uploads/2016/04/1-allctins-16-17-tech-guid-formulae.pdf. Accessed 25 May healthcare need and the location of the facilities. 17. Health & Social Care Board (HSCB). Proposed Changes To the Northern Abbreviations 2 Ireland Weighted Capitation Formula. 2015. GP: General practitioner; Km: Kilometre; km : Square kilometres; 18. Luo W, Qi Y. An enhanced two-step floating catchment area ( E2SFCA ) Max: Maximum; MDM: Multiple deprivation measure; Min: Minimum; method for measuring spatial accessibility to primary care physicians. Health SD: Standard deviation; SOA: Super output area Place. 2009;15:1100–7. 19. Hare TS, Barcus HR. Geographical accessibility and Kentucky’s heart-related Availability of data and materials hospital services. Appl Geogr. 2007;27:181–205. The data used in this study is publicly available at the references included in 20. Hawthorne TL. Using GIS and perceived distance to understand the the text. unequal geographies of healthcare in lower-income urban neighbourhoods. Geogr J. 2012;178:18–30. Authors’ contributions 21. Jordan H, Roderick P, Martin D, Barnett S. Distance, rurality and the need for BG carried out all tasks. The author read and approved the final manuscript. care: access to health services in South West England. Int J Health Geogr. 2004;3:21. https://doi.org/10.1186/1476-072X-3-21. Ethics approval and consent to participate 22. Andersen R. Revisiting the behavioral model and access to medical Care : Not Applicable. does it matter? J Health Soc Behav. 1995;36:1–10. 23. Rosero-Bixby L. Spatial access to health care in Costa Rica and its equity: a GIS-based study. Soc Sci Med. 2004;58:1271–84. Competing interests 24. Furler JS, Harris E, Chondros P, Davies PGP, Harris MF, Young DYL. The The author declares that he has no competing interests. inverse care law revisited: impact of disadvantaged location on accessing longer GP consultation times. Med J Aust. 2002;177:80–3. Publisher’sNote 25. Adams J, White M. Socio-economic deprivation is associated with increased Springer Nature remains neutral with regard to jurisdictional claims in proximity to general practices in England: an ecological analysis. J Public published maps and institutional affiliations. Health (Bangkok). 2005;27:80–1. Graham BMC Health Services Research (2018) 18:398 Page 10 of 10 26. Asthana S, Gibson A. Deprivation, demography, and the distribution of 49. Yen IH, Michael YL, Perdue L. Neighborhood environment in studies of general practice: challenging the conventional wisdom of inverse care. Br J health of older adults. A Systematic Review. Am J Prev Med. 2009;37:455–63. Gen Pract. 2008;58:720–8. https://doi.org/10.1016/j.amepre.2009.06.022. 27. Northern Ireland Health and Social Care Business Services Organisation. 50. Riva M, Gauvin L, Barnett TA. Toward the next generation of research into Minor Ailments Scheme. http://www.hscbusiness.hscni.net/2055.htm. small area effects on health: a synthesis of multilevel investigations Accessed 4 Apr 2018. published since July 1998. J Epidemiol Community Health. 2007;61:853–61. https://doi.org/10.1136/jech.2006.050740. 28. Health and Social Care Board. Investment in GP Practices. http://www. 51. Paez A, Mercado RG, Farber S, Morency C, Roorda M. Accessibility to health hscboard.hscni.net/our-work/integrated-care/gps/investment-in-gp- care facilities in Montreal Island: an application of relative accessibility practices/. Accessed 4 Apr 2018. indicators from the perspective of senior and non-senior residents. Int J 29. Northern Ireland Council for Voluntary Action. GP Practices. 2017. http:// Health Geogr. 2010;9:52. data.nicva.org/dataset/gp-practices. Accessed 20 May 2018. 52. Rice DP, Feldman JJ. Living longer in the United States: demographic 30. Health and Social Care Board. General Medical Services - Frequently Asked changes and health needs of the elderly. Milbank Mem Fund Q Health Soc. Questions. 2018. http://www.hscboard.hscni.net/our-work/integrated-care/ 1983;61:362–96. https://doi.org/10.2307/3349863. gps/faqs/. Accessed 4 Apr 2018. 53. Teljeur C, O’Dowd T, Thomas S, Kelly A. The distribution of GPs in Ireland in 31. Potter M. Equality and human rights legislation in Northern Ireland : a relation to deprivation. Health Place. 2010;16:1077–83. https://doi.org/10. review. 2011. http://www.niassembly.gov.uk/globalassets/Documents/RaISe/ 1016/j.healthplace.2010.06.011. Publications/2011/OFMdFM/7511.pdf. Accessed 25 May 2018. 54. Alford-Teaster J, Lange JM, Hubbard RA, Lee CI, Haas JS, Shi X, et al. Is the 32. Department of Health. Equality Scheme for the Department of Health. 2017. closest facility the one actually used? An assessment of travel time https://www.health-ni.gov.uk/sites/default/files/consultations/health/ estimation based on mammography facilities. Int J Health Geogr. 2016;15:8. equality-scheme-consultation-document.docx. Accessed 20 May 2018. https://doi.org/10.1186/s12942-016-0039-7. 33. Abolhallaje M, Mousavi SM, Anjomshoa M, Beigi Nasiri A, Seyedin H, 55. Health and Social Care Board. Developments in GP Services. http://www. Sadeghifar J, et al. Assessing health inequalities in Iran: a focus on the hscboard.hscni.net/our-work/integrated-care/gps/gp-service-developments/. distribution of health care facilities. Glob J Health Sci. 2014;6:33750. https:// Accessed 4 Apr 2018. doi.org/10.5539/gjhs.v6n4p285. 34. Mobley L, Root E, Anselin L, Lozano-Gracia N, Koschinsky J. Spatial analysis of elderly access to primary care services. Int J Health Geogr. 2006;5:19. https://doi.org/10.1186/1476-072X-5-19. 35. Luo W. Using a GIS-based floating catchment method to assess areas with shortage of physicians. Health Place. 2004;10:1–11. 36. Ordinance Survey Northern Ireland. Ordinance survey Northern Ireland open data. Road Networks. http://osni-spatial-ni.opendata.arcgis.com/. Accessed 4 Apr 2018. 37. Healt and Social Care Business Services Organisation. Northern Ireland GP practice lists for professional use. 2017. http://www.hscbusiness.hscni.net/ services/1816.htm. Accessed 17 Aug 2017. 38. Comber AJ, Brunsdon C, Radburn R. A spatial analysis of variations in health access : linking geography, socio-economic status and access perceptions. Int J Health Geogr. 2011;10:44. https://doi.org/10.1186/1476-072X-10-44. 39. Shah TI, Bell S, Wilson K. Spatial accessibility to health care services: identifying under-serviced Neighbourhoods in Canadian urban areas. PLoS One. 2016;11:1–22. https://doi.org/10.1371/journal.pone.0168208. 40. Northern Ireland Statistics and Research Agency. Northern Ireland Multiple Deprivation Measure 2017 (NIMDM2017). Northern Ireland Statistics Research Agency. 2017. https://www.nisra.gov.uk/statistics/deprivation/ northern-ireland-multiple-deprivation-measure-2017-nimdm2017. Accessed 4 Apr 2018. 41. NISRA. 2016 Mid year population estimates for Northern Ireland. Northern Ireland Statistics and Research Agency. 2016. https://www.nisra.gov.uk/ publications/2016-mid-year-population-estimates-northern-ireland. Accessed 25 May 2018. 42. Northern Ireland Statistics and Research Agency. Northern Ireland Super Output Areas. 2011. https://www.nisra.gov.uk/support/geography/northern- ireland-super-output-areas#toc-0. Accessed 4 Apr 2018. 43. Northern Ireland Statistics and Research Agency. Census 2011. 2011. http:// www.ninis2.nisra.gov.uk/public/Theme.aspx?themeNumber= 136&themeName=Census. Accessed 4 Apr 2018. 44. Northern Ireland Statistics and Research Agency. Geography Fact Sheet. 2013. https://www.ninis2.nisra.gov.uk/public/documents/ NISRA%20Geography%20Fact%20Sheet.pdf. Accessed 5 Nov 2017. 45. The R project for statistical computing. 2017. https://www.r-project.org/. Accessed 5 Nov 2017. 46. Wood D, Clark D, Gatrell AC. Equity of access to adult hospice inpatient care within north-West England. Palliat Med. 2004;18:543–9. https://doi.org/10. 1191/0269216304pm892oa. 47. Jokela M. Does neighbourhood deprivation cause poor health? Within- individual analysis of movers in a prospective cohort study. J Epidemiol Community Health. 2015;69:899–904. https://doi.org/10.1136/jech-2014- 48. Tøge AG, Bell R. Material deprivation and health: a longitudinal study. BMC Public Health. 2016;16:747. https://doi.org/10.1186/s12889-016-3327-z. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png BMC Health Services Research Springer Journals

Population characteristics and geographic coverage of primary care facilities

Free
10 pages

Loading next page...
 
/lp/springer_journal/population-characteristics-and-geographic-coverage-of-primary-care-7aOUTETwPe
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s).
Subject
Medicine & Public Health; Public Health; Health Administration; Health Informatics; Nursing Research
eISSN
1472-6963
D.O.I.
10.1186/s12913-018-3221-8
Publisher site
See Article on Publisher Site

Abstract

Background: The location of General Practitioner (GP) facilities is an important aspect in the design of healthcare systems to ensure they are accessible by populations with healthcare needs. A key consideration in the facility location decision involves matching the population need for the services with the supply of healthcare resources. The literature points to several factors which may be important in the decision making process, such as deprivation, transportation, rurality, and population age. Methods: This study uses two approaches to examine the factors associated with GP accessibility in Northern Ireland. The first uses multinomial regression to examine the factors associated with GP coverage, measured as the proportion of people who live within 1.5 km road network distance from the nearest GP practice. The second focuses on the factors associated with the average travel distance to the nearest GP practice, again measured using network distance. The empirical research is carried out using population and geospatial data from Northern Ireland, across 890 Super Output Areas and 343 GP practices. Results: In 19% of Super Output Areas, all of the population live within 1.5 km of a GP practice, whilst in 24% none of the population live within 1.5 km. The regression results show that there are higher levels of population coverage in more deprived areas, smaller areas, and areas that have more elderly populations. Similarly, the average travel distance is related to deprivation, population age, and area size. Conclusions: The results indicate that GP practices are located in areas with higher levels of service need, but also that care needs to be taken to ensure rural populations have sufficient access to services, whether delivered through GP practices or through alternative services where GP practices are less accessible. The methodology and results should be considered by policy makers and healthcare managers when making decisions about GP facility location and service provision. Keywords: Accessibility, Facility location, Primary care location, Coverage, General practice, Healthcare need Background Facility location is also important in maximising cost and Primary care plays an important role in improving the utilisation efficiencies in the healthcare system [9], which health of surrounding populations [1–3]. However, GP is crucial in a climate of public funding constraints and practices must be accessible to facilitate utilisation and for the business efficiency of GP practices. Ensuring equity better health outcomes [4–7], thus highlighting the im- of access to facilities across the population is also import- portance of considering accessibility in the design and ant, and can be particularly challenging in areas with a management of healthcare systems. An important com- rural geography, deprivation inequalities, and other social, ponent of accessibility is related to geography, and in religious and political divides. particular to the distance that must be travelled to reach This study focuses specifically on the location of GP the facility, which can impact on service utilisation for practices, which are crucial components of the health- both preventative and curative interventions [3, 4, 8]. care system in providing both primary care and func- tioning as an access point to other parts of the healthcare system. In addition, GP’s provide a range of Correspondence: byron.graham@qub.ac.uk other services such as vaccinations and smoking Queen’s Management School, Queen’s University Belfast, Riddel Hall, 185 Stranmillis Road, Belfast BT9 5EE, UK © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Graham BMC Health Services Research (2018) 18:398 Page 2 of 10 cessation [10]. Although the importance of primary care such as cost effectiveness, the location of existing facil- accessibility is recognised, the inverse care law proposes ities, planned developments, and government and ad- that areas which are most in need of healthcare tend to ministrative policies. Despite this, much of the literature have poorer levels of access, particularly where market on the location of healthcare facility location only in- forces are at play [11], although the empirical evidence cludes a small subset of the possible factors influencing for this proposition is mixed [12–14]. the location decision, or focuses only on achieving max- Deciding on the location of primary care facilities is com- imal coverage regardless of the population characteris- plex, with accessibility and utilisation influenced by factors tics influencing the service need. in addition to geographical proximity, such as wait time, af- This study draws on the existing literature on health- fordability, healthcare knowledge, and the availability of care need, accessibility, and utilisation to examine the transportation [4, 15]. When studying access to services it factors associated with geographical location of GP prac- is important to take into consideration the factors associ- tices. Specifically, this study will seek to answer the fol- ated with population need for the service, such as age, gen- lowing research question: der, deprivation, and measures of health [15]. These are What factors are associated with the location of pri- considered in both the academic literature [10, 13], and by mary healthcare facilities? healthcare policy and planning authorities, through for ex- To answer this question, multivariable regression ample, capitation formulas [16, 17], which are used in the models are developed to examine the factors related to allocation of funding based on the population need. The the proportion of each Super Output Area (SOA) residing ability to access the service should also be considered, along within 1.5 km of the nearest GP practice, and the average with the need for the service and any predisposing risk fac- travel distance for each SOA. In contrast to some previous tors [15]. Moreover, these factors may also be important de- studies which have used Euclidian distance [10, 23], this terminants of the service mix required by the population. study uses the more accurate network distance. These Healthcare decision makers and practice managers should measures are combined with other sources of area level therefore consider these wider factors when making loca- data which may be related to the location of the GP prac- tional decisions and decisions around the services provided, tice, such as multiple deprivation, health deprivation, such as opening hours and specialist services. population age, area, and religion. This facilitates a de- The accessibility of healthcare facilities can take into tailed examination of the relationships between accessibil- account both spatial and non-spatial factors, and can be ity and population characteristics. measured and interpreted in a range of ways. These can This study makes several theoretical and practical con- include the distance between the population and facility, tributions to the literature. The first is the use of a novel facilities per population of geographical region, the num- dataset to study the factors related to the location of ber of staff and waiting times, satisfaction, utilisation, healthcare facilities. The paper adds to our understand- cost of access, transportation links, catchment areas or ing of the range of factors that could shape the location coverage, and gravity models [3, 15, 18, 19]. Studies such of GP practices, providing policy makers with insight as Hawthorne [20] have added to these accessibility about current levels of provision and the factors related measures by creating a measure of access based on dis- to accessibility. In particular, this study will contribute to tance and satisfaction with the service, to take account our knowledge of the population characteristics in which of the potential for people to travel further to access GP practices are located in Northern Ireland. This re- higher quality care. mains an important topic, with previous studies report- Despite the breadth and depth of research into the ac- ing conflicting findings around the existence of an cessibility of healthcare facilities, challenges remain in inverse care law, with some studies finding evidence in both the scientific literature and in healthcare location support of an inverse care law [12, 13, 24], and others policy and decision-making. One challenge is that acces- not [10, 25]. However, in general distance based studies sibility is a multifaceted concept, with no consistent def- tend to show that GP practices are more accessible in inition of ‘poor access’ to healthcare services [21]. more deprived areas [10]. Moreover, the relationship be- Moreover, the literature highlights a wide range of pre- tween deprivation and accessibility is made less clear disposing, enabling and need factors important in deter- due to differences in the age profile of areas, with more mining healthcare utilisation [15, 22]. This points deprived areas consisting of more younger people, com- towards the complexity of the facility location problem, pared with less deprived areas consisting of more older and makes it challenging to objectively evaluate and plan people and consequently more age related healthcare service provision. needs [26]. This study makes a further contribution to The multifaceted nature of accessibility makes the fa- this debate, by considering both deprivation and age. cility location decision particularly challenging due to From a practical perspective, this study will contribute the need to take into consideration a range of factors, to planning and policy decisions on the location of Graham BMC Health Services Research (2018) 18:398 Page 3 of 10 primary healthcare facilities through the measurement variety of measures such as straight line distance [10], of facility coverage, and its associated factors. The results drive time [13], availability of bus services [13], floating presented in this study will also help policy makers to catchment areas [18, 34, 35], and population to GP ra- examine the extent to which GP practice location is tios. For example, Todd et al. [10] measure the Euclidian matched to population characteristics. distance between population centroids and the nearest Identification of areas where accessibility is low is also GP practice. Rosero-Bixby [23] also uses a distance important to healthcare decision makers in planning the measure, but also creates an additional weighted index provision of services. This is of current policy relevance in measure taking account other relevant factors. primary care as pharmacies take on some of the roles his- As the focus of this study is on the factors related to torically reserved for GP’s such as minor ailment schemes distance accessibility, two measures of geographic acces- [27], and GP practices expand to take on additional roles sibility were developed for use as dependent variables. [28]. Incorporation of the insights gained in this study The first measures practice coverage as the proportion could therefore add to the planning and policy discussions of people in each SOA that live within 1.5 km of a GP around the provision of wider healthcare services. As this practice. The second measure of accessibility is the aver- study is based on the analysis of publicly available data, it age distance between the population of each SOA and also demonstrates the usefulness of open data in studying the nearest GP practice. healthcare geography and highlights the importance of re- Both measures were calculated based on network dis- cent government open data initiatives. tance using road network data from Ordinance Survey Northern Ireland [36]. Network distance and subsequent Methods mapping was carried out using ArcGIS (version 10.5). The Study setting 2017 GP list [37] and the 2017 central postcode directory This studyfocuses on thelocation of GP practices in were used to identify the postcodes and latitude and longi- Northern Ireland. In Northern Ireland there are 343 GP tude of the facilities and postcodes in Northern Ireland. practices and 1710 GP’s serving a population of over 1.9 The postcode location of the GP practice was identified by million registered patients [29, 30]. GP practices are run as combining the GP postcode with the central postcode small businesses, with practices responsible for day to day lookup, which contains the latitude and longitude of every management and recruitment of staff [30]. Practices are postcode in the United Kingdom. Postcodes have been funded by the Northern Ireland Health and Social Care widely used to calculate distance in previous GIS studies Board under the current GP contract, but remain as separ- e.g. [10, 38]. On average there are 31 people living at each ate businesses [30]. The initial location decision is therefore postcode in Northern Ireland, and there are over 58,000 likely to be crucial, as it is likely to be easier to expand an postcodes. Distance was measured by calculating the net- existing practice by employing additional staff than it would work distance between the GP postcode and every other be to open an entirely new facility. There may also be sce- postcode in Northern Ireland. narios where alternative services are more appropriate. The proportion of people within the coverage area was The locational decision of GP practices in Northern calculated by dividing the sum of the usually resident Ireland is based on criteria that aims to ensure accessi- population at each postcode location within 1.5 km of a bility, taking into consideration factors such as demand, GP by the total population of the SOA, resulting in a travel distance, new housing developments and difficul- value between 0 and 1. This measure acts as a proxy for ties registering on existing GP lists. Equality of access to individual level travel distance to the nearest GP prac- public services is a particularly important issue in North- tice. Due to a high proportion of areas having either full ern Ireland due to community divides on political and or no coverage, a new variable was computed for use in religious grounds. Ensuring equality of provision of pub- the subsequent modelling, consisting of four categories: lic services is also written into Northern Ireland legisla- no coverage, low coverage, high coverage, and full cover- tion [31], with Section 75 of the Northern Ireland Act age. This variable was based on percentage of the popu- 1998 requiring public services to consider equality of lation within 1.5 km of a GP practice, with no coverage provision across the areas of religion, political opinion, representing 0%, low coverage greater than 0% and less race, age, gender, marital status, sexual orientation, dis- than 50%, high coverage between 50% and less than ability, and dependants [32]. Equality of access to health- 100%, and full coverage indicating 100% of the area is care services in general has been widely studied in the within 1.5 km of a GP practice. literature across various geographical areas [33]. Similar to Todd et al. [10] who used a distance of 1.6 km, a distance of 1.5 km coverage was chosen as Dependent variables representing around a 20 min walk from the persons There is no universally agreed measure of coverage or house to the GP practice. Sensitivity analysis was also accessibility in the literature, with studies adopting a carried out based on other distances, as discussed Graham BMC Health Services Research (2018) 18:398 Page 4 of 10 further in the analysis section. The average travel dis- The final set of variables included in the analysis was de- tance to the nearest GP practice was also based on the termined by the literature reviewed, the data available, and network distance calculated for the coverage proportion significance in the models. All data included in the study measures. This was calculated by deriving the mean net- was measured at the SOA level in Northern Ireland, which work travel distance for the population of each SOA, is the lowest level at which most of the statistical measures again based on the usually resident postcode population of interest are made available. Northern Ireland is broken and the distance to their nearest GP practice. down into 890 SOAs, with an average of 2000 people or 700 households in each area [44]. Independent variables Data analysis In addition to geographic measures of accessibility, The initial analysis of the dataset involved producing de- previous studies have considered a range of other fac- scriptive statistics and visualisations, and investigating tors which may be important in the location of GP the correlation between variables. Due to the ordered practices. These include access to transportation, nature of the coverage dependent variable ordinal logis- healthcare needs, rurality, deprivation, and demog- tic regression would be the best choice as it utilises the raphy [10, 13, 14, 26, 38, 39]. Drawing on the wider information contained in the order of the variables. literature, the independent variables included in this However, the test of parallel lines assumption was vio- study were the Northern Ireland Measure of Multiple lated. Multinomial logistic regression was therefore used Deprivation (MDM), health deprivation, the SOA level to examine the relationships between coverage and the population aged 0–15 and over 65, the proportion of population characteristics. Ordinary least squares regres- the area who report being protestant religion, and the sion was used to examine the relationships between size of the area. In total six publicly available datasets population characteristics and the average travel distance were combined to form the final dataset. The 2017 within each SOA. Separate models were built to include MDM was used in the final models, which is a com- multiple deprivation and health deprivation due to high posite measure made up of 8 sub domains, which levels of correlation between these two variables. provides a deprivation ranking to each area [40]. The Sensitivity analysis was carried out by rerunning the health deprivation domain was also included individu- multinomial regression models based on different net- ally due to the particular relevance of this domain in work distances of 750 m, 1 km, 1.25 km, 1.75 km and the location of GP practices. 2 km. Each model was examined for differences in the The 2016 mid year population age groups [41] were significance of the variables compared to the focal used as the measure of age. Earlier models included four 1.5 km model. All data pre-processing, variable calcula- age bands: 0–15, 16–39, 40–64 and 65+, but the inclu- tion, visualisations and initial statistical analyses were sion of all four bands introduced high levels of multicoli- undertaken using the R statistical programming language nearity into the model, probably because of children [45], with multinomial regression analysis undertaken in aged 0–15 cohabiting with their parents or guardians SPSS Version 22. aged 15–64. The final models therefore only include the age bands of 0–15 and 65+. The SOA area was mea- sured in square kilometres as presented in the most re- Results cent Northern Ireland SOA’s[42]. Table 1 summarises the dependent variable and the Religion was captured in the models using the 2011 Northern Ireland SOA level measures that are included census measurement of the percentage of people report- in the study. Across all SOA’s the mean network distance ing their religion or religion brought up in was Protest- between the population and the nearest GP practice is ant and other Christian [43]. In the 2011 census, 93.5% 2684.51 m. This is displayed visually in Fig. 1, which of the population selected either this option, or Catholic, shows a map of Northern Ireland with the mean travel with other religions accounting for less than 1% of the distance within each SOA, along with the location of population, and no religion accounting for 5.6% of the each GP practice. In 18.7% of SOA’s all of the population population. Because of the very small proportions out- live within 1.5 km of a GP practice, and in 24.4% of side of Protestant and Catholic, the models focus on the SOA’s none of the population live within 1.5 km of a GP proportion of the population reporting as Protestant and practice. This highlights the differences in GP coverage Other Christian. Car ownership and rurality were in- levels across the region. SOA’s have a mean size of cluded in earlier models, but are not included in the 15.882 km . Across all SOA’s, a mean of 436 people are final models as these measures are likely to be captured aged 0–15, 654 are aged 16–39, 668 are aged 40–64, and by the multiple deprivation measure, and SOA area mea- 335 are aged 65 and over. The mean MDM and health sures respectively. deprivation rank is 445.5. Graham BMC Health Services Research (2018) 18:398 Page 5 of 10 Table 1 Descriptive statistics Statistic N Mean / (%) SD Min Max Average Distance to GP 890 2684.51 2561.73 132.93 17,455.07 MDM Rank 890 445.50 257.065 1 890 Health Deprivation Rank 890 445.50 257.065 1 890 Area (km ) 890 15.882 29.54 0.11 205.11 Persons Aged 0–15 890 435.95 176.96 48 1663 Persons Aged 16–39 890 654.12 250.942 94 2362 Persons Aged 40–64 890 667.68 200.58 110 1575 Persons Aged 65+ 890 334.55 123.57 6 1066 Protestant or Other Christian (%) 890 48.96 29.14 1.5 91.26 Full Coverage (%) 167 18.7 High Coverage (%) 260 29.2 Low Coverage (%) 246 27.7 No Coverage (%) 217 24.4 For subsequent multinomial models the proportion coverage. The results show that the most deprived of people in each SOA living within 1.5 km of a GP areas tend to have the highest level of coverage, practice was converted to four levels, ranging from across both MDM and health deprivation ranks. no coverage to full coverage. Table 2 presents the de- Areas with lower coverage also tend to be much lar- scriptive statistics broken down by the four levels of ger in terms of geographical area. No consistent Fig. 1 Mean travel distance to the nearest GP practice, and the location of GP practices in Northern Ireland (source: Authors’ own work) Graham BMC Health Services Research (2018) 18:398 Page 6 of 10 Table 2 Descriptive Statistics by Coverage Level. Mean with standard deviation in parentheses Full High Low None Average Distance 615.24 (252.14) 1495.02 (1782.43) 3205.39 (2076.71) 5111.68 (2649.21) MDM Rank 355.64 (295.17) 431.36 (259.25) 494.38 (244.02) 476.18 (216.19) Health Deprivation Rank 306.57 (256.47) 401.12 (245.67) 529.39 (244.28) 510.48 (227.65) Area (km ) 0.59 (0.788) 5.27 (12.19) 27.84 (39.15) 26.81 (32.79) Persons Aged 0–15 354.71 (127.08) 421.34 (167.73) 477.07 (201.51) 469.38 (168.12) Persons Aged 16–39 619.93 (255.23) 659.02 (267.42) 679.09 (262.81) 646.26 (208.13) Persons Aged 40–64 558.87 (139.07) 644.75 (192.15) 728.76 (210.96) 709.65 (200.41 Persons Aged 65+ 303.83 (104.89) 337.75 (124.20) 359.30 (133.89) 326.30 (118.56) Protestant or Other Christian 46.63 (30.68) 47.57 (28.87) 49.30 (28.83) 52.05 (28.52) pattern seems to emerge from the descriptive statis- because of children and their parents cohabiting. The re- tics around the population age and religion. sults also show a significant negative relationship be- tween the proportion of people who report being Regression results protestant, and the level of GP coverage. Smaller areas The results of the regression analyses are presented in also have higher levels of GP coverage. Tables 3 and 4. Model 1 shows the results of the multi- Model 2 (Table 3) focuses on the relationships with nomial regression, focusing on factors related to the pro- the average network distance that must be travelled by portion of the area level population living within 1.5 km the population to reach their nearest GP practice. This of a GP practice – grouped into full coverage, high model shows that areas with higher levels of health coverage, low coverage, and no coverage. This model deprivation have lower travel distances to their nearest shows that, compared to areas with no coverage, areas GP practice. There is a significant positive relationship with high and full coverage are more health deprived. between the number of people aged 0–15 and the aver- The results also show that areas with full coverage have age distance to the nearest GP practice. However, the a lower number of people aged 0–15 when compared number of people aged over 65 is not significant, nor is with areas with no coverage. Conversely, areas with a religion. A significant positive relationship is found be- higher number of people aged over 65 have higher levels tween the size of the area and the distance that must be of coverage. In earlier models, other age bands were in- travelled to reach the nearest GP practice. cluded, but were removed in the final models due to Model 3 focuses on the proportion of the population high levels of multicolinearity, probably occurring of each area within 1.5 km of a GP practice, but unlike Table 3 Model 1 shows the multinomial regression including Age, area, religion, and health deprivation influences on coverage. Reference category is no coverage. Model two shows the relationships with average travel distance Model 1 Model 2 Low coverage High coverage Full coverage Average distance Health Deprivation and Disability (reverse coded) .000 .001** .002*** −1.143*** (.000) (.000) (.001) (.306) Population aged 0–15 .000 −.001 −.002*** 1.387** (.001) (.001) (.001) (.437) Population aged 65+ .003*** .004*** .006*** −.855 (.001) (.001) (.001) (.630) Protestant −.010** −.012*** −.012** −.346 (.004) (.004) (.005) (2.756) Area (km ) −.004 −.051*** −1.272*** 45.741*** (2.653) (.003) (.007) (.224) Intercept −.079 .002*** .448*** 2165.628*** (337.347) (.472) (.476) (.567) N 890 Log likelihood 2010.943*** 2 2 Nagelkerke R : .412 R :.352 *p < 0.10; **p < 0.05; ***p < 0.01. Unstandardised coefficients with standard errors in parentheses Graham BMC Health Services Research (2018) 18:398 Page 7 of 10 Table 4 Model 3 shows the multinomial regression including Age, area, religion, and multiple deprivation influences on coverage. Reference category is no coverage. Model 4 shows the relationships with average travel distance Model 3 Model 4 Low Coverage High Coverage Full Coverage Average Distance Multiple Deprivation (reverse coded) .000 .001** .002*** −0.616** (.000) (.000) (.000) (0.296) Population aged 0–15 .000 −.001 −.002** 1.329*** (.001) (.001) (.001) (0.440) Population aged 65+ .003 .004*** .006*** −0.747 (.001) (.001) (.001) (0.637) Protestant −.009** −.013*** −.012** 1.351 (.004) (.004) (.005) (2.736) Area (km ) −.003 −.055*** −1.312*** 48.574*** (2.567) (.003) (.007) (.225) Intercept −.127 .053** .528 1791.466*** (328.601) (.465) (.466) (.550) N 890 Log likelihood 2011.011*** 2 2 Nagelkerke R :.412 R :.345 *p < 0.10; **p < 0.05; ***p < 0.01. Unstandardised coefficients with standard errors in parentheses model 1, it focuses on multiple deprivation rather than when different distances are used. This section reviews specifically focusing on health deprivation. Compared to the patterns in the differences that were found. For the low coverage areas, high and full coverage areas tend to purposes of this section, significance is considered at the have higher levels of deprivation. The number of people 5% level. Although, the population aged 0–15 was only aged over 65 is significantly positively related to higher significant in the full coverage models presented previ- levels of coverage, whereas the number of people aged ously, significant positive relationships were also found under 15 is significantly negatively related to full cover- in all sensitivity models at the high coverage level, and at age. There is a significant negative relationship between the low coverage level in the two models based on a dis- the proportion of the area level population reporting as tance of 1.25 km. Health deprivation and MDM were Protestant and Other Christian and higher levels of found to be significant in the 0.75 km and 1 km models, coverage. The results also show that smaller areas have but MDM was not significant in the model based on higher levels of coverage. 1 km distance when considering full coverage. The SOA Model 4 focuses on the relationships with average area was also found to be significant in the health travel distance, including multiple deprivation rather deprivation models when considering distances of than health deprivation. The results of this model show 0.75 km and 1 km. Some differences also emerged when a significant negative relationship between deprivation focusing on religion. In contrast to models 1 and 3, the and travel distance. The proportion of the population number of people reporting protestant religion was aged under 15 is significantly positively related to the found not to be significant in the models for full cover- average travel distance, but no evidence is found for a age based on 0.75 km and 1 km, nor in the low coverage relationship between the proportion of the population models based on distances of 1.75 km and 2 km. More- aged over 65 and travel distance. The proportion of over, the overall variance explained by the models in- people who report as Protestant and Other Christian is creased as the distance used to calculate the dependent not found to be significant. Evidence is also found for a variable was increased. Across all models the lowest positive relationship between the size of the area and the Nagelkerke R-squared was 0.266 in the model based on average travel distance. a distance of 0.75 km, including health deprivation. The highest R-squared was 0.513, which resulted in both Results of the sensitivity analysis models based on the 2 km distance. Sensitivity analysis was carried out by constructing a series of dependent variables based on different network Discussion distances of 0.75 km, 1 km, 1.25 km, 1.75 km and 2 km, This study focuses on the relationships between primary and comparing the results of these models with the focal care accessibility and population characteristics in 1.5 km model. Although the overall patterns were simi- Northern Ireland. The results provide insight into the lar, this analysis revealed some differences in the models relationships between the population characteristics and Graham BMC Health Services Research (2018) 18:398 Page 8 of 10 the location of GP practices, adding to our understand- across models. These differences highlight the importance ing of the populations in which GP practices are located. of considering different levels of coverage when examining The healthcare planning assumption underpinning the accessibility. The population aged 0–15 may bemoreim- study is that GP practices should be located closest to portant for other distances than the focal distance pre- areas with the greatest need, and therefore relationships sented here. Religion may be less important for other are expected between population need and accessibility. distances. One potential explanation for the differences in The results of the regression analysis support the the levels of significance could be the movement of people proposition that the most deprived areas should have from one category to another as the distance changes, the highest levels of GP coverage. This is the case for therefore changing the number of people in each category. both overall multiple deprivation and for health Healthcare planners may therefore want to consider mul- deprivation, even when controlling for area size, popula- tiple distances when examining GP practice coverage. tion age, and religion. This relationship exists when ac- Moreover, the larger the distance considered in the cessibility is measured using coverage and average dependent variable the more variance the model explains. distance. The only models where deprivation is not sig- Although this study contributes to the current body of nificant are those comparing no coverage with low levels literature on the geography of healthcare facilities, it is of coverage. This finding suggests that GP practices are not without limitations. This study focuses specifically located in areas with highest levels of deprivation related on coverage as one measure of accessibility, but does need. The relationship between deprivation and facility not consider which facilities people choose to attend. Al- location observed in this study is consistent with find- though there is some evidence from other contexts to ings from the wider literature [10, 25]. This importance suggest people tend to choose one of the closest facilities of this finding is further highlighted by the negative rela- [54], future studies could consider whether people tionship between deprivation and population health dis- choose to attend the facility that is closest to home. This cussed in the literature [46–50]. study also considers two specific measures of accessibil- The population age is an important consideration in the ity, and there are other potential measures that future location of healthcare facilities as older people tend to studies could incorporate to build on this research. Fu- have increased mobility problems [51], and also suffer ture research could consider the specific population from increased health problems [52]. The findings pre- needs and service profiles of the GP practices, as well as sented here show that areas with older populations also the impact of need and accessibility on utilisation. This have higher levels of coverage even when controlling for research also focuses specifically on factors that may in- other factors such as deprivation, area size, and religion. fluence the need for primary care services, rather than However, no evidence is found for a relationship between focusing on the factors related to the utilisation of ser- the population aged 65 and over and average travel dis- vices. Future work could consider which factors drive tance. Areas with more people aged under 15 are associ- need, accessibility, and utilisation of services, and ated with higher average travel distance. Moreover, areas whether these factors are the same. Consideration of with full coverage have lower numbers of people in this other healthcare services could also provide additional age group. These findings suggests that there is not neces- insight for healthcare decision makers. Moreover, the sarily a trade-off between coverage in deprived areas and measures considered here do not consider the ability of coverage in areas with an older population. the healthcare provider to deliver the care. Future stud- Smaller areas tend to have higher levels of coverage, ies could therefore incorporate factors such as the num- which shows that GP’s are located closer to population ber of GP’s and other staff working at the practice, the centres, where they can serve more dense populations. practice list size, specialist services, and opening hours Although this make sense conceptually, consideration matched to population need. also needs to be given to how the primary healthcare needs of more rural areas are met, whether this is Conclusions through GP practices or alternative services. These find- This study focuses on the relationship between popula- ings are consistent with previous research showing sub- tion characteristics and GP practice accessibility. The stantially better access to GP practices in urban areas in findings of the research show that GP practices are lo- England [10] and Ireland [53]. The results also show cated in areas with higher levels of need, as measured by some evidence of differing access to GP practices by reli- older populations and deprivation. This research con- gion. Historically, religious divides have been prominent tributes to the existing methodological and empirical lit- in Northern Ireland, with certain areas continuing to re- erature through the examination of multiple variables main predominantly Protestant or Catholic. which should influence the location decision making Although the overall general trend is similar across the process. From a policy perspective, the findings indicate models, the sensitivity analysis does show some differences some of the factors that are associated with GP location. Graham BMC Health Services Research (2018) 18:398 Page 9 of 10 However, a significant proportion of the population live Received: 8 November 2017 Accepted: 22 May 2018 outside of the 1.5 km coverage area, which is consistent with the rural geography of Northern Ireland. Planners should consider how best to serve this population, for References 1. Starfield B, Shi L, Macinko J, Starfield B, Shi L. Contribution of primary care example through other means of care, such as pharma- to health systems and health. Milbank Q. 2005;83:457–502. cies and district nursing services. Long travel distances 2. Thorlby R. Reclaiming a population health perspective. Future challenges for in rural areas also highlight the need for either private primary care. Nuff Trust; 2013. p. 1–24. https://www.nuffieldtrust.org.uk/ research/reclaiming-a-population-health-perspective. Accessed 27 May 2018. or public transportation to reach primary care services. 3. Guagliardo MF. Spatial accessibility of primary care: concepts, methods and From a methodological perspective, this study also high- challenges. Int J Health Geogr. 2004;3:3. https://doi.org/10.1186/1476-072X-3-3. lights the importance of considering multiple measures 4. Arcury TA, Gesler WM, Preisser JS, Sherman J, Spencer J, Perin J. The effects of coverage and accessibility when considering the loca- of geography and spatial behavior on health care utilization among the residents of a rual region. Health Serv Res. 2005;40:135–55. tion of GP practices. 5. Nemet GF, Bailey AJ. Distance and health care utilization among the rural The incorporation of the data and analysis presented elderly. Soc Sci Med. 2000;50:1197–208. here into a planning tool could assist with the location 6. Tanser F, Gijsbertsen B, Herbst K. Modelling and understanding primary health care accessibility and utilization in rural South Africa: an exploration decision-making process, as well as the wider provision using a geographical information system. Soc Sci Med. 2006;63:691–705. of healthcare services. This is particularly important in 7. Perry B, Gesler W. Physical access to primary health care in Andean Bolivia. the current Northern Ireland healthcare system, where Soc Sci Med. 2000;50:1177–88. 8. Bright T, Felix L, Kuper H, Polack S. A systematic review of strategies to the role of GP’s and pharmacies are evolving to meet increase access to health services among children in low and middle healthcare needs. For example, current GP practice de- income countries. BMC Health Serv Res. 2017;17:252. https://doi.org/10. velopments include practice based pharmacists, advance 1186/s12913-017-2180-9. 9. Afshari H, Peng Q. Challenges and solutions for location of healthcare nurse practitioners, telephone triage, and online services facilities. Ind Eng Manag. 2014;3(1):12. [28]. The implementation of these innovations should be 10. Todd A, Copeland A, Husband A, Kasim A, Bambra C. Access all areas? An driven by evidence, and informed by the local population area-level analysis of accessibility to general practice and community pharmacy services in England by urbanity and social deprivation. BMJ Open. characteristics. Moreover, this evidence should be con- 2015;5:e007328. https://doi.org/10.1136/bmjopen-2014-007328. sidered alongside alternative services, such as the phar- 11. Tudor Hart J. The inverse care law. Lancet. 1971;297:405–12. macy minor ailments scheme [27]. GP’s could also use 12. Mercer SW, Watt GCM. The inverse care law: clinical primary care encounters in deprived and affluent areas of Scotland. Ann Fam Med. 2007; this information to plan their own staffing requirements 5:503–10. based on population need. This is likely to become a 13. Lovett A, Haynes R, Sunnerberg G, Gate S. Car ravel time and accessibility more prominent concern in Northern Ireland which has by bus to GP services: a study using patient registers and GIS. Soc Sci Med. 2002;55:97–111. an ageing GP workforce [55]. In addition, the analysis of 14. Pearce J, Witten K, Hiscock R, Blakely T. Regional and urban-rural variations accessibility and population characteristics could be used in the association of neighbourhood deprivation with community resource retrospectively to evaluate the performance of past loca- access : a national study. Environ Plan. 2008;40:2469–89. 15. Aday LA, Andersen R. A Framework for the study of access to medical care. tional decisions. If locational decision making processes Health Serv Res. 1974;9:208–20. https://doi.org/10.3205/psm000089. are functioning effectively high levels of correlation 16. National Health Service England. Technical Guide to Allocation Formulae would be expected between the factors which are driving and Pace of Change. 2016. https://www.england.nhs.uk/wp-content/ uploads/2016/04/1-allctins-16-17-tech-guid-formulae.pdf. Accessed 25 May healthcare need and the location of the facilities. 17. Health & Social Care Board (HSCB). Proposed Changes To the Northern Abbreviations 2 Ireland Weighted Capitation Formula. 2015. GP: General practitioner; Km: Kilometre; km : Square kilometres; 18. Luo W, Qi Y. An enhanced two-step floating catchment area ( E2SFCA ) Max: Maximum; MDM: Multiple deprivation measure; Min: Minimum; method for measuring spatial accessibility to primary care physicians. Health SD: Standard deviation; SOA: Super output area Place. 2009;15:1100–7. 19. Hare TS, Barcus HR. Geographical accessibility and Kentucky’s heart-related Availability of data and materials hospital services. Appl Geogr. 2007;27:181–205. The data used in this study is publicly available at the references included in 20. Hawthorne TL. Using GIS and perceived distance to understand the the text. unequal geographies of healthcare in lower-income urban neighbourhoods. Geogr J. 2012;178:18–30. Authors’ contributions 21. Jordan H, Roderick P, Martin D, Barnett S. Distance, rurality and the need for BG carried out all tasks. The author read and approved the final manuscript. care: access to health services in South West England. Int J Health Geogr. 2004;3:21. https://doi.org/10.1186/1476-072X-3-21. Ethics approval and consent to participate 22. Andersen R. Revisiting the behavioral model and access to medical Care : Not Applicable. does it matter? J Health Soc Behav. 1995;36:1–10. 23. Rosero-Bixby L. Spatial access to health care in Costa Rica and its equity: a GIS-based study. Soc Sci Med. 2004;58:1271–84. Competing interests 24. Furler JS, Harris E, Chondros P, Davies PGP, Harris MF, Young DYL. The The author declares that he has no competing interests. inverse care law revisited: impact of disadvantaged location on accessing longer GP consultation times. Med J Aust. 2002;177:80–3. Publisher’sNote 25. Adams J, White M. Socio-economic deprivation is associated with increased Springer Nature remains neutral with regard to jurisdictional claims in proximity to general practices in England: an ecological analysis. J Public published maps and institutional affiliations. Health (Bangkok). 2005;27:80–1. Graham BMC Health Services Research (2018) 18:398 Page 10 of 10 26. Asthana S, Gibson A. Deprivation, demography, and the distribution of 49. Yen IH, Michael YL, Perdue L. Neighborhood environment in studies of general practice: challenging the conventional wisdom of inverse care. Br J health of older adults. A Systematic Review. Am J Prev Med. 2009;37:455–63. Gen Pract. 2008;58:720–8. https://doi.org/10.1016/j.amepre.2009.06.022. 27. Northern Ireland Health and Social Care Business Services Organisation. 50. Riva M, Gauvin L, Barnett TA. Toward the next generation of research into Minor Ailments Scheme. http://www.hscbusiness.hscni.net/2055.htm. small area effects on health: a synthesis of multilevel investigations Accessed 4 Apr 2018. published since July 1998. J Epidemiol Community Health. 2007;61:853–61. https://doi.org/10.1136/jech.2006.050740. 28. Health and Social Care Board. Investment in GP Practices. http://www. 51. Paez A, Mercado RG, Farber S, Morency C, Roorda M. Accessibility to health hscboard.hscni.net/our-work/integrated-care/gps/investment-in-gp- care facilities in Montreal Island: an application of relative accessibility practices/. Accessed 4 Apr 2018. indicators from the perspective of senior and non-senior residents. Int J 29. Northern Ireland Council for Voluntary Action. GP Practices. 2017. http:// Health Geogr. 2010;9:52. data.nicva.org/dataset/gp-practices. Accessed 20 May 2018. 52. Rice DP, Feldman JJ. Living longer in the United States: demographic 30. Health and Social Care Board. General Medical Services - Frequently Asked changes and health needs of the elderly. Milbank Mem Fund Q Health Soc. Questions. 2018. http://www.hscboard.hscni.net/our-work/integrated-care/ 1983;61:362–96. https://doi.org/10.2307/3349863. gps/faqs/. Accessed 4 Apr 2018. 53. Teljeur C, O’Dowd T, Thomas S, Kelly A. The distribution of GPs in Ireland in 31. Potter M. Equality and human rights legislation in Northern Ireland : a relation to deprivation. Health Place. 2010;16:1077–83. https://doi.org/10. review. 2011. http://www.niassembly.gov.uk/globalassets/Documents/RaISe/ 1016/j.healthplace.2010.06.011. Publications/2011/OFMdFM/7511.pdf. Accessed 25 May 2018. 54. Alford-Teaster J, Lange JM, Hubbard RA, Lee CI, Haas JS, Shi X, et al. Is the 32. Department of Health. Equality Scheme for the Department of Health. 2017. closest facility the one actually used? An assessment of travel time https://www.health-ni.gov.uk/sites/default/files/consultations/health/ estimation based on mammography facilities. Int J Health Geogr. 2016;15:8. equality-scheme-consultation-document.docx. Accessed 20 May 2018. https://doi.org/10.1186/s12942-016-0039-7. 33. Abolhallaje M, Mousavi SM, Anjomshoa M, Beigi Nasiri A, Seyedin H, 55. Health and Social Care Board. Developments in GP Services. http://www. Sadeghifar J, et al. Assessing health inequalities in Iran: a focus on the hscboard.hscni.net/our-work/integrated-care/gps/gp-service-developments/. distribution of health care facilities. Glob J Health Sci. 2014;6:33750. https:// Accessed 4 Apr 2018. doi.org/10.5539/gjhs.v6n4p285. 34. Mobley L, Root E, Anselin L, Lozano-Gracia N, Koschinsky J. Spatial analysis of elderly access to primary care services. Int J Health Geogr. 2006;5:19. https://doi.org/10.1186/1476-072X-5-19. 35. Luo W. Using a GIS-based floating catchment method to assess areas with shortage of physicians. Health Place. 2004;10:1–11. 36. Ordinance Survey Northern Ireland. Ordinance survey Northern Ireland open data. Road Networks. http://osni-spatial-ni.opendata.arcgis.com/. Accessed 4 Apr 2018. 37. Healt and Social Care Business Services Organisation. Northern Ireland GP practice lists for professional use. 2017. http://www.hscbusiness.hscni.net/ services/1816.htm. Accessed 17 Aug 2017. 38. Comber AJ, Brunsdon C, Radburn R. A spatial analysis of variations in health access : linking geography, socio-economic status and access perceptions. Int J Health Geogr. 2011;10:44. https://doi.org/10.1186/1476-072X-10-44. 39. Shah TI, Bell S, Wilson K. Spatial accessibility to health care services: identifying under-serviced Neighbourhoods in Canadian urban areas. PLoS One. 2016;11:1–22. https://doi.org/10.1371/journal.pone.0168208. 40. Northern Ireland Statistics and Research Agency. Northern Ireland Multiple Deprivation Measure 2017 (NIMDM2017). Northern Ireland Statistics Research Agency. 2017. https://www.nisra.gov.uk/statistics/deprivation/ northern-ireland-multiple-deprivation-measure-2017-nimdm2017. Accessed 4 Apr 2018. 41. NISRA. 2016 Mid year population estimates for Northern Ireland. Northern Ireland Statistics and Research Agency. 2016. https://www.nisra.gov.uk/ publications/2016-mid-year-population-estimates-northern-ireland. Accessed 25 May 2018. 42. Northern Ireland Statistics and Research Agency. Northern Ireland Super Output Areas. 2011. https://www.nisra.gov.uk/support/geography/northern- ireland-super-output-areas#toc-0. Accessed 4 Apr 2018. 43. Northern Ireland Statistics and Research Agency. Census 2011. 2011. http:// www.ninis2.nisra.gov.uk/public/Theme.aspx?themeNumber= 136&themeName=Census. Accessed 4 Apr 2018. 44. Northern Ireland Statistics and Research Agency. Geography Fact Sheet. 2013. https://www.ninis2.nisra.gov.uk/public/documents/ NISRA%20Geography%20Fact%20Sheet.pdf. Accessed 5 Nov 2017. 45. The R project for statistical computing. 2017. https://www.r-project.org/. Accessed 5 Nov 2017. 46. Wood D, Clark D, Gatrell AC. Equity of access to adult hospice inpatient care within north-West England. Palliat Med. 2004;18:543–9. https://doi.org/10. 1191/0269216304pm892oa. 47. Jokela M. Does neighbourhood deprivation cause poor health? Within- individual analysis of movers in a prospective cohort study. J Epidemiol Community Health. 2015;69:899–904. https://doi.org/10.1136/jech-2014- 48. Tøge AG, Bell R. Material deprivation and health: a longitudinal study. BMC Public Health. 2016;16:747. https://doi.org/10.1186/s12889-016-3327-z.

Journal

BMC Health Services ResearchSpringer Journals

Published: Jun 1, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off