eMJA: Bamford et al, Accessibility to general practitioners in rural South Australia

eMJA The Medical Journal of Australia

Rural Healthcare

Accessibility to general practitioners in rural South Australia

A case study using geographic information system technology

MJA 1999; 171: 614-616

Errol J Bamford, Lyle Dunne, Danielle S Taylor, Brian G Symon, Graeme J Hugo and David Wilkinson

Abstract - Introduction - Methods - Results - Discussion - References - Authors' details -
Make a comment - Register to be notified of new articles by e-mail - Current contents list - More articles on Administration and health services

Abstract

Objective: To demonstrate the potential of GIS (geographic information system) technology and ARIA (Accessibility/Remoteness Index for Australia) as tools for medical workforce and health service planning in Australia.
Design: ARIA is an index of remoteness derived by measuring road distance between populated localities and service centres. A continuous variable of remoteness from 0 to 12 is generated for any location in Australia. We created a GIS, with data on location of general practitioner services in non-metropolitan South Australia derived from the database of RUMPS (Rural Undergraduate Medical Placement System), and estimated, for the 1170 populated localities in South Australia, the accessibility/inaccessibility of the 109 identified GP services.
Main outcome measures: Distance from populated locality to GP services.
Results: Distance from populated locality to GP service ranged from 0 to 677 km (mean, 58 km). In all, 513 localities (43%) had a GP service within 20 km (for the majority this meant located within the town). However, for 173 populated localities (15%), the nearest GP service was more than 80 km away. There was a strong correlation between distance to GP service and ARIA value for each locality (0.69; P < 0.05).
Conclusions: GP services are relatively inaccessible to many rural South Australian communities. There is potential for GIS and for ARIA to contribute to rational medical workforce and health service planning. Adding measures of health need and more detailed data on types and extent of GP services provided will allow more sophisticated planning.

Introduction

Australia's inequitable distribution of health services and health professionals is well documented, with areas outside the capital cities underserved.¹ As part of the effort to redress these imbalances, health workforce data have been gathered,² but a problem with analysis of workforce data is the classification of areas as "rural" and "remote". The categorical RRMA (rural, remote and metropolitan areas) classification, developed in 1994, although widely used, has limitations.² These include the large and varying size of the statistical areas forming the unit of analysis in RRMA, and the use of straight-line distance measurements which do not reflect the reality of travel by road.³

In May 1998, the Commonwealth Department of Health and Aged Care commissioned a new measure of remoteness using modern geographic information system (GIS) technology.⁴ Specific aims were to compile a GIS database of road, locality and service information (health, education, retail facilities, banking) for all Australia; to measure remoteness as a continuous variable; to produce an index of remoteness; and to calculate and map remoteness for Australia. The new index is called ARIA (Accessibility/Remoteness Index for Australia).⁵

Here, we describe the use of GIS technology and ARIA methodology (Box 1) in a case study that displays and quantifies the distribution of GPs in non-metropolitan South Australia.

Methods

Source of data on GP services
The Rural Undergraduate Medical Placement System (RUMPS) is a database designed to facilitate placement of medical students in rural communities. It was developed by a consortium comprising the Department of General Practice, the University of Adelaide; the Centre for Rural Health, Monash University; the Rural Health Training Unit, South Australia; the Department of Evidence Based Care and General Practice, Flinders University; and the South Australian Centre for Rural and Remote Health. Funding was provided by the Commonwealth Department of Health and Aged Care. The system defines the location of all rural GPs in South Australia and those in large centres close to the South Australian border.

All rural GPs were surveyed in late 1998. Comparison with Census data on GP location⁶ and a database held by the South Australian Rural and Remote Medical Support Agency confirmed the completeness of the RUMPS database.

Analysis
For this case study, we used the simplest version of an index based on the ARIA methodology: the road distance from a populated locality to the nearest GP service. No account could be taken of whether the GP service was full or part time. We generated hypothetical journeys whose origins were the 1170 populated localities in non-metropolitan South Australia and whose destinations were the 109 locations with a GP service in non-metropolitan South Australia. GP services included a small number beyond the SA border (Alice Springs, Broken Hill), as some South Australian residents use these services.

The GIS calculated the minimum distance from each community to the closest service using the road network, and values were interpolated onto a regular 1 km grid for the whole of South Australia. The correlation coefficient between distance from locality to service, with the ARIA value for each locality, was calculated using Microsoft Excel.

Results

The distance from a populated locality to the nearest GP service in South Australia ranged from 0 to 677 km, with a mean distance of 58 km. As illustrated in Figure 1, the distribution of access to GPs is highly skewed. In all, 513 localities (43%) had a GP service within 20 km, and for the majority this meant located within the town. However, for 173 (15%) populated localities the nearest GP service was more than 80 km away.

Assuming travel at 80 km/h, estimates of the relationship between travel time and remoteness from GP services were calculated (Table): 28% of populated localities were 30 minutes' or more travel from a GP service. We further measured a strong correlation of 0.69 (P < 0.05) between ARIA values for the populated locality and the measure of access to GP services.

Figure 2 shows the distances from GP services for the whole of South Australia in the form of a "contour map". This allows the distance between any point in the State and a GP service to be determined. This map can be interrogated by computer to report the actual distance for any point, and can also be used to model the impact on accessibility of placing new GP services in defined places.

Figure 3 shows the populated centres in South Australia, the location of GP services, the main road network and the distances that need to be travelled between the population localities and GP services.

Discussion

This case study demonstrates the potential utility of GIS technology and the new ARIA measure in documenting and analysing accessibility to health services in Australia. The results can potentially be related to population need by estimating the number of people in each of the distance zones shown in Figure 2. The pattern of future need can then be estimated from population projections.

Our data indicate a substantial lack of accessibility to GP services by rural and remote communities in South Australia, and we are unaware of any previous study that has applied a measure of inaccessibility to these services. This inaccessibility to GP services is correlated closely with remoteness from other services. In this case study we did not include medical officers working in Aboriginal communities or the Royal Flying Doctor Service, but this will be possible in the future as data on location and services provided become more readily available. It would also be of value to consider access to telehealth facilities when considering access to services in general.

A limitation of our study is that the RUMPS database does not provide any detail of the availability of GP services: a site staffed by a GP one day each week is recorded in the same way as a full-time clinic. Neither does RUMPS identify the number of practitioners at a particular site. Similar limitations of public access databases of medical workforce have been reported.^7,8 The strength of the RUMPS database is that it is up-to-date (as of late 1998), and is accurate when compared with two other data sources.

The availability of more detailed workforce data would enable further analyses using ARIA and GIS. Details of full- or part-time work and types of GP services rendered would allow more rational and cost-efficient medical workforce and health service planning at the local level, based on population distribution, population need and accessibility.

Our case study shows how ARIA and GIS technology could be used to support the process of recruiting doctors to rural areas by targeting areas of particular need, and by providing a ready means of evaluating the effectiveness of recruitment interventions. ARIA is now available as a flexible, Web-enabled spatial information decision support system <http://pc137.gisca.adelaide. edu.au/aria/home.html>, giving decision makers, practitioners, researchers and the public access to this information. ARIA can be used to develop models to assist in the provision of a more equitable distribution of health services and health professionals in Australia.

References

Cameron I. Retaining a medical workforce in rural Australia. Med J Aust 1998; 169: 293-294.
Australian Medical Workforce Advisory Committee. Medical workforce supply and demand in Australia: a discussion paper. Canberra: Australian Institute of Health and Welfare, 1998. <http://amwac.health.nsw.gov.au/> Accessed 2 November 1999.
Australian Institute for Health and Welfare. Health in rural and remote Australia. Canberra: Australian Institute for Health and Welfare, 1998.
Clarke K, McLafferty S, Tempalski B. On epidemiology and geographic information systems: a review and discussion of future directions. Emerg Infect Dis 1996; 2: 85-92.
Department of Health and Aged Care. Accessibility/ Remoteness Index of Australia (ARIA). Canberra: The Department, March 1999. (Occasional Papers Series No. 6) <http://www.health.gov.au/pubs/hfsocc/ hacocc6.pdf> Accessed 2 November 1999.
Wilkinson D. Inequitable distribution of general practitioners in Australia: analysis by State and Territory using Census data. Aust J Rural Health 1999. In press.
Hays R, Veitch C, Franklin L, Crossland L. Methodological issues in medical workforce analysis: implications for regional Australia. Aust J Rural Health 1998; 6: 32-35.
Gill G, Thomson A, Pilotto L. Identifying the active general practice workforce in one division of general practice: the utility of public domain databases. Med J Aust 1997; 166: 208-210.

(Received 29 Apr, accepted 8 Sep, 1999)

Authors' details

National Key Centre for Social Applications of Geographic Information Systems, University of Adelaide, Adelaide, SA.
Errol J Bamford, BE, Senior GIS Consultant;
Danielle S Taylor, BA, GradDip Appl Remote Sensing, GIS Research Officer;
Graeme J Hugo, BA, PhD, Director; and Professor, Department of Geographical and Environmental Studies.

Information Section, Department of Health and Aged Care, Canberra, ACT.
Lyle Dunne, BSc, Scientist.

South Australian Centre for Rural and Remote Health, University of Adelaide, and University of South Australia, Whyalla and Adelaide, SA.
Brian G Symon, MB ChB, FRACGP, Director of Multidisciplinary Teaching Practices.
David Wilkinson, MB ChB, MD, Head; and Professor of Rural Health.

Reprints will not be available from the authors.
Correspondence: Professor D Wilkinson, SACRRH, c/- University of South Australia - Whyalla Campus, Nicolson Avenue, Whyalla Norrie, SA 5608.
david.wilkinsonATunisa.edu.au

Home | Issues | eMJA shop | Terms of use | Classifieds | More... | Contact | Topics | Search

The Medical Journal of Australia eMJA

Readers may print a single copy for personal use. No further reproduction or distribution of the articles should proceed without the permission of the publisher. For permission, contact the Australasian Medical Publishing Company.
Journalists are welcome to write news stories based on what they read here, but should acknowledge their source as "an article published on the Internet by The Medical Journal of Australia <http://www.mja.com.au>".
<URL: http://www.mja.com.au/> © 1999 Medical Journal of Australia.
We appreciate your comments.

Geographic information System (GIS) technology and the Accessibility/ Remoteness Index for Australia (ARIA)

GIS
A GIS is a digital, computerised map with associated databases so that a variety of data can be linked to places on the map. The GIS can be changed and updated simply by updating the database. For example, residential addresses of patients with malaria can be marked on a map, displaying the spatial distribution of the disease. To this can be added the distribution of water sources, and the statistical relationship between dwellings, water sources and malaria can be explored. Climate data can then be added, together with data on population migration, and the distribution of the malaria vector, adding further complexity. By updating, these analyses can then be repeated annually to demonstrate trends. To date, GIS has been widely applied to the study of infectious diseases, but more examples of its use in health service planning, as in our article, are now emerging.

ARIA
The fundamental assumption underlying ARIA is that the degree of remoteness of a community is determined by the level of its accessibility to a range of services, some of which are available in smaller and others in larger centres. To develop ARIA, we defined a populated locality as any of the 11338 populated centres in Australia. Service centres are the 201 populated centres with a population of 5000 or more at the time of the 1996 Census. Services available in a centre with 5000 people might include a GP, a bank, a post office and basic retail outlets. Accessibility was measured along the existing road network from the 11338 populated localities to four categories of service centres: Category A (>250000 persons), Category B (48000-249999), Category C (18000- 47999) and Category D (5000-17999). The four categories were determined so that towns within each group were alike in service provision, while there was a difference in level of service availability between the groups. Distances entirely within service centres were disregarded in calculating the index. The four values for distances from each populated locality to the nearest Category A-D service centres were each converted to a ratio of the mean distance to each of the A-D centres (by dividing by the mean for that distance category) and summed to give the ARIA value. This produced a continuous variable with values between 0 (high accessibility) and 12 (high remoteness). Values for populated localities were then interpolated to a 1km grid, and averages calculated for larger areas. Thus, each populated locality has an ARIA value, and so do any chosen geographic units of analysis (eg, Whyalla=2.45; Coober Pedy=10.98). (See Figure 3 for the location of these towns.)