Primary Health Care Accessibility for Rural Otago: “A Spatial Analysis”

• Abstract
• Introduction
• Background
• Methods
• Data Issues and Otago Study Area
• WHO “Rules” and NZ “Guidelines” for Access to PHC
• WHO “Rules”
• New Zealand Guidelines to Improve Access to PHC
• Modelling Accessibility
• Travel Time
• Methodology for Determining Accessibility
• Results
• 30- and 60-Minutes Threshold Travel Time
• The Number of GP per Population in Rural Otago Area
• Accessibility Index at Meshblock Level
• Discussion and Conclusions
• Acknowledgements
• References
• Footnotes

Abstract
This research aims to develop a new approach for measuring spatial accessibility to primary health care (PHC) services. New Zealand and World Health Organisation (WHO) rules were used to determine optimum levels of minimum travel time to the closest PHC facilities via a road network. This analysis was applied to 1200 census areas (meshblocks) defined according in the 2001 Census for rural Otago, New Zealand.

A two-step floating catchment area (FCA) method was used to calculate spatial accessibility based on travel time. The mean centre of population distribution within each meshblock polygon was used as patient locations The initial results of this study have shown that some parts of north and central Otago do not meet WHO rules and New Zealand health guidelines (during business hours). The paper illustrates the use of the “accessibility index” as a tool to model the level of accessibility of people to PHC. The spatial accessibility index ranged from 1 to 10, with 1 representing very low accessibility and 10 representing high accessibility.

Introduction
Primary health care (PHC) is an important strategy for improving and maintaining population wellbeing.^[1] World Health Organisation (WHO) and UNICEF^[2] outlined the importance of PHC in the 1978 Declaration of Alma-Ata and PHC is the first level of contact between individuals, the family and community and the national health system, bringing health care as close as possible to where people live and work, and constitutes the first element of a continuing health care process.

Four important principles of a successful PHC strategy are:

1. Universal access to and coverage of health services based on health needs.
2. Assurance to health equity as part of development oriented to social justice.
3. Community involvement in defining and implementing health programmes.
4. Intersectoral approach to health.

Universal accessibility^[2] is the availability of facilities and services for everyone regardless of where they work or live. Therefore, WHO encourages all countries to provide universal accessibility as the main guideline for planning and delivering health care. Access is a multidimensional concept that describes people’s ability to use the necessary health care, immediately wherever they are.^[3]

Access to PHC is known as one of the indexes to achieving “health for all” and it was expressed as one of the important principles of PHC (see Declaration of Alma-Ata). There are two major dimensions for access:

1. Potential; and
2. Actual.

Potential accessibility implies the service is actually available within the vicinity of a potential user. Actual access follows when all barriers or impediments to access for PHC are removed.^[4] However, the number and the type of barriers to accessibility of PHC differs from country to country and time to time. Penchansky and Thomas^[5] categorised barriers into five types: availability, accessibility, affordability, accommodation and acceptability. The first two types are considered spatial in nature. The last three types are known as non-spatial and temporal.^[5-7] The five types of barriers in the PHC accessibility context are:

1. Availability in the context of PHC refers to the number of health care services from which needy people can choose.
2. Accessibility is travel impedance (distance or time) between residential or demand area and health care facilities or destinations.
3. Affordability refers to the price of services in regard to people’s ability to pay. Income level and insurance coverage are crucial aspects of affordability.
4. Accommodation identifies the degree to which services are organised to meet patient’s need, including hours of operations, waiting time and application procedure.
5. Acceptability describes people’s views about health care services and how service providers interact with patients.^[5-7]

This research has attempted to determine and map a spatial accessibility index. In this study, availability and accessibility are considered as spatial variables and the aim is to model (map) accessibility based on WHO “rules” and New Zealand’s guidelines.

Background
The use of geographic information systems (GIS) in health care has increased considerably in recent years. GIS has provided new solutions for measuring spatial access to health care, such as, modelling the health need to indicate and map the accessibility level of people with high health need to PHC in the study area. Potential and actual access must be considered together in assessing and locating health care services in a community.^[8] There are a number of questions that GIS could answer, for example, “Where should health providers be located to best serve a population with high health needs?”.

Accessibility applied to health care has been an area of ongoing research and each of these applications has focused on one dimension of accessibility. This research considers spatial and temporal access into one framework. Hall & Bowerman^[9] have to date produced the most elegant approach with their AccessPlan software. AccessMod, a tool developed by WHO, was designed to analyse physical accessibility to health care and population coverage.^[10,11] The important feature of AccessMod is to determine catchment areas around health care facilities and produces data to support evidence based policy assessment corresponding to the coverage offered by active health care services. This research and AccessPlan software both uses vector based analysis for measuring accessibility but with different method and constraints.

Accessibility to health care in New Zealand is an active area of research. Ultimately the government of the day wants to achieve the highest level of care within budget constraints. The research has not focused on health economics but on people, infrastructures and providers. However, it is thought that the tools and modelling techniques that come out of this research on accessibility could help some health economics decision.

Figure 1: New Zealand population more than 30 minutes from a GP^[12]

Research into accessibility to primary and secondary health care in the New Zealand context^[12-15] has focused on the relationship between secondary health care utilisation and distance to care. The method of Slack et al^[15] for studying accessibility of secondary health care was spatially simplistic and they acknowledge advances in GIS tools, together with accurate data on general practitioner (GP) locations would provide more sophisticated results. Brabyn and Barnett used GIS technology for modelling population need and spatial access to GP in rural New Zealand, however, they did not use accurate data on GP locations and active GP figures.^[13]

This research agrees that rural areas are those where it is most likely there might not be equitable access to PHC. However, unlike Brabyn and Barnett,^[12] this study used accurate data on GPs, including where GP practices are located. The study also did not aggregate provinces to give a general scale; rather a micro-scale analysis was used for rural Otago. Figure 1^[12] shows 500–2000 people are more than 30 minutes from a GP, but the present study indicates this rate is 4100 and also there are still some people that have more than 60 minutes’ travel time to a GP or PHC centre in Otago region (figure 4). Both studies used 2001 census dataset for population information.

Methods

Data Issues and Otago Study Area
Spatial data such as PHC locations (health care centres), road networks, coverage of census blocks and core record system addresses (this dataset contains the cadastral information for NZ) were obtained from land information New Zealand (LINZ) that provides New Zealand’s authoritative land and seabed information. Core record system addresses were used to calculate the mean centre of population residence as the patient location. Socio-demographic data (such as age, gender, deprivation index, ethnicity, mean income) and meshblock boundaries were taken from the Statistics New Zealand 2001 Census dataset. More information about PHC team location and the number of GPs came from the New Zealand Ministry of Health (MoH) data.

Figure 2: Urban / rural profile categories: Otago region^[16]

The study area is the rural areas of the Otago province at meshblock level (see Figure 2). The meshblock is the smallest geographic unit for which statistical data is collected by Statistics New Zealand.^[16] A meshblock is a defined geographic area, varying in size from part of a city block to large areas of rural land. The size of a meshblock depends primarily on the number of people and type of area covered. There were approximately 1200 meshblocks in rural Otago at the 2001 census, each with an average population of 76 people. Meshblock data information and road network files were used to calculate the minimum travel time (based on road type, topography, sinuosity, surface) from population location inside each meshblock to PHC locations.

The mean centre is a point constructed from the average x and y values for the input feature centroids. Calculations are based on either Euclidean distance (straight line distance) or Manhattan distance (the distance between two points measured along axes at right angles) and require projected data to accurately measure distances. To calculate the mean centre population, the geo-coded population at meshblock level was required and this was extracted from Core Record System (CRS) address dataset. Since population is not uniformly distributed within the meshblocks, particularly in rural areas with large meshblocks, the use of simple geographic centroid is not an appropriate representative of population location and potentially produces error in distance calculations. The use of mean centre overcomes the error.^[15]

WHO “Rules” and NZ “Guidelines” for Access to PHC
One of the most important characteristics of a tool is that it must apply effectively in different regions according to different policies. In this study the key elements of the tool are the WHO and New Zealand guidelines for people’s accessibility to PHC. This causes the tool is simply adapted to different rules in any where. In addition, it can compare the results of local guidelines with global standards.

WHO “Rules”
WHO’s main objective is to ensure that all people to have a high level of health care. One of the important functions of WHO is to determine guidelines and rules for supporting individual countries to attain this goal. WHO’s guidelines and rules can be used as indicators to assist countries in evaluating their activities and situations in the context of health both globally and locally.

In WHO’s guidelines and rules touching on accessibility to PHC are:

1. Universal access: This is a concept that, when applied to environments, ensures that health facilities can be accessed by all people regardless of where they live or work.
2. Focus on population with “high health needs”: High-need population groups usually experience more access difficulties than the rest of the population. Therefore, the priority is to ensure that they have appropriate access to PHC.
3. 1000 people per general practitioner: The availability of human resources for health care is an important indicator in the quality of the health care. However, sometimes a high number of GPs is not necessarily a good indicator. Some WHO documents highlight as an optimal ratio one GP per 1000 head of population.^[2]

New Zealand Guidelines to Improve Access to PHC
The key priority for implementation of the PHC policy is to reduce barriers to high need groups’ access to health care. The New Zealand Ministry of Health’s guidelines are:^[17]

1. To provide additional services to improve access to PHC among high need groups. eg:

• Maori and Pacific people; and
• Those living in NZ Deprivation index 8-10 deciles areas.

2. PHC services must be available for 95 percent of population in New Zealand during:

• Normal business hours = within 30 minutes car travel time
• After hours = within 60 minutes car travel time
  

Modelling Accessibility
This study uses WHO rules and New Zealand policy to model people’s accessibility and their need for accessibility to PHC.

Travel Time
In regard to spatial accessibility to PHC services, studies have used Euclidean distance, travel time and Thiessen polygons^[a] for assessing access to health care.^[18-20]

This research used drive time between any pair of origin (demand) and destination (PHC) locations that was calculated using network analysis tools. For measuring spatial accessibility, the best route (shortest travel time) from population residence area to PHC services was also computed using network analysis (figure 3). There are many factors such as road type, topology, time of day, urban or rural road, land use, different seasons, slope and sinuosity that affect road travel time. Collecting a reliable dataset for all these variables is the beyond the scope of this research. As a result, this study estimated drive time for the road network based on significant variables like the type of roads, surface, number of lanes, sinuosity index, rural and urban land use and the New Zealand’s road speed limits.

Figure 3: Best route (shortest travel time) from population location to PHC services

To calculate the sinuosity index, two variables are required: the observed length; and the expected direct or straight length.^[21,22] The observed length is easily calculated for the actual road network. However, to compute the straight length, we had to generalise the original road layer by removing vertices from the road segments until only one vertex remained per 500-metre tolerance. After simplifying the road layer, the lengths were calculated and called direct length. The new lengths joined to the original road data set (see Table 2).

The sinuosity index is defined as observed length / expected direct length. ^[21] The sinuosity index ranged from 1 to 6.54: roads with sinuosity of 1 are straight and sinuosity increases with the “tortuousness” of a road; an index of 6.54 typically indicates a very bendy road (see Table 1). Therefore, roads with a high sinuosity index require low speed values for travel time estimation.

Travel time was estimated for each road segment (calculated as an average speed) as:

• Sealed urban roads 35 km/h
• Un-metalled, rural, bendy roads 35 km/h
• Metalled, 1 lane, rural, bendy road 40 km/h
• Un-metalled, rural, straight roads 45 km/h
• Metalled, 1 lane, rural, straight roads 60 km/h
• Metalled, 2 lanes, rural, bendy roads 60 km/h
• Sealed, 2 lanes, rural straight roads 80 km/h
• High-way 80 km/h.

This categorisation of road speeds would alter, based on speed limit rules. The values in this study do not take into account different times of day, traffic congestion, different seasons and different modes of transportation. If these are important, then considerable effect would be required to collect this volatile parameter to impact on the friction surface calculations.

Methodology for Determining Accessibility
There are several methods to measure spatial accessibility to primary hearth care and these can be grouped into five main categories:^[1]

1. Provider to population ratio: A crude analysis of accessibility that refers to supply ratios, calculated inside catchment areas.
2. Travel time or distance impedance to nearest provider: Used from a patient’s residence (or population centre) to the PHC supply point; calculated in units of Euclidean distance or actual travel time.
3. The gravity model: A modified version of Newton’s Law of Gravity, this is a combined indicator of accessibility and availability. Joseph and Bantock improved this model by adding a population demand adjustment factor.^[23]
4. Two-step floating catchment area (FCA) method: Sometimes, health provider data are not suitable for the spatial resolution needed for the gravity method and this leads to loss of resolution and more useful detail information. A method developed by Radke and Mu^[24] which overcomes this problem is to repeat the process of floating catchment twice, once on population location and once on health services. This method was used for first time by Peng^[25] for job accessibility.
5. Kernel density method: A method that uses the “Gaussian kernel” approach to calculate the density value of each cell.^[26]

A study made by Luo and Wang for measuring spatial accessibility to health care in the Chicago region has proven that the gravity model and two-step FCA method have a similar theoretical approach.^[27] However, the gravity model needs some weights to calculate the travel friction coefficient for road networks. These weights are not usually available and need to be computed through complex calculations. So this research used the two-step FCA approach, which does not need the travel friction coefficient computed and can be comfortably modified to meet the New Zealand MoH policy. Furthermore, it easily compares the provider and population to each other as nominator and denominator, respectively. Hence, it can simply calculate how many residents are cared for by PHC teams within a defined area. In addition, the FCA method is very suitable for finding areas with low accessibility in the community because it repeats the process of floating catchment twice, once on population location and once on health care services and, unlike the gravity-based method, does not ignore the local pockets of low or very low accessibility of PHC.^[27]

For the purposes of this research, PHC is considered to be administered by a team with multi-disciplinary staff, such as GP, practice nurse, midwife, dentist and pharmacologist. Therefore, this study uses PHC team location instead of PHC location. The implication of this choice is that the research determined the accessibility of people to the actual PHC teams providing the first level of health care in the community, rather than the PHC location, which implies a building or centre where the PHC teams provide their services.

A PHC team is usually led by a GP. The GP co-ordinates and directs the team towards achieving the final goal of “health for all” in their community.^[2] To implement this method of measuring accessibility, the origin or patient locations, destination points and travel time must be known. For this purpose, the Core Recorded System address (CRS) was used to extract the mean centre of population as patient location for each meshblock. The New Zealand MoH guideline was used to define the radius of the catchment area. The 30-minutes, 60 and over 60-minutes threshold travel times were calculated from network analysis. The 30-minutes travel time was considered, based on New Zealand MoH guideline, as a radius of the catchment area for PHC teams and residential locations for day time (business hour) criteria.

Equations 1 and 2 represent the FCA method. Equation 1 computes, first, for each PHC team or surgery location j, searches all population locations (k) that are within a 30-minutes threshold travel time from location (j) and computes the PHC team to population ratio within the catchment area:

Equation 1: Formula for calculation of PHC teams per population within the catchment area

The next step (Equation 2) takes account of each population location (i), searches all PHC locations (j) that are within the 30-minute threshold travel time from location (i) and sums up the PHC teams to population ratios at these locations.

Equation 2: Formula for calculation of accessibility index

The result of equation 2 indicates the accessibility at resident location based on the FCA method between (i) and (j).

Based on the value of accessibility index, the access of people to PHC classified in a study area fell into four domains:

1. High accessibility area
2. Moderate accessibility area
3. Low accessibility area
4. Very low accessibility area.

A high value in the accessibility index represents better access to PHC. In practice, the first phase was to assign a primary ratio to each PHC team service (catchment area centred at a PHC team location). Then, the initial values were calculated in the overlapped service area to measure the PHC accessibility for the population location, where the residents have access to multiple PHC locations. Indeed, the method supports interaction between people and PHC services within a meshblock based upon travel times and calculates an accessibility index that varies from one meshblock to another. Equation 2 indicates the regional accessibility and is, fundamentally, the ratio of PHC team to population that, filtered by travel time, can be explained as the same as Equation 1. To implement the method the steps are:

1. Calculate the travel time between PHC team locations and population locations using spatial network analysis. Then, split the travel time into three categories: 30 minutes, 30–60 minutes and over 60 minutes threshold travel time.
2. Match the tables of population and PHC team locations to the table of 30-minutes travel time (“Time 30”).
3. Generate a new table based on Time 30, by summing population by PHC team locations and calculating the PHC team to population ratio for each PHC location (updated Time 30).
4. Calculate the PHC team to population ratios by population location (meshblocks area), constructing a new table (“MeshAcc”). This summarises the availability of PHC teams accessible from a residential location and indicates the accessibility index in equation 2.
5. Finally the MeshAcc table is joined to the population table by meshblock for mapping and further analysis of accessibility to primary health care services (see figure 5).

As mentioned earlier, this process shows the accessibility index for PHC during business hours, based on New Zealand health policy (30-minutes threshold travel time). This approach was also applied to after-hours criteria.

Results

30- and 60-Minutes Threshold Travel Time
Figure 4 shows the results for the 30-minutes and 60-minutes travel time categories of access. The green polygons represent areas where PHC services are accessible within 0- to 30-minutes travel time. The red polygons indicate regions that have access to a PHC facility within 30- to 60-minutes travel time. Blue areas are highly remote and the time needed for people in these areas to access a PHC centre is over 60-minutes drive time. These polygons show initial results measuring the accessibility the people of rural Otago have to PHC. It is necessary to consider some socio-demographic analyses and population information to better explore which are the accessible and inaccessible areas in the Otago region.

Figure 4: 30- and 60-minutes drive time limit in study area

The Number of GP per Population in Rural Otago Area
The rate of occurrence of a GP per 1000 people in this study area is 0.5. The rate for whole of New Zealand is 2.4 GP per 1000 population. ^[28] This research shows that the rate of GP to population in the studied area falls below the WHO’s guideline (one GP per 1000 people).

Accessibility Index at Meshblock Level
Figure 5 displays rural Otago people’s accessibility to PHC services at the meshblock level. The accessibility index ranged from 1 to 10. Those areas with an accessibility index of 1 to 2 are known as the very high accessibility regions and areas with an accessibility index of 9 to 10 are very low accessibility regions. The pattern of PHC accessibility is shown in the map based on 30- and 60-minutes travel time in the Otago area. This figure also shows 30- and 60-minutes travel time limits in rural Otago.

Discussion and Conclusion
The emphasis in this paper is on developing a tool used to measure the accessibility of PHC against the optimum, based on WHO’s rules and New Zealand’s health policy. The results highlight that the difficulties of access to PHC continue in rural Otago. For some parts of the region, particularly in the more remote areas, people still have long travel times to access health care. Network analysis produces new information at the meshblock level. It calculates travel distance from population location in each meshblock to the closest PHC facility. This analysis can be adapted simply to New Zealand guidelines for accessibility of health care and the meshblocks that do not meet New Zealand guideline will be defined.

Measuring spatial accessibility to PHC and identifying geographical variations that affect people’s access to it are important steps in accessibility studies. As stated in the introduction, accessibility has many dimensions. When exploring the question of whether PHC is distributed equitably in a rural community, the geographical accessibility might be misleading and we need to examine the influence of some non-spatial factors such as high health need people as a surrogate analysis for spatial accessibility to PHC.

This research used New Zealand road network and focused on important features of roads that affect estimating of the optimum travel time from population location to PHC services. The study does not compare the cost of improving quality of roads in New Zealand with other possible health expenditures and it is beyond the scope of this research. The paper illustrates the accessibility index as a tool to model the level of accessibility to PHC and also shows the poor accessibility “hot spots” in rural Otago.

To conclude, WHO provides guidelines and policies at global level directed to protecting and promoting the health of all the people of the world and supports, where possible and appropriate, all governments and countries to meet these policies. Criteria for achieving these goals will differ, however, at the local level, according to the socio-demographic and economic situations of each country. Thus, development of an accessibility tool to better explore any inequity is seen as a means towards better access to PHC for all.

Acknowledgements
The authors gratefully acknowledge the useful comments of reviewers. We would like to thank Dr Roy Morris, Primary Care Advisor for the Otago DHB, for his helpful advice and supplying data on Geo-coded GP locations and the MoH and Statistics NZ for preparing Geo-coded Health centres in Otago region and census 2001 data, respectively. We would like to thank Dr Pat Farry, Medical Director of the Te Waipounamu Rural Health Unit, for his advice and intimate knowledge of the data. We also thank Floss Caughey, senior analyst at the NZ MoH, and Dr Steeve Ebener at WHO in Geneva for their helpful assistance to provide us with the New Zealand guidelines and WHO rules respectively.

References

1. Guagliardo MK.(2004). Spatial accessibility of primary care: concepts, methods and challenges [Online]. Available from: http://www.pubmedcentral.gov/picrender.fcgi?artid=394340&blobtype=pdf. Acessed 3 March, 2005.
2. WHO, UNICEF.(1978). Declaration of Alma-Ata. International Conference on Primary Health Care, Alma-Ata [On line]. WHO. Available from: http://www.who.int/chronic_conditions/primary_health_care/en/almaata_declaration.pdf.  Acessed 20 December 2005.
3. Aday L, Anderson R. Equity of access to medical care: a conceptual and empirical overview. Med Care 1981;19(12):4-27.
4. United Nations. Access to social services by the poor and disadvantaged in Asia and the Pacific: Major trends and issues. New York: United Nation; 2002.
5. Penchansky R, Thomas JW. The concept of access: Definition and relationship to consumer satisfaction. Med Care 1981;19(2):127-140.
6. McLafferty S, Grady S. Prenatal Care Need and Access: A GIS Analysis. J Med Syst 2004;28(3):321-333.
7. Cromley EK, McLafferty SL. GIS and Public Health. New York: The Guildford Press; 2002.
8. McLafferty SL. GIS and Health Care. Annu Rev Public Health 2003;24(1):7-25.
9. Hall GB, Bowerman RL. Modelling access to family planning services using GIS: A case study of the Central Valley of Costa Rica. ITC Journal 1996;1:38-48.
10. Ebener S, Ray N, Black M, El Morjani Z.(2004). Physical Accessibility to health care: from isotropy to anisotropy [Online]. GIS@development. Available from: http://www.gisdevelopment.net/magazine/index.htm. Acessed 15 June 2005.
11. Ebener S, El Morjani Z, Black M, Alexakis E, Martin DA, Vanasse A, Hemiari A. Measurement of Accessibility to Health Care, Part II: GIS development. In: Poster presented at the 6th GEOIDE Annual Scientific Conference; 2004 May 30 - June 1; Ottawa, Canada; 2004.
12. Brabyn L, Barnett R. Population need and geographical access to general practitioners in rural New Zealand. N Z Med J 2004;117(1199):1-13.
13. Atmore C. Letter: Population need and geographical access to general practitioners in rural New Zealand. N Z Med J 2004;117(1201):1-3.
14. Slack A, Cumming J, Mare D, Timmins J. Variations in Secondary Care Utilisation and Geographic Access: Initial Analysis of 1996 Data. In: the 14th Annual Colloquium of the Spatial Information Research Centre; 2002; Dunedin, New Zealand: University of Otago; 2002. p. 105-114.
15. Bagheri N, Benwell GL, Holt A. Measuring spatial accessibility to primary health care. In: Whigham PA, editor. The 17th annual Colloquium of the Spatial Information Research Centre; 2005 November, 24-25; Dunedin, New Zealand: University of Otago; 2005. pp 103-108.
16. Statistics New Zealand.(2005). Defining Urban and Rural New Zealand [Online]. New Zealand Statistics. Available from: http://www.stats.govt.nz/urban-rural-profiles/defining-urban-rural-nz/default.htm. Acessed 5 August 2005.
17. New Zealand Ministry of Health.(2001). Primary Health Organisation Agreement [Online]. Ministry of Health. Available from: http://www.moh.govt.nz/pho. Acessed 10 April 2005.
18. Kohli S, Sahlén K, Sivertun A, Löfman O, Trell E, Wigertz O. Distance from the primary health centre: a GIS method to study geographical access to health care. J Med Syst 1995;19(6):425-436.
19. Lovett A, Haynes R, Sunnenberg G, Gale S. Car travel time and accessibility by bus to general practitioner services: a study using patient registers and GIS. Soc Sci Med 2002; 55(1): 97-111.
20. Wang F, Luo W. Assessing spatial and non spatial factors for healthcare access: towards an integrated approach to defining health professional shortage areas. Health and Place 2004;11(2):62-63.
21. Haggett P, Chorley R. Network Analysis in Geography. London: Edward Arnold; 1969.
22. Brabyn L, Skelly C.(2002). Modeling population access to New Zealand public hospitals [Online]. Available from: http://www.ij-healthgeographics.com/content/1/1/3. Acessed 24 December 2004.
23. Joseph AE, Bantock PR. Measuring potential physical accessibility to general practitioners in rural areas: A method and case study. Soc Sci Med 1982;16(1):85-90.
24. Radke J, Mu L. Spatial decomposition, modelling and mapping service regions to predict access to social programs. Geographic Information Sciences 2000;6(2):105-112.
25. Peng Z. The jobs housing balance and urban commuting. Urban Studies 1997;34:1215-1235.
26. Guagliardo MF, Ronzio CR, Cheung I, Chacko E, Joseph JG. Physician accessibility: an urban case study of paediatric providers. Health and Place 2004;10(3):273-283.
27. Luo W, Wang F. Measures of spatial accessibility to health care in a GIS environment: synthesis and a case study in Chicago region. Environment and Planning B 2003;30:865-884.
28. WHO.(2001). World Health Statistics [On line]. World Health Organization. Available from: http://www3.who.int/whosis/core/core_select_process.cfm. Acessed 9 June 2006.

Footnotes

(a). Polygons whose boundaries define the area that is closest to each point relative to all other points.