7.3 Site availability and suitability

One commonly used methodology in many GIS applications is Multiple Criteria Evaluation (MCE). MCE is used to identify suitable locations for a given activity, such as identifying sites suitable for new housing, plantation forestry, or landfill sites. This methodology is also relevant when deciding where to locate new health facilities.

In MCE, suitable locations are identified by examining a set of criteria, which can be represented as map layers. These criteria fall into one of two categories:

  • Constraints : Constraints are hard ‘yes/no’ criteria that determine whether or not a given site is available (also known as Boolean criteria after the mathematician Boole). For example, in locating health facilities, certain types of land use (such as water bodies and land set aside for military use) would be unsuitable for developing as health facilities.
  • Factors : Other criteria are not so clear-cut, are ‘softer’ and describe the relative suitability of different sites. In MCE, these are known as factors. For example, we may wish to locate a new health facility within easy reach of public transport. Whilst we know that a site close to a bus terminus would be more suitable than a site with only limited public transport on the edge of town, we would not automatically rule out locating the facility on the outskirts of town. Using factors such as accessibility via public transport, we can grade sites on a sliding scale from most to least suitable, but (perhaps with the exception of the remotest sites) not eliminate potential sites altogether.

Health Professional Shortage Areas and Medically Underserved Areas:

In the US, a prominent example of suitability indices is used to identifying Health Professional Shortage Areas [HPSAs] and Medically Underserved Areas [MUAs]. HPSAs, for example, are identified from a composite score based on:

  • Population: primary care physician ratio
  • infant mortality / low birthweight
  • poverty
  • travel time to the nearest health facility.

Today, these scores are used to design incentives for staff moving to areas with a shortage of medical staff (see references for further details).

 

It is relatively straightforward to identify the area that meets all of the constraints within a GIS by overlaying map layers representing each constraint. However, taking into account the factors affecting suitability is more complex. For example, some criteria can act both as factors and constraints. It may be clear that a site with an area of less than a hectare may be unsuitable for a new birthing unit, for example and a site where only 0.5 Ha is available for building is unsuitable. In this sense, the area of a plot available for building is a constraint – a ‘yes / no’ criterion. However, we may seriously consider two sites where 1.5 and 2 Ha of land respectively are available for building. The site with 2 Ha may be more suitable than the 1.5 Ha site, but there is a ‘sliding scale’ of suitability here and neither site would be ruled out completely. It is therefore often important as a first step to decide exactly how suitability changes in line with a characteristic like land availability for building at a particular location.

Furthermore, different factors are typically measured in different units. For example, the availability of land for building might be measured in hectares, whereas the distance to the nearest bus terminus might be measured in metres. Across all of the plots being considered, the land available for building might vary from 0.5 Ha to 5 Ha, whereas the distance to the bus terminus might vary from 200 metres to 5,000 metres. Unless we change the scale of measurement for each factor here, when we combine the different factors, the larger numbers for distance to bus terminus will completely drown out those for land availability. One solution to this problem is to rank each site from highest to lowest before combining them. Thus, we might rank the short-listed sites in terms of land availability, again rank them in terms of public transport access, and so on for all of the other factors. We would then combine the ranks, rather than the original data, to obtain an overall picture of suitability. It should be noted that there are also other methods apart from ranking for bringing different factors to a common numerical range.

Even after taking into account the different measurement units and statistical properties of different factors, those with a good knowledge of what constitutes a suitable site (e.g. hospital managers, physicians, and representatives of patient groups) will have opinions about the relative importance of the different factors. For example, for a publicly-funded hospital targeting a relatively poor population, good access to public transport may be considered more important than car parking capacity. The relative importance of each map layer is normally represented through a series of weights. Each weight is a number reflecting the relative importance of a different site characteristic (see Rinner and Taranu (2005) for greater detail). The map layer representing a site characteristic (such as distance to a bus terminus) is multiplied by its weight, and then the resultant maps are added together for all characteristics to give an overall relative suitability map. This map shows the relative suitability of different sites, but the units have no meaning in themselves. They are not, for example, the number of disease cases averted per $ spent, but simply reflect how one site compares to others.

Often, deciding on an appropriate set of weights and constraints is an iterative process. A group meets to decide on an appropriate set of weights, each map layer is multiplied by its weight, and the resultant relative suitability map is created and circulated back to the group. The group then examines the map output, looks at the most suitable sites, and if they feel that the current weights are not selecting the most suitable sites, the weights are modified. The process begins again, until the group is satisfied that the suitability maps being produced by the weights genuinely reflect their view of site suitability.


Activity

You are identifying potential sites for the development of a new general practice surgery, which will provide out-of-hours (weekend and late night) care in Bristol , UK . The new out-of-hours facility will be developed as a new facility and run as a publicly funded (rather than private) facility.

You have already been provided with post codes for a short-list of plots of land that are currently for sale within your region.

You have at your disposal the following map layers / attributes for each plot:

  • the location of existing primary care facilities providing out-of-hours care;
  • the area of land available for development at each site;
  • census data on Limiting Long Term Illness in the general population, referenced to super output areas (census polygons comprising around 1,000 households);
  • census data on levels of multiple deprivation in the general population, again referenced to super output areas
  • prices for each plot of land;
  • the boundaries of designated Conservation Areas and Greenbelts, where any new building work is heavily constrained by planning laws and must first go through a lengthy planning approval process with the local government authorities;
  • the locations of water bodies and floodplains near rivers, known to be liable to flooding;

Post a message to the course discussion board, outlining:

  1. Which of these map layers you think represent constraints and which represent factors?
  2. Is there anything else that you might like to include in your criteria that is not covered in this list?
  3. For those map layers that you have identified as factors, rank these in order of importance from most important to least.

References (Essential reading for this learning object indicated by *)

The Bureau of Health Professionals in the US uses a suitability scoring system to design incentives for medical staff wishing to move to areas with limited healthcare provision. For example, see the Health Professional Shortage Areas initiative and the Medically Underserved Areas initiative at https://datawarehouse.hrsa.gov/topics/shortageAreas.aspx

There are some examples of GIS and MCE being used to identify potential locations for health facilities here:

Ohtaa, K., Kobashib, G., Takanoc, S., Kagayad, S., Yamadaa, H., Minakamia, H., and Yammer, E. (2007) Analysis of the geographical accessibility of neurosurgical emergency hospitals in Sapporo city using GIS and AHP. International Journal of Geographic Information Systems 21 (6), 687-696

Vahidnia, M., Alesheikh, A., and Alimohammadi, A. (2009) Hospital site selection using fuzzy AHP and its derivatives. Journal of Environmental Management 90 (10), 3048-3056.

This article uses MCE and GIS to identify areas with a high demand for health services (though it does not consider the other factors involved in assessing potential hospital locations):

* Rinner, C., and Taranu, J. (2005) A Geographic Visualization Approach to Multi-Criteria Evaluation. In Proceedings of Geoinformatics Conference 17.-19.8.2005, Toronto, Canada. http://www.ryerson.ca/~crinner/pubs/F131.pdf

This article also looks at site suitability for hospitals with GIS, but uses a human-guided approach rather than the MCE approach described above:

Kofi, R. Y., and Moller-Jensen, L. (2001) Towards a framework for delineating sub-districts for primary healthcare administration in rural Ghana: a case study using GIS. Norsk geogr. Tidsskr. 55, 26–33

Comments are closed.