More associated with finding solutions to planning and social problems than health problems, the use of geographic information sciences within the NHS has been steadily increasing.
Geographical information sciences, otherwise known as geographical information systems, can be defined as the handling of spatially referenced data – its collection, storage, integration and analysis, as well as how it is displayed.

The Public Health Geographical Information Sciences Unit at the University of Sheffield is one department specialising in the use of geographic information sciences in this field. It was established in 1999, and since then it has provided training programmes in the analysis of geographic data with regards to public health. They have attracted over 150 health and local authority professionals over the last three years. Members of the academic staff work particularly closely with health authorities, primary care and NHS trusts, universities, local authorities, government offices and public health observatories and in 2001 hosted the first European conference on this issue, which was attended by 252 professionals from 19 countries.
In the context of public health, the process of geographical information science involves examining regional variations in the state of a population’s health and the local healthcare provision. This can include travel time analysis, which measures travel time and distance as a variable in modelling treatment uptake rates across the country. Health inequalities, such as variations in life expectancy, are also calculated, as are such factors as environment – an example of this would be to model cardiovascular and respiratory mortality rates against pollution levels in a particular area.

The distribution of mortality can also be assessed through disease distribution analysis. This involves plotting incidences of a particular condition on surface maps and, using census enumeration districts (EDs) it can be determined whether that particular disease or condition is more prevalent in one area of the country than another. Accuracy can be ensured through small area statistical analysis, involving special analysis techniques to model small area data, where the figures concerned are normally considered too small to generate meaningful results.
Practical applications for the technology and techniques include service planning and evaluation, which can be used to determine the location of a particularly expensive piece of medical equipment, such as screening units, to maximise their impact on public health. A health impact assessment can predict the outcome of changes to local areas, such as the increase in pollution resulting from a new road.

But there are problems with applying this technology to public health situations. The use of spatial data in this context proves very difficult when, as is true in the vast majority of cases, it is prohibited to publicly identify a person’s health status. Indeed, there have only been two recent cases where identification has been allowed – Megan’s Law in the United States, which allows the release of residential information of known child sex offenders, and the mapping of SARS cases in Hong Kong. To avoid complications and direct, or indirect, patient identification, aggregated data can be used, but this creates “spatial uncertainties”, or inaccuracies. Use of the technology is also hampered by a need to present complex information in a simple way, as data can be misinterpreted, misunderstood or misused.  Indeed, it has been argued that in some cases, it is essential that cartographers involved in mapping the health information “lie”, removing some information to allow the critical information to be displayed in a simple, easy-to-understand fashion.

Another pitfall is what has become known as the “gee whiz” effect, where a hypothesis is formed on the basis of a trend seen in the mapping of the data that has not yet been confirmed. An ecologic fallacy can occur, where users can infer causation of a particular trend from correlation in the data and, from this, make conclusions about an individual using population data. Conversely, an atomistic fallacy is also possible, where an individual’s behaviour is not considered in its broader context. A patient’s health outcomes may be affected by environmental variables not considered in the study, as their case is mapped at the place of diagnosis and not their daily activity space – the places in which they live and work. As a result, patterns in the occurrences of conditions can be missed. And all these traps that analysts can fall into are in addition to data problems and errors in the mapping of cases.1
The use of geographical information sciences has the potential to bring enormous benefits to public health analysis. But there are many hurdles to overcome before any useful, meaningful and accurate conclusions can be drawn from its findings.

Footnote:  1 Towards Evidence-Based, GIS-Driven National Spatial Health Information, Infrastructure and Surveillance Services in the United Kingdom by Maged N Kamel Boulos, School of Health, University of Bath. Published in the International Journal of Health Geographics, January 28 2004