The saying, ‘a picture paints a thousand words’, is never more accurate than in the context of maps. Maps are unrivalled in their power to make geographical data more comprehensible.
While in some areas of the health sector, such as epidemiology or population health, the use of maps for spatial representation of data is common and has a long tradition (from John Snow’s map of cholera outbreaks in London in the 1850s to contemporary atlases such as the Atlas of US Mortality and the Dartmouth Atlas of Health Care), in many other domains of medical research the use of maps is far less common.
In the NHS, mapping datasets such as hospital episode statistics and geodemographic and public health data can expose inequalities in health service provision and inform the commissioning of services.
Many managers responsible for health service provision may not be familiar with public health and epidemiology, yet they need to manage and commission health interventions regionally. The use of proper cartographic representation is therefore critical in ensuring that commissioning decisions, based on maps created from the relevant data, are reasonable.
A survey of public health, performance and information analysts, carried out by Dr Foster Intelligence and University College London, revealed fascinating insights into how map design can impact on the interpretation of data. Some results are shown below. Colour is often used to portray differences in percentages, rates and other intensity measures (for instance, the prevalence of type 2 diabetes). The survey found that red shades were popular among NHS data analysts and single-colour gradients were preferred to display intensity. This is because gradients with two or three colours offer no logical ordering from low to high intensity.
For mapping of count data (for example, the incidence of type 2 diabetes), it might be more appropriate to use graduated point symbols; this would also avoid the problem of large areas being disproportionately dominant on maps. The survey found that scaling the points by size and colour assisted interpretation of data.
Data classification can also influence interpretation. Look at the two maps below and notice how data is grouped or “classified” differently. Each map gives a different picture of the data. The one on the left would assist in assessing overall performance, but if the map was created to assess the local area, the one on the right might be more useful. Classification depends not only on the data distribution but the question the map is created to answer, as well as providing an appropriate level of detail. Wide ranging issues surround the mapping of health data because the maps must accurately represent the data and be easy to interpret.