4.4 Demand characteristics

In deciding where to locate any new health facilities, we need to understand likely patterns of demand for services among the population. The number of out-patients making use of a new hospital, for example, will depend on factors such as the health profile of the population it serves, its age-sex structure, household income levels, and on household access to transport. These population characteristics – and the way that they are distributed geographically – together form demand characteristics for health facilities.

The population attributes affecting demand can be represented with differing degrees of sophistication. In terms of attribute data, for example, there are several approaches that can be taken:

• Population counts : At its simplest, demand can be represented through population counts, either for the population as a whole or for specific age/sex cohorts. For example, counts of the population under 5 years and under 1 years could be used in planning immunisation services. Typically, these population counts are derived from census data.
• General measures of health and disease : As a broad proxy measure of demand, several censuses now include questions about an individual’s general health. For example, the 2001 UK census included a question about Limiting Long-Term Illness (LLTI) – ‘Do you have any long-term illness, health problem or disability which limits your activities or the work you can do?’. Shaw and Dorling (2004), for example, have used responses to this question to assess demand for healthcare. Although such measures may not be useful in predicting demand for specific diseases (such as diabetes or stroke care), they are of use in predicting general consultation patterns across a range of different services.
• Inferred measures of demand : Demand for health services is sometimes inferred from other population characteristics, particularly levels of multiple deprivation. More recently, attempts have been made to infer the demand characteristics of small areas by statistical analysis of information about larger areas and the results of health-related sample surveys. For example, Twigg et al. (2000) have estimated smoking prevalence for small areas of the UK. Smoking is strongly associated with many diseases, most notably Chronic Obstructive Pulmonary Disorder (COPD), and demand for associated services.
• Measures of demand derived from patient records : In some instances, patient records may be available for assessing demand levels. For example, registered drug user records could potentially be used in planning drug rehabilitation services. However, it should be noted that data protection and confidentiality legislation may mean that in many cases patient records can only be used for clinical purposes and not for health service planning.

There are also different ways of representing demand characteristics spatially. These methods of spatial representation include:

• Centroids for census areas : One of the most widely used methods for representing demand spatially is to use the centroids (centre points) of census polygons. Using population centroids has two advantages. Firstly, centroids are relatively easy to calculate from existing census records. Secondly, centroids make any subsequent calculations to estimate the best configuration of a health service network much less computer-intensive.
• Population surfaces : Less commonly, population is also sometimes represented as a surface or raster grid, in which each grid cell contains a population count. Whilst this is intuitively appealing, this can make subsequent calculations to optimise the health service network more time-consuming. Furthermore, many grid cells on a population surface may contain zeros, since population tends to be concentrated in settlements.
• Geocoded patient records : In some contexts, instead of relying on census records, it may be possible to use patient or other hospital records to plan health services. For example, the postal or zip codes of emergency call-outs for ambulances can be used for such purposes. However, in many cases, confidentiality and data protection legislation may prevent individual patient data being used for such purposes, unless consent has been granted explicitly.

 

---

 

---