5.1 Overview of clustering issues

A disease cluster may be defined as an aggregation of disease in time, space or both that is either real or perceived. Historically, the identification of clusters has led to some major breakthroughs in understanding of diseases. The most well known example is John Snow’s investigation of a cluster of cholera deaths in 19th century London, which was centred around a contaminated water pump. Further examples of both successful and unsuccessful attempts to identify clusters are described in CDC (1990).

In identifying disease clusters, it is important to identify whether a perceived cluster is real, due to chance, or simply the product of data collection methods. Some examples of possible data quality issues are given in the articles by Forand et al (2002), Gregorio et al (1999) and Rogerson and Han (2002). If a disease cluster does appear to be real, subsequent investigation of the causes of the cluster is usually necessary. For example, the water from the suspect pump could be tested for cholera bacteria in a follow-up study and this would provide a plausible explanation for the cluster of cholera deaths.

Another important aspect in assessing disease clusters is the management of public perceptions of disease risk. For example, in the UK a cluster of leukaemia cases around the Sellafield nuclear processing plant received considerable media attention during the 1980s and 1990s. The CDC (1990) article below describes some of the difficulties in managing public perceptions of risk as well as guidelines for doing so.

For an overview of the issues involved in cluster identification and interpretation, view the flash animation below and read through the references.

 

---

---