California Department of Public Health logo: three likenesses of people colored blue, green, and orange  
Last Edited: 8/3/2010

Clues to the Etiology of Disease Clusters -- A Practical Method
Written: 2009

Investigating a community disease outbreak or cluster, such as a cancer excess where environmental factors are suspected, can be time-consuming if not focused. We describe simple approaches to thinking about risk factors and their prevalence that can guide the investigator looking for plausible explanations for an observed excess.

The calculation of the observed number of cases in the index community and the expected number, based on rates in a reference area, is typically the first step in a cluster investigation. The ratio of observed to expected cases, the standardized incidence ratio (SIR), reflects a combination of the relative risk (RR) of disease for a given risk factor and its prevalence in the index community. Therefore, given the SIR and plausible RRs for suspected causes, one can calculate the proportion of the index community that would have to have been exposed to have resulted in that SIR.

For example, consider an index community that has twice the disease incidence as a reference area (SIR = 2). In this case, if a risk factor under consideration as a candidate for causing the cluster increases disease risk three-fold (RR = 3), and is normally present in 10% of the population, then 70% of the individuals in the index community would have to have been exposed to this risk factor to result in the observed doubling of the overall SIR.

The usual next step in an investigation examines the series of cases.  We can screen for candidate exposures by calculating the proportion of cases that would be required to have a risk factor if that factor were responsible for the observed community SIR.

We illustrate these methods with examples, and show how they can be used to help judge whether certain combinations of risk factors and exposure frequencies are plausible, and to rule in or out potential risk factors in a case series investigation. 

Full Report (Size:  207KB)

Suggested Citation

  • Smith D, Hoshiko S. Clues to the etiology of disease clusters: a practical method. American Journal of Epidemiology 2009;169:S111