A cluster is an observed excess of cases in a defined population and period. Formal cluster investigation applies statistical analysis to determine whether the excess exceeds what random variation would produce.
Most investigated clusters are not statistically significant.
Alternative Names:
Disease Cluster, Cancer Cluster
Why it Matters?
Cluster claims are emotionally compelling and epidemiologically weak, because random variation produces apparent clusters in any distributed population and investigating after observing an excess introduces selection bias. Public health agencies investigate most reported clusters and rarely establish an environmental cause. That record is the substantive defense, along with the multiple comparisons problem inherent in searching for excesses across many locations and diseases.
Frequently Confused with
Related terms
Frequently asked questions
Are disease clusters usually attributable to contamination?
What is the statistical problem?
