SaTScan

SaTScan is a software package designed for the statistical detection of spatial, temporal, and space-time clusters of disease and other events. It implements Kulldorff’s scan statistics and likelihood ratio testing to identify areas and time periods where observed cases exceed (or fall short of) what would be expected under a chosen null model, accounting for population at risk.

The program supports several statistical models, including Poisson, Bernoulli, and Gaussian, to accommodate different types of data. It can conduct purely spatial analyses, purely temporal analyses, or space-time analyses that search for clusters across both space and time. Analyses are conducted using scanning windows, typically cylindrical in shape, that move across the study region and time period to test for elevated risk within each window.

Analysis workflow involves inputting case data, population at risk, and optional covariates; selecting parameters such as maximum cluster size, scanning window parameters, and the number of Monte Carlo replications for significance testing. For each potential cluster, SaTScan computes a likelihood ratio statistic and uses Monte Carlo methods to generate p-values, helping to distinguish true clusters from random variation.

Outputs from SaTScan include identified clusters with geographic location, time frame, radius, observed and expected case counts, relative risk estimates, and associated p-values. Results are presented in tables and maps, and can be exported for reporting and visualization. The tool is widely used in epidemiology and public health for surveillance, outbreak investigation, and research on disease clustering.

SaTScan was developed by Martin Kulldorff and collaborators and is available for non-commercial use from its official website. It has become a standard reference in cluster detection studies and has inspired related methodologies and software in spatial epidemiology.