Health surveillance around prespecified locations using case-control data

There are several approaches one may use to model or test for potential risk around point sources of interest. These approaches have been developed almost universally to (a) fit model parameters to estimate the nature and significance of decline in risk as one moves away from the point source, or (b) assess the significance of a test statistic based upon the null hypothesis of no raised incidence around the source. In this paper, I assume that the data on the locations of cases and controls often used for these questions may be arranged in temporal order (for example, data might consist of the date of diagnosis for both case and control diseases). I then illustrate how conventional modeling approaches may be adapted to use the dataset observation by observation, to detect as quickly as possible a change from one set of model parameters to another.