MULTIPLE CHANGE-POINT ANALYSIS OF DISEASE INCIDENCE RATES

Monitoring disease frequency in the livestock industry has become feas ible by use of the recently developed health monitoring databases, Bas ic problems in analysing incidence rate (incidence density) over calen dar time are to detect changes and to estimate the times and amounts o f change. We discuss a number of plotting techniques and associated st atistical hypothesis tests for retrospective detection of single and m ultiple change-points, The tests are modifications of well-known tests either for testing uniformity of a sample, or for comparing two Poiss on variates by a likelihood-ratio test or a Pearson chi-square test. F irstly, tests for a change-point at a fixed time are modified to conse rvative tests for one change-point in a given finite set of possible c hange-points in a single interval. For the problem of finding multiple change-points we then study sequential strategies similar to procedur es for selecting regressor variables in multiple linear regression. In particular, a modified forward selection technique is shown to perfor m well in two examples with disease incidence data from a Danish pig h ealth and production monitoring system. In one of the data sets there is a large number of cases and fairly small changes in the rates, whil e the other data set has a smaller total number of cases but large cha nges in the rates.