The Effect of Uncertain Locations on Disease Cluster Statistics

Disease clusters are space-time aggregations of adverse health events such as birth defects and cancer. Because of incomplete information the exact locations of health events are often unknown, but to date the effect of uncertain locations on disease cluster statistics has not been assessed. This research demonstrates that uncertain locations, if not accounted for, lead to test statistics, null distributions and p-values that differ substantially from those based on actual locations, and increases the probability of type I1 error (false negatives). Further research is needed to develop statistical methods and software to account for uncertain locations in disease cluster investigations.