When is an apparent excess of neurologic cases epidemiologically significant?

To conserve clinical and public health resources, it is necessary to screen reports from the community of an excess or cluster of cases of chronic neurologic diseases for potential epidemiologic significance, ie, to identify those clusters that may have occurred owing to the operation of an underlying etiologic cause rather than to chance alone. Traditionally, the probability that such a cluster has occurred by chance within the reporting community is computed, ignoring the fact that many other similar communities have not reported a cluster. We propose a modified approach that takes this larger universe of communities into account, thereby raising the threshold for potential epidemiologic significance. As a result, the lowest value for the ratio of observed-to-expected cases that may be considered of epidemiologic significance should be increased by factors of 2 for small clusters (one to three expected cases), 1.5 for intermediate clusters (seven to 30 expected cases), and 1.3 for large clusters (60 to 90 expected cases). Consequently, case ascertainment and full field investigations can be reserved for only those reports that, if confirmed, would represent a cluster not due to chance alone. We illustrate this with a case from our own experience.