The relative efficiency of the sets and the cusum techniques in monitoring the occurrence of a rare event

Abstract

The efficiency of each of two methods for monitoring the occurrence of a rare event, the cusum, carried out monthly, and the sets techniques, was assessed by use of simulated datasets. The monthly cusum parameters used in this study were those recommended for monitoring of birth defects in a monthly analysis. The sets technique parameters, recommended for monitoring of low-level epidemics, were adjusted to match the average interval between false alarms to the corresponding values observed for the monthly cusum technique. The results indicate that the relative efficiency of the two techniques depends on the baseline frequency of diagnoses. The sets technique is shown to be more efficient than monthly cusum when the number of cases expected in a year is no greater than five, but less efficient otherwise. In the former case, lengthening the interval between cusum analyses can make the cusum technique more efficient than the sets technique. This possibility deserves further exploration and could affect future recommendations.