A review on statistical monitoring of rare events and their necessity in healthcare area

There are a variety of quality characteristics in healthcare which are needed to be monitored and controlled. Of course, every quality characteristic needs to apply a special and exclusive statistical method. There are some special quality characteristics which are considered as rare health events. The incidence rates of these events are such small that one might guess monitoring is not necessary or even possible. In such cases, conventional p and np control charts are not efficient, since over dispersion occurs. It should be noted that various statistical methods have been developed for monitoring such quality characteristics. In recent years, the problem of monitoring health events with small incidence rates has been as an attractive subject. Rare health events, such as congenital malformation has received considerable attention in the literature. Thereafter, miscellaneous statistical methods have been developed to monitor such rare events. The most important methods of such type are the "Sets method", "CUSCORE", "SHDA" and the "Bernoulli CUSUM".In this paper, performance of aforementioned methods, are compared with each other based on average run length (ARL). Comparing the aforementioned methods regarding the ARL measure shows the descending performnces for the Bernoulli CUSUM, the CUSCORE, the Sets method and the SHDA. That is, the Bernoulli CUSUM has the highest and the SHDA has the lowest performance. Science and technology developments, especially in healthcare area has improved quality characteristics levels, such that some quality characteristics (like failures rates) are known as rare health events. Various statistical methods show that either such rare events can be out of control because of some assignable causes. Developing and applying such methods is a necessity to quality improvement in healthcare. Among such methods, Bernoulli CUSUM chart has the highest performance.