Surveillance of birth defects: the Multicommunity Sets Technique tested by computer simulation.

Using a computer simulation for a series of births subject to various congenital malformations, the surveillance performance of the Multicommunity Sets Technique (MST) was compared to that of the Cumulative Sum Technique (CUSUM). Increases in malformation frequencies were simulated in (i) 6 out of 6 centres and (ii) only 3 out of 6 centres. MST was neither more sensitive nor more specific than CUSUM, and signalled increases with greater delay. One type of CUSUM procedure showed to accumulate more false alarms than MST in long periods of surveillance at the baseline malformation rate. The advantage of CUSUM was also shown for a single malformation with a very low baseline incidence, which according to Chen et al. (1983) should have been particularly suitable for MST surveillance.