[The impact of changing critical values of the early epidemic detection system for infectious disease surveillance in Japan].

INTRODUCTION

It is important to examine how critical values for initiation/termination affect the trend and frequency of epidemic/pre-epidemic warnings with the early epidemic detection system in Japan. Here we looked at the number of epidemic warning/pre-warning weeks and the influence of changing the criteria values for infectious diseases surveillance.

METHODS

An epidemic warning is initiated if the number of cases per week per sentinel medical institution exceeds a critical value. A pre-warning for an epidemic is initiated if cases per week per sentinel medical institution exceed a critical value and there is a non-epidemic warning. To determine effects of the criteria values for epidemics/pre-epidemics for warning onset and termination, we set different values and compared the number of weeks of epidemic warning, the proportion of the total observed weeks. Also, pre-epidemic warning measurements were compared. Data from the infectious diseases surveillance system were analyzed from fiscal years 1999 to 2003.

RESULTS

When the critical value for warning onset was lowered, the warning week started sooner and ended later. When the critical value was raised, the opposite occurred: the number of weeks with a warning status decreased. When the critical value was changed within a certain range, the number of weeks with a warning status became 0.5 to 2 times larger than those with the defined value. Similar trends were observed when the pre-warning was examined: the number of warning weeks was 0.4 to 2 times (for influenza and chickenpox) and 0.3 to 3 times (for measles and mumps) larger than those with the defined value. Except for pertussis and rubella, the proportion of warning weeks was approximately 5% for all diseases listed in the early epidemic detection system. In addition, there was no distinct issue with the critical values themselves.

CONCLUSION

The present examination of linkage between trends and frequencies of epidemic warnings/ pre-warnings and the critical values in the early epidemic detection system of infectious disease surveillance in Japan confirmed the expected increase with lowering of the threshold. Except for pertussis and rubella, there was no distinct issue with the critical values themselves.