Evaluation of a method for detecting outbreaks of diseases in six states

A new statistical method, developed for detection of changes in reporting, has proved useful in analysis of provisional data reported by state health departments to the National Notifiable Diseases Surveillance System (NNDSS). In this system, data from the current four-week period can be compared with data from the previous, same, and subsequent four-week periods from each of the preceding five years, and reports exceeding historical limits are highlighted in a horizontal bar graph. To evaluate the usefulness of this method at the state level, we applied it to weekly reports of seven notifiable diseases in six states over a four-month period. Participating state health departments investigated all events exceeding historical limits and reported known outbreaks that were not identified by the method. During the four-month period, the method identified 27 episodes of disease reports exceeding historical limits. Of these, 14 (52%) represented outbreaks. None was detectable by analysis of aggregate national surveillance data. Five outbreaks known to state health department officials were not identified by the method, because of increased disease activity during the baseline period or lack of timely provisional reporting of outbreak-related cases. Methods for detection of increases in reporting at the state level may identify events of public health importance that are obscured in aggregate national data and may supplement other local sources of information available to state health departments in the recognition of significant public health events.