Evaluation of Temporal Aberration Detection Methods in New York City Syndromic Data

The New York City (NYC) syndromic surveillance system has been monitoring syndromes from city emergency department (ED) visits for over a decade. We applied four aberration detection methodologies to a time series of ED visits in NYC spiked with synthetic outbreaks. Among the methods tested, performance varied by outbreak type and size; sudden large one-day spikes in cases were the most commonly detected, although sensitivity was low.  The methods tested did not perform well; variability in method performance by outbreak type suggests multiple methods may be ideal for detecting different outbreak features.