Which Methods of Syndromic Surveillance Are Most Effective?
The researchers also compared four specific detection algorithms. The first used hospital admissions data from a single day; the second used a moving daily average that gave greater weight to more recent data; the third used cumulative deviations from a constant expected value; and the fourth used cumulative deviations from an expected value that is adjusted for seasonal variation in flu symptoms. The analysis found that all of the algorithms were equally effective in detecting a fast-spreading agent (one in which all simulated new cases were spread over three days — see Figure 1). However, the more sophisticated statistical methods had a higher probability of detecting a slow-moving attack (in which simulated new cases were spread over nine days — see Figure 2). Conceivably, this performance could be improved by monitoring a less common syndrome, pooling data across multiple hospitals, analyzing more indicators or hospitals, or studying geographic patterns. However, additional data or
