The road to last month’s cookie dough recall started when CDC scientists reviewed information collected through PulseNet, a national network of laboratories that perform DNA “fingerprinting” of foodborne bacteria like E. coli O157:H7, Salmonella, and Listeria. These fingerprints are plugged into a database that CDC and its state partners routinely scan. I’m a PulseNet database manager at CDC and one of my jobs is to identify “clusters” – groups of illnesses that share the same fingerprint.
On May 14, 2009, I performed a routine weekly cluster search of the E. coli national database and found one of our “common” patterns rearing its head as it does each week. There’s lots of E. coli out there and some strains are more common than others. In these cases, we have to make an educated guess with each review – is there something different about this common occurrence that might signal an outbreak? In other words, “Ignore this or report it?”
In this case, we were seeing an increase in cases. It wasn’t extremely high, but for May it was higher than usual. At the same time, we got an e-mail notification from our state lab colleague in Massachusetts stating they had patterns that matched the potential cluster. At this point, I decided to report it.
Once our epidemiologists started looking into this cluster our work really began. The PulseNet database team started receiving requests for updated case counts with demographics and pattern frequency graphs. Our lab was asked to perform additional tests to fine tune the distinction between cases, due to the common nature of this pattern.
Little did we know that this common pattern cluster of E. coli O157:H7 would turn into one of the more challenging outbreaks so far this year! Soon after the investigation began, it appeared that raw cookie dough may be the source of the cluster, something that still puzzles the experts!
PulseNet is a cluster detection system that helps us know quickly if something is not right with our food supply out in the public. Improvements to surveillance can always be made, and one thing that would be great to improve is how to more easily identify true clusters of our common patterns as many of these may be outbreaks that go unnoticed. Next generation methods are in development and it is going to be interesting to see what the future has in store for PulseNet, surveillance, and cluster detection!