Since the ISDS annual meeting is in December, every year I bring home Christmas ornaments for my kids representing the city I visited. The other night we hung 6 years worth of those ornaments on the tree–a moose from Park City, St. Louis Cathedral in New Orleans, the Constitution from Philadelphia–bringing back quite a few memories in the process.
Ornaments aren’t the only thing I’ve collected from these trips. My first few trips, I tried to track down every analytical method available. Of course there are the classic control chart methods at the core of syndromic surveillance. There’s the slew of Bayesian spatial scans. And then a whole passel of time series regression techniques have been presented from groups all over the world.
As time went on, there was also a growing interest in new data sources. Internet search terms, Twitter posts, outbreak prediction markets, and, most recently, Wikipedia edits have all been evaluated. Can they tell us things about influenza season and other known events that we can’t get elsewhere? Do they reveal signals not apparent in other sources?
This year, the collection that grew the most was health domains where syndromic surveillance data was found useful. Mental health seemed to be a particular point of emphasis. I heard several presentations on using syndromic surveillance data to monitor mental health in the wake of mass casualty events like the Boston Marathon bombing, devastating natural disasters such as Hurricane Sandy, and periodic occurrences like the anniversaries of tragedies or certain holiday seasons.
Other sectors of public health were represented as well. Syndromic surveillance data was used to understand the health impact of environmental exposures such as the smoke from wildfires, or storm systems that disrupt normal healthcare delivery. There was a talk on gun injuries, and another on motor vehicle collisions. Another speaker looked at trends in opioid abuse among veteran populations. Beyond specific conditions, matters of healthcare policy were also discussed, including my talk on the Impact of Demographics on Healthcare Utilization.
It was heartening to see such widespread use of syndromic surveillance data. Significant resources have been leveraged to establish and maintain the necessary data connections, and to build the systems to analyze that data. It is a rich data set, and we should absolutely be doing everything we can to maximize the return from that investment.
Yet for all the diversity in the use of syndromic surveillance data, the emphasis was still on acute conditions. All of the research I mentioned used some combination of chief complaints and diagnosis codes from emergency department, and in some cases urgent care, visits. This data naturally lends itself to acute conditions, since they are predominantly what are treated in those settings. In the United States, however, some of our greatest public health challenges are related to chronic conditions.
Meanwhile, at Health Monitoring Systems we have begun receiving data from ambulatory practices that desire to send to a syndromic surveillance system as part of their meaningful use public health objectives. So far, our experience with that data suggests that it is better matched to chronic disease surveillance than emergency department data is. I presented some of our initial findings on using ambulatory care data for chronic disease surveillance at this year’s ISDS, but this is just the most preliminary attempt at scratching the surface.
Looking ahead to ISDS 2015 and beyond, my kids are wondering what Christmas ornaments I’ll bring home from Denver and elsewhere. What I am looking forward to is seeing an expansion of the use of ambulatory data for a whole range of chronic health conditions in the way that emergency department data is now being used to cover the whole spectrum of acute conditions.