Every year around this time, we anticipate news coverage on the flu season, the time when influenza is at its most prevalent in the U.S., and something we’ve tracked in past years. But this year has been quite different: First, a broad and severe outbreak of enterovirus has changed the conversation about infectious disease, the virus leading to high rates of hospitalizations since it first surfaced in the Midwest in August; second, the first confirmed case of Ebola in the U.S. was reported last week in Dallas.
With these developments, coordinated and accessible data is now more vital than ever for understanding the patterns of these diseases; where they’re rising, ages that are most at risk. While officials at the Centers for Disease Control and Prevention (CDC), working with local public health departments, are vigilant in monitoring and providing guidance for these and other outbreaks, data from electronic health records (EHRs) can and should play a more integral role. You see this starting to happen with Meaningful Use, the government incentive program designed to increase adoption of EHRs. The program requires that providers be able to “submit syndromic surveillance data to public health agencies.”
However, the ability to implement and develop the capabilities to feed this data into one streamlined network for public health agencies may be several (or more) years away. This is simply because the adoption of EHRs and the “connection” of data is a massive undertaking. There are existing infrastructures to monitor and manage disease outbreaks — the National Notifiable Diseases Surveillance System being the most notable one. As data from EHRs become more widely centralized and accessible, the hope is these officials will be able to spend more time analyzing the data versus collecting it.
What could the future of disease surveillance look like with real-time EHR data from ambulatory care settings being fed to public health agencies? The athenaResearch team examined the latest disease trends for enterovirus and influenza, based on diagnoses across our network of more than 55,000 health care providers. These providers all use either athenahealth’s cloud-based billing, or its EHR applications, both of which document diagnoses and procedures from patient encounters. And since these applications are “single instance” (everyone uses the same software version), the data is centralized in a common database. In other words, our providers and their data are already connected on a common network.
Figure 1 shows the rate of upper respiratory infection (URI) during the first 39 weeks of the year (week 39 ended Saturday, September 27) in the Midwest states of Illinois, Indiana, Michigan, Missouri and Ohio. These five states were chosen because the recent enterovirus outbreak roughly started in this band of states; URI was selected since it is a common diagnosis related to the virus.
Comparing 2014 diagnosis rates to those in 2012 and 2013, we can see a sharp peak in week 37 (week ending September 13) that coincides with the Midwest outbreak. Now, URI is a much broader diagnosis and doesn’t necessarily indicate enterovirus. However, the point is that this data is available from our cloud-based network practically upon demand — and calculating alternative measures that may indicate enterovirus (short of confirmed lab tests) is a straightforward process.
Given that flu season is near, we’d also like to share our tracking of influenza-like illness (ILI). athenahealth has been participating in flu tracking over the past year, primarily to determine whether our data reflects national health care trends, and to fill in as a supplemental data source during the government shutdown last fall.
Figure 3 shows rates of ILI from two sources: the CDC’s outpatient illness surveillance network (ILINet), and estimates from athenahealth’s cloud-based network (athenaNet). Although our estimates do not exactly match those from ILINet, the patterns are fairly striking, with “peaks” and “valleys” occurring in parallel.
Our goal over the next year is to improve the accuracy of our estimates and continue to learn how this type of data can be valuable for disease surveillance. We are currently conducting a series of pilot partnerships with public health agencies to provide data that is more detailed than is available through other sources. For example, we are working with four agencies in Ohio — Ohio Department of Health, Columbus Public Health, Cuyahoga County Board of Health and Delaware General Health District — to provide them an additional data source for ILI, URI and flu vaccination rates, at the county and age group levels.
We compile simple summary statistics on a weekly basis. But these statistics come from a network of data from millions of patient visits; data that we are quickly accessible and available at scale — providing data for one county health department is as simple as providing data to multiple counties.
More important, it is data that health care officials and researchers can access for quick and actionable insight into disease outbreaks. The key to effective disease surveillance is the connected, open flow of data between sites of care and public health authorities.
For more information, see the athenaResearch team’s post on disease surveillance.
October 6, 2014