In a Jan. 13 presentation to the federal Health IT Policy Committee, Annie Fine, M.D., a medical epidemiologist in the New York City Department of Health and Mental Hygiene, described both the sophisticated software used to track disease outbreaks such as Ebola, as well as how better integration with clinicians’ electronic health records (EHRs) would improve her department’s capabilities.
“In New York City, every day we are on the lookout for unusual clusters of illness. And we receive more than 1,000 reports a day just in my program,” Fine said. Epidemiologists run a weekly analysis to detect clusters in space and time, and use analytics and geocoding to compare current four-week periods with baselines of earlier four-week periods.
“We get a large number of suspect cases reported, and they may be way out of proportion to the number of actual cases,” Fine said. Epidemiological investigations require hundreds of phone calls to providers and labs. “That could be made much less burdensome and efficient if we could have improved integration with EHR data.”
Fine used recent and past outbreaks, including West Nile virus, SARS and flu epidemics, as examples of how New York City gathers and shares information. Because the meeting was prompted in part by the recent Ebola outbreak, she started there.
“We were obviously aware of what was going on in West Africa, so we did have time to prepare for cases here, anticipating how they would present to the medical setting,” she said. “We actively alert and reach out to clinicians and hospitals with the key features they should be looking for and reporting to us.” In this case, it was fever, with compatible symptoms and travel to the effected areas.
The challenges from a surveillance point of view were that the symptoms were very nonspecific and the margin of error was essentially zero. New York City uses an email system called the Health Alert Network to notify clinicians. “We hope as many providers as possible see it,” she said, “but we know we don’t reach every provider through that network.”
In the case of Ebola, the city also conducted unannounced drills with fake patients just to be sure procedures were in place. It is now conducting active surveillance of all returning travelers and care workers who care for people with Ebola for fever and symptoms.
For surveillance and outbreak investigation, New York City uses Consilience’s Maven disease surveillance and outbreak management software, which enables monitoring of more than 90 communicable diseases. “We use the same system when people become ill to track their illness and to receive their lab data,” Fine explained. “When one of these folks becomes ill, we need to work closely with evaluating hospitals to rule in alternative diagnoses and rule out Ebola, either by testing or on clinical grounds.
Rapid access to clinical and lab data is critical for evaluating persons under investigation, especially with the enormous amount of pressure from the media and city leaders, she said. “The key needs in this outbreak were for rapid diagnosis at our local public health laboratory and the ability to launch a rapid contact investigation to identify all contacts, assess their risk, and quarantine or monitor them as needed,” Fine added. “When we had our case in New York City, we needed to create a timeline, and we did use cell phone records to track where the person had spent time. That was extremely helpful.”
She described a current outbreak of Legionella bacteria in a densely populated area of the Bronx. “We are feverishly working to determine the source,” Fine said. Epidemiologists suspect the source is the power plant serving the largest apartment complex in the United States. “We are using the Health Alert System to alert physicians to test for Legionella in patients who meet the case definition and report immediately to us,” she said.
They detected this outbreak because it is a reportable disease, reported by electronic laboratory reporting (ELR) systems. They took advantage of the fact that data is geocoded automatically to recognize two cases in one building using a building ID number.
“You can see how critical it is to get good lab data and patient addresses,” Fine said. “Now we need to nail down the source of the outbreak using molecular epidemiology and work with the facility to stop the contamination at its source.”
Much of the epidemiological data comes from ELR, which is very high volume in New York City. Some of the data is standardized with LOINC (Logical Observation Identifiers Names and Codes) and SNOMED (Systematized Nomenclature of Medicine) codes, but much is nonstandardized and requires a lot of manual effort to clean up and make useful, she said.
“It requires IT systems at the laboratory or hospital with many integration points along the way. Because all the data is electronic, there are many opportunities for failure and data problems at integration points. We need good quality assurance systems to make sure data aren’t dropping out,” Fine said. “We rely on provider reporting for clinical data to complement ELR data. We have a Web-based provider reporting system, as well as paper and phone, which we still definitely need.”
So what could better integration with EHRs do? “When we have larger or more complex outbreaks, we have a demand for real-time data, making integration more critical,” Fine said, adding that her department is currently defining the use case for bidirectional communication with providers and labs. “We are working with an outpatient network in the Bronx on inserting alerts from us into the EHR system to provide information and prompt reporting in some situations."
EHRs could also provide improved demographics and contact information for patients. “Our work requires we know where patients live and be able to interview them. To do that, we need the patient demographics,” she said. “We need the provider information and how to reach them as well. EHRs have a lot of that information and it is not available from labs reporting to us.”
Once there is a known outbreak, Fine said it would be fantastic to query EHRs using a case definition to assess medical risk factors and to help target populations at risk.
“Many times the discussions about outbreak management focus on detection — the fire alarm,” she said. But once an outbreak is detected, the need for ongoing information to manage it is far from over. “EHRs could help us with finding all the cases and confirming them.”
EHRs could assist with provision of specimens for laboratory diagnostics, determining who is at risk from a medical point of view, tracking the course of the outbreak, evaluating interventions, determining the outcome of those interventions.
“All of these are critical to putting out the fire,” Fine said, “and restoring the population health.”