Hurley Medical’s EHR analysis is playing a key role in Flint’s water crisis
Without the power of electronic health records, the discovery of high levels of lead in the water supply of Flint, Mich.—and subsequent proof that its children were accumulating high levels of lead in their blood—is a tragic story that might not even be known to this day
Acting on a hunch, a pediatrician at Hurley Medical Center ordered an analysis of blood test records that compared blood lead levels in small children before and after the city switched its water supply from Lake Huron to the Flint River in a cost-saving move.
The results, calculated from hundreds of specific records that were identified and then geographically mapped, took just two weeks. The stark evidence showed a doubling, and in some places a tripling, of lead levels in 1- and 2-year-olds—ages in which the neurotoxin can cause significant developmental damage. Those results led to a shutdown of the Flint water supply and reestablished the lake water source.
“We routinely check blood lead levels, and there is no safe level of lead exposure,” says Mona Hanna-Attisha, M.D., the pediatrician who acted on the hunch that bared the findings. “But especially underserved minority communities have a pre-existing disparity in regards to lead exposure.” The Medicaid-mandated checks created the data, and their availability in an EHR from Epic that went into operation in 2011 made the pinpoint analysis possible.
The story is far from over. The next challenge is to isolate and tag the records of 9,000 to 10,000 children under age 6, whose brains are still developing, and arrange to track them long term to try to lessen the ill effects that are likely to occur.
That’s where the limits of the current healthcare IT infrastructure will be felt, says Michael Roebuck, M.D., Hurley’s chief medical information officer. “This is a great case example of the current abilities and the future deficiencies of medical records.”
Everyone knew there was something in the water from the start. Days after the switch, complaints about the color, odor, and increases in skin rashes all erupted. “But we were reassuring our families, because we were being reassured by government, that the water was safe,” says Hanna-Attisha.
Then a research crew from Virginia Tech took measurements at various sites along the river and found high levels of lead. Hanna-Attisha took the report as a template for mapping records of blood levels geocoded to determine where they were highest, as well as the greatest before-and-after increases. “Where the water [lead] levels were the highest was the exact same place as (where) the blood levels were the highest.”
Overall, the percentage of children with elevated blood lead levels--defined as 5 micrograms per deciliter or higher--went from 2.4 percent in 2013 to 4.9 percent in 2015. The water source was switched in April 2014. In one ward, the percentage spiked from 4.8 percent to nearly 16 percent. The Virginia Tech results recorded the highest water lead levels in that same ward--30 percent of the tested samples exceeded 15 parts per billion, which is the threshold for action by the Environmental Protection Agency.
This was the first increase in water lead levels in many years, says Hanna-Attisha. “We set ourselves back decades in terms of where our blood lead levels are.”
The findings were even more significant when contrasted with a trend of steadily decreasing lead exposure in Flint and around the state. This was the first increase in many years, says Hanna-Attisha. “We set ourselves back decades in terms of where our blood lead levels are.”
Unlike the evidence of dirty water, lead exposure has no immediate symptoms. Only by doing population-level investigation, “which is robust in systems like Epic,” did the trends show up, she emphasizes. “If we had still been on paper, we probably would still be poring through records.”
The power of the modern EHR is its ability to “grab data from multiple databases and combine data in a way that lets you do population-level analysis,” says Roebuck. “As an example, it’s one thing to get a lab system to spit out a bunch of lead levels--I think that’s pretty easy. It’s another thing to get a lab system to spit out lead levels and addresses and primary care doctor and phone number . . . that then you can use to do population-level analysis.”
“If you want to geo-map them, you need their addresses. If you want to contact them, you need their PCPs and phone numbers. And modern-day data analytics makes that very easy,” he adds.
The clear benefit of population health analysis goes further. The quick work staved off other threats that would have continued, such as public health advisories to boil water following episodes of high amounts of bacteria found in periodic samplings. For lead, it was the worst thing they could do.
“Even today, there are people who think, and rightly so, that boiling the water is the safest thing,” says Hanna-Attisha. But boiling evaporates water while leaving behind all the lead. “That lead is going to concentrate in whatever you’re boiling.” There were three boil advisories just in the period before the lead was discovered.
Although the lead discoveries kept things from getting worse, damage is already done. Exposure lowers IQ and causes attention deficit, hyperactivity, impulse and conduct disorders. Nutrition choices can lessen the effects, and services such as universal preschool and mental health access can counter some of the known impacts on cognition and behavior, says Hanna-Attisha. “We’re definitely looking at a 20-year lifespan tracking these children.”
For Roebuck, “that’s been the challenge. The health IT infrastructure nationally isn’t in a spot yet that tagging these kids for the duration of their lifetime is easy.”
Locally, it may be doable in conjunction with the regional health information exchange, he says. “As long as they stay in the city of Flint, we can manage the majority of that tagging; but as soon as some kid graduates high school and goes to college, or as soon as some family member gets a new job in a different state and the child leaves the state, maintaining that data connection with that person at that time is not something that’s easily done.”
The tracking of developmental milestones and lead-linked health outcomes will be the easy part, at least under the aegis of Hurley Medical Center and its EHR system. One essential set of data will come from problem lists, those computerized details required by the HITECH Act’s Meaningful Use program. The EHR also will have medical history and diagnoses that may include the cognitive and behavior changes most concerning to healthcare providers, says Roebuck. “It’s pretty simple to get those out of the record.”