Sarasota institute mines patient records for clues and cures

Aug. 26--Named in honor of a retired schoolteacher who died 15 years ago, a small Sarasota nonprofit is going public with an outsize ambition: to conduct original medical research that has potential to revolutionize the treatment of age-related problems like frailty, dementia and multiple chronic illnesses -- and to share this new information with the world.

In a nondescript building tucked behind a Tamiami Trail strip mall, the F.A.R. Institute is home to data analysts working on the principle that hundreds of thousands of medical records, collected over time, contain significant patterns too large for any one team of doctors to see. Building on a diagnostic system already used in major hospitals, known as the Rothman Index, these researchers are mining patient histories to answer a most basic but elusive question: What causes an individual human organism to fail or thrive?

The Rothman Index had its beginnings in failure. Like too many 87-year-olds who undergo heart surgery, Florence A. Rothman came home from the hospital, then went downhill fast and never recovered. After her death in 2003, her son Steven approached Sarasota Memorial Hospital and asked for permission to explore behind the scenes for reasons why his mother's declining condition had escaped the notice of the medical teams charged with her care.

Remarkably, the hospital's CEO said yes.

So Steven Rothman and his brother Michael used their expertise in computer science, and the hospital's extensive patient database, to discern key indicators in electronic medical records that can predict whether a patient will get better or worse. In 2012, after successfully developing and marketing the Rothman Index to hospitals nationwide, Steven Rothman established the nonprofit F.A.R. Institute. And that former hospital CEO, longtime Sarasota physician Duncan Finlay, is now the institute's president.

Among the questions its researchers seek to answer: Is it possible to tell from sifting the clues hidden in 18 years' worth of hospital records which patients are likely to develop dementia -- long before they receive a formal diagnosis?

"Almost no one is in the hospital because of Alzheimer's," Steven Rothman explains. "They're there because they broke their leg, or they have pneumonia or something." But their detailed treatment records, he believed, contain clues that could help identify the onset of dementia, before symptoms have progressed enough to warrant a formal diagnosis.

So Rothman designed a feasibility study -- based on some 380,000 Sarasota Memorial patient records dating back to 1999 -- to determine how many of the people who were hospitalized with an actual diagnosis of Alzheimer's disease had been in the same hospital during the previous five years, for some other reason. The astonishing answer, given the number of retirees who move in and out of Sarasota: 30 percent. These patient records, with multiple hospital visits, provided what Rothman calls "an Alzheimer's time machine."

F.A.R. Institute researchers analyzed records for two groups of Sarasota Memorial patients -- those who visited the hospital without a dementia diagnosis two years before coming back with one; and an equal number of patients hospitalized twice within that timeframe who showed no signs of serious cognitive decline on either visit. Preliminary results, Rothman says, are intriguing: The team can predict who would fall into the first group with 40 percent accuracy, and can achieve 80 percent accuracy on the second group -- those without dementia symptoms.

"I think we can do even better when we get a full data set and have even better parameters," Rothman says.

If he's right, the payoff for such predictability could be life-changing. Evidence is mounting that the drugs now available to treat cognitive loss are far more effective when taken as close as possible to the onset of Alzheimer's disease, before any symptoms are noticeable. With a measurable way to identify those patients at risk, it may become cost-effective to screen them for the disease and begin an early drug regimen.

From that point, the possibilities absolutely flower, Rothman observes. The data could be studied retrospectively, to determine all the different treatments given to patients who do or don't develop dementia.

"The hospital has done a thousand natural experiments; it's up to us to ferret it out," Rothman notes. "One blood pressure drug may delay the onset of Alzheimer's. And it works both ways -- maybe if you chose the wrong drug, it advances Alzheimer's. But no one knows until you look, in a big data way, for this kind of thing. We're looking for signals across everything."

A different dimension

F.A.R. Institute, formed in 2012 and recently attaining retroactive federal nonprofit status, was supported for years solely by Rothman and his wife Barbara. The original aim was simply to publish insights gained from the Rothman Index in medical journals, to help convince doctors and hospitals that data mining, done correctly, can predict those subtle, unexpected turns for the worse that so often result in mortality.

PeraHealth, the company licensed to sell the Rothman Index to hospitals, cites impressive outcomes documented by its customers: a 29 percent reduction in deaths from sepsis infections at Yale New Haven Health System; a 43 percent drop in unplanned transfers to intensive care at Oconee Medical Center in South Carolina; a 22 percent cut in "failure to rescue" events at Houston Methodist Hospital.

Yet many health care professionals, Rothman says, remain resistant to making changes in their protocols in response to systems analysis -- "unlike the airline industry, where a near-miss is really examined to make sure it doesn't happen again."

Back in 2004, when Steven and Barbara Rothman first asked for five minutes of Finlay's time, they encountered a hospital CEO who also happened to be a physician and an early, passionate advocate for electronic medical records. Finlay gave the Rothman brothers access to those records, and they located a clear signal that others had missed: a correlation between nurses' routine "head-to-toe" assessments of their patients' stability, and lab test results that could indicate risk for an acute change in that condition -- if only those results were interpreted in an entirely different way.

Steven Rothman says this was the first time he and Michael had ever collaborated.

"That's one of the things that my mother really would have loved about this," he says, smiling gently. "She always wanted the brothers to work together. The fact that we were not medical doctors was, I think, looking back, an essential part. We were looking at it from a completely different dimension. And we had no preconceived notions of what was going to be important."

The brothers, Finlay, and another colleague -- the late Alan B. Solinger -- published a paper in 2012 that validated the uncanny accuracy of nursing assessments, which all hospital nurses perform and most doctors ignore. An essay this month in The New York Times, "How to Quantify a Nurse's 'Gut Feelings,'" offers testimony for the effectiveness of the Rothman Index at Yale New Haven.

So, at least in part, the original intent behind the F.A.R. Institute has been realized. Now, with its scope broadened dramatically, the institute is seeking donors to underwrite the cost of multiple research projects.

"To me, there are some things that should be tempting to the right person," says Finlay, who served as medical adviser to the Rothmans throughout the process of refining their system. Now he is engaged in community outreach on the institute's behalf. "There's a way for Sarasota donors to be involved in the work -- because we're here. I've been talking to everybody I know."

Sarasota Memorial Healthcare Foundation has already helped to fund the Alzheimer's initiative, and F.A.R. Institute analysts are working even more closely with the hospital on database analysis and quality control. With additional donations, Rothman says, the institute can gain access to valuable patient data from other hospitals, and tap expertise at medical schools.

The mid-range goal, Finlay adds, is to publish a sufficient body of work that will generate further research.

"We are going to be able to do some preliminary studies, which will then allow us to apply for government grants," he explains. "We think we'll be self-sustaining with the grants."

Although he was originally motivated to help the Rothmans by a desire to improve outcomes at his own hospital, Finlay says he remains active in their work because he sees the potential for much-needed progress in the science of caring for a rapidly aging population.

"Tied into this is trying to understand frailty," Finlay says. "What's the difference between this 78-year-old who's in a nursing home somewhere, and me -- here I'm sitting, doing just fine. That's a fascinating business."

___

(c)2018 Sarasota Herald-Tribune, Fla.

Visit Sarasota Herald-Tribune, Fla. at www.heraldtribune.com

Distributed by Tribune Content Agency, LLC.