Predicting the sick through personal trails of health data

    Listen
    (Todd Bookman/WHYY)

    (Todd Bookman/WHYY)

    How Independence Blue Cross is using algorithms to better target care. 

    The first thing out of John Iovine’s mouth is an apology.

    “You got to forgive me if I don’t remember too much. I had a stroke.”

    Signs of that stroke are everywhere in his spotless Frankford home, like the bed in the dining room, a shower installed in the pantry. John sits, thin, in blue pajama pants, in a blue wheelchair that he uses to get around.

    The memory issues, though, may be overstated.

    “We went to Harding…that’s the school right up here. And I seen this girl, in this long red sweater, and her red hair, and I said, that’s the girl for me,” he says of his wife, Carol, who he met in 1952.

    “I came out on top.”

    Carol’s on a plastic covered couch next to John. His stroke, she explains, came in the middle of a bad run of health. First, he developed an ulcer, then he needed a bowel resection, then the stroke.

    “And he had a pneumonia, jaundice, sepsis, clot in the right lung,” all between October, 2013 and January, 2014.

    John, a former house painter, spent 79 days in the hospital, some of that unconscious, nearly all of it stuck in a bed.

    “Aw man, it was hell,” he says.

    Sink or swim 

    John Iovine, his health still a work in progress, was finally sent home in April of last year, after several months in a rehab facility.

    And this moment in a patient’s recovery–when they seemingly have to sink or swim on their own–is the one that everyone in the health system is paying attention to right now. For too long, too many Johns sunk, ending up right back in the hospital.

    The industry calls these preventable readmissions, and they are a huge drain, costing Medicare alone $15 billion annually. That’s why it launched a program a few years ago that penalizes hospitals that see too many patients come back in the door. And in turn, that spurred many hospitals to pay more attention to the issue, and try new programs.

    Now insurance companies are also taking a stab at the problem.

    “So we are trying to identify which patients are likely to be hospitalized in the next three months, so that’s our target,” says Somesh Nigam, Chief Informatics Officer for Independence Blue Cross (IBX).

    IBX is working to identify customers most at risk, not just of being readmitted, but everyone sick or frail enough to be on the edge of hospitalization.

    To do so, they run algorithms on the huge amounts of health data at their disposal: billing claims, lab readings, medications, height, weight, family history. Plus, they throw in information about neighborhoods, including poverty rates.

    “The health care data we provided to build these algorithms is equivalent, I think, to five Wikipedias,” says Nigam.

    The computer algorithm sifts through all that information and pops out a score, identifying those highest at risk.

    IBX assigns these high scorers a staff member, called a ‘health coach,’ who then works with patients to see what extra services may be helpful.

    “This coordinated effort then works for the patient to try to get them the right level of education, physician appointments, medication issues, perhaps transportation benefits to go to a doctor,” explains Nigam.

    “Even home care nurses, who can work with the member. And all of that is beginning to show a pretty significant drop in hospitalization rates in our region.”

    Of those identified at highest risk, IBX says there’s been a 40-50 percent reduction in expected hospital admission rates.

    That statistic includes the Iovines.

    Picture Carol’s situation: she’s managing new medications, helping John shower, get to the toilet. They need to hire a wheelchair van for each appointment and therapy session, of which there are many.

    She could use some help, and she gets it, in the form of a letter from her new health coach.

    “It is dated May 1st, and it came May the 8th, and I called the number, and it was Donna Crockett, and boy did we bond right off the bat. I love her. She’s our angel, you know. We call her our angel. And anything that went on, she was right there for us.”

    That’s strong praise to lay on an insurance company employee. She offers an example.

    “Oh, he was supposed to get blood work, and they wanted me to take him to the ER to get blood work. Uh uh, I said, no way…called Donna, ‘they want me to take him there.’ And the next thing, a nurse was here taking blood,” says Carol.

    Big picture, the pennies spent on Crockett arranging a visiting nurse or streamlining appointments are nothing compared to the dollars a hospital readmission would cost. 

    Writing the rules 

    That promise of savings has others paying attention to the power, and possible drawbacks, of these predictive computer formulas.

    “I think there is a lot of interest in the area right now, and it is a great coming together of the healthcare world and the computer science world, as well as the patient experience world,” says I. Glenn Cohen, a professor at Harvard Law School, who has written about the legal and ethical concerns raised by the collision of health care and big data.

    He believes there are still gray areas, though, about ownership and control of the information.

    “Here there are questions of whether people whose data is going to be used to build the engine have the right to opt out, do they have to affirmatively opt in? Do they have to even be notified it’s being used?”

    Cohen says this whole field is too new for there to be accepted standards for how this information should be handled. For its part, IBX says it follows HIPAA guidelines regarding anonymity, and is only using the information to better serve its members. But it doesn’t ask people to opt in.

    “The data is only used to improve or coordinate care. And that is something that you would agree is our role,” says Somesh Nigam.

    IBX says early results from its hospitalization efforts are so promising that the company is expanding its predictive analytics program. It is partnering with NYU Langone Medical Center on its next target, Type 2 diabetes, trying to spot those most at risk before they start showing symptoms. And then, trying to do something about it. 

    Disclosure: Independence Blue Cross is an underwriter of The Pulse. 

    Want a digest of WHYY’s programs, events & stories? Sign up for our weekly newsletter.

    It will take 126,000 members this year for great news and programs to thrive. Help us get to 100% of the goal.