When John Halamka, M.D., CIO of Beth Israel Deaconess Medical Center in Boston, and his wife, Kathy, found out she had stage 3 breast cancer in 2011, they turned to big data analytics in healthcare to find the best treatment plan.
Fortunately for Halamka and his wife, the Boston area is home to 17 Harvard-University-affiliated hospitals, including Beth Israel, that have opened their data for queries via a free open source, Web-based application called i2b2. Any medical record system in the country can connect to i2b2's database and EHR, Halamka said.
However, in most places in the United States, such capabilities -- querying a network of hospitals' databases to learn which care paths prove most successful -- are not possible yet.
"Harvard is somewhat unique because the National Institutes of Health did fund the Clinical [and] Translational Science awards a few years ago, and part of that meant that the 17 Harvard hospitals made their data query-able," Halamka said. I couldn't ask about, 'Oh what was this person's condition?' but as a population I can ask questions like, 'With this diagnosis and this age and this gender, what was the outcome?' Things of that nature that wouldn't compromise an individual's privacy."
John Halamka, M.D.CIO, Beth Israel Deaconess Medical Center
Halamka explains that with i2b2, questions come in and answers go out but no data actually leaves the queried hospital's firewall and no patient privacy is compromised.
He queried the databases to learn about treatment plans patients similar to his wife -- at the time, a 50-year-old Korean female with stage 3 breast cancer -- underwent and what the outcomes were. He talks more in the accompanying video about the options they considered.
Halamka believes that new discoveries concerning treatments or drugs should be shared with the rest of the country's healthcare organizations and hospitals instantaneously. The use of i2b2 is a step in the right direction, he said. Taking advantage of big data in healthcare not only helped to deliver precision medicine to his wife and cure her of breast cancer, it also helped his family make more informed decisions.
"Big data really helps us understand our options," Halamka said. "So sure, she probably would've had a similar outcome but it was so much easier to understand as a family what our choices were and make rational choices because we had data and examples of other patients like us."
Transcript - John Halamka uses big data analytics in healthcare to fight wife's cancer
Kristen Lee: This is John Halamka, CIO at Beth Israel Deaconess Medical Center, health IT expert, mushroom expert, farmer, and beer and wine maker. Halamka, who has been a CIO at Beth Israel for 18 years, invited me to his office and home to not only talk about matters in health IT, but also to get to know about him more as a person. He and his wife Cathy own and run a 15-acre farm in Sherborn, Mass where they take care of geese, guinea fowl, ducks, chickens, alpacas, one llama, two Great Pyrenees dogs, 250,000 bees, and 55 different kinds of apples. He told me he only sleeps about two to three hours a night and that farming allows him to use his CIO and technology skills in a different way. He also told me how big data was to help his wife when she was diagnosed with stage three breast cancer.
John Halamka: So in December of 2013 my wife was diagnosed with stage 3A breast cancer that had genomic markers, HER2 negative, estrogen positive, progesterone positive, at the time she was 50, she's Korean. So, of course, in America where we have 330 million people, you should be able to push a button and be able to say, "Of the last 10,000 50-year old Korean females with stage 3 A breast cancer, with these genomics, what were the treatments and what were the outcomes?" Although that's an aspirational goal, it turns out in the United States, that's not really possible in most places.
Harvard is somewhat unique because the National Institute of Health did fund the Clinical Translational and Science Awards a few years ago, and part of that meant that the 17 Harvard hospitals made their data queriable -- not by person. I couldn't ask about it or, "What was this person's condition?" But as a population, I can ask questions like, "With this diagnosis and this age and this gender, what was the outcome?" Things of that nature that wouldn't compromise an individual's privacy.
Lee: The Harvard-affiliated hospitals in Boston came together to create a web-based application called I2B2. It's a free downloadable open source tool that any medical record system in the country can connect to their database, as well as to their EHR.
Halamka: Questions come in and answers go back, so no data actually leaves the firewall. If I were to say, "How many patients with stage 3A breast cancer do you have, hospital?" The hospital reports "42." There is no privacy compromise and no data that leaves. It's just an answer to a question. And that's I2B2, which is today used in over a hundred healthcare organizations in the United States.
Lee: Halamka used I2B2 to query the hospital databases connected to the tool and learned about treatment plans patients with similar cases to his wife's underwent and what the outcomes were. One medication that many patients in similar situations took was Taxol, which can cause neuropathy. Halamka told me his wife is a visual artist and therefore needs fine motor movements with her hands.
Halamka: So that means you might have cancer cured but your hands will be dead. And so her issue was, "How do I get cure but at the same time retain function?" And have there been clinical trials done in 50-year-old Korean women as to the right dose of Taxol? No. So what you could say is there's been clinical trial, which says nine doses of Taxol is a reasonable number, but no one's tried four, so we did. And so with her therapy she got the best practice protocol for women of her kind, but we reduced the Taxol dose because her precision medicine was to regain function in her hands while also achieving cure, and it all worked.
Lee: It's important that healthcare organizations learn from each other. But that is a tall order in the United States today, Halamka told me. For example, Halamka said a discovery at Mass General Hospital today would take 20 years to reach a hospital, in say, South Dakota.
Halamka: Shouldn't it be instantaneous? If we discover that there's a new drug, Vioxxx, and Vioxx causes heart attacks and stroke, shouldn't we be able to discontinue the use of Vioxx within a day across every hospital in America? So, unfortunately, because electronic health records are mostly silos of data created for episodic care within an institution, they don't have these learning characteristics.
Lee: In Halamka's opinion, the sharing nature of I2B2 is a step in the right direction and big data in general not only helps the patient but also the patient's family and loved ones.
Halamka: So the big data really helped us understand our options. It was so much easier to understand as a family what our choices were and make rational choices because we had data and examples from other patients like us.