Big data shows the promise to transform how health care is delivered, utilizing real-time information to assist doctors and patients in making more informed decisions at the point of care, a transformation that could arrive in the next 5 years.

That time prediction was offered by Dr. Harlan Krumholz, a professor at Yale University, New Haven, Conn.

Big data could eventually use real-time information to help treat patients in a clinical setting, but the technology still needs work, Dr. Krumholz said.

"We are on the cusp of dramatic change in the way in which we are thinking about leveraging data that’s being generated in everyday clinical practice and in everyday daily life," Dr. Krumholz said in an interview. "I think there’s this recognition that the current clinical scientific enterprise is inadequate to keep up with information needs."

But getting to that point is going to require a shift in the research culture and how that information is brought to doctors as they are treating patients.

"One of the biggest problems with our current research approach is it’s very reductionist," Dr. Krumholz observed. "We try to pursue the scientific method with testing hypotheses. We try to pull out all the complexity around the question we are trying to answer and simplify everything so we can cleanly look at a particular effect of one thing on patients. ... By the time [research results] come out, they are very anachronistic because medicine’s progressed and the patients that they’re studying aren’t exactly the people who are being seen in practice, so the studies aren’t necessarily having the effect they could have."

Dr. Krumholz also is a member of the Patient-Centered Outcomes Research Institute board of governors. The organization is building a research network that will facilitate the gathering and usage of big data to help inform patient decisions.

But it’s more than just health-specific data that need to be incorporated into big data analysis.

"It’s not just the pill, but it’s the pill for a specific patient who’s got a specific profile, who is in a certain social situation, who is taking other medication, maybe has certain access to physicians, and a lot of complicated issues coming together," Dr. Krumholz continued. "We need to be able to learn from everyday experience and we need to embrace the complexity, not reject it, so that our studies are taking into account the complex aspects of medicine, not trying to get rid of them so we have clean studies that may not well-relate back to the real-world situations doctors and patients face everyday."

The next step will be to find something that can truly demonstrate the power of big data.

"I think that once we break the dam open and allow the water to flow and turn the turbines of knowledge generation, people are going to demand better and better information in order to be in a stronger position to make decisions," he predicted. "I think the decision support enterprise, that community of people who are working on decision support tools, is going to start competing to develop and create these things that are going to be at the bedside, that is going to make a whole lot of difference from where we are now."

A number of pilots and areas where big data could have an immediate impact were highlighted in a recent article in the July issue of Health Affairs (doi: 10.1377/hlthaff.2014.0041). For example, Kaiser Permanente Northern California; Harvard University, Boston; and the University of California, San Francisco and Santa Cruz, are pilot testing a two-step protocol aimed at reducing antibiotic prescriptions for newborns by using objective maternal data to determine a preliminary probability for early-onset sepsis. The second step uses a set of clinical findings combined with the estimate based on maternal data to yield a new posterior probability for risk of sepsis following birth. Kaiser anticipates that the combination of these two steps could lead to as many as 240,000 fewer U.S. newborns being treated with systemic antibiotics each year.

Lead author Dr. David Bates, chief of the division of general medicine at Brigham and Women’s Hospital, Boston, and others identified several key areas where big data could have an effect at the point of care, including identifying high-cost patients to more efficiently care for them; helping to reduce readmissions; lowering the risk of complications when a patient is first admitted to a hospital; determining the risk that a patient’s condition will worsen; and understanding the risk of adverse events for patients.

"The better we are at getting to good information about options and consequences [regarding] who’s about to get into trouble, the more people are going to demand [the use of big data,] and the race will be on to create the products that are underneath the hood that are leveraging large data," Dr. Krumholz said.

gtwachtman@frontlinemedcom.com

Big data destined for the bedside within 5 years