In late September 2014 a man showing symptoms of the Ebola virus came into the United States from West Africa, was examined by doctors at a hospital in Dallas — and was then released back into the community. There he came into contact with other people before finally being readmitted to the hospital, where he died within a matter of days.
If the entry of the Ebola virus into the US wasn’t itself shocking enough, one doctor claims that this will not be the last time such an error occurs — due to what she calls “a simmering crisis in medical data management.”
Writing in the New York Times on October 14, Dr. Abigal Zuger warns, “Even scarier than that mistake is the certainty that similar ones lie in wait for all of us who cope with medical information stored in digital piles grown so gigantic, unwieldy, and unreadable that sometimes we wind up working with no information at all.” (My emphasis).
Dr. Zuger claims that “hospital servers store great masses of trivia mixed with valuable information and gross misinformation, all cut and pasted and endlessly reiterated. Even the best software is no match for the accumulation. When we need facts, we swoop over the surface like sea gulls over landfill, pick out what we can, and flap on. There is no time to dig and, even worse, no time to do what we were trained to do.”
Elsewhere I’ve described a condition called data blindness, and this sounds like a textbook example of it. This occurs when there is so much data that there is neither time nor inclination to analyze it and convert it into productive sense-making. It’s not just a medical records problem, every type of organization faces it — it’s just that in a hospital, it can literally be a life-and-death problem.
Dr. Zuger goes on to imply that the answer to this is to have more patient-doctor conversations that give a richer, ultimately more accurate picture of a patent’s condition and medical history than would check boxes on a data input screen. In these conversations, the doctor would take notes as an aide-memoire, rather than just poring over screens of data. Inquiries would be interactive and iterative, and narrative threads could be pursued.
I cannot deny the appeal of the classic picture of the professional in deep and productive discussion with his or her client, unfettered by the pressures of time, money, and data collection.
That said, I cannot honestly say I think that richer professional-client interaction is a solution to the medical records problem — mainly because I fear those pressures are here to stay. Most likely they will be joined by new and even greater pressures for performance standardization, accountability, and cost-effectiveness.
The use of checklists, for example, has been described as one of the pillars on which the modern practice of medicine will redeem itself. This comes from Atul Gawande, who is himself a surgeon (not just a sideline critic like this humble scribe). Gawande has commented that the solution to many of the problems of modern health care is to make its practices even more like those of an industrial organization — a modern business.
The solution is better quality data, and better use of the data we have. And data systems and processes that are flexible and adaptive enough to respond to rapidly emerging conditions like a potential viral outbreak — and can even anticipate future such occurrences.
The glossy TV and print ads speak with awe of the miracles that await us with Big Data. Yet in many cases (like medical records) that is at best an aspirational goal, completely at odds with current the on-the-ground reality.
In other cases, Big Data is simply a pipe dream. The size of the data is fine, thanks — it just doesn’t work right.
We need data that works for us, solving our problems.