What should doctors do to make medicine safer? The numbers of deaths from errors, by most accounts, are not improving. The medical establishment bared its soul in a recent issue of the Journal of the American Medical Association. Instead of original research papers it consisted almost entirely of 22 opinion pieces written by a heterodox group of authors about controversies related to achieving the goal of doing no harm.
There was emphasis on a recent flap over how much didactic preparation, i.e., book, journal and lecture learning, doctors need. The American Board of Internal Medicine, by its own admission, had overstepped its bounds. Internists and all other specialists used to get certified for life after completing a residency and taking rigorous exams. Although it began with a ten year recertification cycle in 1990, the internal medicine board in 2014 ramped up its requirements to a two year cycle, a relentless, continuous process of didactic learning that took up huge amounts of time and money on the part of the board’s diplomates in order for them to be recertified in their specialty. Thousands of diplomates balked and wrested an apology from the board’s leader. The certified internists pointed to several pieces of scientific evidence that this type of learning did not improve patient outcomes.
The urge to maximize book learning derives from concerns for patient safety enunciated almost 16 years ago in To Err is Human, a well-meaning book about how medical errors happen compiled by the Institute of Medicine, now known as the National Academy of Medicine. Its findings were weakly supported by scientific evidence. Certainly few things in the profession of medicine should demand more of our attention than how to limit the number of harmful errors. However, the number of scientifically valid papers on the subject is small.
Recent research should steer the profession in another direction than trying to fill doctors’ heads with more information. It shows that the worst kind of doctor’s mistake, the one that produces the highest incidence of serious harm is more directly related to deficiencies in observation rather than in knowledge. It is the diagnostic error. The technical error, such as injuring a normal structure during surgery, is more common. However it can often be corrected, either on the spot or during a subsequent operation. Diagnostic errors often lead to irreversible failures to treat the patient.
Two years ago, Dr. Hardeep Singh and his associates at Baylor University told us that among primary physicians, the most common sources of mistakes, 79 percent of them, were taking an inadequate history, missing findings on physical exam and ordering the wrong tests, the blocking and tackling of the profession. Singh et al’s paper was an echo of the great Hopkins physician of a century ago, William Osler, who repeatedly told us that book learning was important but nothing took the place of careful observation both for healing the patient in front of us and for life-long learning. He said that for every error that came from not knowing, we make ten of not observing.
Singh and his co-authors eschewed the use of differential diagnoses—a hierarchical list of possibilities-- generated by computer programs which are sometimes added to the software that generates electronic health records. They concluded that no technology exists to help us reduce the number of diagnostic misses. They lamented that in many of the cases where the diagnosis was missed the physician had failed to write out a differential diagnosis of his or her own. This implies strongly that people still do this better than computers.
There is an aspect of the medical error problem that the JAMA special issue did not explore. Even rigorous investigators of errors like Singh et al did not tell us was whether there were outlier physicians who missed things much more frequently than their colleagues, thereby adding disproportionately to the total of faulty evaluations. If this were so, it would be consistent with research in the United States that shows that 2% of physicians are responsible for 50% of payouts in medical malpractice cases. Only 12.3 percent of the 2 percent have ever had even one adverse licensure action. Only 6.2 percent have ever had even one adverse action taken by the physician governance of their own hospital. In a parallel study done in Australia, researchers found that 3% of their doctors prompted 49% of complaints to a central agency.
There was an essay among the 22 in that special issue in which the executive directors of the AMA bemoaned that medicine had gotten wrong the performance incentives for doctors. The mistake, their essay implied, was to acquiesce to paying doctors by the procedure or examination. There is, of course, a considerable body of thought that says that this rewards physicians for volume, thereby encouraging unnecessary and questionable procedures. For doctors whose primary responsibility is to diagnose, this also means less time for patients, increasing the risk that examinations will be too brief and miss important findings.
The AMA directors cited a survey of their members that showed that American physicians were mostly motivated by wanting to do the right thing for their patients. This is not hard for me to believe. The vast majority of the physicians I know are ethical, careful and conscientious. The other side of the coin of the 2 or 3 percent being responsible for half the problems, is that the other 97 or 98 percent make mistakes at a much lower rate, around one-fiftieth. One of the great difficulties in making medicine safer has always been in trying to identify and weed out the outliers while not making life miserable for the good doctors who only occasionally err and are the pillars of the profession. Thus far we are failing in that responsibility.