A funny thing happened on the way to the first Diagnostic Errors in Medicine Conference, from which Bob Wachter’s thoughtful commentary in the September issue of Health Affairs is drawn.
A group of us struggling to “get respect” for the problem of diagnosis errors conspired to bring together several important streams of work in order to attract attention to this neglected problem and make progress in its solution. Our planning committee members (including Mark Graber, Eta Berner and Pat Crosskerry) decided that two important sources of work needed to be brought together: advances from the cognitive sciences and lessons from the emerging field of patient safety research. We selected keynoters to represent each of these two paradigms: Art Elstein, a psychologist whose career has been dedicated to studying and improving how physicians make diagnostic decisions, and Bob Wachter, who had established the AHRQ patient safety web site and was an articulate proponent of systems solutions to patient safety.
When each gave their presentation, a remarkable shift happened. While this may be a bit of an over-simplification, I believe it is accurate to say that each speaker surprisingly ended up staking out the opposite territory from what we had expected. Instead of extolling cognitive approaches to reducing errors resulting from known psychological biases, Elstein argued that humans seem “hard-wired” to make various errors, stating that “overconfidence and premature closure seem to be deeply ingrained in the way we think.” Efforts to “de-bias” physicians or train them to be more aware of the limitations of their mental heuristics (so called metacognition-training) may not be terribly effective, Elstein said. He suggested that other fields might be more fertile to plow, such as systems redesign to prevent or minimize errors.
Meanwhile Wachter warned those gathered that diagnostic errors were somehow different and much less amenable to systems’ solutions. Owing to their cognitive nature, addressing errors with the type of simple technical fixes or system solutions that had worked for other types of medical errors was unlikely to succeed. Because physicians were making “highly nuanced, cognitively complex decisions” under “conditions of great uncertainty and time pressures,” there were no solutions analogous to the bar-coding and computerized physician order entry (CPOE) that had reduced medication errors. Adding this to the other “disadvantages” he enumerates in his expanded Health Affairs commentary, he argues that diagnostic errors need be viewed and tackled differently.
An unproductive dichotomy. This “systems vs. cognitive” polarity, so prevalent in analyses of error causation, strikes me as an highly artificial and false distinction. If I am an emergency room physician taking a history to sort out a patient’s back pain and fail to elicit a key piece of information or consider a critical diagnosis because I am interrupted by a cardiac arrest in the next bed, is this a cognitive or system error? If only the junior intern with insufficient experience sees a patient and fails to recognize subtle signs of serious necrotizing skin and soft tissue infection, and thus fails to make this life-threatenng diagnosis, is it her brain or the oversight system this produced the error? If I fail to consider hypothyroidism in my differential diagnosis, but would have surely made the diagnosis if I was hit over the head with a recent abnormal thyroid test result sitting in a lab computer that does not connect to my electronic medical record (EMR), is it my thinking or the computers’ interoperability that is deficient? I have seen each of these errors on multiple occasions and am convinced not only that we can not tease these “cognitive” and “systems” factors apart, but that solutions lie only in recognizing their interconnectedness.
That is why we recently advocated for a new emphasis on leveraging the capabilities of health IT to aid clinicians; electronic documentation functions are particularly important as a promising but insufficiently developed foundation to support more reliable diagnosis. Electronic clinical documentation has the potential to address multiple weaknesses in our current approach to diagnosis accuracy and reliability, including poorly wired (and often nonexistent) feedback loops; overreliance on human memory; barriers to timely and efficient collaboration (among care teams and between PCPs and specialists); and failure to adequately order, track, and follow-up on diagnostic tests.
However, as Wachter correctly points out, there is little empirical evidence to judge either the value of — or the best approaches to — supporting electronic documentation functions. And as he has written elsewhere, implementation is harder than it looks and unintended consequences need to be considered.
Getting to true “meaningful use.” Should evidence of the effectiveness of electronic documentation in reducing diagnostic errors emerge, Wachter urges incorporating any proven features of health IT into “meaningful use” regulations. I should probably be excited that the government is finally cracking the whip to get vendors and purchasers in line with meaningful benchmarks, especially before frivolously doling out carrots to reward clinicians and hospitals for purchasing EHRs.
However, there is something unsettling in both Wachter’s and the Obama administration’s approach to ratcheting up quality and accountability in this way. Do I need to be rewarded for using Google, or accountable to regulators regarding what percentages of my searches I do using Google? Do physicians need to be incentivized and scored on their use of cell phones or beepers, or for that matter on their use of Up-to-Date or Epocrates?
These are IT solutions that are for the most part thoughtfully designed to fill a need and efficiently fit into workflow. The fact that physicians have been slow to adopt proposed “meaningful uses” of existing health IT tools speaks more about the weakness and limitations of these products than physicians’ putative backwardness. Perhaps it would be more cost effective to use the scores of billions of dollars to design more ideal systems, ones that more thoughtfully aided the diagnostic process, and place them in the public domain for clinicians to inexpensively adopt.
The neglected importance of empowering patients. Even more important than supporting clinicians in diagnosis-related functions is recognizing and supporting patients in their role as co-producers of diagnosis. This largely unexplored aspect of diagnosis improvement represents another essential pillar in making diagnoses more timely, accurate, reliable, and efficient. Patients can make valuable contributions by seeking timely access for worrisome symptoms; providing accurate and thorough histories; sharing their hunches about possible exposures or etiologies; helping ensure that test results are reported back; following up with feedback about expected improvement; adhering to empirical treatment trials permitting accurate re-assessment of preliminary diagnoses; respecting limits on staff time and societal resources; and in some cases even getting involved with disease-specific or generic patient safety advocacy organizations.
In contemplating such patient contributions to timely and accurate diagnosis, it is important that we not shift our responsibilities onto sick patients. Instead we need to imagine and facilitate an enhanced role for patients to play. The key question is: what will it take at the provider and institutional level to empower patients to take on these roles and help them flourish in them?
The journey continues: the upcoming third annual conference on diagnostic errors. Let’s hope we don’t have to wait for the “high-profile errors” Wachter cites to kill more patients for diagnostic errors to get their due respect. Enough “low-profile” diagnostic delays, needless morbidity, inefficiencies, missed diagnoses, and even deaths are already happening every day. Hopefully the next highly profiled event will instead be the upcoming 3rd Annual AHRQ-sponsored International Conference on Diagnosis Errors in Medicine next month in Toronto. Here we will again bring together engaged clinicians, researchers, academics, quality improvement and risk management professionals and patients—all dedicated to a common goal—to better understand and reduce medical errors and to reduce the harm from missed and delayed diagnoses.
Thanks, Bob, for being there with us at the launch; I hope many others will attend this year to further continue this important journey.