|
More on Biases
Lisa Rosenbaum, M. D., has written a multi-faceted analysis of bias as it impacts industry-physician relationships.[1] In these articles she makes several points that are relevant to all physicians and the practice of medicine. As the work of social psychologist Robert Zajoine helped establish, feeling precedes cognition, rather than vice versa. Even when we think we are thinking, almost nothing we perceive is emotionally neutral. In a narrative vignette, Dr. Rosenbaum described the sorts of conflicts she encountered while being on-call as a cardiology fellow, where she was expected to triage calls from outside hospitals wanting to refer patients with ST segment elevation myocardial infarctions. Whether our judgments are motivated by fatigue, hunger, institutional norms, the diagnosis of the last patient we saw, or a memory of a patient who died, we all biased in countless subtle ways. Teasing out the relative effects of any of these other biases is nearly impossible…The difficulty of measuring these other motivations, however, creates the problem that plagues many quality-improvement efforts: we go after only what we can count. Finally, she notes There’s an old adage that if you haven’t done any negative appendectomies, you aren’t operating enough. Though high-sensitivity CT scans may render its application to appendectomy obsolete, the principle that an approach that increases the likelihood of benefit often also confers increased risk of harm remains pertinent. In evaluating interventions, particularly those that elicit strong emotional reactions, we tend to assume that risk and benefit move in opposite directions. Positive feelings toward an intervention can make us assume that a high likelihood of benefit means low risk. And when we find a risk particularly noxious, we may believe that eliminating it will inevitably increase benefit. We don’t evaluate trade-offs and then develop a feeling based on that analysis; our feelings guide our evaluations. I find all of these observations resonate with my own experience. Consider three recent patient consultations. The first was an elderly man who presented with an acute myocardial infarction and who also had valvular heart disease. His acute problem was managed in “standard” fashion. Since he was a functional individual, the cardiovascular surgeon was consulted. He was counseled about the risk and the benefit of surgery and decided he wanted to proceed rather than “just wait for something bad to happen.” Unfortunately, he developed a stroke and renal failure in the post-operative period, and so I was consulted. When I talked to his daughter, she had difficulty understanding that the nature of the question had changed. Something bad had happened, and now we had to decide how much more we should do. In her mind, though, she had decided the great benefit was associated with reduced risk. (The surgeon did not share that opinion.) Fortunately, he stabilized and did not require urgent dialysis. The second patient was a middle aged woman who was known to have diabetes, but who did not believe in going to the doctor and did not take any medications for it. She presented with a soft tissue infection and rapidly developed multi-organ dysfunction, including renal failure. The cause was not immediately apparent, but it seemed likely that it was cardiac in origin. The cardiologist thought a diagnostic left heart catheterization was indicated and was anxious to proceed. She met all of the “evidence-based” criteria for the procedure, but it was not clear what her attitude toward any of this would be, nor what we might do with the information if she was not planning on follow up, so I counseled delay. In the end she improved and was able to give informed consent to the procedure. The third patient was a young woman who had an acute myocardial infarction and developed cardiogenic shock. She was air-lifted to our hospital and was taken immediately to the catheterization laboratory in an effort to open her occluded vessel. Unfortunately, she had a series of unsuccessful interventions, and left the laboratory on a ventricular assist device. By the time I saw her she had been in refractory shock for more than 24 hours. The cardiologist was still at the bedside trying to figure out what else he could do. I talked to him later about ghosts and he admitted he could not shake the image of her five year old son from his mind as he looked at her. She died about 18 hours after I saw her. Did we provide the scientifically correct care to these three patients? I really don’t know. We did try to provide care that was likely to help, and tried to be alert to the possibility of bad outcomes, and we did try to explain all of these issues in ways the patients and their families could comprehend. But I am sure both the doctors and the patients and their families brought their biases to these encounters. I see this as a problem, though, only if we view the physician as an unemotional computer who applies the scientifically correct answer to get the best possible outcome. Perhaps the real issue is the notion that practicing medicine is a science. In a scientific study, the investigator attempts to control for all of the factors that might influence the result and to isolate the variable of interest. In clinical practice, patients present as discrete bundles of variables which are generally not modifiable and certainly not controllable. If we use the notion of statistical process control, which is the mathematical basis for quality improvement, we view the variable of interest as showing either “common cause” or “special cause” variation, depending upon how far an individual result is from the mean result. In this language, all of the biases, known and unknown, that undergird decision making at the bedside are subsumed under the “common cause” variation. Modifying common cause variation requires making major changes to the inputs, which is always difficult. Special cause variation might trigger a root cause analysis, but the conclusion may be that it was a chance event. In other words, deciding if there is common cause or special cause variation is mathematical. Deciding what, if anything, to do about it requires decision making by someone, which means bias is controlled only in the sense that once a decision is made, statistical process control methods can be used to see if the changes induced a special cause variation in the desired direction. This method, then, assumes bias is inevitable, and simple tries to account for it at a systemic level. Since we cannot be bias free, I think we should strive to understand our biases as clearly as possible, so we are free to explore the patient’s biases and craft a recommendation that reflects not only our understanding of the “evidence,” but also our understanding of what the patient desires. Unfortunately, doing this well requires time and often longitudinal relationships with our patients. The norm now seems to be short-term transactional encounters with neither side recognizing the subtle psychological underpinnings of the doctor-patient decision-making process. 29 May 2015 [1] Rosenbaum L. Reconnecting the Dots—Reinterpreting Industry-Physician Relations. N Engl J Med 2015;372(19):1860-64.; Rosenbaum L. Understanding Bias—The Case for Careful Study. N Engl J Med 2015;372(20):1959-1963; Rosenbaum L. Beyond Moral Outrage—Weighing the Trade-Offs of COI Regulation. N Engl J Med 2015;372(21):2064-2068. They are also available online at http://www.nejm.org/medical-articles/commentary. Accessed 29 May 2015. |
Further Reading
Activating Patients - The Achilles Heel of Healthcare Reform? Studies show 25% of the population is not involved in their healthcare, but reform efforts assume wider application of evidence-based medicine is the key to better value. Perhaps it is the Achilles' Heel? Are We Too Task Oriented? The number of tasks doctors must complete grows exponentially. Have we become too task oriented at the expense of our patients? Medical Evidence Medical evidence is a four-source: guidelines, registries, data mining and " in my experience". Different clinical situations use different types of evidence and have different implications for provider behavior. These implications are considered in detail. Patient-Centered Care A consideration of the interactions of patient preferences, evidence-based medicine and peer review. The Anchoring Heuristic Businessmen and health policy experts fail to recognize the limits imposed by the experiential nature of medical practice, both of which impact achieving the "triple aim." |