08.07.2015
by Craig Knoche, MS, MBA, and Judith Kalinyak MD, PhD
“First do no harm,” a phrase found in Epidemics, Book I, of the Hippocratic school, is a core principle in the education of medical students and is the belief of the patients being treated by their physician. Unfortunately, misdiagnosis results in an estimated 40,000 to 80,000 U.S. hospital deaths annually, and approximately 5% of autopsies identify lethal diagnostic errors for which a correct diagnosis with proper treatment could have averted the death. In the outpatient setting, one in every 20 U.S. adult patients is misdiagnosed every year. Half of these misdiagnoses result in some form of morbidity. Upon further analysis, deficiencies in patient assessment and diagnostic reasoning account for 17% of all adverse medical events.
These failures in healthcare delivery not only cost lives, but also money. In 2011, anestimated $102 billion to $154 billion was alleged to be due to diagnostic failures. All of these statistics suggest significant shortcomings in our current physician training process, and highlight the need for improvement in clinical education and evaluation of diagnostic reasoning. So, how is diagnostic reasoning learned, and how is our educational process addressing the shortcomings mentioned above?
Diagnostic reasoning has traditionally been taught by apprenticeship through a series of conversations surrounding patient encounters that are intended to reveal the complex clinical reasoning that forms the basis of decisions made by experienced clinicians. Much of this instruction occurs indirectly as there is rarely explicit instruction in or description of the diagnostic process. In his book “Outliers,” Malcolm Gladwell argues that it takes roughly 10,000 hours of practice to achieve mastery in a field.
However, this traditional clinical apprenticeship approach has serious limitations:
• ACGME resident work hour limitations have resulted in a 56% decrease in duty hours and resident-student contact time as well as a 40% decrease in student observation of clinical history and physical examinations. These limitations aim to avoid sleep-deprived residents from caring for patients, but they also reduce resident-student mentoring. For example, consider a new sub-intern who joins a medicine team post-call and has to report to the intern since the resident is off duty due to hour restrictions. Although the student is still in a learning situation, the mastery level of the mentor is lower.
• The random array of patients that students encounter may not provide them with the optimal mix of clinical material to develop the necessary breadth of knowledge or experience with even common clinical presentations. This problem is emphasized when medical schools use a variety of sites for clerkship rotations. Since clinical problem-solving skills are at least to some degree content-specific, this lack of clinical exposure may lead to serious deficiencies in evaluating even simple clinical presentations. Clerkship programs are now looking to objective structured clinical examination (OSCE) simulations to rectify these deficiencies, but they can be expensive and difficult to schedule.
• Students and residents often have short shifts and rotations, and may not have the opportunity to observe the outcomes of patients they have evaluated and treated. This is exacerbated by a decrease in patient availability and diversity because there are fewer hospital admissions for diagnostic evaluation than for treatment, and the average length of stay in U.S. hospitals has declined from an average of 7.8 days in 1970 to 4.6 days in 2013. In the absence of follow-up, clinicians are prone to assume their diagnostic and therapeutic decisions were correct, potentially fostering overconfidence.
• Finally, many clinician-educators are expert diagnosticians, but lack the training and skill to explicitly outline their thought processes. These physicians may quickly recognize complex patterns and subtle clues but have difficulty articulating the complex cognitive processes that led them to their conclusions. But the medical community is taking steps to combat this issue. The American Medical Association has established a broad initiative to accelerate change in medical education. Dr. Scott Stern and Drs. Rencic, Durning, and Trowbridge have published new textbooks — “Symptom-to-Diagnosis: An Evidence-Based Guide, 3rd Edition” and “Teaching Clinical Reasoning,” respectively — that focus on clinical reasoning and incorporate more diagnostic reasoning into medical school education. Additionally, many college of medicine educators are collaborating to create computer-simulated case scenarios to further facilitate self-directed learning among students at all levels of medical education.
Another major factor contributing to diagnostic reasoning errors might be the focus medical schools still place on diseases and disease progression with insufficient time spent on identifying and diagnosing the presenting clinical symptom. The disease focus in medical education is reinforced in most medical textbooks, which frequently use a body systems or disease process approach.
With this traditional approach, how do students learn diagnostic reasoning? How do they learn what types of disease present with certain symptoms, so they start with a broad differential diagnosis that minimizes diagnostic errors of omission? How do they learn the pivotal questions that can narrow the differential? Where do they learn how to search for disease-specific clues, risk factors, associated symptoms, and signs to further modify the likelihood of those diseases? How do they learn a cost-effective and efficient testing strategy to confirm their diagnostic suspicions?
Do medical schools spend enough time discussing test accuracy and test interpretation, particularly when tests are discordant with the clinical assessment? Although more clinical skills and diagnostic reasoning are being introduced in the first two years of medical education, pressure to restrict contact hours to less than 25 hours per week makes any expansion of clinical time challenging.
Medical school curricula have long focused on diseases, and especially the classical presentation of diseases, to the detriment of students. Dr. Scott Stern, in writing his book “Symptom to Diagnosis,” found that in 101 diseases only 20% of 337 “classical findings” were more than 80% sensitive. The result is that students are being taught to look for archetypal signs and symptoms of diseases that in fact patients rarely exhibit. This may explain some of the errors being made in diagnosis or clinical reasoning.
Differential diagnosis development and clinical reasoning are competencies that, if taught and assessed correctly, could reduce the number of diagnostic errors. But how can medical schools ensure their students are equipped with these skills when they come face to face with a real, live patient for the first time and have to start asking questions?
It requires the combination of knowledge with clinical diagnostic skills in the education process. After all, it isn’t physicians’ encyclopedic knowledge of afflictions that serve them in treating patients, but their ability to play Sherlock Holmes, ask the right questions, interpret physical exam findings, order only the right tests, and put the findings together into the correct diagnosis. These critical thinking skills are the core of medical education reform targeted to meet the U.S.’s expanding need for health care with a brigade of doctors who have finely honed their diagnostic reasoning skills.
Medical educators are renewing their focus on the history and physical exam, long advocated for by Dr. Abraham Verghese as a way to determine which tests to order for a patient. Communication skills must be emphasized — many physicians interrupt patientsonly 23 seconds into their description of their ailment, and one study found that 37% of physicians made no inquiry of the patient’s complaint at all in the first five minutes of the visit. How can these physicians diagnose what they don’t understand?
Medical schools across the country are beginning to incorporate online patient simulation into their curricula to overcome many of the challenges associated with traditional medical education, and allow students to supplement their classroom and exam room experience with self-directed learning. Recent studies demonstrate that symptoms-based e-learning modules can improve students’ OSCE performance.
Student testing and assessments will also have to change, shifting from rote memorization to more dynamic patient encounters in a clinical setting, whether through standardized patients, simulation, or other means.
All of these efforts help young doctors hone their history taking and diagnostic reasoning. Mastery takes practice — 10,000 hours of practice, if we believe Malcolm Gladwell. For medical students to become competent physicians able to quickly formulate accurate diagnoses, order appropriate tests in a cost-efficient manner, and interpret the results with accuracy, they will need to begin their practice long before they see their first real patient. To accomplish this, medical schools will need to expand their educational and student assessment approach to focus on the application of clinical knowledge, use of diagnostic skills, and interpretation of laboratory tests to better understand and monitor students’ clinical reasoning skills. As medical knowledge continues to expand and therapies become more personalized and complex, more effort and time will be required to master the art of medicine.
Craig Knoche, MS, MBA, is president of i-Human Patients, a company that provides virtual simulated patient encounters for medical education. Judith Kalinyak, MD, PhD, is the company’s director of clinical curriculum integration.