When we go to see a doctor with an ailment or symptoms of any kind, the first thing we want to know is the diagnosis of our condition — what is it that I have? An accurate diagnosis is the first step in finding the right treatment but it is not always given at the point of care as patients may hope.
In March, independent healthcare research organization ECRI Institute published its 2018 patient safety report where diagnostic errors ranks as the №1 concern, outpacing opioid safety and internal care coordination. In fact, approximately 18 million diagnostic errors occur every year and nearly every person will experience one in their lifetime, making diagnostic errors an understated medical crisis.
Unfortunately, general medicine practitioners are the most likely to make diagnostic errors because they are usually the first step in providing care for patients and it is impossible to be an expert across all of medicine. For instance, generalists receive an average of 21 hours of training in dermatology, but they see more than 65% of skin conditions. The inevitable consequence is that an astonishing 48% of skin presenting diagnoses made by non-dermatologists are incorrect.
For example, recently a middle-aged woman requested an appointment with an urgent care physician about a red rash on her leg. The urgent care physician thought it was early cellulitis and prescribed oral cephalosporin to treat a presumed staph skin infection. The patient actually had Lyme disease which requires Doxycycline. This is an example of premature closure, one type of cognitive bias.
Why Doctors Make Diagnostic Errors — Cognitive Bias and Efficiency-Thoroughness Trade-Off
Many arguments have been made in defense of doctors for misdiagnosing conditions, chief among them being “we are all humans and humans make mistake.” It is certainly true that we all make mistakes and I don’t think anyone’s expecting the impossible out of doctors, but is there something we can do to avoid making quite as many mistakes especially in medical practice, a field where mistakes often are irreversible and can do great damage to patients?
First things first, we need to find out the cause of diagnostic errors. When making a diagnosis at the point of care, doctors look for classic presentations of the disease instead of the variations, but disease symptoms can vary drastically from one patient to another. While four years of medical school and three years of residency might seem a very long time, it is simply not enough to master the broad, ever-evolving study of medicine considering the volume of knowledge in the field. Future doctors most likely focus on learning the representative, most common presentations of the diagnoses and have little opportunity to explore the rare, and unusual presentations.
Skin rashes are usually attributed to allergic reactions, forms of dermatitis, tinea, psoriasis and common infectious diseases such as herpes zoster, but at times a rash can be a symptom of a life-threating disease. A young, healthy patient visited an ER in Texas with a chief complaint of a chronic rash which surprisingly turned out to be a rare presentation of a disseminated gonococcal infection (DGI), also known as gonococcemia or arthritis-dermatitis syndrome. Prior to the ER visit, she had seen at least three care providers and all of them failed to provide the accurate diagnosis, causing increasing pain and a worsening rash for the patient.
Additionally, doctors are often caught in the dilemma of efficiency-thoroughness trade-off. According to the 2018 Medscape National Physician Burnout & Depression Report, 42% of doctors are experiencing burnout and in practices such as emergency medicine, the burnout rate is close to 60%. This is an issue largely caused by the overwhelming workload on both the administrative side and in the clinic.
Doctors, in an attempt to see more patients and improve efficiency in their practice, frequently use heuristics in making diagnoses at the point of care. Although heuristics have proven efficient and sufficient under most circumstances, the empiricism-based decision-making approach doesn’t always yield the optimal result.
The list of reasons for misdiagnosis does not stop here. As ECRI Institute’s Patient Safety Analyst Gail M. Horvath pointed out,
“Diagnostic errors is a multifactorial problem.”
But to get one step closer to diagnostic accuracy, we need to first tackle the factors discussed earlier.
How We Improve Diagnostic Accuracy: Man + Machine = Augmented Intelligence
Artificial intelligence is both a new buzzword and something truly exciting. But does that mean machines can replace human doctors and solve every problem we have? While AI is certainly revolutionizing medicine, I’m afraid it’s not the ultimate solution for the top patient safety concern.
Machine learning lacks contextualized knowledge regarding a patient’s specific case history and faces challenges in communicating diagnosis, not to mention the accountability for patient safety. The industry has been amplifying the competition between human intelligence and artificial intelligence, whereas the answer lies right in the middle.
Combining machine’s unparalleled thoroughness and precision in detecting problems and man’s profound knowledge and empirical practice in solving problems, we’re looking at a different kind of intelligence — augmented intelligence. Machine learning, when applied appropriately, can free up doctors from administrative work and reduce the cognitive burden in clinical practice.
In the past year or two, we are beginning to see a growing number of doctors adopting clinical decision support (CDS) tools to get a second opinion on diagnoses, taking a step further in ensuring patient safety. These tools can be integrated into the clinical workflow and connect with patients’ existing health records for a more thorough look into diagnostic possibilities.
Diagnostic error has moved to the center stage as a critical patient safety issue and is a matter of paramount importance for every caregiver. Doctors need to use computer-based knowledge to expand their cognitive reach and to bring evidence to their decisions. Clinical decision support is not the only solution to the problem but can largely augment doctors’ cognitive capability and assist in making diagnostic decisions at the point of care.