Medical decision-making and bias
Key Points: Bias infiltrates every step of a clinician’s approach to diagnosis and treatment. A sensitivity to this bias can lead to a greater ability to consider other possibilities to affect decision-making and move toward determining the cause. Doctor-patient collaboration builds a bridge to understanding and reducing the affects of bias on managing health.
Case I: A 64 yo male is seen in the emergency room for lower abdominal pain. He stated that he noticed when he woke up to get out of bed. The ER physician raised the concern of possible diverticulitis and orders a computed tomography (CT) study. He receives antibiotics for diverticulitis after a “soft call” on CT after the suspicion was conveyed from the doctor to the radiologist beforehand. The radiologist subsequently dictates “there is the possibility of early diverticulitis”. On a follow-up clinic visit, after a thorough history and physical, a groin pull was confirmed on the exam and the antibiotics are stopped.
Case II: A 50 year female who is morbidly obese is seen by her primary physician. She mentions that she had woken up and noticed her lower leg was reddened. She remembers possibly bumping it while walking around her bed. She denied any fevers or chills. He notes that her leg is swollen, warm and red. The doctor orders a white blood cell count, which is found to be 12,000 (normal <10,000). She was referred to the ER for possible cellulitis. She is admitted to the hospital and started on intravenous antibiotics, with a gradual improvement of her leg coloring and swelling. She is discharged with oral antibiotics for 1 week. She returns to the clinic after this, stating that the redness returned, and he readmits her to the hospital. She is then given a central line and arranged to received IV antibiotics for several weeks with a different antibiotic. He refers her to the infectious diseases physician. In the office, I examine her leg. She never had any fever or chills at onset and both legs demonstrate swelling , with the side of concern with redness and warmth. I remove her line in the office and discontinue the antibiotics. She was found to have post-traumatic lymphedema.
* * *
A patient comes to the doctor for advice and consultation to better understand the “why” behind his/her disease. Someone may have a general idea of what is going on and would like confirmation or they suspect one cause but it doesn’t completely fit what they read online. They may have googled their symptoms and “the C word” (cancer) came up among the 200 possible causes. Unfortunately, finding answers in the clinic can be more complicated than it may seem.
Determining the cause of a process relies on knowledge base, experience, clinical acumen, medical decision-making and a healthy dose of curiosity and self-doubt. A physician spends approximately seven to ten years mastering the practice of medicine through medical school and residency training. During this time, s/he makes the transition from a passive role as student learner to an active clinician. Much of the training is familiarizing oneself with the common presentations of common illnesses. Physicians are taught early that “when you hear the sound of hoofs, think horses not zebras”. Given the volume of patients, physicians in training will often see several uncommon presentations of common conditions. Perhaps a few times a month at larger teaching hospitals during the average training of three times, they may even see a common presentation of an uncommon illness, such as malaria or dengue fever. Infrequently during the span of a physician’s career, s/he may encounter an uncommon presentation of an uncommon illness, sometimes without even realizing it. They may still be trying to fit a square peg in a round hole.
Types of Presentation in the Diagnostic spectrum
Type I: Common presentation of common illnesses: Typical cold, pneumonia, heart failure
Type II: Uncommon presentation of common illnesses: MRSA abscess presenting like cellulitis, influenza myocarditis, pneumonia with empyema. C. difficile colitis with toxic megacolon rather than diarrhea.
Type III: Common presentation of uncommon illnesses: Malaria, dengue fever and any other tropical illness presenting like the textbook description
Type IV: Uncommon presentation of uncommon illnesses: Miller-Fisher variant of Guillain-Barre syndrome, cerebral malaria, Fitz-Hugh-Curtis syndrome from Chlamydia. See National Organization for Rare disorders (NORD).
Factors affecting diagnostic accuracy
Location of training and practice. Illnesses sometimes have a geographic distribution. One practice location may see more cases of Lyme disease, that there is greater influence on this disease earlier. Malaria, a disease of the tropics, is always the first consideration in a returned traveler with fever. Though the majority of physicians in smaller towns or cities in the US haven’t seen a case. Travel to the malaria belt in sub-Saharan Africa and most physicians haven’t seen many patients with diabetes.
Scope of practice. In the field of medicine, there are many specialties which are divided by systems. A specialist is trained additional years (2-5) in one specific field, such as a nephrology (kidney), infectious diseases, cardiology (heart) or pulmonology (lung) and rheumatology (joint and inflammation). Although this focused training allows for more cases in one discipline, there is still overlap with conditions that are presenting as one problem but are another, such as a rheumatologic problem – lupus – presenting with fever.
Experience base. A physician who has seen and managed many cases of one particular diagnosis is often more able to accurately diagnose, understand nuances of the condition and appropriately treat a person. This could also be subjected to bias, when overly relies on this to make a decision that may need to be adjusted for a particular case.
Knowledge base. This may allow a physician to generate a more expansive differential diagnosis – a listing of potential causes that are ruled in or out on the basis of the history, physical examination, laboratory and diagnostic information. Nevertheless, experience helps to apply a weight to specific possibility over a range.
Diagnostic Tools available. Having additional tools, such as ultrasound, radiographs, CT and MRI as well as laboratory tests can undoubtedly enhance diagnostic accuracy. They still need to be assimilated by an experienced physician, to weigh likelihoods. This goes back to the example of the patient with a CT showing “early diverticulitis”.
Some degree of medical decision-making is based on one’s previous experience of a condition and one’s ability to detect patterns that are consistent with that condition. A physician generates a differential diagnosis e.g. a patient presenting with “shortness of breath” can have pulmonary embolism, acute congestive heart failure, acute renal failure, pneumonia. A clinician then must rank order the possibilities based on the elements of the history and physical examination and work toward a plan, including further tests and treatments, e.g. order a chest radiograph, order a 2-D echo to evaluate the heart valves, etc.
As a physician sifts through a barrage of data from patient encounters and lab and imaging studies, s/he must weigh the impact of bias in medical decision making. Even ordering a test can affect medical decision-making and lead to further tests that may not have been necessary. Although the experience base of a physician provides a foundation in which to consider multiple diagnoses while evaluating a patient, some habits develop that can alter perception of importance and lead a physician down the wrong path. For instance, a physician’s misattribution of a process as infectious (Case 2) had implications on the treatment chosen, such as antibiotic use and duration.
A few common forms of bias include 1) anchoring bias and 2) diagnosis momentum, 3) confirmation bias and 4) omission and 5) commission. These biases can originate from physicians and patients alike and can occur simultaneously.
Anchoring bias relates to settling on a diagnosis from early on and following through with treatment toward this potential diagnosis. A physician may reflexively base his/her diagnosis and decision making too strongly ruling out the worst-case scenarios, even though even as further tests come in that contradict the original fear. For example, a patient who is complaining to a physician about shortness of breath is sent for a CT angiogram, a CT that requires a dye load to evaluate for pulmonary embolism, or blood clots to the lungs, even after history and testing (e.g. negative D-dimer) do not clearly suggest it.
Diagnosis momentum can occur after a patient receives a diagnosis, such as cellulitis as described above, without questioning further the original diagnosis.
Confirmation bias occurs when a physician looks for symptoms and signs that confirm his/her original suspicion, rather than for things that discredit it.
Omission is a lack of action or care to reduce harm, but as a disease process develops, it may cause more harm, for instance not working up further a patient with headaches, who ultimately has meningitis.
Commission is providing care toward the benefit of the patient, rather than inaction. Occasionally, a diagnosis is not completely clear, even after review and studies, such as with the patient who was treated for diverticulitis who only had a groin pull or the patent with “cellulitis” coming in from the clinic – with the continued diagnosis.
A physician must consider multiple potential diagnoses while evaluating a patient. Even the fact-finding task of taking a history can be marred by the patient’s confirmation bias. For instance, a patient coming in stating, “doctor, I have this urinary tract infection,“ without considering other causes. A patient may consider more uncommon diagnoses and read about chronic Lyme disease and believe that they have this condition, even with a negative Lyme titer and western blot. The office visit may not be long enough to consider other possibilities or obtain a more detailed history. The patient may not recall some aspects of the history or recall it a different way based on his/her concerns or pre-reading of the possibility, inviting another bias, recall bias.
Ways to Avoid Bias Traps
There are several ways a patient and clinician can avoid the pathway of bias and possibly missing a diagnosis or making an incorrect diagnosis. Below are some suggestions:
Start with the history and physical and end with the history and physical. The median time it take a physician to interrupt a patient explaining his/her symptoms is 11 seconds. Start by actively listening to the patient without interrupting for several minutes. ACTIVE listening implies eye contact, nodding, verbal confirmation, summarizing back and NOT by cell-phone checks, eyes fixed on the keyboard and cutting the patient off. There will be a time when the patient starts to circle back or become tangential. If there is anything that isn’t clear, go back to the patient and ask more questions. Sometimes their concerns may drive the words they choose to describe the symptoms. If you sense that this is biasing them, ask the patient what they think they have or fear they may have? Oh – and don’t forget to do a proper physical examination. As a patient, I have had to remind the doctor, “aren’t you going to examine me?” A proper physical examination can be very telling. We sometimes joke about the “internists physical examination” being a full body CT scan.
Try to at least think of 2 or 3 other possibilities. Avoid quick decisions that try to explain the problem backwards. Let the symptoms guide the differential, not the other way around.
Order baseline blood tests only when needed and let the severity of the condition dictate the timing of further testing. There is nothing wrong with appropriate watchful waiting. I always like to think that the best exam is one that includes a follow-up visit.
Collaborative care in consults or curbsides is reasonable to get a second opinion and avoid biased conclusions. Though, when a doctor discusses a case with his/her peers, there is an inherent bias in what is relayed to the other as germane. Most of the time, it is better to have a formal consult.
Always be one’s own skeptic, look for bias traps and keep the dialogue open to the patient if a diagnosis is not clear. The patient may help the physician in providing additional information or concerns. Being direct with the patient about the level of uncertainty allows the patient to partner with the physician and usually improves rapport. We are all humans and to err is human – start with doing no harm and then collaborate to explore possible answers.
Write the line sequence of the events when they occurred before the visit. This is usually more helpful than just providing a list of signs and symptoms in no particular order. This can help state the concerns more clearly.
Prepare a list of concerns and questions for the visit in advance. I usually recommend 1-2 concerns and 1-2 additional questions for a 20 minute visit. Some things can be saved for a routine follow-up or full physical examination.
Avoid using a diagnosis in your description of the symptoms. Instead of “Doctor, I have a UTI.”, try “doctor it hurts when I urinate and I have increase need and frequency to urinate.
Discuss you concerns with the clinician after you have stated the symptoms and he/she has examined you. The clinician may weigh the concerns and discuss likelihood and a plan of addressing the condition.
Go through plan with physician and contigencies.
Schedule a follow-up visit in 1-2 months that same day to ensure that the problem can be readdressed. This keeps the issue at the forefront and ensures that things get addressed promptly and not put aside. I once saw a patient who had anemia and I wanted to see him back – I gave him a referral to the GI doctor. Unfortunately, he never followed through — he came back 2 years later to see me with severe anemia, apologized for not following back. I examined him that day and sent him to the hospital after a found a mass on rectal exam. I sent him to the hospital he was diagnosed with colorectal cancer.
“Oh, and by the way, doctor”
I am always particularly attentive – or perhaps biased – when a patient brings up a concern the moment that you are saying goodbye. Maybe it is a last chance brainstorm for them to bring up something. Maybe it is after opening up with the physician during the visit, that they can confide in him/her what they are about to say. Maybe it is something that they were keeping an eye on for a while and didn’t think much of it and wanted the doctor’s opinion.
Whatever the reason, it is often prefaced, “Oh, by the way…” I recall one patient who brought up the “Oh, by the way, what is this lump on my chest, doctor?” I had my hand on the doorknob when he stated this. I turned around, agreeing that this didn’t have to wait until the next visit. His chest exam revealed a lump the size and density of an acorn – a worrisome finding that prompted a referral to the surgeon and turned out to be breast cancer.