Teaching about cognitive error Part 1
I started thinking about this topic during a presentation on the ROME II Criteria for IBS. An internal medicine resident said with great sarcasm, “If she’s overweight and depressed, she probably has irritable bowel syndrome.” No one in the audience challenged him on this statement.
Subsequently, I had several discussions with clinical teachers about how hard it was to give feedback to students whose medical knowledge was extensive, but who were unable to accurately diagnose or treat patients because they were making irrational inferences about or from their knowledge. When a resident has invested time, ego and energy into a particular diagnosis, it can be difficult to move them in a different direction. In some cases, the resident is so invested that s/he will respond inappropriately to suggestions they have made a mistake.
Thanks to Dr. Keith White’s presentation on medical error at the Rural and Remote Conference, I realized that many clinicians were unaware of how to diagnose cognitive errors in their students clinical reasoning. A 2005 study published in Archives of Internal Medicine found that cognitive error is the single most common cause of diagnostic errors. This post hopes to begin to help clinical teachers help students correct errors by examining four common types of cognitive error.
1. Confirmation Biases
A 37-year-old woman is severely anemic. She has previously been diagnosed with celiac disease, which causes malabsorption, and she is told to increase her iron intake. Several months later when she doesn’t improve, she is discovered to have a small tear in her esophagus. She had mentioned the pain when swallowing previously, but this symptom had been under investigated.
Confirmation biases cause you to look for proof of your diagnosis while ignoring factors that might disagree with it. They are the result of errors in cognition combined with the fast paced decision making required in many family practices and emergency rooms. These errors are frequently based on:
When you see a lot of X or have studied a particular diagnosis recently, you tend to be looking for it. Ex. You have seen 6 cases of flu this week and here’s another one. You only see the symptoms of flu you are looking for and ignore the symptoms of food poisoning.
When a patient has been diagnosed with A previously, the assumption is that A was a correct diagnosis and that present symptoms are probably related to the previous diagnosis. This is more likely to occur when the initial diagnosis came from an authority figure, but has also occurred when a patient self diagnoses (I have a migraine).
Lack of Experience
Lack of experience may cause the student who does not understand the variability of human biology to depend on textbook knowledge or medical studies that are incomplete. On the other hand, the student may have used the same diagnosis successfully in another situation and were hoping for the same results.
The student may stop searching when one diagnosis is made and not look for other problems.
The student is so invested in proving themselves right that the patient may be at risk. This is linked to the tendency to believe that one’s previous decision-making was better than it was. Hindsight bias, the assumed ability to see how errors were made in the past can contribute to overconfidence.
2. Attribution Errors
A 65-year-old man arrives in emergency on a very busy Friday night. Joe is unconscious after being found in an alley outside the local bar. He is well known in the ER as being severely alcohol dependent, and he is placed in a bed “to sleep it off”. When the doctor goes to send him home in the morning, Joe isn’t responding and upon examination he is discovered to be the victim of a car accident.
When stereotypes about a race, gender, religion, age, addictions etc. result in misdiagnoses, the underlying assumption is frequently that this person is judged unworthy of full attention because they are … Other examples I’ve heard are: “He lied about his military service, I don’t want him (a patient with a history of psychiatric illness) in my hospital”; “Here’s another woman with mysterious abdominal pain (a woman with a perforated uterus).” In each case, attribution errors led to misdiagnoses.
A subsection of attribution errors is liking the patient too much and not wanting to cause them pain or embarrassment, so you don’t ask them questions about their sexual history, don’t examine them for prostrate cancer, don’t ask about sleeping pill use etc.
3. Commission Bias
A 45-year-old man with terminal bone cancer is unresponsive when the resident enters the room. He successfully resuscitates the man and then remembers the DNR order.
Commission bias is the result of overwhelming internal or external pressure to do something NOW rather than wait. Doctors want to help patients and their fearful families. Antibiotics, painkillers and sleep aids have all been over administered because of commission bias. Aggregate bias or the ordering of tests and x-rays when the guidelines don’t recommend them is a form of commission bias. Omission bias is the other side of this coin and results in the student doing very little in the hope of avoiding errors.
4. Investigation Errors
A 56-year-old woman comes to emergency with pain in her left, back shoulder and a feeling that something isn’t right. The patient is given medication for indigestion and send home. Two hours later, she returns because of heart failure.
Our job as clinical teachers is to help students learn from and correct mistakes so they don’t developing habits of thinking that are ineffective and unhealthy. Investigation errors are primarily the result of asking the wrong questions because of the following factors:
The tendency to rely too heavily, or "anchor," on one trait or piece of information when making decisions, can lead to search satisfaction and other confirmation biases.
Base Rate Neglect
Under or over estimating how common a disease is in a community, gender, ethnic group etc.
The way a patient is described influences clinical reasoning. Some students may be unable to pick up key words and nonverbal cues from nurses, patients and family members to make judgments about severity, frequency and urgency because of fatigue, coming from a different culture or inexperience. Others may rely too heavily on the frame and fail to look at the bigger picture. For example, when a patient is labeled as having fever, shortness of breath and cough, the student may jump to a diagnosis of pneumonia in a patient with pulmonary embolism.
Fear of death, fear of failure, fear of uncertainty can lead to avoiding patients with possible unpleasant outcomes. Some students become overly rational/clinical (as if the patient is an experiment) to cover the irrational fear. See the movie Wit for an example. Others become overly dependant on guidelines/evidence because their fear of failure is pushing them to the safety of outside expertise/authority.
The Last Bad Experience
If a student makes a serious error, they can run away, cover up, bluff it out, avoid or they can fight, be overly vigilant, obsess.
For more information, see previous posts on
Teaching Issues of Diversity http://blogs.usask.ca/medical_education/archive/2006/04/teaching_issues.html
Preparing Students to Work with Addiction Issues http://blogs.usask.ca/medical_education/archive/2006/08/preparing_stude.html
How Doctors Think Introduction
Croskerry, P. (2003) When diagnoses fail, The Canadian Journal of CME: 79-87
Crosskerry, P. (2003)The importance of cognitive errors in diagnosis and strategies to minimize them. Academic Medicine: 78(8):775-780
Groopman, J. (2007) How doctors think, Houghton Mifflin
Mazor, K. et all (2005) Teaching and medical errors Medical Education Journal:39:982-990
Redelmeir D. (2005) The cognitive psychology of missed diagnoses Annals of Internal Medicine Volume 142 Issue 2 | Pages 115-120
Wade, M. (2007) 26 Reasons What You Think is Right is Wrong http://www.healthbolt.net/2007/02/14/26-reasons-what-you-think-is-right-is-wrong/
Wikipedia Cognitive Distortion http://en.wikipedia.org/wiki/Cognitive_distortion