Weighed Down by Words: How to Recognize and Avoid Anchoring Bias

0

“Chief complaint: flank pain.” I started thinking about my differential: nephrolithiasis, urinary tract infection, pyelonephritis, pelvic inflammatory disease, musculoskeletal pain. I read the triage note: “30yo F, 2 days of worsening left flank pain. Patient is heroin user on methadone.” Great, I thought, mentally adding drug seeker to my differential. As the tech pushed her past me in a wheelchair, her hands were covering her face and she was rocking back and forth, moaning. The tech rolled his eyes; I shook my head.

I turned my attention back to her chart. Afebrile, normal heart rate, normal blood pressure. Boring.

I walked into the room, introduced myself, and asked how I could help her. She continued rocking in the gurney, yelling “It hurts, it really hurts! I need pain medicine!” I made my way through the history. Two days of worsening left flank pain, nausea, and vomiting. No urinary symptoms, vaginal discharge, or trauma. She denied recent IV drug abuse but wouldn’t let me look at her arms for track marks. She was very hesitant with answers, very difficult to examine, and resistant to questioning. I was growing frustrated.

As I started to place her orders, I was apprehensive about giving a significant amount of pain medication to a heroin user. 4 mg of morphine it is. After two doses, it was clear the morphine was not working. She requested “that medicine that starts with a D.” I shook my head again. Great. The nurse notified me she was throwing up — in fact I could hear her. Maybe it was the morphine, I thought.

I gave more pain medicine, anti-emetics, fluids, and waited. Her labs came back: a mildly elevated white blood cell count, some white cells and some red cells in her urine, normal hemoglobin and electrolytes. Nothing exciting. As I saw her being wheeled back from her CT scan, I moved on to my other patients.

The phone rang. I looked at the caller ID: Radiology. Oh great, she has a stone.

“Are you taking care of room 28?” asked the radiologist.

“Yeah, how big is her stone?” I replied.

“Umm, can you tell me more about her story, like what happened?” was the response.

“I’m not sure, just two days of worsening flank pain, sounds like a stone.” I said.

He then began to tell me about the spleen and liver lacerations with active signs of extravasation, the left perinephric hematoma, the free fluid pelvis, and the multiple rib fractures at various stages of healing. My stomach was in knots.

I went back to the patient’s room to notify her of the findings and, after adequate pain control, I really saw the patient for the first time. I could see faint bruises over the left side of her eye and bruises over her legs.

“I fell out of bed.” was her answer when I asked what had happened. I asked if she had been assaulted, hit, abused. “…I fell out of bed,” she repeated in a defeated tone as if she knew it was a lie but was not letting herself admit the truth.

I felt sick. I felt I had let myself down. More important, I had let my patient down. Why did I not grab the ultrasound machine and look at her kidneys at the bedside like I usually do? Why did I not heed the feeling that something was not adding up in her story and her symptoms? Why did I not take her complaint of pain more seriously? Why had I anchored my feelings on her drug addiction and ultimately let it affect my feelings toward the patient and my treatment? Where did it start?

I read the triage note. “Patient is heroin user, on methadone.”

It amazed me how just a simple sentence could have altered how I thought about a patient, how a tech’s comment added to my own doubts and skepticism, how a nurse telling me about her pain medicine request “confirmed” my suspicions. Before I had even evaluated the patient, she had become a victim of anchoring bias.

“Anchoring bias” is the term used to describe the mental error of maintaining one’s initial impression despite evidence pointing to the contrary.1 Ultimately, anchoring bias can lead to incorrect diagnoses or a delay in arriving at the correct diagnosis. Anchoring bias commonly results from paying too much attention to one finding, not listening to the patient’s full story, not reassessing the patient when information does not correlate with their symptoms, or simply being in too much of a hurry. In addition, emotions, previous experiences, personal beliefs and values, time, pressure, and peer influence can all lead to a biased perception of the patient. It is the clinician’s responsibility to not only be aware of, but to overcome these internal and external biases, because they impede our ability to provide maximum care and compassion to our patients.

How to Combat Anchoring Bias

Being aware of bias is not enough to eliminate it. Research suggests that while there is no way to eliminate bias, it is possible to develop strategies to help reduce its likelihood of occurring. These include limiting words that might introduce bias, reporting only factual (not assumed) information, being careful to separate professional decisions from personal feelings, and developing cognitive “walk through” strategies for scenarios in which bias is more likely to be present, such as in my patient.2,3

To reduce my own biases, now when I am receiving sign out on a patient, I ask only for the facts of their current medical complaint. I discourage disparaging terms such as “frequent flyer,” “malingerer,” “crazy,” or “drug-seeking” and do not use them personally. I frequently reassess my patients when the clinical exam does not correlate with the laboratory data, forcing myself to think of alternative diagnoses from new information I have gathered. Additionally, I treat acute pain adequately regardless of past history.

So what happened to my patient? Her bleeding was controlled through embolization, she was admitted to the ICU, and ultimately transferred to the floor. Through her entire hospital stay, “heroin abuser” was plastered all over her chart. They never found a definitive case of her injuries, although assault and abuse was strongly suspected. She was discharged in “good condition,” a term that has lost much of its meaning for me. And while most of the people who cared for her surely have moved on, she is a patient I will never forget.

References

  1. Wellerby C. Flaws in clinical reasoning: a common cause of diagnostic error. Am Fam Physician. 2011;84(9):1042-1048.
  2. Scott IA. Errors in clinical reasoning: causes and remedial strategies. BMJ. 2009;338:b1860.
  3. Croskerry P. The importance of cognitive errors in diagnosis and strategies to minimize them. Acad Med. 2003;78(8):775–780.
Zac Hafez, MD

Zac Hafez, MD

Emergency Medicine Resident, Washington University in St. Louis, St. Louis, MO
Zac Hafez, MD

Latest posts by Zac Hafez, MD (see all)

Share.

Leave A Reply