Journal Logo

Trial & Error by Michael Mouw, MD

Mistakes, near-misses, and bouncebacks. Chances are we mere mortals who have been at this a while have a case or three we aren't proud of. The trouble is that the medical culture doesn't exactly encourage admitting one's human errors, at least not outside M&M peer review.

This blog, moderated by Michael Mouw, MD, aims to provide a collegial forum for emergency physicians to share lessons they learned from mistakes made in the practice of emergency medicine.

Submit your case about errors you've learned from, and we will publish it here with a brief analysis by Dr. Mouw. All cases will be published anonymously, though you may choose to include your name, as Dr. Mouw has done with the first case.

Submit your case to

A few rules:

  • Mind HIPAA! No names (patients or otherwise) or identifying characteristics.
  • Cases should not exceed 500 words, and will be edited for style, grammar, length, and clarity.
  • Photographs, clinical images, and lab reports are welcome, and should be submitted as separate attachments. Images should be 300 dpi and in jpg, tif, gif, or eps format.
  • Authors are responsible for obtaining consent from patients, family members, and health care professionals depicted in images, and must attest to EMN that consent was obtained.
  • Authors should provide their full name and contact information for verification purposes. This is strictly to ascertain identity and will not be published without permission.
  • Include all relevant information about the patient in the case as well as a full description of the patient's symptoms, diagnosis, treatment, and outcome.
  • Only post cases outside the statute of limitations. Please provide your retrospective insights and analysis, and any pertinent references you've found helpful.
  • Comments about the cases posted are also welcome. Be kind. Submit yours to Word limit: 200 words.
  • Submission grants permission to publish in EMN, its website, enews, and other formats.

Friday, May 1, 2020

The EM lecturer closed with: "So if you see a heavy-set person with nontraumatic muscle pain, think about hypothyroidism."

"Hmm," I thought, "Wonder if that might apply to that guy we saw a few days ago?"

A burly construction worker returned to the ED with persistent bilateral arm pain and swelling. He had been seen two days earlier for the same complaint, and had been worked up for rhabdomyolysis and compartment syndrome, which had revealed a creatine kinase (CK) in the 400s. He had been told to return for a recheck and trending of his CK. The resident on the case repeated his CK at his most recent visit, which was trending lower, so we let him go home to follow up in the clinic.

On my way home from that lecture, however, I swung through the ED, where the same resident was on duty again. "You remember that guy we saw with muscle pain who we repeated the CK on? We should call him back to test his thyroid function." He returned later that day, and the test showed a markedly elevated thyroid-stimulating hormone (TSH) level of 12 mU/L.

We researched the issue further and discovered that the incidence of clinical hypothyroidism in the United States isn't low: one in 300, which means most EPs will see at least one case in their career. My threshold for ordering screening TSH levels for patients with vague complaints definitely went down after this case. As did my threshold for calling a patient back if I may have missed something.

Wednesday, April 1, 2020

A 16-year-old football player presented to the pediatric ED with a severe headache. The attending physician was a first-year pediatric emergency physician.

The patient reported a constant, severe headache that he attributed to a hard collision in practice the day before. He had not lost consciousness, and his vital signs were normal.

A head CT was normal, and he was discharged with a diagnosis of post-concussion syndrome and given appropriate instructions.

The patient returned to the ED two days later (day three of symptoms), still alert and ambulatory. His headache had worsened, and was now severe. He presented with normal vital signs and was afebrile. He did report malaise and subjective fever.

He had taken ibuprofen an hour before ED arrival, and was still attributing his symptoms to the football collision. He said, however, that his headache had not started immediately after the collision but approximately 15 hours later. The collision itself didn't involve his head.

He had subjective posterior neck pain with flexion, but his neck was supple. Neurologic and skin exams were normal. I reviewed the initial head CT, and agreed that it appeared normal. I felt the presentation was more consistent with meningitis than a TBI, and after discussing my thoughts with the patient and his mother, they consented to a lumbar puncture.

An examination of cerebrospinal fluid revealed aseptic meningitis, and he was admitted for observation. Antibiotics were withheld because nothing suggested bacterial meningitis (afebrile, normal peripheral WBC count).

Case Lessons

The "consider the worst first" rule was broken. Anchoring on a TBI occurred when the physician followed the patient down the wrong path and didn't take a detailed history about symptom onset or consider alternative and potentially more dangerous explanations.

Although listening to patients and considering their ideas is good, it can lead to error if we take the path of least resistance and accept an incorrect self-diagnosis. In a case like this, thinking for ourselves and disagreeing with them is essential.

Absence of fever due to antipyretic administration can also lead to error. This patient endorsed subjective fever, but didn't have a fever in triage. A history of fever should be treated the same as an objective fever recorded in the ED.

Monday, March 2, 2020

It was the mid-'90s, I was eight years out of residency, and it was before the Haemophilus influenzae vaccine, so meningitis was not uncommon. I was on a busy flu-season shift in the pediatric ED supervising a third-year emergency medicine resident.

He presented the case of a 9-year-old boy who had been transferred to us from another ED for meningitis. A lumbar puncture done four hours earlier that day at the transferring ED was traumatic, and the boy had gotten his first dose of antibiotics before the transfer. He was alert with normal vital signs, but had a severe headache and stiff neck. We decided to repeat the LP for microbiologic diagnosis because it had been only four hours since antibiotic administration. We were swamped, so I went on to the next case while the resident attempted the LP.

The resident returned a short time later, sheepishly reporting that his attempt was also traumatic and that he had only gotten blood. He held up three tubes of bloody CSF.

This was definitely going to take more time than anticipated. I went to talk to the family, and a more detailed history revealed that his headache had started gradually two days before, had gotten progressively worse, and became severe that morning. His neck was definitely stiff, but there were no petechiae, and he had not had a fever at home, at the transferring facility, or at our ED. I considered repeating the LP myself to be sure, but mom wasn't down with that idea. I'd never seen a spontaneous head bleed in a child, but we sent him for a head CT, which showed significant subarachnoid hemorrhage from a previously undiagnosed arteriovenous malformation! Neurosurgery was consulted, and an arteriogram was obtained. He eventually did well.

Case Lessons

Resident supervision: There is a balance between giving residents some autonomy so they can learn from mistakes and protecting patients from the consequences of those mistakes. In retrospect, even though this resident was a third-year, I should have done my own history and exam before coming to a plan of action. I likely would have keyed in on the absence of fever, and might have done a CT before another LP attempt. So minus one for not doing my own H&P right away before repeating the LP.

Anchoring: This case was in a busy pediatric ED at a time when meningitis wasn't uncommon. It would have been tempting to accept the transferring diagnosis and admit the child without repeating the LP. I honestly thought we would get a nontraumatic (and hopefully diagnostic) CSF sample, and we would have missed the diagnosis if we hadn't. Sometimes luck steps in where specific intent is absent. So plus one for recognizing this case didn't fit the typical pattern for meningitis and pursuing the repeat LP.

Friday, January 31, 2020

I was covering the fast track near the end of a day shift when I picked up a chart with a triage note that said "third visit for knee pain." The patient's vital signs were normal except for a heart rate in the 120s.

I found a healthy-appearing man in his 40s seated in a wheelchair with his right leg propped up. He seemed uncomfortable, but he didn't appear ill. He had been seen twice by an excellent nurse practitioner in the preceding three days, and was being treated presumptively for gout. He returned a third time because his pain was worsening despite hefty doses of naproxen and Norco.

I noticed an obvious and quite impressive knee joint effusion when I pulled back the sheet covering his leg. It was his third visit and he had never had gout before, so I recommended that we proceed with arthrocentesis to confirm that it was gout, not a septic joint. He agreed. A short time later, I was shocked to retrieve 70 cc of frank, thick pus from his knee.

I did a more detailed history as I was putting the specimens in tubes for the lab. He reported some subjective fever and chills, and admitted to a cough that was new (he hadn't volunteered this information to the NP or me). It was now obvious that he needed admission, so I expanded his workup to include all the usual labs and a chest x-ray. This was in the early days of "surviving sepsis" when serum lactates were becoming standard of care, so I sent one of those too.

His chest x-ray showed a large lower lobe infiltrate, which was surprising because he had no respiratory complaints and a normal oxygen saturation. I nearly fell out of my chair when I saw his lactate: 9 mmol/L. He was transferred to the critical care area for a central line, SvO2 catheter, and fluid resuscitation. Cultures of his blood and synovial fluid both grew pneumococcus. He was admitted and did well.

This case represents a common pitfall of contextual bias in our practice: We don't expect to find septic patients in the fast track, and pneumococcal sepsis presenting as knee pain is a pretty rare bird in my experience. Most triage protocols today would have flagged his heart rate as a sepsis alert and yielded a different ED bed assignment, but those protocols weren't in place at the time. I honestly can't say that I tapped every new nontraumatic knee effusion before this case, but I do now.

Thursday, January 2, 2020

Humans, in general, are much less likely to see their own mistakes than those of others. This is called the blind spot bias. That's why it's good that we train under supervision.

Once during a slow critical care shift, I was down the hall in our charting area. A first-year EM resident was assigned to me at the time. The charge nurse rang my belt phone: "We need you in room three. EMS brought a woman in respiratory distress. Your resident's in there." It was July, so I hustled right over. From the doorway, I noticed a young woman sitting up on the EMS stretcher in obvious respiratory distress. She was puffed up like the Michelin Man. Something bad was going on here.

The resident was giving verbal orders: "Give her 50 of Benadryl and a milligram of epi IV."

"What's going on?" I wondered aloud as I walked over to assess her.

"She's having an allergic reaction."

"OK, what do we think she's allergic to?"

Puzzled, the resident turned and looked to EMS for the answer as they moved her to our stretcher. The medics looked at each other. I asked again.

"We don't know."

By now I have my hand on her chest. "OK, let's back up. Why were you called, and what was the scene?"

"She was in a car accident. We found her ambulatory outside the vehicle in respiratory distress."

Feeling subcutaneous air or crepitation, I turned to the patient: "Ma'am, it looks like it hurts you to breathe. Which side hurts?"

Unable to speak, she pointed emphatically to her left chest. No breath sounds. The same subcutaneous crepitation could be felt anywhere you touched, thus her Michelin Man appearance. Three minutes, one anterior intercostal 14-gauge dart, and a loud hiss of air later, we were prepping her for a left chest tube. A chest x-ray before formal thoracostomy tube confirmed a residual 50% pneumothorax. She improved rapidly.


  • Pattern recognition: I had a big advantage. I had seen the pattern of respiratory distress and subcutaneous air after trauma, so I knew right away this wasn't an allergic reaction. The resident hadn't. I was confident she would recognize it in the future.
  • Anchoring/diagnostic momentum: The medics hadn't seen the pattern either. When presented with a swollen patient in respiratory distress, they anchored on allergic reaction while ignoring the elephant in the room—a trauma mechanism. The newbie first-year resident understandably fell victim to the (system 1) diagnostic momentum started by EMS without engaging her system 2 to stop and think.
  • Telephone game: A common source of error with transports is when crucial information from the scene is omitted or conveyed incorrectly. I've been burned by this more than once, so I try to be present when EMS delivers the patient or to track down the run sheet to clarify why they were called. System 1 is fine if everything fits into an obvious, easily recognizable pattern, but if not, stop and engage system 2!