First a case…

An elderly man reported to the emergency room after a fall down 15 stairs at home. He was more confused than his baseline but otherwise did not have any specific complaints. Vitals and labs were within normal limits. A non-contrast CT of the brain and C-spine was performed and read as normal. See the image below.

Without a cause for confusion detected, he was diagnosed with a concussion. He had a great deal of difficulty with ambulation, so he stayed overnight in the ED to be observed and to see physical therapy in the morning for assessment. A long discussion took place between PT, the physicians, case management, and the patient’s wife who collectively decided that the patient would go home with services.

He had progressive confusion throughout the remainder of the day at home so he presented to an outside hospital where they performed another non-contrast CT of his brain, shown below.

The patient hadn’t fallen again and delayed bleeding is rare (around 0.3%) [1], so let’s look at the first CT again, this time in a different orientation.

As you can see, marked by the arrows, bilateral subdural bleeds were present on the initial CT, only visible in the coronal slices. One ED-based study with 213 patients showed that 14% of intracranial hemorrhage foci were only visible on coronal slices and not on axial slices.[2] This case represents the most common type of radiological error – a missed diagnosis. Why diagnoses are missed is multifactorial and the main discussion of this blog post, in addition to other common radiological errors.

Diagnosis

Bilateral Subdural Hematoma - Missed Diagnosis

Discussion

Whenever a medical error is made it is important at this point to review it in context with an understanding of the root cause. Here we will review the numerous types of errors which have been found to contribute to radiological errors.

Scanning Error: When the interpreter does not look over the region with the abnormality.

Location Error: Similar to above, specifically when an abnormality is missed because it falls outside of the area of focused interest (e.g. missing lung cancer on a humeral film).

Recognition Error: The interpreter does look over the region but fails to recognize the abnormality.

Decision-making Error: The interpreter looks over the region and misinterprets the abnormality (e.g. calling a pneumonia on chest x-ray when it is in fact a pulmonary contusion)

Satisfaction of Search (SOS): Missing a diagnosis because the eyes are drawn to a second abnormality.[3] This is why reading chest X-rays is commonly taught as ABCDE, so that one moves through all portions systematically.

These errors can be broken up further into cognitive errors, in which the clinician is lacking knowledge that ultimately leads to the misinterpretation of the radiographs, or perceptual (which account for >80% of errors) in which the abnormality is not perceived.

There are of course other types of errors, but these are by far the most common. Other factors that contribute to misinterpretation of images, laid out well by a review article include:

“the quality of images and the [appropriateness of] views obtained, quantity of clinical information, the absence of previous imaging studies, the reading room atmosphere, the level of alertness of the interpreter, errors of speed, failure of perception, the lack of knowledge…errors due to multitasking, increased workload” [4]

Moreover, specifically with CT scans, dye loads must be timed properly. Some pathologies are difficult to diagnose on CT such as diaphragmatic injuries. Certain patterns emerge with both false-positives and false-negatives within CT reads due to the nature of the images.[5] Image artifacts from movement may make interpretation more difficult. Scout images are often overlooked and can have meaningful pathologies not seen on CT in about 2% of cases.[6]

The list goes on and on far beyond everything listed above… The main rationale for introducing all of these causes of error is to show that the imaging technicians who collect the images and the radiologists who officially read them have challenging jobs.

What can ED providers do to reduce errors

Order the correct study. Certain diagnoses such as the brain bleed earlier can only be made with certain imaging modalities. Building off of that – try and order only what is necessary and don’t hit “panscan” simply because it’s easier for you – it makes a lot of other people’s lives more challenging and can lead to missed diagnoses.

Give as much clinical context as possible to the radiologist. They don’t get to see and examine the patient! Writing “pain” is not nearly as helpful as “fell onto outstretched hand, has tenderness at the distal radius.”

Communicate with radiology. Building off the last point – if you don’t agree with an interpretation or something doesn’t sit right, such as this patient’s persistent altered mental status, call the radiologist and discuss the case. Perhaps you will speak to a new radiologist who can view the images with fresh eyes in the morning.

Have a checklist when reviewing images. Be systematic so that you don’t get distracted by satisfaction of search.

Bringing it back to the first case…

After review it is unlikely the first EM team or radiology committed any egregious errors but with the information provided above, a few things could have been done better. Nowhere in the history provided to radiology did the EM provider mention that the patient was significantly altered, tuning the radiologist to look for intracranial pathology. When the patient remained persistently altered, the images were not reviewed over the phone with radiology. It is impossible to know, but unlikely that the radiologists reviewed the CT brain in all planes or the bleed likely would have been seen, representing a scanning and/or recognition error. This perfect storm of common human errors related to radiology combined and unfortunately led to a missed diagnosis. Hopefully understanding these potential pitfalls will help providers going forward to identify and avoid radiological errors.

Author: Austin Quinn, MD is a third year resident.

Faculty Reviewer: Alexis Lawrence, MD, is an Assistant Professor of Emergency Medicine.

References:

1. Chenoweth JA, Gaona SD, Faul M, Holmes JF, Nishijima DK. Incidence of Delayed Intracranial Hemorrhage in Older Patients After Blunt Head Trauma. JAMA Surg. 2018;153(6):570–575. doi:10.1001/jamasurg.2017.6159

2. Wei SC, Ulmer S, Lev MH, Pomerantz SR, González RG, Henson JW. Value of coronal reformations in the CT evaluation of acute head trauma. AJNR Am J Neuroradiol. 2010;31(2):334-339. doi:10.3174/ajnr.A1824

3. Samuel S, Kundel HL, Nodine CF, Toto LC. Mechanism of satisfaction of search: eye position recordings in the reading of chest radiographs. Radiology. 1995;194(3):895-902. doi:10.1148/radiology.194.3.7862998

4. Pinto A, Reginelli A, Pinto F, et al. Errors in imaging patients in the emergency setting. Br J Radiol. 2016;89(1061):20150914. doi:10.1259/bjr.20150914

5. West OC, Anderson J, Lee JS, Finnell CW, Raval BK. Patterns of diagnostic error in trauma abdominal CT. Emerg Radiol. 2002;9(4):195-200. doi:10.1007/s10140-002-0225-8

6. Johnson PT, Scott WW, Gayler BW, Lewin JS, Fishman EK. The CT scout view: does it need to be routinely reviewed as part of the CT interpretation?. AJR Am J Roentgenol. 2014;202(6):1256-1263. doi:10.2214/AJR.13.10545