Misuse and over-diagnosis in POCUS
As Steve Socransky recently observed, talking about errors in POCUS is a sign of the growing maturity in the field. In its infancy, we wanted to promote the success stories, the wins, the need to do it right with rigorous training and standards.
There were numerous logistical and political obstacles to overcome so this positive attitude was understandable. There was the fear that POCUS would be taken away from us if we weren’t careful about minimizing bad outcomes. However, this puts a damper on discussing our errors and misses and learning from what went wrong. Now we feel comfortable enough that it is time to explore errors and misuse.
Rather than dwell on technical errors specific to ultrasound, I thought it would be interesting to discuss more broad-based errors that are common to most areas of practice.
There is still a paucity of good research on POCUS errors. A meta-analysis: Sources of error in emergency ultrasonography by Pinto, et al. tried to analyze common errors. Unfortunately much of the research dealt with scans done by radiologists and obstetricians, but there were some findings that are relevant to all practitioners of POCUS. Poor patient communication, inadequate quality of images, failure to do a proper history and physical, lack of good differential diagnosis, and overestimation of skill were all found to be implicated in poor outcomes.
Before we get concerned about the medicolegal ramifications of exploring our errors, it’s comforting to recall the article by Blaivas that looked at lawsuits related to POCUS. Emergency physicians were not sued for performing POCUS, but failure to do a scan was implicated in litigation.
This does not mean we should be complacent. A study by Moon, et. al. looked at Incidence of posterior vessel wall puncture during ultrasound-guided vessel cannulation in a simulated model. The results were concerning as even with ultrasound guidance approximately one third of cannulation attempts punctured the posterior wall of the vein. If the artery were to lie posterior to the vein, which is most often the case with internal jugular anatomy, then a high risk of carotid injury exists.
IJ anterior to carotid is a poor arrangement for cannulation that you can fix.
Anecdotally, there was a rise in consultations to vascular surgery noticed at one teaching centre when POCUS was first mandated for central line insertion. This is not surprising as education in use of ultrasound in early days was minimal and variable, combined with an invasive, higher risk procedure rarely done by most physicians. Instruction on proper technique, maneuvers to adjust the probe and needle approach so the carotid doesn’t lie posterior to the IJ, and other tricks to better visualize the needle can minimize this error.
A common mistake with broader implications deals with the lack of an adequate history and physical, along with a poor differential diagnosis being generated prior to applying a probe to the patient. POCUS is best used to narrow a differential, rule out certain pathology, and risk stratify patients in their workup. When a probe is applied as part of a fishing expedition, there is a threat of over diagnosis or following the wrong treatment path.
For example many studies have emerged regarding POCUS for diagnosing pneumonia. Jones et. al found that ultrasound was more sensitive than plain chest Xray for diagnosing pneumonia in children. However, Laursen, et. al. found that the early and improved rate of diagnosing pneumonia did not translate into improved patient outcomes.
There are several issues with these studies but there is a real risk to finding infections that do not require intervention or further investigation. This doesn’t mean we should not be considering POCUS for respiratory illness, but we should be careful not to perform the test in the wrong patients.
Getting routine CXRs on children with mild viral illnesses will find a lot of questionable consolidations that get over diagnosed as bacterial pneumonia. Ultrasound is at risk of finding even more occult infections if used indiscriminately in these same patients.
Another example is routinely scanning gallbladders in patients with abdominal pain but failing to correlate whether the pain is biliary in nature. As roughly 10-20% of people will have gallstones but only one third become symptomatic, it is extremely important to correctly incorporate the data points generated by POCUS into the entire clinical picture.
Thus having a clear clinical question to answer, having a thorough differential, and having a plan to incorporate findings into the care plan is important before applying the probe. Don’t let POCUS make the decision for you.
Great post, I think this is something we should discuss very early on in any PoCUS teaching.
My biggest problems with PoCUS have been premature closure – finding something like a AAA, assuming that is the problem, and not looking further. I have found a porcelain GB and missed a RLL PE, found severe hydronephrosis and missed acute appendicitis…
Agree whole-heartedly that a little knowledge is dangerous and protocols are protective. Our job is to define, refine and teach beginner-proof protocols.
Finally, every practitioner to expect and provide back up plans for errors.
I agree Kylie that premature closure is a big deal with POCUS. You can find a lot of incidental things when scanning an entire abdomen. I think it’s important to have learners ask themselves what clinical question they are trying to answer before applying the probe and how they plan to incorporate any incidental findings. Just like ordering a radiologist interpreted ultrasound for abdominal pain, there should be a differential diagnosis the scan is trying to narrow. Big challenge for beginners to assimilate all the data points POCUS adds to the decision process which requires a lot of supervision and feedback.
So far as I know(?), I haven’t missed a diagnosis due to POCUS, but I have ordered imaging based on my POCUS diagnosis and found a different diagnosis on the formal imaging (similar to the above comment). I’ve also ordered formal imaging where it wasn’t “needed” in the end, such as CT for what turned out to be parapelvic renal cysts (although the CT would have likely been ordered anyway).
Also the POCUS can provide an overwhelming amount of information about the CVS/Lungs/LKKS/Visci/Vascular, etc. Some of this information may be important in the short term (hydration), some like renal cysts, cardiomyopathy may be more of an outpatient problem. I find it’s easy to miss obvious things if you’re bomabarded with U/S findings. Maybe finish the appropriate (history guided) exam and then do focused POCUS.
On the other hand without POCUS, I probably would have stopped at the Zoster diagnosis for the R lower thoracic pain which on POCUS proved to be calculous cholecystitis.
No free lunch with (or without) POCUS, unfortunately. As Greg says, we’re going to have to grow up with POCUS around.
Yes Lloyd, I think we are facing a learning curve in how to incorporate a POCUS enhanced exam into decision making now that we are mastering the ability to generate the images. In some respects image generation has outpaced problem-solving.
I was just having a conversation with a colleague about this topic yesterday. Specifically, we were talking about the misapplication of the FAST exam to stable trauma pts. Like any other ‘test’ POCUS has to be applied in tjhe right patient population to have the desired sensitivity and specificity. With this tool so readily available I worry we apply it too broadly.
Good point Stew.
FAST tutorials often ask for examples of false positives and include ascites or physiologic free fluid in the answer. I have always argued that these are true positives for free fluid, it is the clinical context of that fluid which must be incorporated into the diagnosis. Scanning a cirrhotic patient who fell down is going to generate findings that can’t necessarily rule in a bleed. Scanning someone with a completely benign abdomen on clinical exam means an increased risk of finding a benign cause of free fluid because of the low pre-test probability of bleeding.
I think part of the problem is that we have learned to incorporate the results of our insensitive and nonspecific physical exam (mildly tender to palpation) and history (low risk mechanism) better than POCUS.
Understand the need/desire for protocols, though the cost is never clear. I agree with comments regarding fishing, but juniors need to scan routinely and see non correlating cases in context.
Agree.
Just as learners are asked to do very complete history and physicals initially to develop their physical exam skills and history taking ability, I think it is reasonable to do the same with POCUS. However, supervisors must be constantly teaching how to use the data generated and the limits of that data just like they do with physical exams.