I think we all can agree that the subtleties of the thoracic cavity go far beyond the diagnostic capabilities of our standard two-view chest x-ray. We have robust data that demonstrates the superb diagnostic prowess of bedside ultrasound (US) when compared to the mediocrity of plain films (1, 2,3,4). And yet more information is not always better. Studies examining diagnostic characteristics of a test cannot predict the downstream effects its use will have on patient care. To uncover such consequences, these diagnostic tools have to be tested in the clinical arena. A recent trial did just that.
Published in Chest in 2016, Jones et al examined the effect bedside US had on the diagnosis and management of pneumonia in a pediatric population presenting to the Emergency Department. The authors randomized 191 well appearing children (< 21 years old) in whom the treating physician was concerned for pneumonia, to either a standard work up using a two-view chest x-ray (CXR) or bedside US. These sonographic studies were performed by Pediatric Emergency attendings or fellows, all of whom underwent a 1-hour training course in lung ultrasonography. Patients randomized to the bedside US group were permitted to receive a CXR if the treating or admitting physician felt it was clinically necessary (5).
The authors found a statistically significant reduction (38%) in their primary endpoint, CXR utilization in the patients randomized to the bedside US group. When authors excluded CXRs that were determined to have been performed for non-clinical reasons (requested by admitting physician, parent, guarding or referring physician), bedside US was found to have the potential of reducing CXR utilization by 67%. The authors claim this reduction produced $9,200 in savings. All this with no events of missed pneumonia during the follow-up period (5).
Given US’s superiority to CXR for the diagnosis of pneumonia, these findings are far from surprising (1,2). I imagine similar results would be obtained if the diagnostic capabilities of plain films were compared to CT scan. This trial does provide insight into how the integration of bedside US in general practice may affect clinical management. The authors observed a significant increase in the rate of pneumonia diagnosed in patients randomized to bedside US when compared to the standard treatment arm (28.2% vs 18.1%). This 10% absolute increase consisted completely of diagnoses made by bedside US. In fact, the number of pneumonias diagnosed by CXR in each group was identical (13.6%). Unsurprisingly, this increase in diagnoses coincided directly with a 10% absolute increase in the rate of antibiotic use (37.9% vs 27.3%) in patients randomized to bedside US (5). And yet despite this increase in diagnostic yield, the authors found no difference in clinical outcomes.
This study underscores the subset of sonographically defined pneumonias, that would otherwise go undiagnosed by CXR. In the past these would have been considered bronchitis or URI of viral etiology and treatment with antibiotics would not have been recommended. Given that the clinical outcomes between the two groups were identical, it is reasonable to assume the majority of these would resolve with symptomatic treatment alone. Certainly without bedside US we have not been stricken with a plague of cryptogenic pneumonias undetectable by plain films. US positive, CXR negative pneumonia is a radiographically defined disease that lacks the clinical mass to support its diagnostic reputation. In short, over-diagnosis.
This is not a statistical lark. It is simply the end result of applying a test with more diagnostic accuracy than the one which was previously employed. Furthermore, it is not the first time we have seen the application of bedside US lead to an increase in downstream testing and treatment. In 2014 Laursen et al published an article in Lancet Respiratory Medicine which examined the use of bedside US to evaluate patients presenting to the Emergency Department with shortness of breath. I have discussed this trial in detail in a previous post, but in brief the authors randomized 320 patients presenting to the Emergency Department with respiratory complaints to either a standard work as per the treating physician or the addition of a bedside sonographic assessment of the cardiovascular, pulmonary and deep venous anatomy. Similar to Jones et al, the authors found more pathology in the group randomized to bedside US. These findings led to increased downstream testing and interventions. Despite these changes in management, patients randomized to the bedside US group had a noticeable trend to worse 30-day mortality (12% vs 7% p= 0.13) (6).
I’m sure to some this will appear as an attack against ultrasound, an argument against its use in the Emergency Department. This is of course not the case. Rather this post should be viewed as an examination of the unintended consequences of replacing the old standard with a new more diagnostically robust alternative. Some may argue that the proper response should be to decipher which sonographic findings require antibiotics and which do not. They may very well be right, but I would argue time and time again we have proven ourselves incapable of making such clinical distinctions and will always rule in favor of action.
A test’s true value is not found in its diagnostic accuracy. It cannot be explained simply through its sensitivity or specificity. Rather a test’s worth lies in how it interacts with the surrounding environment, how its is utilized by the clinical operator. Ultrasound is a powerful tool that has already improved many aspects of patient care in the Emergency Department. But its diagnostic prowess comes at the price of information. Information in the form of radiographic disease whose clinical relevance is not always clear. While such findings do not mean we should abandon this valuable bedside tool, it does remind us that it is a test. And like any test we should be mindful in whom it is applied. We owe it to ourselves and our patients not only to hone our skills in the technical aspects of ultrasonography but to understand the downstream consequences of these new found abilities.
Sources Cited:
- Chavez MA, Shams N, Ellington LE, et al. Lung ultrasound for the diagnosis of pneumonia in adults: a systematic review and meta-analysis. Respir Res. 2014;15:50.
- Pereda MA, Chavez MA, Hooper-miele CC, et al. Lung ultrasound for the diagnosis of pneumonia in children: a meta-analysis. Pediatrics. 2015;135(4):714-22.
- Al Deeb M, Barbic S, Featherstone R, Dankoff J, Barbic D. Point-of-care ultrasonography for the diagnosis of acute cardiogenic pulmonary edema in patients presenting with acute dyspnea: a systematic review and meta-analysis. Acad Emerg Med. 2014;21:(8)843-52.
- Ding W, Shen Y, Yang J, He X, Zhang M. Diagnosis of pneumothorax by radiography and ultrasonography: a meta-analysis. Chest. 2011;140(4):859-66.
- Jones BP, Tay ET, Elikashvili I, et al. Feasibility and Safety of Substituting Lung Ultrasound for Chest X-ray When Diagnosing Pneumonia in Children: A Randomized Controlled Trial. Chest. 2016;
- Laursen et al. Point-of-care ultrasonography in patients admitted with respiratory symptoms: a single-blind, randomised controlled trial The Lancet Respiratory Medicine – 1 August 2014 ( Vol. 2, Issue 8, Pages 638-646)
- EM Nerd-The Case of the Partial Cohort - May 24, 2020
- EM Nerd: The Case of the Sour Remedy Continues - January 20, 2020
- EM Nerd-The Case of the Adjacent Contradictions - December 23, 2019
I agree. But I might state it a little differently. Lung ultrasound is an amazing test. But it needs to be applied to patients in a way that you as the doctor are asking a question of the test and have a pretest probability of disease to understand the lung findings. We all have blines at some point. For example 1. If you are in the ED and someone calls an ambulance because they cant breathe. They arrive hypoxic and tachypneic. If you see a few blines per scan zone only this is probably not enough to cause acute decompensation.… Read more »
Brilliant as always! Thank you for your response Vicki! So here is my question/concern and since I have learned approximately 95% of my lung US knowledge from you, I bow to you superior wisdom. Jones et al seemingly did everything right. They included patients who were undergoing a work up for pneumonia, patients we would otherwise obtain a CXR. They Sonographically defined pneumonia as “sonographic finding of lung consolidation with air bronchograms”. The discounted B-lines, confluent B-lines and small subpleural consolidations (with no air bronchograms) as viral etiology. What this suggest to me is what they found was not “false… Read more »
I was going to reply, but then read your amazing summary – couldnt have said it better myself. US provides data, it adds data to what is already known from history and physical exam (as do labs, other imaging, etc), and that data must be interpreted correctly given known limitations (of US or of the patient). The doctor must take all data, including US, and make an informed decision on diagnosis and management. Lung US is still in the discovery process. Sure, lots have been done, it has been shown to be superior to CXR for quite a few pathologies,… Read more »
Excellent points raised. Similar to the situation of U/S finding clinically insignificant pneumothoraces not requiring treatment. However let’s not assume that positive CXR findings equates to bacterial pneumonia, or more specifically pneumonia that improves with interventions such as antibiotics. Perhaps well appearing children should not receive either CXR or U/S for pneumonia initially. It is likely that most of them have a viral etiology on CXR as well as U/S and thus the failure to show benefit of using more antibiotics in the U/S group. If we are comparing two tests of varying sensitivity for detecting a disease that doesn’t… Read more »
Thanks Greg,
Great points and all I can say, I completely agree!
Thanks for the kind words
Amen Greg
Hi Rory
After your excellent post I have responded with my thoughts on the blog here:
http://broomedocs.com/2016/03/pneumonia-ultrasound-diagnostic-strategy/
As Vicki has stated, there is a lot to the concept of meditating on the question: “what do I expect to see here?” before applying the probe. If the findings are not what you expect then either your clinical assessment is wrong or your scan is trying to tell you something?
Change the plan
Cheers
Casey.
” Despite these changes in management, patients randomized to the bedside US group had a noticeable trend to worse 30-day mortality (12% vs 7% p= 0.13) (6).”
Mmmmm
Trend = there is a difference that doesn’t reach statistical significance.
I would deem any trend “to be actively ignored” since a trend is more likely to be a random observation than an effect of the intervention.
Thank you so much for your feedback. While I appreciate your point I take an alternative view. Technically these findings state there is a point estimate of 5% absolute increase in mortality which cannot be statistically distinguished from a 1.6% mortality benefit or a 12% absolute harm. To determine if “a trend” is more likely to be a random observation or an underpowered representation of reality one has to take a more holistic view of the data. The likelihood that any finding is representative of the truth is only partly based of the statistical strength of those observations. It is… Read more »