The use of point of care ultrasound (POCUS) in the Emergency Department is a Bayesian playground for those willing to indulge. Take for example the performance of POCUS in the diagnosis of pulmonary embolism (PE).
The poor sensitivity of bedside echocardiography to identify all-comers with pulmonary embolism is well documented. Most series cite a sensitivity and specificity of 32% to 50% and 90% to 98% (1,2), respectively. Traditional thinking would presume these performance characteristics remain stable independent of the prevalence and acuity of the disease in question. By this logic one would assume in a high risk group of patients presenting with hemodynamic compromise, the sensitivity of bedside echocardiography would not be sufficient to safely rule out thromboembolic disease as the source of a patient’s hypotension. But as Nazerian et al demonstrated in their recent publication in Internal and Emergency Medicine, the diagnostic performance of bedside ultrasound is highly dependent on the clinical context in which it is used (3).
This was a prospective observational study which enrolled patients in two university hospital Emergency Departments. The authors enrolled consecutive patients presenting in shock (defined as a systolic blood pressure (SBP) less than 90 mmHg or a drop of SBP more than 40 mmHg for more than 15 min, with signs of end-organ hypoperfusion), in whom the treating clinician had a suspicion for PE. Patients were excluded if they presented with a clear alternative source of their hypotension.
The patients underwent a focused cardiac and LE venous ultrasound which was performed by one of eight sonographers, who were blinded to the results of the remainder of the patient’s diagnostic workup. Sonographic evaluations had to occur within three hours of enrollment and before any secondary confirmatory studies (CTPA, V/Q, etc.).
The authors sought to evaluate the diagnostic accuracy of right heart strain (as defined by the presence of right/left ventricular end-diastolic diameter ratio 0.9, or right ventricular end-diastolic diameter > 30 mm) in this subset of critically ill patients. They also examined whether LE venous US provided any added diagnostic value. Patients were considered to have PE confirmed as the source of their shock, when PE was was found on CTPA, V/Q scan, subtraction digital angiography or autopsy.
Of the 105 patients included in the final analysis, in 43 (40.9%) PE was determined to be the etiology of their shock. Unsurprisingly, other causes of shock were sepsis (32.4%), acute heart failure (13.3%), severe hypovolemia (4.7%), acute aortic syndrome (2.9%), advanced neoplastic cachexia (2.9%), and cardiac tamponade in one case (1%).
The bedside echo demonstrated notable diagnostic prowess when employed in this subset of patients, sensitivity (91%), specificity (87%), –LR (0.11), +LR (7.03). The sensitivity and –LR were further augmented when the venous US of the LE was included (sensitivity of 95% and –LR of 0.06) in the diagnostic workup.
What this paper illustrates is the concept of spectrum bias, or spectrum effect (4). Spectrum effect represents a shift in a diagnostic test’s performance depending on the population in which it is studied. More specifically, the more obvious a disease state becomes, the better a test functions at differentiating true positives from true negatives. This phenomenon is particularly noticeable with bedside US, whose test characteristics seem to be acutely sensitive to the change in a patient’s acuity.
But it is more than just a shift in the spectrum of disease, it is also a shift in the clinician’s expectations of their sonographic tool. In the undifferentiated population we ask the question, are these patients symptoms caused by a pulmonary embolism. And in that case a bedside echocardiographic assessment provides an insufficient answer. But when the question becomes, “is this patient’s hypotension, caused by a pulmonary embolism?”, bedside US is far more capable of answering such an inquiry.
This effect is not limited to the assessment of PEs. US has demonstrated its performance improves with an increase in the disease acuity, and a shift from the general, to a more specific set of questions and expectations. The FAST exam, has consistently demonstrated its inability identify all intra-abdominal injuries in an undifferentiated set of trauma patients. But when used to identify whether the source of a patient’s hypotension is due to blood in the abdomen it becomes far more accurate (5). Similarly, a two-point assessment for lung sliding will fail to identify a significant number of pneumothoraces when applied to the general population (6), but becomes fairly accurate when answering the simple question, “is this patient’s shock due to a pneumothorax?(7)” Even in patients with abdominal pain in whom the treating physician has a concern for appendicitis, bedside US has demonstrated increasing diagnostic prowess with increasing likelihood of the disease state in question (8).
Many diagnostic modalities, and certainly bedside US, are not the stable diagnostic pillar traditionally thought to be impervious to variations in prevalence and acuity of the disease state in question. Not only should we be aware of the accuracy of the answers provided by our bedside sonographic tests, but the complexity of the questions they are capable of addressing.
- Dresden S, Mitchell P, Rahimi L, et al. Right ventricular dilatation on bedside echocardiography performed by emergency physicians aids in the diagnosis of pulmonary embolism. Ann Emerg Med. 2014;63(1):16-24.
- Nazerian P, Vanni S, Volpicelli G, et al. Accuracy of point-of-care multiorgan ultrasonography for the diagnosis of pulmonary embolism. Chest. 2014;145(5):950-957.
- Nazerian P, Volpicelli G, Gigli C, Lamorte A, Grifoni S, Vanni S. Diagnostic accuracy of focused cardiac and venous ultrasound examinations in patients with shock and suspected pulmonary embolism. Intern Emerg Med. 2017;
- Usher-smith JA, Sharp SJ, Griffin SJ. The spectrum effect in tests for risk prediction, screening, and diagnosis. BMJ. 2016;353:i3139.
- D Stengel et al. Systematic review and meta-analysis of emergency ultrasonography for blunt abdominal trauma. Br J Surg 2001 July; 88(7):901-12.
- Sauter TC, Hoess S, Lehmann B, Exadaktylos AK, Haider DG. Detection of pneumothoraces in patients with multiple blunt trauma: use and limitations of eFAST. Emerg Med J. 2017;
- Helland G, Gaspari R, Licciardo S, et al. Comparison of Four Views to Single-view Ultrasound Protocols to Identify Clinically Significant Pneumothorax. Acad Emerg Med. 2016;23(10):1170-1175.
- Bachur RG, Dayan PS, Bajaj L, et al. The effect of abdominal pain duration on the accuracy of diagnostic imaging for pediatric appendicitis. Ann Emerg Med. 2012;60(5):582-590.e3.
University of Georgetown
Resuscitation and Critical Care Fellowship Graduate