Scott Sparks, MD and Mark Favot, MD, FACEP
Imagine the following phone call:
ED Physician: “Hey, I need to admit an elderly female for fever.”
Hospitalist: “What’s the source?”
ED Physician: “Hmmm … I don’t really have one. Urine and Chest X-Ray are negative, and the mental status is normal.”
Hospitalist: “Can’t they go home?”
ED Physician: “No, too high of a risk.”
I am certain that many of us have had the same conversation while working. Read through Tintinalli, Rosens, etc, and the cited mortality rates in the elderly are between 7 to 10%. With these poor outcomes, when do we order three blood cultures for the fever of unknown origin and presumptively treat for endocarditis?
Traditional thinking about the diagnostic capabilities of transthoracic echocardiography (TTE) for infective endocarditis (IE) is that TTE is helpful when it demonstrates a vegetation (ie, it has specificity); however, when there is a high clinical suspicion for the disease, a trans-esophageal echocardiogram (TEE) is necessary. Previous studies showed a poor sensitivity for TTE, ranging from 44-75%, with current guidelines from multiple specialty societies recommending TEE if the pre-test probability for IE remains intermediate following a negative TTE (2). This month’s cardiac ultrasound journal watch brings us an article (1) that, although retrospective in nature, attempts to throw a wrench in traditional thinking and demonstrate a higher sensitivity for TTE.
Joe Sivak, MD and the folks at Duke University Medical Center performed a retrospective analysis of the Duke Echocardiography Lab Database. They searched for adult patients who underwent TTE followed by TEE within 7 days. Patients with prior valve repair/replacement, complex congenital heart disease, prior heart transplantation, or left ventricular assist devices (LVADs) were excluded.
Two separate analyses were performed to evaluate the diagnostic characteristics of TTE. In the standard definition, positive or negative results on TTE were used, based solely on the presence of absence of vegetation. In the strict definition, studies not meeting these criteria were considered to have positive or indeterminate findings; however, negative findings on TTE were if they met the following criteria:
During the study period 3495 TTE studies were performed to evaluate for IE, 790 met the inclusion criteria and were followed by TEE within 7 days. Among the 790 paired TTE & TEE exams, 104 of the TTE’s met the strict negative criteria (13.2% of al included TTE’s). The strict negative criteria group was significantly younger, with lower rates of congestive heart failure, coronary artery disease, chronic kidney disease and hypertension. The proportion of patients with positive blood cultures was similar in each group, as was the distribution of infectious organisms.
Using the standard approach, 661/790 TTE’s were read as negative, and 129 were read as positive for vegetation. Of the 661 negative TTE’s, 89 had vegetations on subsequent TEE. The sensitivity was (68/157) 43% (95% CI, 36-51%) with a negative predictive value of (572/661) 87% (95% CI, 84-89%). The negative likelihood ratio was 0.627 (95% CI 0.546-0.721).
Using the strict negative criteria, 104 TTE’s had negative results (meeting all of the strict negative criteria), and 686 had positive results (did not meet all of the strict negative criteria). Of the 104 patients meeting strict negative criteria, 3 had evidence of vegetation on subsequent TEE. The sensitivity was (154/157) 98% (95% CI 95-99%), with a negative predictive value of (101/104) 97% (95% CI 92-99%). The negative likelihood ratio was 0.120 (95% CI 0.038-0.373).
Use caution when interpreting the results. Good quality TTE using the Duke Guidelines may help lower the probability of missing endocarditis early in development. When grandma needs to be admitted for fever of unknown origin, PoCUS users can better evaluate for acute infectious processes and feel better that the hospitalist can monitor cultures until her recovery.