Accuracy of Emergency Physicians in Grading Diastolic Dysfunction Using Visual Estimation of Waveforms

TL;DR

Emergency physicians can visually estimate heart diastolic dysfunction with reasonable accuracy, especially for identifying severe cases.

Contribution

This study evaluates the accuracy of emergency physicians in grading diastolic dysfunction using visual estimation of ultrasound waveforms.

Findings

Overall accuracy for exact grading of diastolic dysfunction was 54.8%.

Emergency physicians had high specificity (90.3%) for identifying severe diastolic dysfunction.

Ultrasound attendings had the highest grading accuracy (70.0%).

Abstract

Diastolic dysfunction occurs when the ventricular walls of the heart stiffen and fail to relax appropriately. Early recognition in the emergency department (ED) enables identification of heart failure with preserved ejection fraction, guides antihypertensive and diuretic therapy, and facilitates timely cardiology referral to reduce morbidity and readmissions. Prior studies show emergency physicians (EP) can diagnose diastolic dysfunction with point-of-care ultrasound using mitral valve inflow velocities and tissue Doppler indices, although quantitative measurements are time-consuming. This study evaluates whether EPs can accurately diagnose and grade diastolic dysfunction based solely on visualization of mitral valve inflow velocities and tissue Doppler wave forms.

After a focused training session, EPs (postgraduate year 1–3 residents, ultrasound fellows, and attendings) were randomized to review archived echocardiograms obtained by certified technicians. The EPs visually assessed echocardiograms for diastolic dysfunction (grades I–III) and whether they were considered “severe” (grade III). Their interpretations were then compared with a cardiologist’s gold-standard readings.

Twenty-three EPs interpreted 100 echocardiograms containing 25 of each grade. Overall accuracy for exact grading was 54.8%. Ultrasound attendings scored highest (70.0%), followed by non-ultrasound fellows (55.0%), attendings (54.0%), and residents (52.9%). For identification of any diastolic dysfunction, the EPs had a sensitivity of 84.6% (95% CI, 78.5–89.5%), specificity of 44.8% (95% CI, 31.7–58.5%), positive likelihood ratio (+LR) 1.53 (95% CI, 1.21–1.95), and negative likelihood ratio (-LR) 0.34 (95% CI, 0.22–0.54). For identification of severe diastolic dysfunction, the EPs’ intrepretations had a sensitivity of 59.4% (95% CI, 46.4–71.5%), specificity of 90.3% (95% CI, 85.0–94.3%), +LR 6.15 (95% CI 3.75–10.09), and -LR 0.45 (95% CI, 0.33–0.61).

Emergency physicians can visually estimate diastolic function using mitral valve inflow velocities and tissue Doppler morphology with good sensitivity for detecting dysfunction and high specificity for identifying severe cases.

Figures