Echo Assessment of Diastolic Function

Abstract & Commentary

By Michael H. Crawford, MD, Editor

Source: Unzek S, et al. Effect of recommendations on interobserver consistency of diastolic function evaluation. JACC Cardiovasc Imaging 2011;4: 460-467.

The echocardiographic assessment of diastolic function of the left ventricle (LV) involves five measurements derived from two-dimensional imaging, pulsed Doppler, Color M-mode, and tissue Doppler. Not only is it complicated, but sometimes the measurements are discordant. Thus, the American and European echo societies have put out a joint algorithm to follow, but its impact on observer concordance is unknown. Thus, these investigators from the Cleveland Clinic selected 20 patients undergoing echocardiographic LV function evaluation with interpretable studies (no atrial fibrillation, etc.) and a brain natriuretic peptide (BNP) level on the same day for analysis. The studies were interpreted by 18 experts from seven countries without clinical data except for age and sex. They scored diastolic function stage 0, I-III, and estimated filling pressure as high or normal. Their readings were compared to a reference standard of two experienced readers using the new algorithm for diastolic function stage and BNP for the filling pressure estimation (> 100 pg/mL was elevated filling pressure).

Agreement among readers for diastolic class was modest (kappa = 0.62; 1.0 would be perfect) as was estimation of filling pressure (k = 0.61). The sensitivity and specificity of raised filling pressure vs the reference read were 66 and 88%, respectively, and vs BNP > 100 pg/mL were 69 and 93%, respectively. Among the diastolic dysfunction classes, stage I (delayed relaxation) had the best concordance at 92%, stage III (restrictive) was next at 65%, and stage II (pseudo normal) was least at 58%. Normal function was 77%. The authors concluded that estimations of diastolic dysfunction and elevated filling pressure are concordant among readers in the majority of patients, but considerable variability exists.


Reading a paper like this one should make all of us who struggle with measuring and classifying diastolic function feel better. Even international experts have difficulty with this and do not agree at a frequency that inspires confidence. Also, the new American/European society's algorithm did not seem to help.

What is going on here? Assessing diastolic function involves synthesizing the results of several measures, each with their own limitations. Many are affected by age, loading conditions, conduction and rhythm abnormalities, mitral valve disease, and systolic LV function. Also, many of these measurements are technically challenging. In addition, diastole is complex, involving three stages, each with a different physiology.

Why didn't the new algorithm help? Basically it did not go beyond what these expert readers already knew. So it might help less advanced readers, but that remains to be shown. What would help expert readers is an algorithm that factored in known variables such as age and gave a hierarchy of measures to resolve conflicting results. For example, if E/E' is normal, but the left atrium is clearly enlarged, filling pressures are probably increased.

I suspect this is not the last word on the topic as this study had its own limitations. The reference read and BNP standards are imperfect. The number of subjects was small and there were not many with E/E' in the 8-15 range. Expect more to come.