Design: Diagnostic accuracy study against invasive exercise haemodynamics, with external multicentre validation
Population: 482 patients with chronic unexplained breathlessness (386 HFpEF, 96 non-cardiac)
Primary outcome: Sensitivity and accuracy for invasively confirmed HFpEF
Effect: Current approach sensitivity 55% to 60%; dual cut points 95% to 99% in classified patients, invasive testing needed in about 30% vs 60%

Exercise stress echocardiography is offered as a non-invasive alternative to invasive haemodynamic testing for heart failure with preserved ejection fraction (HFpEF). This study ran both at once in 482 patients with chronic unexplained breathlessness, 386 of whom had HFpEF on invasive testing, and validated the findings in an international cohort.

Using the current scores (H2FPEF, HFA-PEFF, HFpEF-ABA) with standard exercise echo, sensitivity was only 55% to 60%. Adding resting left atrial compliance (reservoir strain divided by E/e') raised sensitivity to about 85% but increased false positives to 31% to 43%.

The workable answer was two sets of cut points. HFpEF was ruled in by exercise E/e' of 13.8 or more, or resting left atrial compliance of 1.6% or less; it was ruled out only when exercise E/e' was below 7.2 and compliance above 4.4%. Everyone in between went on to invasive testing. This classified 95% to 99% of definite cases correctly and cut the share needing invasive testing from about 60% to about 30%.

A normal exercise echo does not exclude HFpEF in a breathless patient with a suggestive profile.
Ask your echo lab whether left atrial reservoir strain can be reported at rest.
Patients in the indeterminate middle still need invasive exercise haemodynamics for a firm diagnosis.
Record E/e' at peak exercise, not only at rest, when HFpEF is the question.

Why it matters

A test widely treated as a substitute for invasive haemodynamics turns out to miss many of the patients it is used to find.

The statistics, in plain English

Sensitivity of 55% to 60% means the current approach labelled only about half to three in five of true HFpEF cases. The two-threshold approach gets accuracy up by declining to classify the middle group, so its high sensitivity applies only to those it did classify.