Design: prospective diagnostic accuracy study against invasive exercise haemodynamics, with international multicentre validation
Population: 482 patients with chronic unexplained dyspnoea; 386 with HFpEF, 96 with non-cardiac dyspnoea
Primary outcome: sensitivity and accuracy for HFpEF against invasive exercise testing
Effect: current scores plus exercise echo 55-60% sensitivity, 61-67% accuracy; optimised dual cut points 95-99% sensitivity among definitively classified patients, invasive testing needed in ~30% rather than ~60%

Four hundred and eighty-two patients with chronic unexplained dyspnoea underwent invasive haemodynamic exercise testing with echocardiography recorded simultaneously, the design that lets you score the non-invasive test against the reference standard in the same breath. HFpEF was present in 386 and non-cardiac dyspnoea in 96.

The recommended algorithms - H2FPEF, HFA-PEFF and HFpEF-ABA - combined with exercise echocardiography detected HFpEF with a sensitivity of 55 to 60% and accuracy of 61 to 67%. Adding abnormal resting left atrial compliance, defined as LA reservoir strain divided by E/e', pushed sensitivity to 84 to 85% but drove the false-positive rate to 31 to 43%. The framework that worked used two different thresholds rather than one: rule in on exercise E/e' of 13.8 or more, or resting LA compliance of 1.6% or less; rule out only when exercise E/e' is under 7.2 and LA compliance is above 4.4%; everyone in between remains indeterminate and goes for invasive testing. That produced 95 to 99% sensitivity among the patients it did classify and reduced the proportion needing catheterisation from about 60% to about 30%. It replicated in an international validation cohort.

The honest conclusion is the authors' own: exercise echocardiography cannot supplant invasive testing. What it can do, with strain added, is halve the number of people who need it - provided the report is allowed to say indeterminate rather than being forced to a binary.

Add left atrial reservoir strain to the resting study in any dyspnoea protocol - it is the variable doing the work here
Report a third category: rule-in, rule-out, indeterminate, and say which patients need invasive testing
Do not read a negative exercise echo under the current algorithms as excluding HFpEF
Check your laboratory can measure LA strain reproducibly before adopting the cut points
Where invasive exercise haemodynamics are not available locally, say so in the report rather than over-calling the echo

Why it matters

It reverses the assumption that exercise echo has replaced invasive haemodynamics - it has not, and the pathway many units adopted on that basis is missing cases.

The statistics, in plain English

Sensitivity of 55 to 60% means the current pathway misses roughly four of every ten patients who truly have HFpEF - in a clinic where most referred patients do have it, that is a lot of wrongly reassured breathlessness. The 95 to 99% sensitivity figure applies only to patients the new framework classified definitively; it is not the sensitivity across everyone tested, because the indeterminate group is excluded from that denominator. Note also that HFpEF prevalence here was 80%, far higher than in a general dyspnoea clinic, and predictive values move with prevalence even when sensitivity does not.