Abstract
Background
Controversy exists regarding how atrial activation mode and heart rate affect optimal AV delay in cardiac resynchronisation therapy. We studied these questions using high‐reproducibility haemodynamic and echocardiographic measurements.

Methods
20 patients were hemodynamically optimized using non‐invasive beat‐to‐beat blood pressure at rest (62±11 bpm), during exercise (80±6 bpm) and at 3 atrially‐paced rates: 5, 25 and 45 bpm above rest, denoted Apaced,r+5, Apaced,r+25 and Apaced,r+45 respectively. Left atrial myocardial motion and transmitral flow were timed echocardiographically.

Results
During atrial‐sensing, raising heart rate shortened optimal AV delay by 25±6 ms (p < 0.001). During atrial pacing, raising heart rate from Apaced,r+5 to Apaced,r+25 shortened it by 16±6 ms; Apaced,r+45 shortened it 17±6 ms further (p < 0.001).

In comparison to atrial‐sensed activation, atrial pacing lengthened optimal AV delay by 76±6 ms (p < 0.0001) at rest, and at ∼20 bpm faster, by 85±7 ms (p < 0.0001), 9±4 ms more (p = 0.017). Mechanically, atrial pacing delayed left atrial contraction by 63±5 ms at rest and by 73±5 ms (i.e. by 10±5 ms more, p < 0.05) at ∼20 bpm faster.

Raising atrial rate by exercise advanced left atrial contraction by 7±2 ms (p = 0.001). Raising it by atrial pacing did not (p = 0.2).

Conclusions
Hemodynamic optimal AV delay shortens with elevation of heart rate. It lengthens on switching from atrial‐sensed to atrial‐paced at the same rate, and echocardiography shows this sensed‐paced difference in optima results from a sensed‐paced difference in atrial electromechanical delay.

The reason for the widening of the sensed‐paced difference in AV optimum may be physiological stimuli (e.g. adrenergic drive) advancing left atrial contraction during exercise but not with fast atrial pacing.