Effect of non-invasive calibration of radial waveforms on error in transfer-function-derived central aortic waveform characteristics

Transfer function techniques are increasingly used for non-invasive estimation of central aortic waveform characteristics. Non-invasive radial waveforms must be calibrated for this purpose. Most validation studies have used invasive pressures for calibration, with little data on the impact of non-invasive calibration on transfer-function-derived aortic waveform characteristics. In the present study, simultaneous invasive central aortic (Millar Mikro-tip catheter transducer) and non-invasive radial (Millar Mikro-tip tonometer) pressure waveforms and non-invasive brachial pressures (Dinamap) were measured in 42 subjects. In this cohort, radial waveforms were calibrated to both invasive and non-invasive mean and diastolic pressures. From each of these, central waveforms were reconstructed using a generalized transfer function obtained by us from a previous cohort [Hope, Tay, Meredith and Cameron (2002) Am. J. Physiol. Heart Circ. Physiol. 283, H1150-H1156]. Waveforms were analysed for parameters of potential clinical interest. For calibrated radial and reconstructed central waveforms, different methods of calibration were associated with differences in pressure (P<0.001), but not time parameters or augmentation index. Whereas invasive calibration resulted in little error in transfer function estimation of central systolic pressure (difference -1+/-8 mmHg; P=not significant), non-invasive calibration resulted in significant underestimation (7+/-12 mmHg; P<0.001). Errors in estimated aortic parameters differed with non-invasively calibrated untransformed radial and transfer-function-derived aortic waveforms (all P<0.01), with smaller absolute errors with untransformed radial waveforms for most pressure parameters [systolic pressure, 5+/-16 and 7+/-12 mmHg; pulse pressure, 0+/-16 and 4+/-12 mmHg (radial and derived aortic respectively)]. When only non-invasive pressures are accessible, analysis of untransformed radial waveforms apparently produces smaller errors in the estimation of central aortic systolic pressure, and other waveform parameters, than using a generalized transfer function.