Is there any advantage to using an arterial transfer function?

(typed double-spaced) in length and may be subject to editing or abridgment. To the Editor: We read with interest the paper by Millasseau et al, 1 which raises the question of whether the application of an arterial transfer function to noninvasively acquired radial artery pressure waveform data is necessary for the estimation of central aortic waveform characteristics. Although the conclusion that this approach offers no advantage over the simple analysis of untransformed radial artery pressure waveforms is valid, it should be remembered that the approach of these authors demonstrates only comparability and does not add to the debate as to whether arterial transfer function techniques enable accurate estimation of central aortic waveform parameters, because no directly measured central aortic waveforms were available for comparison. Our own findings, derived from noninvasive radial waveforms compared with directly measured central aortic waveforms, have demonstrated close relationships between several radial and measured central aortic waveform parameters proposed to be of potential clinical value, such as systolic and diastolic pressure time integrals. 2,3 However, the radial artery augmentation index (AI) was unrelated to, or at best weakly correlated with, the directly measured central aortic AI. 2,3 The poor correlation and wide limits of agreement between directly measured central aortic and transfer function– derived AI observed by most authors is explained by the persisting close correlation between the radial AI and that derived by the application of a generalized transfer function to this data, which has been confirmed by Millasseau et al 1,3– 6 Those who have reported correlation between measured and derived values have done so using combined data sets including repeated data from each individual before and after an intervention. 7,8 Given that such repeated values are correlated within the individual unpublished observations), this practice will falsely strengthen any underlying correlation or indeed create an apparent statistical correlation where no such correlation exists in the baseline data. Millasseau and colleagues have demonstrated changes in the radial AI with interventions that are correlated with the AI of the waveform derived by the application of an arterial transfer function to the same data. 1 As we have previously noted and as is reiterated in the editorial accompanying this paper, given that the radial artery waveform contains all the data utilized to synthesize the transfer function– derived waveform, this is not surprising. 9,10 These findings do not, however, indicate that measured central aortic AI is necessarily changing …