Respiratory transducers requiring a mask or mouthpiece generate volumetric data that is notionally accurate, but they are unsuitable for prolonged use and also cause significant artefactual changes in ventilatory parameters (Perez & Tobin, 1985; Weissman et al. 1984). Accordingly a variety of indirect respiratory transducers has been developed (Sackner & Krieger, 1989; Sartene et al. 1990). One such is the Pneumotrace transducer (PT), based on a piezoelectric device incorporated into an elastic belt (UFI, 545 Main C-2, Morro Bay, CA 93442; UK suppliers: AD Instruments Ltd., Unit 56, Monument Business Park, Chalgrove, Oxfordshire OX44 7RW). There is little or nothing by way of published evaluation of this particular type of transducer.

In order to evaluate the PT’s signal characteristics, a mechanical spring-loaded device was constructed, capable of imposing step extensions on the PT, of variable magnitude and from a range of starting lengths. The transducer showed a phasic response to this type of input, with exponential die-away of time constant 2.4 s. Peak output was linearly related to step magnitude for any fixed starting length of the transducer, but the proportionality constant decreased markedly with increasing starting length. The resulting experimentally-derived expression for the step response was differentiated to give an impulse response function, and Fourier transform of the latter yielded the system response in the frequency domain. This analysis revealed that, if presented with a sinusoidal input of constant amplitude, the magnitude and phase of the PT output would be, respectively, directly and inversely related to input frequency.

In vivo measurements were then made in human subjects (n = 5), using simultaneous recording from a pneumotachometer and PT transducer. The study had local ethical committee approval and the written consent of the participants. The subjects performed both spontaneous resting breathing as well as breaths having exaggerated and suppressed volumes. The results confirmed that the PT imposed a phase change on the signal (under 5 % of average breath cycle time), although this was less than predicted by the sinusoidal modelling data. Integrated pneumotachometer output and PT peak height showed Pearson correlation coefficients of 0.85 to 0.97. Estimates of the error associated with prediction of individual tidal volumes from the PT data indicated confidence intervals of the order of 20-30 %, in line with other non-invasive techniques (Sackner & Krieger, 1989).

In steadily-breathing subjects, the PT is capable of generating useful data for breath cycle time. Breath volume data are more approximate and, in addition, the device yields a less accurate record of ventilation when breathing pattern is unstable. This results both from its phasic responsiveness and also from the dependence of signal magnitude on initial transducer extension.