Portal de Periódicos da CAPES

Commentary: Attenuation in invasive blood pressure measurement systems

2006; Elsevier BV; Volume: 96; Issue: 5 Linguagem: Inglês

10.1093/bja/ael070

1471-6771

Ari Ercole,

Blood Pressure and Hypertension Studies

Poor fidelity invasive arterial blood pressure (IABP) traces are a frequent practical problem. It is common practice to describe any such trace as being 'damped'; the resonance behaviour of IABP measurement systems having been extensively described in the literature. However, as poor quality arterial blood pressure signals are seen even with optimal pressure transduction circuits, this cannot be the sole mechanism. In this commentary the classical lumped-parameter Windkessel model is extended by postulating an additional impedance proximal to the site of IABP measurement. This impedance represents any mechanical obstruction to laminar flow. Equations are presented relating measured and actual arterial blood pressures in terms of the model impedances. The reactive properties of such a partial obstruction may lead to an IABP trace that is superficially similar in appearance to the case of an over-damped measurement system. However, this phenomenon should be termed 'attenuation' rather than 'damping' and is probably more common. The distinction is of practical importance as the behaviour of the measured systolic and diastolic pressures is different—both are systematically underestimated and the mean arterial pressure is thus not preserved. Furthermore, this error varies inversely with the peripheral vascular resistance of the tissues distal to the measurement point, therefore apparently magnifying the effect of vasodilatation on blood pressure or derived quantities. Poor fidelity invasive arterial blood pressure (IABP) traces are a frequent practical problem. It is common practice to describe any such trace as being 'damped'; the resonance behaviour of IABP measurement systems having been extensively described in the literature. However, as poor quality arterial blood pressure signals are seen even with optimal pressure transduction circuits, this cannot be the sole mechanism. In this commentary the classical lumped-parameter Windkessel model is extended by postulating an additional impedance proximal to the site of IABP measurement. This impedance represents any mechanical obstruction to laminar flow. Equations are presented relating measured and actual arterial blood pressures in terms of the model impedances. The reactive properties of such a partial obstruction may lead to an IABP trace that is superficially similar in appearance to the case of an over-damped measurement system. However, this phenomenon should be termed 'attenuation' rather than 'damping' and is probably more common. The distinction is of practical importance as the behaviour of the measured systolic and diastolic pressures is different—both are systematically underestimated and the mean arterial pressure is thus not preserved. Furthermore, this error varies inversely with the peripheral vascular resistance of the tissues distal to the measurement point, therefore apparently magnifying the effect of vasodilatation on blood pressure or derived quantities. Invasive arterial blood pressure (IABP) measurement by means of an intra-arterial cannula is a key monitoring technique in high-risk patients both intra-operatively and on the intensive care or high-dependency unit. In addition to giving beat-to-beat blood pressure, the IABP system is increasingly being utilized as the basis of a variety of real-time haemodynamic monitoring systems based on pulse pressure1Linton NWF Linton RAF Estimation of changes in cardiac output from the arterial blood pressure waveform in the upper limb.Br J Anaesth. 2001; 86: 486-496Crossref PubMed Scopus (112) Google Scholar or contour.2de Vaal JB de Wilde RBP van den Berg PCM Schreuder JJ Jansen JRC Less invasive determination of cardiac output from the arterial pressure by aortic diameter-calibrated pulse contour.Br J Anaesth. 2005; 95: 326-331Crossref PubMed Scopus (62) Google Scholar IABP systems use a fluid-filled column to transmit the arterial pulsation to a pressure transducer. The phenomenon of oscillation in this arrangement has been extensively studied and is described in great detail in most texts discussing IABP measurements (see for instance 3Brown BH Smallwood RH. Medical Physics and Physiological Measurement. Blackwell Scientific, Oxford1981Google Scholar). A degree of damping is required to compensate for overshoots attributable to resonance in the measurement system. However, it is well known that excessive damping in the circuit leads to falsely low systolic blood pressure (SBP) and high diastolic blood pressure (DBP) readings in such a way that the mean arterial pressure (MAP) is preserved because of energy conservation considerations. With increasing damping, the pressure trace becomes increasingly sinusoidal, losing fine detail. Such appearances are very common in clinical practice and the traces are often colloquially described as being 'damped'. The causes of excessive damping such as air bubbles and excessively tortuous arterial line circuits are well known and are intrinsic to the pressure transmitting system. However, sub-optimal recordings are often seen even when short, bubble-free equipment is used particularly when the arterial line is 'positional' (i.e. the recorded blood pressure varies with, for instance, wrist position in the case of radial artery lines). This suggests a different mechanism. It is usual in the literature to describe the behaviour of IABP systems in isolation from the arterial tree to which it is coupled; this is a simplification. There is little bulk fluid flow within the IABP circuit and its behaviour is well described as a classically resonant system. Damping is caused by energy dissipation by imperfectly elastic elements of the system. The dynamics of the arterial tree are very different as there is a (pulsatile) bulk flow of blood past the arterial cannula. In this system energy can be dissipated as a result of frictional forces arising from this flow. This is manifested as pressure differences between different parts of the vascular system. The arterial tree is not a resonant system, at least to a first approximation. In particular, what is not generally considered is the effect of any resistance or disturbance to laminar flow proximal to the cannula. In this case a flow-dependent pressure reduction can be expected, and thus the pressure recorded by the IABP system will be reduced. The postulated obstruction may take the form of a physical narrowing through arterial spasm, thrombus secondary to endothelial trauma or any change in vessel geometry especially if it leads to turbulent flow (even locally) as is presumably the case when the arterial lines are found to be 'positional'. Indeed, even without any physical abnormality, the small lumen of the artery from which the pressure is being measured constitutes a resistance in accordance with the Hagen–Poiseuille law. Clearly these considerations apply only to peripheral arterial cannulae such as those in the radial artery but these are the most commonly used. Figure 1a shows an electrical analogy of the two-element Windkessel representation of the vascular system originally conceived by Frank.4Frank O Die Theorie der Pulswellen.Z Biol. 1926; 85: 91-130Google Scholar Although this model is clearly a simplification, it grossly reproduces physiological behaviour and thus remains a useful pedagogical paradigm. Pulsatile flow [ILV(t)] generated by the left ventricle develops a time-dependent voltage Vo(t) across the systemic vascular impedance ZSVR which has resistive (lumped systemic vascular resistance, SVR) and capacitive (from lumped vascular compliance) contributions as indicated. Vo(t) represents the true arterial blood pressure to be faithfully measured. When analysing such circuits, the continuous (direct current or DC) and pulsatile (alternating current or AC) behaviour of the circuitry can be examined independently. Consideration of the DC case leads to the well-known conclusion that the MAP is related to the SVR multiplied by the cardiac output (by analogy with Ohm's law). As all vascular beds are in parallel the peripheral vascular resistance (PVR) of the limb in which the arterial cannula is situated can be separated from the other lumped parameters (Fig. 1b). The vascular resistance of this periphery (e.g. hand in the case of a radial cannula) is represented as an impedance, ZPVR, to indicate that it may also have some reactive components. The SVR and compliance of the remaining vascular beds are adjusted accordingly. An additional arterial impedance to flow (Zart) proximal to the measurement point is postulated to represent the obstruction described above. Examining the behaviour of the combination of Zart and ZPVR, these components behave as a potential divider. Such circuits are used in electronics to reduce voltages or attenuate signals. The total voltage applied Vo(t) (which represents the true IABP) appears across both of these elements in series. The fraction of this voltage which appears across ZPVR is denoted by V(t) (which represents the measured IABP) and depends on the value of this impedance as a fraction of the total, namely: V(t)=Vo(t)ZPVRZPVR+Zart.(1) This equation shows that the measured blood pressure differs from the true blood pressure in a way which depends on both the peripheral vascular tree in which the arterial cannula is sited, and also the impedance of the proximal obstruction. It is instructive to examine the following limiting cases. If Zart is negligible compared with ZPVR (i.e. there is no proximal obstruction) then, V(t) ≈ Vo(t), and the IABP system records a faithful blood pressure. On the other hand, if we compare this with the other extreme, where Zart >> ZPVR, we can ignore ZPVR and see that equation (1) simplifies to give: V(t)≈Vo(t)ZPVRZart.(2) Although the practical situation is likely to lie between these extremes, two important implications of the model are evident from equation (2). Firstly, the recorded IABP is less than the true blood pressure by a factor of 1/Zart. This is different to the behaviour seen with excessive damping as both SBP and DBP (and thus MAP) will be reduced. Secondly, the error in the recorded IABP is sensitive to ZPVR. In effect, under such circumstances the measured blood pressure is not only systematically less than the real value but also artificially sensitive to ZPVR, and by inference to changes to the SVR. This is in addition to the effect of the SVR on true arterial blood pressure from the Windkessel model in equation (1). This situation is mathematically analogous to the precipitous decrease in true blood pressure observed with a reduction in SVR, in the presence of an aortic resistance, as is seen if vasodilators are given to patients with aortic stenosis. As IABP measurement is usually undertaken in circumstances where maintenance of blood pressure is either of particular importance or is likely to be difficult (e.g. in sepsis), it is important to realize that the apparent decrease in blood pressure in response to vasodilators such as anaesthetic agents will be exaggerated in this case. This may lead to sub-optimal management. Thus far only the DC case has been considered as this governs tissue perfusion. A qualitative analysis of the AC behaviour is required in order to understand the likely effect of the addition of a proximal obstruction on the shape of the arterial waveform. It is likely that Zart has a reactive component. This is expected to be inductive as an obstruction moving under the constraint of elastic forces in a pulsatile flow will exhibit mechanical inertia. The magnitude of inductive reactance increases with frequency and |Zart| would thus be expected to similarly increase. If the arterial pressure waveform is regarded as being composed of different Fourier components, the combination of ZPVR and Zart behaves like an electrical low-pass filter. As the fine detail (such as the systolic peak pressure and dicrotic notch) in any signal is represented by its high frequency components, such features are lost, giving rise to a 'smeared out' sinusoidal appearance. Such an appearance is superficially similar to that of damping but the distinction is important, as the behaviour of this system is very different as is described above. It is suggested that a better term would be an 'attenuated trace'. The Windkessel model is certainly a simplification, and a great deal of work by numerous authors has suggested that transmission line networks or lumped parametertransmission line hybrid models better describe the dynamic behaviour of the vascular tree. However, the simple potential-divider approach adopted here is likely to be a valid first-order approximation as the length scales involved are small compared with the wavelengths over which interference effects are likely to be seen. Thus, the assumption that the arterial system is not a resonant one and that damping does not occur outside the pressure transduction system seems reasonable. In summary, it is suggested that attenuation is a more clinically significant problem than over-damping although it has not received attention in the literature. When attenuation is significant, the IABP waveform may appear similar to the case of over-damping but both the measured SBP and DBP will be erroneously low. Furthermore, the haemodynamic response to vasodilatation will appear to be exaggerated which may lead to excessive treatment with vasopressors or inotropes or to inappropriate reduction in administered anaesthetic concentrations intra-operatively. It is important for clinicians to realize this when interpreting beat-to-beat blood pressures or any parameters derived from such measurements. The author is greatly indebted to Dr C. N. Adams for his encouragement and for numerous indispensable discussions. He would also like to thank Drs M. Palmer, E. A. Bright, P. Mills and J. Hall for their helpful comments on the manuscript.