EDITOR:
While a universally accepted measure of volume status remains elusive, measurement of central venous pressure (CVP) remains widely utilized. The inherent requirement for the insertion and maintenance of a catheter within the thorax however garners a wide range of potential morbidity and occasional mortality. Recent reports of correlation between CVP and peripheral venous pressure (PVP) have prompted interest in PVP substitution for CVP. To date, correlation has been reported in the operating theatre and critical care settings for both adult [Reference Tugrul, Camci and Pembeci1–Reference Sahin, Salman and Salman5] and paediatric [Reference Anter and Bondok6] populations, without consensus as to whether reliance on PVP alone can be endorsed.
Investigators of isolated vascular beds have described a ‘waterfall effect’ to explain blood flow through collapsible tubing deformed by external pressures. Inherent in this concept is the summative effect of distending intraluminal, and compressive extraluminal pressures on vessel cross-section, and thereby resistance. We hypothesized the existence of vascular waterfall phenomena in the peripheral venous conduits of the upper limb that may impact the relationship between measured PVP and CVP. Specifically, we postulated elevated and erroneous PVP estimation of CVP at low CVP values secondary to venous collapse and associated elevation in conduit resistance. We therefore determined to compare CVP and PVP in mechanically ventilated subjects with emphasis on the observed differential.
The investigation was approved by the Northern Y Regional Committee of the New Zealand Health and Disability Ethics Committee. Thirty-four patients (27 male, 7 female) undergoing elective cardiac surgery (27 coronary artery grafting (CAG), 3 aortic valve replacement (AVR), 1 atrial septal defect repair, 3 combined CABG/AVR) were studied.
Preoperatively, 16-G peripheral catheter-over-needle cannula were sited in left antecubital fossa veins. Central venous catheters (7 Fr, 3 lumen, 16 cm; Arrow International, Reading, MA, USA) were introduced following induction of anaesthesia via the right internal jugular vein. Truwave pressure transducers (Edwards Lifesciences, Irvine, CA, USA) connected to Solar 9500 bedside monitors (Marquette Medical Systems, WI, USA) were used for acquisition of CVP and PVP.
CVP recordings were made via the distal lumen of the central venous catheter following mechanical flushing and visual confirmation of the CVP waveform. PVP recordings were performed following transfer of the CVP pressure transducer to the antecubital fossa catheter with the arm in the mid-thoracic position. Mean pressure values were recorded at end-expiration. Central and peripheral transducers were room air calibrated contemporaneously at the estimated phlebostatic axis in addition to undergoing inverted ‘U-tube’ levelling to eliminate zero error difference. Pressures recorded represented a time-weighted average of venous pressure waves as per manufacturers software.
Patients were studied on the first postoperative day in ICU while intubated and sedated with propofol infusion. Ventilator settings provided tidal volumes of 5–7 mL kg−1 in the absence of positive end-expiratory pressure. Maintenance fluid was 5% dextrose infused at constant rate of 1 mL kg−1 h−1. Inotropic and vasoactive infusions (epinephrine in 1, milrinone in 3, norepinephrine in 11) were continuing at the time of data acquisition.
CVP/PVP parings were recorded on return from theatre and subsequently before and after undergoing volume expansion. Volume expansion (250–500 mL Hemohes® (Pentastarch 10 g 100 mL−1 in 0.9% saline)) was undertaken to maintain haemodynamic parameters (mean arterial pressure (MAP) > 65 mmHg) according to local protocol, or at the discretion of the attending intensivist.
Statistical analysis of all variables was performed using SPSS for Windows (version 10, SPSS, Chicago, IL, USA). Linear regression analysis was performed to establish correlation between CVP and PVP. t-Test was used when comparing mean data. Kruskal–Wallis analysis of variance was used for subgroup analysis of PVP less CVP according to CVP. All data is presented as mean ± SD.
Fifty-four initial and pre-infusion paired CVP/PVP measures were obtained with 37 further parings recorded following volume expansion. Seventeen of these were in response to a MAP ⩽ 65 mmHg. At the time of initial sampling, pulse rates and MAP were 78 ± 11 bpm, and 72 ± 9 mmHg, respectively. Mean volume infused was 382 ± 180 mL (4.4 ± 1.9 mL kg−1). CVP was 10 ± 3.6 mmHg prior to volume expansion rising to 12.6 ± 3.5 mmHg following Pentostarch infusion (P = 0.001). Similarly PVP rose from 11 ± 3.7 to 13 ± 3.3 mmHg (P = 0.007).
PVP was found to be greater than CVP by 1.13 ± 1.9 mmHg (P < 0.001). Mean increments in CVP and PVP recordings (∂CVP and ∂PVP) were 2.6 ± 2.6 and 1.9 ± 3.3 mmHg, respectively. There was a trend towards change in PVP being less then change in CVP (∂CVP − ∂PVP was 0.64 mmHg, P = 0.084).
Linear regression analysis of the initial paired recordings revealed good correlation between PVP and CVP (r = 0.864, P = 0.01). Regression of these PVP/CVP pairings yielded the equation PVP = 2.88 + (0.84 × CVP). The regression equation for PVP − CVP was 2.88 − (0.16 × CVP), (r 2 = 0.22, P = 0.01). Bar plot demonstrating distribution of PVP − CVP for CVP groupings are shown in Figure 1.
Figure 1 Distribution of (PVP − CVP) for CVP groupings. Group 1 CVP ⩽ 8 mmHg. Group 2 CVP 9–12 mmHg. Group 3 CVP ⩾13 mmHg. PVP: peripheral venous pressure; CVP: central venous pressure.
Discussion
These data demonstrate a negative correlation between central and peripherally acquired venous pressure differential and absolute CVP. Subgroup analysis proporting reduction in PVP − CVP with rising CVP, and demonstration of a trend towards volume infusion resulting in lesser PVP increment lends additional support to the study hypothesis that PVP – CVP is a decreasing function CVP. An alternative statement of this relationship is that the correlation between PVP and CVP improves at higher CVP values. This finding is not new having been previously reported in a series of hepatic transplant patients undergoing investigation for PVP/CVP correlation [Reference Hoftman, Braunfeld and Hoftman4]. These authors, likewise, postulated that venous collapse may have contributed to observed PVP/CVP divergence.
We have additionally observed good correlation between PVP and CVP over a wide range of CVP values consistent with a number of recent reports demonstrating statistically significant correlation [Reference Tugrul, Camci and Pembeci1–Reference Sahin, Salman and Salman5]. The limits of agreement in this study are however insufficient to advocate substitution of PVP for CVP in routine clinical practice.
The conceptual framework of the present study – investigating factors that impact on difference between CVP and PVP – represents a break from previous work, which has sought correlation alone. That we have reported positive results within this construct has implications for future research. Adopting a similar approach may be gainfully employed in designing future studies to identify subgroups of patients in which PVP and CVP are clinically interchangeable.
Earlier work [Reference Kamm and Shapiro7] describing flow in compressible tubes as a function of transmural pressure described three states of resistance: (1) Collapsed conduit, where resistance is high but constant and flow occurs through open channels at the lateral aspects of the collapsed tubes only; (2) Partially open conduit, where the relationship between driving pressure (PVP − CVP) and flow is complex and non-linear. Increasing intraluminal pressure dilates the tube leading to declining resistance; and (3) Open conduit, where resistance is constant. We suggest the variation in resistance in the peripheral veins with differing transmural pressures (i.e. classic vascular waterfall effect) accounts for the relationship between PVP − CVP and CVP observed in this and other studies. Such consideration permits the cogent but unexplored hypothesis that the accuracy of PVP estimation of CVP increases markedly when unimpeded venous patency is achieved i.e. at a point wherein CVP exceeds the compressive effect of tissue pressure on venous conduits.
This study has a number of limitations. Investigators were not blinded to knowledge of the CVP before assessing PVP potentially introducing observer bias. A number of patients were at the time of data collection receiving infusions of inotropic or vasoactive medications. The effect of these on peripheral venous tone may have influenced the recorded PVP measurements. Additionally, temperature, which is known to affect correlation between CVP and PVP perhaps through an effect on venous tone [Reference Sahin, Salman and Salman5], was not recorded. Tissue oedema may furthermore have affected the relationship between CVP and (PVP − CVP). As has been hypothesized tissue pressure may represent a critical opening pressure at the ‘top of the waterfaIl’ below which the relationship between PVP − CVP and CVP inherently non-linear. Given the small numbers in this study however, analyses other than simple linear regression to seek such a relationship were not undertaken. Finally, this study was undertaken in a population of intubated post-cardiac surgical patients. As such, the findings may not be generalizable to the more heterogeneous ICU or ward population.
In conclusion we have demonstrated (PVP – CVP) to reduce with increasing CVP, which we suggest is due to variable resistance in collapsible veins at differing transmural pressures. The clinician contemplating use of PVP in volume state assessment may find the approach of considering potential for a large PVP − CVP useful.
