INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

THERE IS A GROWING INTEREST in monitoring physiological parameters noninvasively, both in research and in clinical situations. Blood pressure can now be measured noninvasively and continuously in the finger by using Finapres, which is based on the volume-clamp method of Peñáz (15, 22). One of the attractive derivatives of this method is the calculation of stroke volume (SV) on a beat-to-beat basis, a technique originally developed for use with the intra-arterial blood pressure wave (aorta or brachial artery). Several models (1, 7, 17, 19, 23) can be used to calculate SV from pulse contour analysis. We have many years of experience with the method of Wesseling and co-workers (23), which is based on a transmission line model of the circulation. In this model, corrections are used for age-dependent, dynamic changes in the aortic impedance due to changes in arterial pressure and heart rate (HR), and also for reflections of the pressure waveform from the periphery. Any pulse contour method will give only relative changes in SV. Calibration against an absolute method is necessary when absolute values are required.

In an earlier study (18), in which cardiac output (CO) was influenced by changes in volume status of the test subjects (by intravenous saline loading and lower body negative pressure), we demonstrated that the method of Wesseling et al. (23) can be successfully applied to the noninvasively measured pressure wave of the finger in a close-to-supine [6° head-down tilt (HDT)] position. Acetylene (C₂H₂) rebreathing (CO_rebr) was used as the absolute reference method for CO measurement. The study showed the calibration factor (CO_rebr-to-CO_pc ratio), where CO_pc is CO measurement by pulse contour analysis, to be stable over a period of weeks.

A major drawback of earlier validation studies (10, 18, 23) is that comparisons are made in only one, mostly supine, body position. Finapres is used in the upright and supine positions during the same session, e.g., to track beat-to-beat blood pressure changes during orthostatic tolerance testing (9, 20). Because body position tends to alter CO and the pulse wave at the same time, an evaluation explicitly testing the reliability of pulse contour analysis at different body positions seemed indicated. We therefore tested the pulse contour method applied to the Finapres waveform during orthostasis, using CO_rebr as a reference.

We used two different protocols: 1) standing vs. a close-to-supine (6° HDT) position and 2) four different tilt angles on a tilt table.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Six healthy male volunteers (age 23-30 yr) participated in the tilt-table study (subjects T1-T6). Written informed consent was obtained from all subjects. The study was approved by the Medical Ethics Committee of the Academic Medical Center of the University of Amsterdam.

Study Protocols

Standing vs. 6° HDT. Data were collected from subjects in the standing position and during a brief (20-min) episode of 6° HDT before the actual 10-day, extended HDT period, and on the first and fifth day of recovery. Rebreathing measurements were performed, using a bag-in-box system with a 3-liter rubber anesthesia bag. A gas mixture of 1.5 liters [0.3% C¹⁸O-0.5% C₂H₂-1.0% He-98.2% air (78.1% N₂-21% O₂-0% CO₂-0.9% inert gases)] was used. The protocols for the experiments performed at DLR have been described elsewhere (18) in detail. In all subjects a rebreathing frequency of 24 breaths/min was used. Rebreathing measurements were performed in duplicate both in the standing and in the 6° HDT position. The washout period between duplicates was at least 2.5 min.

Tilt table. Rebreathing measurements were performed, using a bag-in-box system with a 2.3-liter rubber anesthesia bag. A gas mixture of 2 liters [1% C₂H₂-10% He-89% air (78.1% N₂-21% O₂-0% CO₂-0.9% inert gases)] was rebreathed for 30 s. Fifteen seconds before rebreathing, a metronome was started for pacing the respiration rate. Then, after a normal expiration, the test subject changed from breathing normal air to breathing the bag-gas mixture by using a hand switch. Gas partial pressures (N₂, O₂, CO₂, He, and C₂H₂) were monitored at the mouth by means of a mass spectrometer (Centronic 200 MGA). Gas partial pressures were corrected to a sum of 99.2%. Argon (not measured) was assumed to account for 0.8%. Respiratory flow was measured by using an Alveo Test (Erich Jaeger, Würzburg, Germany) flowmeter. In subject T1, a rebreathing frequency of 24 breaths/min was used, which was lowered in subjects T2-T6 to 15 breaths/min.

Data Analysis

Rebreathing. Data were evaluated by using the continuously ventilated two-compartment lung model of Hook and Meyer (8). In this model, the equilibration of partial pressures of a soluble perfusion-limited gas (C₂H₂) in two compartments (rebreathing bag and alveolar space) is described by a biexponential process with a fast (k₁) and a slower (k₂) rate constant, where k₁ represents the bag-alveolar gas mixing and k₂ the alveolar-lung capillary blood-gas transfer. Both k₁ and k₂ are derived from a semilogarithmic plot of the end-expiratory partial pressures against time of rebreathing. The time 0 intercept of k₂ is used for calculating the initial C₂H₂ concentration. Knowing the alveolar-to-blood conductance of C₂H₂, the pulmonary capillary blood flow can be derived according to
where c is pulmonary capillary blood flow, _g and _b are the capacitance coefficients of C₂H₂ for gas and blood respectively, V_R is the rebreathing bag volume, S_A is the zero intercept of the slower alveolar exponential component, and k₁ and k₂ are rate constants of the fast and slow alveolar compartments, respectively. Model calculations showed the model to be relatively insensitive to changes in respiration rate, dead space, alveolar volume, tidal volume, and the ratio of tidal volume to bag volume.

Pulse contour analysis. Beat-to-beat values for systolic, diastolic, and mean pressure and HR were derived from the pressure wave. To calculate SV from the pressure wave (SV_pc), the method of Wesseling et al. (23) was used, which relates the pulsatile systolic area (PSA; area under the pressure curve from the start of the upstroke to the incisura) and SV_pc by the effective characteristic impedance of the aorta (Zao). In this method, corrections are used for age-dependent, dynamic changes in Zao because of changes in arterial pressure and HR, which are due to changes in aortic cross section and compliance (both pressure and age dependent) and to reflections of the pressure waveform from the periphery (which is influenced by HR). SV for each beat was calculated from the PSA with the correction formula

where HR is the momentary HR (in beats/min), MAP is mean arterial pressure (in mmHg) of the same beat, and Zao_ini is a first approximation of Zao and is given the value (90 + age)/1,000. As Zao is not actually known, only uncalibrated SV is derived. Beat-by-beat CO_pc was then calculated by multiplying SV_pc by the HR belonging to the same beat. A more comprehensive description of the method is given in the literature (18, 23).

Statistics

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Standing Vs. 6° HDT

We calculated CO_{cal sup} in both standing and 6° HDT positions, before and during rebreathing. On average, the decrease in CO_{cal sup} in response to standing was one-half that of CO_rebr. Standing CO_{cal sup} was significantly different from the 6° HDT position before and on the first day after the HDT period, both during and just before the rebreathing measurement.

As shown in Fig. 1A, when we took 6° HDT as the reference position, we found a large range in the individual change of R_rebr/pc (the ratio between the methods) during standing.

View larger version (19K): [in this window] [in a new window]   Fig. 1.   Ratio of cardiac output (CO) measured by acetylene rebreathing to that measured by pulse contour analysis (Rrebr/pc) for different body postures. A: subjects S1-S6 in both 6° head-down tilt (HDT) and standing postures on 3 different measurement days. Values are means ± SD of duplicate measurements. B: subjects T1-T6 at tilt angles of 30° (30° HDT), supine, +30° [30° head-up tilt (HUT)], and +70° (70° HUT). Values are means ± SD of triplicate measurements.

Subjects S1 and S6 showed almost no change (<6%), and in subject S4 there was only a small decrease (8%) in R_rebr/pc between HDT and standing. In contrast, in subjects S2 and S5 there was a moderate-to-large decrease (14-34%) in R_rebr/pc. In subject S3 there were large changes (>20%) due to standing, but the direction of the change was inconsistent among the three measurement periods.

In the 6° HDT reference position, R_rebr/pc was constant within 9% in subjects S1-S5 and within 12% in subject S6 over the total of six measurements in each subject (3 days, duplicate measurements). In the standing position, differences were larger and ranged from 8 to 15%, with one outlier of 34% in subject S3.

When body position was changed from 6° HDT to standing, average R_rebr/pc for six subjects changed from 1.04 ± 0.09 to 0.96 ± 0.13 (P < 0.05) before HDT, from 1.03 ± 0.10 to 0.90 ± 0.10 (P < 0.01) on the first day of recovery, and from 1.01 ± 0.10 to 0.87 ± 0.16 (P < 0.01) on the fifth day of recovery. Proportional changes in R_rebr/pc when standing were 7 (range +13 to 27%), 12 (range 0 to 28%), and 14 (range +4 to 33%), respectively.

The difference between CO_rebr and CO_{cal sup} for the 6° HDT and standing postures are shown in the Bland-Altman (4) scattergrams of Fig. 2A. The overall differences (all subjects, all measurement days) are 0.0 ± 0.2 l/min in 6° HDT and 0.7 ± 0.8 l/min in the standing position.

View larger version (1118K): [in this window] [in a new window]   Fig. 2.   Bland-Altman scattergrams of all data points. A: measurements during standing and 6° HDT. B: measurements in 4 body positions during tilt experiments. COcal sup and COrebr, calibrated CO measurement by pulse contour analysis at a (near-)supine position and CO measured by acetylene rebreathing, respectively. Horizontal lines, means ± 1.96 SD.

The MAP difference between the 6° HDT and standing position was smallest before the HDT period (14%) and largest on the first day of recovery (31%) (Table 2). The average response in HR was somewhat larger after the HDT period. HR increase during rebreathing compared with prerebreathing was only marginally different and independent of the measurement day.

Tilt-Table Maneuvers

We calculated CO_{cal sup} in all four positions, both before and during rebreathing. During rebreathing, three of six subjects showed decreases in CO_{cal sup} after being tilted to angles of 30 and 70°. In these subjects the decreases in CO_{cal sup} were, on average, one-half those of CO_rebr. In two subjects CO_{cal sup} did not change, and in one subject it even increased. Also during tilt, we found a large range in the individual change in R_rebr/pc when using supine as the reference position (Fig. 1B). Average R_rebr/pc for six subjects changed from 1.27 ± 0.34 in the supine position to 1.06 ± 0.34 (P < 0.01) at 30° and to 0.96 ± 0.39 (P < 0.01) at 70°. Proportional changes in R_rebr/pc during tilt were 17% (range 13 to 29%) and 27% (range 11 to 44%), respectively. Thus in all subjects there was an overestimation of CO_{cal sup} in the HUT positions compared with supine, but of different magnitude in each subject.

Bland-Altman scattergrams (4) of differences between CO_rebr and CO_{cal sup} are shown in Fig. 2B for each tilt position. The overall differences (all subjects, triplicate measurements) are 0.0 ± 0.6, 0.3 ± 0.5, 0.9 ± 0.6, and 1.5 ± 0.8 l/min for supine, 30° HDT, 30° HUT, and 70° HUT, respectively.

In all subjects MAP increased significantly during tilt from supine to 30 and 70° HUT. During tilt from supine to 30°, MAP increased slightly, but this increase was not significant (Table 4).

In five of six subjects HR increased with tilt angle (30 to 70°). Only in subject T3 did HR decrease with tilt angles from 30 to 30° and increase at 70° but did not reach the value at 30°. Averaged values are presented in Table 4. The HR increase during rebreathing compared with prerebreathing was affected only slightly by the tilt angle: 7, 6, 10, and 7 beats/min, respectively, for the four tilt angles.

Prerebreathing Pulse Contour Measurements

In the standing vs. 6° HDT experiments, CO_{cal sup} increased between 0.6 and 0.7 l/min (10-15%) due to the rebreathing maneuver. Results were similar for both the 6° HDT and standing postures (Table 1). In the 6° HDT position this increase was primarily caused by the increase in HR, whereas in the standing position both HR and SV_{cal sup} increased; however, again, differences among subjects were large.

In the tilt experiments, the increase in CO_{cal sup} during the rebreathing period was largest in subject T1, whose rebreathing was performed at a higher respiration rate than in the other five subjects. On average CO_{cal sup} increased by 0.5 l/min (9.4%), 0.4 l/min (7.8%), and 0.4 l/min (9.1%) at tilt angles of 30° HDT, 0°, and 30° HUT, respectively, and by 0.9 l/min (19.0%) at 70° HUT (Table 3). This increase was primarily due to an increase in HR compared with prerebreathing, but it also occurred at 70° HUT in subjects T1 and T2 partly by an increase in SV_{cal sup}.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In the present study we made a direct comparison of CO measurements by using pulse contour analysis at different body positions with an absolute, and noninvasive, reference method.

To investigate the rebreathing maneuver itself, we searched the literature for studies that compared C₂H₂ rebreathing to other CO measurement methods in different body postures. We found none. One might reason that in the HUT compared with supine position a change in distribution of the ventilation-perfusion ratio (A/) is to be expected (24). In a study using C₂H₂ rebreathing in dogs, Friedman et al. (6) did not find a significant difference in estimation of pulmonary blood flow when A/ was changed over a large range by occlusion of either one pulmonary artery or one main bronchus. Using a three-compartment computer model, Burma and Saidel (5) found a small overestimation (<10%) of pulmonary blood flow in the upright position compared with supine, when the A/ distribution in the upright position as described by West (24) was used. A small underestimation was found by Petrini et al. (16), who used a two-compartment model with changing A/. Verbanck and Paiva (21), in a model study, found only a very small influence of A/ on CO estimation. From these studies we concluded that the rebreathing method can be used very well as a reference CO measurement method even under conditions where A/ is changing.

In the standing vs. 6° HDT experiments, we used the 6° HDT position as a reference. This position was part of the HDT study measurement protocol in which the supine position was not planned. As our later tilt experiments indicate only a small but not significant difference between supine and 30° HDT, we consider it admissible to use the 6° HDT position as a substitute reference position. In three of six subjects (S1, S4, S6), differences between the pulse contour and rebreathing methods were small or absent, but in the other three subjects we found a substantial overestimation of CO_{cal sup} compared with CO_rebr in the standing position. Differences between the duplicate measurements within each subject were small.

In the tilt experiments, we found CO_pc in the HUT position to overestimate CO in all six subjects. Among subjects, variability in responses to tilt was high using the pulse contour method, but within each subject the ratio between pulse contour and rebreathing was consistent within the triplicate measurements. In contrast to the first part of the study, in which duplicate measurements were made shortly after the initial measurement and without an in-between change of body position, triplicate measurements during the tilt experiments were made during a random sequence of 12 measurements in 4 positions. This, and the lower rebreathing frequency used, may explain the larger SDs found within the triplicate measurements in the tilt experiments (Fig. 1B).

In the supine position, the calibration factors in four of six subjects (T1-T4) were within the range found in our earlier study: 0.91-1.24 (18). In the other two subjects the calibration factors were 1.58 and 1.77 (Fig. 2B). CO_rebr values in the supine position for these subjects were not different from those for the other four subjects, but the variance in R_rebr/pc was larger at all tilt angles. This was due to a larger variance in the rebreathing measurements. Pulse contour measurements were consistent within each triplicate measurement. Also, in these two subjects there was no further decrease in R_rebr/pc from 30 to 70° HUT.

As already found in our earlier study (18), the rebreathing procedure itself changes the cardiovascular status of the subject: both HR and CO_{cal sup} increased compared with prerebreathing values. In the standing vs. 6° HDT study, which followed the same protocol, all measurements were made by using a rebreathing frequency of 24 breaths/min. The increase in CO_{cal sup} due to the rebreathing maneuver was ~0.7 l/min in both positions, but, especially in the standing position, differences among subjects were large. During the tilt experiments, CO_{cal sup} increased less during rebreathing in the supine reference position compared with the 6° HDT reference position (0.4 vs. 0.7 l/min) but slightly more (0.9 vs. 0.7 l/min) at 70° HUT compared with standing.

Tilt experiments were started in subject T1 by using a rebreathing frequency of 24 breaths/min. We decreased the rebreathing frequency to 15 breaths/min in the other five subjects to minimize the effect of rebreathing on CO_{cal sup}, although a less accurate estimation of the k₂ rate constant can then be expected, because of less available data points for the curve fit. Despite the different breathing frequency and the consequently larger increase in CO_{cal sup} due to the rebreathing maneuver, subject T1 produced differences between CO_{cal sup} and CO_rebr similar to those in subjects T2-T6. Therefore, we included the results from this subject in this study. When subject T1 is excluded, the average increase in CO_{cal sup} during rebreathing is reduced to 0.3, 0.2, 0.2, and 0.7 l/min for the tilt angles of 30° HDT, 0°, 30° HUT, and 70° HUT, respectively.

In the 6° HDT and standing positions the differences between CO_{cal sup} prerebreathing and during rebreathing were approximately equal (0.6-0.7 l/min, Table 1). Therefore, the difference between pulse contour and rebreathing of 0.5-0.9 l/min in the standing position cannot be attributable to the rebreathing maneuver itself. The same is true for the supine, 30° HDT, and 30° HUT positions during the tilt experiments, where the difference between CO_{cal sup} prerebreathing and during rebreathing was 0.4 l/min in all three cases (Table 2). At 70° HUT, however, the increase in CO_{cal sup} due to rebreathing was approximately twice as much (0.9 l/min) as in the supine position. Part of the 1.5-l/min difference between the methods at 70° HUT may therefore be attributable to the larger influence of the rebreathing maneuver on the parameter to be measured (SV) in some of the subjects (T1 and T2). This does not fully explain the large interindividual range of change in R_rebr/pc (11 to 44%) at this tilt angle, as both extreme values were found in subjects in which SV_{cal sup} was not influenced by the rebreathing maneuver.

Why does R_rebr/pc change when the position of subjects is changed from the (near) supine to the upright position? First, this is likely not to be caused by using Finapres as an arterial pressure measurement device. Jellema et al. (11) found no difference between noninvasive finger arterial and intrabrachial CO_pc responses during tilt experiments. However, in both cases the method has been applied to a pressure wave more peripheral than aortic pressure, for which the method was originally developed.

Overestimation of CO in the upright position by pulse contour can be caused by a too-large systolic area (PSA) and/or a too-small estimated characteristic Zao. In the original model study of Wesseling et al. (23), hydrostatic pressure gradients as a result of the upright position are not taken into account. MAP is used in the model corrections to adapt the estimation of Zao for changes in compliance. A possible consequence is that an increasing pressure in the thoracic and abdominal aorta could mean a change toward the properties of a more rigid system with an increase in pulse-wave velocity (2) and/or a change in mismatch of characteristic impedance at the primary reflection sites (12, 13). Consequently, reflections may play a more dominant role in the systolic part of the pulse wave, also at higher HRs when ejection time is shorter. A change in the location of the primary reflection site will have the same effect. When a larger part of the PSA is due to reflections (14, 25), this has to be corrected in the model by assuming a less compliant arterial system, which results in an increase in the estimated Zao, leading to a lower estimated SV.

In our earlier validation study (18), in which saline infusion was used to enlarge SV, only a small increase in HR (+4.1%) and no change in MAP (+1.3%) was seen. During lower body negative pressure, which was used to reduce SV, HR increased at most by +24% (at 40 mmHg lower body negative pressure), whereas MAP decreased by 2.2-4.1%. Furthermore, all measurements were performed in the same body position (6° HDT). In the present study, responses in HR and MAP due to HUT and standing were much larger, and both parameters increased. Maximal responses during standing were +62% for HR and +31% for MAP; at 70° HUT they were +35% for HR and +17% for MAP. By using Wesseling's correction formula (23), these changes in HR and MAP will result in a decrease in the estimated Zao (and hence in a larger SV_pc), which in this situation is probably a correction in the wrong direction.

In conclusion, the present model used to adapt the estimation of Zao to different hemodynamic parameters, i.e., HR, blood pressure, and the influence of age, does not account for situations in which a hydrostatic pressure gradient exists. Responses of CO_{cal sup} to changes in body position are widely different in our subjects and are sometimes nonphysiological. In our earlier study (18), the reproducibility of the pulse contour measurements in one body position was good over a long time period and over a large range of COs. Also, in the present study, pulse contour measurements in any one body position were well reproducible. However, pulse contour analysis using the Finapres waveform needs additional calibrations when changes in body position are involved.