Methodology of method comparison studies evaluating the validity of cardiac output monitors: a stepwise approach and checklist.

The validity of each new cardiac output (CO) monitor should be established before implementation in clinical practice. For this purpose, method comparison studies investigate the accuracy and precision against a reference technique. With the emergence of continuous CO monitors, the ability to detect changes in CO, in addition to its absolute value, has gained interest. Therefore, method comparison studies increasingly include assessment of trending ability in the data analysis. A number of methodological challenges arise in method comparison research with respect to the application of Bland-Altman and trending analysis. Failure to face these methodological challenges will lead to misinterpretation and erroneous conclusions. We therefore review the basic principles and pitfalls of Bland-Altman analysis in method comparison studies concerning new CO monitors. In addition, the concept of clinical concordance is introduced to evaluate trending ability from a clinical perspective. The primary scope of this review is to provide a complete overview of the pitfalls in CO method comparison research, whereas other publications focused on a single aspect of the study design or data analysis. This leads to a stepwise approach and checklist for a complete data analysis and data representation.

The effect of propofol on haemodynamics: cardiac output, venous return, mean systemic filling pressure, and vascular resistances.

BACKGROUND: Although arterial hypotension occurs frequently with propofol use in humans, its effects on intravascular volume and vascular capacitance are uncertain. We hypothesized that propofol decreases vascular capacitance and therefore decreases stressed volume. METHODS: Cardiac output (CO) was measured using Modelflow(®) in 17 adult subjects after upper abdominal surgery. Mean systemic filling pressure (MSFP) and vascular resistances were calculated using venous return curves constructed by measuring steady-state arterial and venous pressures and CO during inspiratory hold manoeuvres at increasing plateau pressures. Measurements were performed at three incremental levels of targeted blood propofol concentrations. RESULTS: Mean blood propofol concentrations for the three targeted levels were 3.0, 4.5, and 6.5 µg ml(-1). Mean arterial pressure, central venous pressure, MSFP, venous return pressure, Rv, systemic arterial resistance, and resistance of the systemic circulation decreased, stroke volume variation increased, and CO was not significantly different as propofol concentration increased. CONCLUSIONS: An increase in propofol concentration within the therapeutic range causes a decrease in vascular stressed volume without a change in CO. The absence of an effect of propofol on CO can be explained by the balance between the decrease in effective, or stressed, volume (as determined by MSFP), the decrease in resistance for venous return, and slightly improved heart function. CLINICAL TRIAL REGISTRATION: Netherlands Trial Register: NTR2486.

Ventilator-induced central venous pressure variation can predict fluid responsiveness in post-operative cardiac surgery patients.

BACKGROUND: Ventilator-induced dynamic hemodynamic parameters such as stroke volume variation (SVV) and pulse pressure variation (PPV) have been shown to predict fluid responsiveness in contrast to static hemodynamic parameters such as central venous pressure (CVP). We hypothesized that the ventilator-induced central venous pressure variation (CVPV) could predict fluid responsiveness. METHODS: Twenty-two elective cardiac surgery patients were studied post-operatively on the intensive care unit during mechanical ventilation with tidal volumes of 6-8 ml/kg without spontaneous breathing efforts or cardiac arrhythmia. Before and after administration of 500mL hydroxyethyl starch, SVV and PPV were measured using pulse contour analysis by modified Modelflow® , while CVP was obtained from a central venous catheter positioned in the superior vena cava. CVPV was calculated as 100 × (CVPmax -CVPmin )/[(CVPmax + CVPmin) /2]. RESULTS: Nineteen patients (86%) were fluid responders defined as an increase in cardiac output of ≥ 15% after fluid administration. CVPV decreased upon fluid loading in responders, but not in non-responders. Baseline CVP values showed no correlation with a change in cardiac output in contrast to baseline SVV (r = 0.60, P = 0.003), PPV (r = 0.58, P = 0.005), and CVPV (r = 0.63, P = 0.002). Baseline values of SVV > 9% and PPV > 8% could predict fluid responsiveness with a sensitivity of 89% and 95%, respectively, both with a specificity of 100%. Baseline CVPV could identify all fluid responders and non-responders correctly at a cut-off value of 12%. There was no difference between the area under the receiver operating characteristic curves of SVV, PPV, and CVPV. CONCLUSION: The use of ventilator-induced CVPV could predict fluid responsiveness similar to SVV and PPV in post-operative cardiac surgery patients.

Non-invasive continuous arterial pressure and pulse pressure variation measured with Nexfin(®) in patients following major upper abdominal surgery: a comparative study.

We compared the accuracy and precision of the non-invasive Nexfin(®) device for determining systolic, diastolic, mean arterial pressure and pulse pressure variation, with arterial blood pressure values measured from a radial artery catheter in 19 patients following upper abdominal surgery. Measurements were taken at baseline and following fluid loading. Pooled data results of the arterial blood pressures showed no difference between the two measurement modalities. Bland-Altman analysis of pulse pressure variation showed significant differences between values obtained from the radial artery catheter and Nexfin finger cuff technology (mean (SD) 1.49 (2.09)%, p < 0.001, coefficient of variation 24%, limits of agreement -2.71% to 5.69%). The effect of volume expansion on pulse pressure variation was identical between methods (concordance correlation coefficient 0.848). We consider the Nexfin monitor system to be acceptable for use in patients after major upper abdominal surgery without major cardiovascular compromise or haemodynamic support.

Predicting cardiac output responses to passive leg raising by a PEEP-induced increase in central venous pressure, in cardiac surgery patients.

BACKGROUND: Changes in central venous pressure (CVP) rather than absolute values may be used to guide fluid therapy in critically ill patients undergoing mechanical ventilation. We conducted a study comparing the changes in the CVP produced by an increase in PEEP and stroke volume variation (SVV) as indicators of fluid responsiveness. Fluid responsiveness was assessed by the changes in cardiac output (CO) produced by passive leg raising (PLR). METHODS: In 20 fully mechanically ventilated patients after cardiac surgery, PEEP was increased +10 cm H2O for 5 min followed by PLR. CVP, SVV, and thermodilution CO were measured before, during, and directly after the PEEP challenge and 30° PLR. The CO increase >7% upon PLR was used to define responders. RESULTS: Twenty patients were included; of whom, 10 responded to PLR. The increase in CO by PLR directly related (r=0.77, P<0.001) to the increase in CVP by PEEP. PLR responsiveness was predicted by the PEEP-induced increase in CVP [area under receiver-operating characteristic (AUROC) curve 0.99, P<0.001] and by baseline SVV (AUROC 0.90, P=0.003). The AUROC's for dCVP and SVV did not differ significantly (P=0.299). CONCLUSIONS: Our data in mechanically ventilated, cardiac surgery patients suggest that the newly defined parameter, PEEP-induced CVP changes, like SVV, appears to be a good parameter to predict fluid responsiveness.

Performance of three minimally invasive cardiac output monitoring systems.

We evaluated cardiac output (CO) using three new methods - the auto-calibrated FloTrac-Vigileo (CO(ed)), the non-calibrated Modelflow (CO(mf) ) pulse contour method and the ultra-sound HemoSonic system (CO(hs)) - with thermodilution (CO(td)) as the reference. In 13 postoperative cardiac surgical patients, 104 paired CO values were assessed before, during and after four interventions: (i) an increase of tidal volume by 50%; (ii) a 10 cm H(2)O increase in positive end-expiratory pressure; (iii) passive leg raising and (iv) head up position. With the pooled data the difference (bias (2SD)) between CO(ed) and CO(td), CO(mf) and CO(td) and CO(hs) and CO(td) was 0.33 (0.90), 0.30 (0.69) and -0.41 (1.11) l.min(-1), respectively. Thus, Modelflow had the lowest mean squared error, suggesting that it had the best performance. CO(ed) significantly overestimates changes in cardiac output while CO(mf) and CO(hs) values are not significantly different from those of CO(td). Directional changes in cardiac output by thermodilution were detected with a high score by all three methods.

A comparison of stroke volume variation measured by the LiDCOplus and FloTrac-Vigileo system.

The aim of this study was to compare the accuracy of stroke volume variation (SVV) as measured by the LiDCOplus system (SVVli) and by the FloTrac-Vigileo system (SVVed). We measured SVVli and SVVed in 15 postoperative cardiac surgical patients following five study interventions; a 50% increase in tidal volume, an increase of PEEP by 10 cm H2O, passive leg raising, a head-up tilt procedure and fluid loading. Between each intervention, baseline measurements were performed. 136 data pairs were obtained. SVVli ranged from 1.4% to 26.8% (mean (SD) 8.7 (4.6)%); SVVed from 2.0% to 26.0% (10.2 (4.7)%). The bias was found to be significantly different from zero at 1.5 (2.5)%, p < 0.001, (95% confidence interval 1.1-1.9). The upper and lower limits of agreement were found to be 6.4 and -3.5% respectively. The coefficient of variation for the differences between SVVli and SVVed was 26%. This results in a relative large range for the percentage limits of agreement of 52%. Analysis in repeated measures showed coefficients of variation of 21% for SVVli and 22% for SVVed. The LiDCOplus and FloTrac-Vigileo system are not interchangeable. Furthermore, the determination of SVVli and SVVed are too ambiguous, as can be concluded from the high values of the coefficient of variation for repeated measures. These findings underline Pinsky's warning of caution in the clinical use of SVV by pulse contour techniques.

A new approach to determine parallel conductance for left ventricular volume measurements.

OBJECTIVES: To determine absolute ventricular volume with the conductance catheter technique, the electrical conductance of tissues and fluids (parallel conductance) around the ventricle should be determined precisely. METHODS: A new objective method to estimate parallel conductance based on analysis of the dilution curve of hypertonic saline was investigated. The parallel conductances obtained with the new method (G(a)(p)) were compared to those obtained with the conventional method (G(l)(p)). The study was performed in the left ventricle of 12 patients. RESULTS: G(a)(p) was not significantly different from G(l)(p). For the G(l)(p) method the average percentage difference between duplicate values, both taken as absolute values, was 15.06% and for the G(a)(p) method it was 4. 01%. Thus the reproducibility of the method is a factor four better than that of the method. This difference appeared to be significant. CONCLUSION: We conclude that a smaller number of injections will be required to obtain the same precision using our method.

The volume-dependency of parallel conductance throughout the cardiac cycle and its consequence for volume estimation of the left ventricle in patients.

OBJECTIVE: To study the hypothesis that the electrical conductance of tissues and fluids (parallel conductance (G(p))) around the ventricle depends on left ventricular volume throughout the cardiac cycle. METHODS: We extended a recently developed method to determine G(p) throughout the cardiac cycle. First, we compared the estimates of parallel conductances obtained with the new method (G(a)(p)) with those of the conventional one (G(1)(p)), both averaged over the cardiac cycles. Secondly, G(a)(p) was determined throughout the cardiac cycle and its volume dependency was assessed. Thirdly, the factor alpha was calculated as the ratio between stroke volume, obtained by the conductance method using G(1)(p), and that obtained by a thermodilution method. Because the non-homogeneous field was indicated to be the reason for the dependency of G(p) on left ventricular volume as well as for the need for alpha, we tested whether the hypothesis implies that a correction with alpha is not needed if G(p) is determined throughout the cardiac cycle. RESULTS: We found a negative linear relation between G(p) and left ventricular volume. This relation appeared to be reproducible within each patient. Furthermore, we found that alpha deviates from 1 primarily due to the dependency of G(p) on left ventricular volume. CONCLUSION: To obtain stroke volume or to determine absolute left ventricular volume continuously within a cardiac cycle, G(p) should be determined throughout each cardiac cycle and if a constant G(p) throughout the cardiac cycle is used a correction with the factor alpha should be made to correct for a possible influence of electrical field heterogeneity.

Random walk type models for indicator-dilution studies: comparison of a local density random walk and a first passage times distribution.

The relative merits of the local density random walk and the first passage times distributions were compared with respect to their practical applicability in cardiovascular research and clinical practice. Open indicator-dilution curves of varying shape were used, and reference values for area and mean transit times were calculated numerically. Curves not perturbed by recirculation were obtained in two different ways. Thermodilution curves were obtained in an animal model at the left and the right side of the heart respectively and conductivity curves with 0.5% NaCl solution as indicator were obtained in a hydrodynamic circulation model. The fits of the two types of distribution were equally accurate for the more symmetrical curves; for very skewed curves the local density random walk fit proved to be more accurate. This result could be related to the greater difference in shape between the first passage times and local density random walk distribution for a large degree of asymmetry. For this reason the local density random walk distribution for fitting indicator-dilution curves was used in a variety of other experimental conditions.

Improvement of cardiac output estimation by the thermodilution method during mechanical ventilation.

The reliability of cardiac output estimation by thermodilution during artificial ventilation was studied in anesthetized pigs at the right side of the heart. The estimates exhibited a cyclic modulation related to the ventilation. The amplitude of the modulation was independent of the level of positive end-expiratory pressure, ventilatory pattern and volemic loading of the animals. However, a non-constant phase relation existed between the ventilatory cycle and the modulation. Single observations at a fixed moment in the ventilatory cycle are therefore not appropriate for estimation of mean cardiac output nor for studying its relative changes. The averaging of estimates spread equally over the ventilatory cycle led to a much larger reduction in the deviation of the averages from the mean cardiac output than an averaging procedure of randomly selected estimates. The accuracy of estimation of mean cardiac output by two estimates equally spread in the ventilatory cycle was equal to the accuracy obtained by averaging five randomly selected estimates. Averaging four estimates, equally spread in the cycle, appeared to be the optimal procedure. For 89% of all averages an accuracy of 5% around the mean was obtained and for 99% an accuracy of +/- 10%.

Alternating versus synchronous ventilation of left and right lungs in piglets.

OBJECTIVE: We tested whether alternating ventilation (AV) of each lung (i.e. with a phase difference of half a ventilatory cycle) would decrease central venous pressure and so increase cardiac output when compared with simultaneous ventilation (SV) of both lungs. THEORY: If, during AV, the inflated lung expands partly via compression of the opposite lung, mean lung volume will be smaller during AV than SV. As a consequence, mean intrathoracic pressure (as cited in the literature), and therefore, central venous pressure will be smaller. DESIGN: The experiments were performed in seven anaesthetized and paralyzed piglets using a double-piston ventilator. Minute ventilation was the same during AV and SV. Starting at SV, we alternated three times between AV and SV for periods of 10 min. RESULTS: During AV, central venous pressure was decreased by 0.7 mmHg and cardiac output was increased by 10 +/- 4.4% (mean, +/-SD) compared with SV. AV also resulted in increased arterial pressure. During one-sided inflation with closed outlet of the opposite lung, a pressure rise occurred in the opposite lung, indicating compression. CONCLUSION: The higher cardiac output during AV than SV can be explained by the fact that central venous pressure is lower during AV. This lower central venous pressure is very probably due to the lower mean intrathoracic pressure caused by compression of the opposite lung during unilateral inflation.

Haemodynamic and respiratory conditions during alternating and synchronous ventilation of both lungs.

OBJECTIVE: We tested the hypothesis that mean thoracic expansion (and mean lung volume) is lower during alternating ventilation (AV), i.e. ventilation of both lungs with a phase shift of half a ventilatory cycle, compared to synchronous ventilation (SV) of both lungs. As a consequence, intrathoracic pressure will be lower, causing lower, central venous pressure and higher cardiac output. DESIGN: In eight anaesthetized and paralysed piglets, differential ventilation was established by fixation of an endobronchial tube in the left main bronchus. SV and AV were sequentially applied for four and three periods, respectively, of 10 minutes each. Minute ventilation was the same during AV and SV and adapted to normocapnia. Two series of observations were performed: series 1 with intact thorax and monitoring of oesophageal pressure; series 2 after perforation of the sternum, airtight closure of the thorax and monitoring of pericardial pressure. RESULTS: In both series, mean lung volume was 16 +/- 4% lower and central venous, oesophageal (series 1) and pericardial pressures (series 2) were 0.5-0.7 mmHg lower during AV compared to SV (all p < 0.001). In series 1, aortic pressure was 5 mmHg and cardiac output 8% higher (both p < 0.001). In series 2, cardiac output was 5% higher during AV (p < 0.001), but aortic pressure did not change (p = 0.07). CONCLUSION: Our data verified the hypothesis. The lower oesophageal (series 1), pericardial (series 2) and central venous pressures during AV compared to SV could be explained by the smaller thoracic expansion due to the lower mean lung volume, which was attributed to compression of the opposite lung by the, expansion of the inflated lung.

Suppression of spontaneous breathing during high-frequency jet ventilation. Influence of dynamic changes and static levels of lung stretch.

Conditions which suppress spontaneous breathing activity during high-frequency jet ventilation (HFJV) were analysed in Yorkshire piglets under pentobarbital anesthesia. The highest PaCO2 at which the animals did not breathe against the ventilator (apnea point) was established during different patterns of ventilation, either by changing the minute volume or by adding CO2 to the inspiratory gas. Arterial oxygen tension was maintained throughout the study above 80 mm Hg. An elevation of ventilatory rate increased the apnea point, suggesting a progressive suppression of spontaneous breathing. This suppression did not depend on the amount of lung stretch during insufflation, because at higher rates lower tidal volumes were used. Suppression also appeared to be independent of insufflatory flow, i.e. the velocity of lung stretch. At higher frequencies end-expiratory airway pressure (PEE) increased and there appeared to be a positive relationship between the apnea point and PEE. In a separate series this positive relationship between the apnea point and PEE was confirmed. A hysteresis effect in this relationship, however, suggests that other than jet frequency, lung volume rather than positive end-expiratory pressure (PEEP) is a major determinant of suppression of spontaneous breathing activity during HFJV.

Suppression of spontaneous breathing during high-frequency jet ventilation. Separate effects of lung volume and jet frequency.

The effect of ventilatory frequency of high-frequency jet ventilation (HFJV) from 1 to 5 Hz, apart from changes in thoracic volume, on spontaneous breathing activity was studied in Yorkshire piglets under pentobarbital anesthesia. The highest PaCO2 at which the animals did not breathe against the ventilator (apnea point) was established either by changing minute volume of ventilation or by adding CO2 to the respiratory gas. The higher the apnea point, the higher the suppression of spontaneous breathing activity was assumed to be. If the apnea point was searched for by changing minute volume a progressive increase of suppression of spontaneous respiratory activity was found at ventilatory rates of 3 Hz or more, concomitantly with a rise in end-expiratory pressure (PEE). In case the tidal volume was kept constant, increase of ventilatory rate resulted in a tremendous increase of lung volume, together with considerably higher levels of PEE. When under these conditions the apnea point was searched for by adding CO2 to the respiratory gas a much higher CO2-drive was needed for spontaneous breathing and therefore a much stronger inhibition of spontaneous breathing was concluded. By placing the animals in a body box in which pressure could be varied, thoracic volume could be kept constant during HFJV. When thoracic volume was kept constant in this way a constant tidal volume at increasing jet frequencies resulted in only a slight increase in suppression of spontaneous breathing. We conclude that the increase in lung volume is a major factor in suppressing central respiratory activity during HFJV. Jet frequency by itself might be an additional suppressive factor.(ABSTRACT TRUNCATED AT 250 WORDS)

Alveolar pressure during high-frequency jet ventilation.

We studied the influence of ventilatory frequency (1-5 Hz), tidal volume, lung volume and body position on the end-expiratory alveolar-to-tracheal pressure difference during high-frequency jet ventilation (HFJV) in Yorkshire piglets. The animals were anesthetized and paralysed. Alveolar pressure was estimated with the clamp off method, which was performed by a computer controlled ventilator and which had been extensively tested on its feasibility. The alveolar-to-tracheal pressure difference increased with increasing frequency and with increasing tidal volume, the common determinant appearing to be the mean expiratory flow. The effects in prone and in supine position were similar. Increasing thoracic volume decreased the alveolar-to-tracheal pressure difference indicating a dependence of this pressure difference on airway resistance. We concluded that the main factors determining the alveolar-to-tracheal pressure difference (delta P) during HFJV are expiratory flow (V'E) and airway resistance (R), delta P congruent to V'E x R.

Tracing best PEEP by applying PEEP as a RAMP.

OBJECTIVE: The aim of this study was to show the feasibility of a slow, continuously increasing level of positive end-expiratory pressure (PEEP) (ramp manoeuvre) in selecting best PEEP and to evaluate whether best PEEP, as defined by maximal oxygen transport, coincides with best systemic arterial oxygenation or best compliance. DESIGN: In 11 anaesthetized piglets, PEEP was increased between 0 cmH2O (zero end-expiratory pressure; ZEEP) and 15 cmH2O (PEEP15) with a constant rate of 0.67 cmH2O x min(-1). This ramp manoeuvre was performed both under normal conditions and after induction of an experimental lung oedema. During the ramp manoeuvre, haemodynamic and pulmonary variables were monitored almost continuously. RESULTS: During the rise in PEEP, cardiac output declined in a non-linear way. In the series with normal conditions, best PEEP was always found at ZEEP. In the series with experimental lung oedema, best PEEP, as defined by maximum oxygen transport, was found at PEEP1-6, as defined by maximal compliance, at PEEP7.5 and by maximal arterial oxygen tension (PaO2) at PEEP10-14. CONCLUSIONS: Best PEEP according to oxygen transport is lower than best PEEP according to compliance and PaO2; the use of PEEP as a ramp might prevent unnecessarily high levels of PEEP.

A stable model of respiratory distress by small injections of oleic acid in pigs.

OBJECTIVE: Development of a stable model of respiratory distress in pigs with oleic acid, fulfilling clinical criteria of the adult respiratory distress syndrome (ARDS). DESIGN: Eight pigs (9.1 +/- 0.7 kg) were anesthetized with pentobarbital, paralyzed with tubocurarine and mechanically ventilated with an FIO2 of 0.6, an I:E ratio of 2:3 and a PEEP of 0.2 kPa. Oleic acid (dissolved 1:1 in 96% alcohol) was administered in a series of multiple injections of 0.1 ml until PaO2 was lower than 8 kPa. MEASUREMENTS AND RESULTS: Careful titration of the oleic acid injections on guidance of the PaO2 established a reproducible respiratory distress (PaO2 = 7.3 +/- 0.8 kPa), in which gas exchange and hemodynamic variables were stable for at least 4 h. The number of oleic acid injections (22 +/- 11, mean and SD) varied between the animals. CONCLUSIONS: With the use of multiple injections of oleic acid, a stable model of early respiratory distress in pigs can be achieved, in spite of individual differences in sensitivity. Such a stable model allows for a diversity of studies on early respiratory distress.

An adequate strategy for the thermodilution technique in patients during mechanical ventilation.

The application of the thermodilution method in conditions associated with variations in blood flow implies a misuse of the Stewart Hamilton equation. Therefore, we studied the reliability of the thermodilution method for the estimation of mean cardiac output (CO) during mechanical ventilation in patients (n = 9). Variation of the injection moment in the ventilatory cycle elicited a cyclic variation of CO estimates. This variation was not the same for all patients neither in phase nor in amplitude. Therefore, no specific phase in the ventilatory cycle could be selected for an accurate estimation of mean CO. Averaging CO estimates randomly distributed in the ventilatory cycle led to an improvement of accuracy with the square root of the number of observations. The averaging of CO estimates spread equally over the ventilatory cycle led to a much better result, e.g., the variation in the average of two estimates equally spread in the ventilatory cycle was similar to the variation in the average of four random estimates. We conclude that averaging of 3 or 4 estimates spread equally over the ventilatory cycle is an adequate strategy to estimate mean cardiac output in patients reliably.

Extrapolation of thermodilution curves obtained during a pause in artificial ventilation.

The feasibility of three mathematical models to extrapolate the tail of thermodilution curves, when flectures are present in the descending limb, was tested in anesthetized pigs. The models were a local random walk model (LDRW), a log-normal distribution, and a two-compartment model. First, the accuracy of the extrapolation of the tail by each model was tested on two undisturbed curves by taking the truncation at five different points on the descending limb. The extrapolated curve area obtained from each model was compared with total area of the undisturbed curve. Next, dilution curves obtained during inspiratory hold maneuvers and characterized by deflection points were analyzed, taking the truncation just before deflection. The estimates of cardiac output by the models were compared with electromagnetically measured flow in the pulmonary artery. The area of the dilution curve was estimated more accurately when more information on the descending limb was available. The LDRW model and the log-normal distribution were superior to the two-compartment model regarding accuracy of cardiac output estimation and root-mean-square errors of the fit. Both models estimated curve area with an error less than 5% when truncation of the descending limb occurred below 60% of the peak value. In circumstances of mechanical ventilation, where only short periods of constant flow will be present, analyses of dilution curves based on the LDRW model or the log-normal distribution are recommended.