Posters_Monday_12 October 2009
Intensive Care Medicine
BACKGROUND. Mathematical coupling may explain why cardiac filling volumes obtained by transpulmonary thermodilution may better predict and monitor responses of cardiac output to fluid loading than pressures obtained by pulmonary artery catheters (PAC). METHODS. Eleven consecutive patients with hypovolaemia after coronary surgery and a PAC, allowing central venous pressure (CVP) and continuous cardiac index (CCIp) measurements, received a femoral artery catheter for transpulmonary thermodilution
... measurements of global end-diastolic volume index (GEDVI) and cardiac index (CItp). One to five fluid loading steps of 250 mL were done in each patient (n = 48 total). RESULTS. Fluid responses were predicted and monitored similarly by CItp and CCIp, whereas CItp and CCIp correlated at r = 0.70 (p \ 0.001) with a bias of 0.40 L min -1 m -2 . Changes in volumes (and not in CVP) related to changes in CItp and not in CCIp. Changes in CVP and GEDVI similarly related to changes in CItp, after elimination of two patients with greatest CItp outliers (as compared to CCIp). Changes in GEDVI correlated better to changes in CItp when derived from the same thermodilution curve than to changes in CItp of unrelated curves and changes in CCIp. CONCLUSIONS. After coronary surgery, fluid responses can be similarly assessed by intermittent transpulmonary and continuous pulmonary thermodilution methods, in spite of overestimation of CCIp by CItp. Filling pressures are poor monitors of fluid responses and superiority of GEDVI can be caused, at least in part, by mathematical coupling when cardiac volume and output are derived from the same thermodilution curve. Bland-Altman plot for cardiac index measurements with help of the transpulmonary (CItp) and pulmonary (CCIp) continuous thermodilution techniques Scatterplots showing relationship between the fluid-induced changes of global end-diastolic volume index (GEDVI) and transpulmonary (panel A) or pulmonary continuous (panel B) thermodilution techniques OBJECTIVES. To evaluate the effects of active mechanical circulatory support on sublingual microcirculation as a model for body tissue perfusion. METHODS. Between May 2008 and January 2009, nine consecutive patients received a mechanical support device (HeartMate II, TandemHeart, or extracorporeal membrane oxygenation) for end-stage chronic heart failure or cardiogenic shock. Microcirculation was investigated using a hand-held sidestream dark field imaging device. Perfused capillary density (PCD) and capillary red blood cell velocity (cRBCv) were assessed before device implantation (T0), immediately after implantation (T1), and one day post implantation (T2). Data are presented as median (interquartile range). RESULTS. Median age of the patients was 44 (37-54) years and 67% were male. Circulatory support devices significantly decreased pulmonary capillary wedge pressure (p = 0.04). Cardiac power index increased [0.31 (0.19-0.35) W m -2 at T0 vs. 0.50 (0.42-0.54) W m -2 at T1, p = 0.008) as well as central venous oxygen saturation [52 (46-61) % at T0 vs. 76 (65-85) % at T1, p = 0.01). There was a fourfold increase in tissue perfusion index, defined as sublingual PCD 9 cRBCv, during mechanical circulatory support [544 (384-671) at T0 vs. 1,932 (1,854-2,842) at T1, p = 0.008; Fig. 1 ). Microcirculatory parameters remained improved at T2. CONCLUSION. Mechanical circulatory support for severe heart failure is associated with a consistent, significant and sustained improvement in tissue perfusion, as measured at the bedside by a two-dimensional microcirculation imaging technique. INTRODUCTION. The second generation FloTrac software has been shown to reliably measure cardiac output (CO) in cardiac surgical patients. However, concerns have been raised regarding its accuracy in vasoplegic states, so that a new software has been developed (third generation). OBJECTIVES. The aim of the present multicentre study was to investigate the value of the third generation software in patients with sepsis, particularly when systemic vascular resistance (SVR) is low. METHODS. We studied 58 septic patients with a pulmonary artery catheter and a radial (n = 32) or femoral (n = 26) arterial catheter. Reference CO was measured by bolus pulmonary thermodilution (iCO) using 3-5 cold saline bolus injected randomly through the respiratory cycle. Simultaneously, CO was computed using the second (CO G2 ) and the third generation FloTrac software (CO G3 ) from the arterial pressure curve recorded on a computer. CO was also measured by semi-continuous pulmonary thermodilution (CCO). A total of 401 simultaneous measurements of iCO, CO G2 , CO G3 , and CCO were available for comparison. RESULTS. The mean bias between CO G2 , CO G3 , CCO and iCO were -12 ± 16% (-1.0 ± 1.2 l/min), -3 ± 15% (-0.2 ± 1.1 l/min), and 8 ± 13% (0.6 ± 1.1 l/min). The percentage errors were 33% for CO G2 , 29% for CO G3 , and 29% for CCO. The bias between iCO and CO G2 was significantly correlated with SVR (r 2 = 0.37, p \ 0.0001). A very weak (r 2 = 0.05) relationship was also observed for the bias between iCO and CO G3 . The bias between iCO and CCO was not correlated with SVR. CONCLUSION. In patients with sepsis, the third generation FloTrac software is more accurate, more precise (percentage error \ 30%) and much less influenced by SVR than the second generation software. INTRODUCTION. The use of cardiac output (CO) values to provide a more detailed overview of the patient's circulation has driven the development of less invasive CO monitoring and allowed applied physiology to guide treatment. The LiDCO TM plus, monitor has been extensively validated in the ICU setting. It combines two systems; (1)The PulseCO system which uses a signal analysis (autocorrelation) to calculate continuous beat-to-beat CO from the arterial waveform (2) lithium dilution calibration to give a stat CO value. A unique calibration factor (CF) is derived from the combination of the software analysis and stat CO values, The CF is patient specific and scales the stroke volumes [vol = CF 9 250 9 (1 -exp(-k 9 P)]. The CF is proportional to aortic compliance which is known to be strongly influenced by the demographic differences between individuals. OBJECTIVES. To explore the relationship between CF and gender, age, weight and height and to determine if the CF can be estimated using this data. METHOD. 22 patients in ITU who had a LiDCOplus in clinical use were studied and relevant data, were recorded. LiDCO calibrations were performed as per our ICU protocol. Two initial calibrations at set up (avgCF) and at 24 h intervals thereafter (more frequently in the face of major haemodynamic changes). Data were collected from the obs chart and by electronic download from the LiDCO monitor. RESULTS. Gender, age, height and weight had moderate to good association to the avgCF. Using multivariable regression an equation was developed to estimate CF. (AvgCF = 0.391-0.176 9 sex -0.006 9 age ? 0.002 9 weight ? 0.511 9 height). Using this equation a Bland Altman graph was plotted which shows good agreement between the estimated and LiDCO generated values for CF. Mean bias was -0.007 and was not statistically significant (p = 0.853). 95% of the data lies between ±0.4-the majority of differences lying between ±0.2. CONCLUSION. Gender, age, weight and height all have a relationship to the average of the initial two CF's. The estimated value of the CF derived from an equation using multivariable regression showed good agreement with the actual LiDCO generated CF. Further analysis using this new data is ongoing and is being used to validate this equation further. REFERENCE(S). 1. Remington JW, Noback CR, Hamilton WF, Gold JJ (1948) Volume elasticity characteristics of the human aorta and prediction of the stroke volume from the pressure pulse. Am J Physiol 153:298-308. Bland Altman: estimated CF versus LiDCO Derived CF INTRODUCTION. Changes of cardiac output (CO) during passive leg raise (PLR) have been proposed as an easy, reproducible and reversible method for selecting patients that would benefit or not from fluid administration. A non invasive bioreactance-based technology (NI-COM) has proved to be accurate, precise and responsive for continuous CO monitoring . OBJECTIVE. This study was designed to compare the PLR-induced change in CO as assessed by the NICOM and by trans thoracic echo-doppler (ECHO). METHODS. We included patients programmed for major (cardiac, vascular and abdominal) elective surgery or during the recovery period of these operations. All patients were in stable normal hemodynamic conditions without any drug infusion. Values of CO were collected: before PLR (baseline), after 2 min of PLR made with a 45°angle (test) and 5 min after test return to baseline. NICOM values were obtained continuously and 3 min were averaged for comparison with ECHO. ECHO analyses were performed by seniors experienced cardiologists and three measurements were averaged. For each patient, predicted CO from physiological normative tables was also collected using a validated computer software. Since all patients were in stable steady state, this predicted CO at basal metabolism was taken as reference value for baseline and return to baseline. RESULTS. We obtained complete data form 50 patients, 39 men 11 women, age 63 ± 28 years, LVEF = 56 ± 9%. NICOM and ECHO values of CO were close at baseline (5.59 ± 1.32 vs. 5.51 ± 1.22, NS) and also very close to predicted values: 5.50 ± 0.91. During PLR and after PLR (return to baseline 2) (NS for all differences) NICOM and ECHO CO values were close (6.28 ± 1.50 vs. 5.96 ± 1.39, NS) and well correlated (NS from the identity line). The mean bias was = 0.08 L/min and the limits of agreement were 1.8 L/min giving a coefficient of variation of 32%. During PLR mean CO absolute increase tended to be higher for NICOM than for ECHO but did not reach significance (p = 0.11) as well as for proportional increase (13 ± 14 vs. 8 ± 6%, p = 0.07). The mean difference in PLR-induced CO changes between NICOM and ECHO was 5 ± 14%. In 68, 86 and 94% of the patients the discordance between the two devices was B10, B20 and B30%, respectively. The discordance was due to higher NICOM CO changes than ECHO during PLR is a vast majority of cases: 94, 98 and 100% for discordances [10, 20 and 30%, respectively. CONCLUSION. In this cohort of 50 volunteers in stable hemodynamic status before major cardiac surgery, CO was acceptably comparable at baseline and during PLR using NICOM and ECHO. INTRODUCTION. Cardiac output (CO) monitoring is limited by the need for invasive, expensive, or time-consuming methods. A new, noninvasive, system for continuous CO monitoring, based on chest bioreactance, has proved to be easy to use, accurate, precise and responsive . OBJECTIVE. The objective of this study was to compare CO and stoke volume (SV) monitoring capabilities of this transthoracic bioreactance-based monitor (NICOM) with those of a pulse contour-based system (PICCO PC) using transpulmonary thermodilution (PICCO TD) as reference method. METHODS. We designed a prospective, single center study in Intensive care unit, including consecutive, post-cardiac surgery, adult patients. Continuous minute-by-minute hemodynamic variables obtained from NICOM and PICCO PC were recorded and compared in 15 patients at baseline, during a lung recruitment maneuver (applying 20 cm H 2 O of PEEP) and following withdrawal of PEEP. PICCO TD measurements were also determined at baseline and during and after PEEP. The NICOM system uses an independent autocalibration process. PICCO TD was used automatically for calibration of PICCO PC. At baseline, we evaluated the accuracy (bias with the reference) and precision (2SD/mean) of these devices to measure CO and SV. During PEEP application and removal, we then assessed time responsiveness, amplitude responsiveness and reliability for detecting expected CO and SV changes. RESULTS. Mean CO values (PICCO TD) ranged from 1.6 to 8.0 L min -1 . At baseline, CO values were comparable for NICOM, PICCO PC and PICCO TD: 5.3 ± 1.2, 5.0 ± 1.5 and 4.8 ± 1.3 L min -1 , respectively (NS). The CO precision was 8 ± 7 and 9 ± 5% for NICOM and PICCO PC, respectively, NS. When PEEP was applied, CO was reduced by 33 ± 13%, 31 ± 15% and 35 ± 13%, for NICOM, PICCO PC and PICCO TD, respectively (NS). Time responsiveness was 3.2 ± 0.7 min for NICOM versus 2.6 ± 0.5 min for PICCO PC (NS). In all patients, the two studied technologies and the reference method showed a decrease in CO. SV results were comparable to CO. When all interpatients averaged points at baseline, during PEEP application and after PEEP removal were plotted together, the correlations NICOM versus PICCO TD and PICCO PC versus PICCO TD were comparable and not significantly different from the identity line. The mean bias of NICOM and PICCO PC was small (0.25 vs. 0.10 L min -1 , respectiveiy) and limits of agreement shown were quite large due to rapid and large changes in CO but comparable for NICOM (1.83 L min -1 ) and PICCO PC (1.93 L min -1 ), despite automatic recalibration of the PICCO PC. CONCLUSION. In this study, bioreactance and pulse contour analysis calibrated by transpulmonary thermodilution have comparable CO and SV monitoring capabilities.