Validation of the USCOM-1A cardiac output monitor in ...

ABSTRACTValidation of the USCOM-1A cardiac output monitorin hemodynamic unstable intensive care patientsL.E.M. Haas, D.H.T. Tjan, J. van Wees, A.R.H. van Zanten,Department of Intensive Care, Gelderse Vallei Hospital, Ede, The NetherlandsIntroductionSince the introduction of the pulmonary artery catheter (PAC) in the 1970s, thermodilution COmeasurement is widely used and considered the gold standard. The use of the PAC has itsinherent risks and has never convincingly been shown to improve the outcome of the criticallyill patient [1]. Therefore, less invasive methods for CO monitoring have been developed.The ultrasonic cardiac output monitor (USCOM Pty Ltd, Coffs Harbour, NSW, Australia) wasdeveloped in 2001 and has been introduced for clinical use in the Netherlands in 2005 (figure1). It measures CO non-invasively by continuous wave Doppler ultrasound with a transducer(2.2 or 3.3 MHz) placed supra- or parasternally (figure 2&3). It has been demonstrated that theUSCOM-1A is accurate in cardiac surgery patients [2].The aim of this study is to evaluate the accuracy of the USCOM-1A (COUSC) in ICU patientscompared to the continuous thermodilution technique (using a PAC; COPAC).MethodsTwenty-five adult ventilated ICU patients on vasoactive drugs, that required a PAC (CCOPAC, 7.5 F, Edwards Lifesciences, LLC, Irvine, CA, USA) to the discretion of the attendingphysician, were included. The protocol was approved by the hospital institutional review board.After a hands-on training five intensive care physicians and five intensive care nursesperformed several series of 5 COUSC measurements on consecutive days. Means of these 5measurements were paired with actual COPAC data. These COPAC data were blinded for theoperator and filed by an independent observer. As the data followed a lognormal distribution, alogarithmic transformation was used. The systematic error and the various random errorcomponents were estimated using a linear random effects model. Finally, the results wereretransformed to percentages.ResultsTwenty-five sedated and mechanically ventilated patients were included and studied betweenSeptember and December 2005. Demographics, co-morbidities and outcomes are shown intables 1, 2 and 3, respectively.We obtained 1315 USCOM measurements and based on means, 263 paired measurements werePresented: February 2006, Annual Intensive Care Society Congress, the Netherlands 1

analysed for bias, precision and correlation. We measured CO in a range of 1.1-11.0 l/min(mean CO 6.0 ± 1.9, median 5.8 l/min). The correlation over the CO range between the twotechniques is shown in figure 4. Correlation of COPAC vs. COUSC is good (correlationcoefficient r = 0,8024 and r2 = 0,6438).The accuracy is depicted in figure 5 using Bland-Altman analysis. Individual comparisons areshown in figure 6. The overall coefficient of variation is 17%. This means that 95% of errors areless than 34%. Random effect model error comprises a inter-operator variability of 7.7%, ainter-patient variability of 8.8% and residual variability of 14%. Further correction for thesystematic error was performed, but did not change results. The error is greatest at high flows,where USCOM underestimated the CO (by up to 30%), but is acceptable in the clinicallyimportant range. Temperature and gender did not influence the correlation. We found asignificant relation of the CO with the body mass index (BMI) (p=0.03). A single point increaseof BMI corresponds with 1% increase of COUSC. Correction for BMI only marginally reducesthe total error from 17% to 16%. In all patients similar circulatory profiles were found withPAC and USCOMPatients (Pts) and Methods:We studied 7 pts with CRT (mean age 70±8) with the mean EF (24±9%) at recruitment. A VOwas first performed by an echo physician using the Ritter’s formula. After determining theoptimal AV interval, the highest Doppler trans-aortic flow determined CO and the highest MDfrom the aortic Doppler at 11 VV intervals (RV80, RV60, RV40, RV20, RV12, 0 LV12, LV20,LV40, LV60 and LV80) in random sequence by the same echo physician and a nurse whowere blind to the sequence.Results:The Doppler and USCOM determined VVI were all within 20ms. 5 pts (71%) had 100%concordant in VV interval, the remaining 2 pts had VVI difference within 20ms. The USCOMdetermined VVI were well correlated with the Doppler determined VVI. (r2 0.92, p

Figure 1: USCOM monitorFigure 2: Anatomic position for USCOM CO measurementsPresented: February 2006, Annual Intensive Care Society Congress, the Netherlands 3

Figure 3: Flow profilePresented: February 2006, Annual Intensive Care Society Congress, the Netherlands 4

Table 2: Co-morbidities (N = 25)Table 1: Demographics (N = 25)Table 3: Outcome (N = 25)Presented: February 2006, Annual Intensive Care Society Congress, the Netherlands 5

Figure 4: Correlation between the COUSC monitor and COPACFigure 5: Bland-Altman plot of errors of COUSC monitor vs. COPACPresented: February 2006, Annual Intensive Care Society Congress, the Netherlands 6

Figure 6: Cardiac Output in individual patients comparing PAC vs. USCOMFigure 7: Diagnosis of shock states comparing clinical suspicion with PAC vs. USCOMPresented: February 2006, Annual Intensive Care Society Congress, the Netherlands 7

References1 Harvey S, Harrison DA, Singer M, Ashcroft J, Jones CM, Elbourne D, Brampton W, Williams D,Young D, Rowan K; PAC-Man study collaboration. Assessment of the clinical effectiveness ofpulmonary artery catheters in management of patients in intensive care (PAC-Man): a randomisedcontrolled trial. Lancet 2005; 366:472-77.2 Tan HL, Pinder M, Parsons R, Roberts B, Heerden PV. Clinical evaluation of USCOM ultrasoniccardiac output monitor in cardiac surgical patients in intensive care unit. Br J Anaesth 2005; 94:287-91.Presented: February 2006, Annual Intensive Care Society Congress, the Netherlands 8