| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
To the Editors:
The recent article by Iregui and colleagues titled "Physicians’ Estimates of Cardiac Index and Intravascular Volume Based on Clinical Assessment Versus Transesophageal Doppler Measurements Obtained by Critical Care Nurses" (July 2003:336–342) deserves comment and very cautious interpretation because of multiple concerns with the study design and statistical analysis.
Accuracy of hemodynamic measurements obtained from transesophageal Doppler (TED) imaging is dependent on training and experience, and error can be easily introduced by inexperienced operators.1 Although nurses in this study received 2 hours of instruction about TED imaging and were supervised by investigators until they demonstrated proficiency, the authors failed to provide objective data to substantiate that proficiency in measuring hemodynamic variables with TED imaging was adequate before the study began. The authors cited research1 that validated specific objective training criteria to assess operator proficiency in obtaining cardiac output using TED imaging to ensure adequate reliability, but did not incorporate such criteria into their study design.
Nurses making hemodynamic measurements using TED imaging were blinded to physicians’ clinical estimates; they, however, were not blinded to pulmonary artery (PA) catheter data, current hemodynamic status, or therapy, all of which are well-known sources for introducing observer bias into measurements made with conventional transesophageal echocardiography.2,3 Seven nurses determined hemodynamic measurements using TED imaging during the study (mean of 15 procedures per nurse), and Doppler flow velocity waveforms were adequate in 99.1% of subjects. This information, however, does not provide data of intranurse and internurse variability of hemodynamic measurements obtained from TED imaging during the study. Owing to the novice experience level and the large number of nurse observers, variability estimates were essential in this study to discern whether variability within or among nurses influenced study results. The poor correlation found between the 2 methods for cardiac output (r = 0.78, P < .001) and an estimate of intravascular volume (IVV) (r = 0.54, P = .01) suggests that large variability within and/or among nurses may have contributed to these results since these values are less than correlation coefficients of 0.9 and greater that have been reported in similar studies.4,5 The authors cited method comparison studies that reported estimates of intraobserver and interobserver variability in measuring cardiac output with TED imaging,4,5 but elected not to incorporate them into their study design. Although training involving approximately 12 patients has been found to ensure reliability of cardiac output measurement with TED imaging,1 this estimate has not been established in nurses who are novices at TED imaging. Not accounting for intranurse and internurse variability casts serious doubt on the validity and reliability of study results for goals 1 and 3 as well as the authors’ conclusions.
Hemodynamic measurements made by novice nurses using TED imaging and measurements obtained with PA catheters (goal 3) should have been compared first to establish whether sufficient agreement existed between methods. These data were essential to establish whether the comparison of hemodynamic measurements made by nurses using TED imaging versus physicians’ clinical estimates (goal 1) was valid. The authors did not state that goal 3 was achieved before goal 1 in their study.
Precision in studies comparing measurement methods is affected by the variability in timing between paired measurements.6,7 Cardiac output varies significantly and spontaneously during brief periods (7-30 minutes) in critically ill patients receiving mechanical ventilation.8 The authors stated that measurements of cardiac index and estimates of IVV by nurses using TED imaging were made "within one hour" of physicians’ clinical estimates. No data were provided as to the actual time that elapsed between the physician’s estimate and TED measurement or whether treatment was added, altered, or deleted during this time. Cardiac index and/or IVV could have been influenced by midazolam (administered during placement of the Doppler probe and after the physician’s estimate), changes in mechanical ventilatory settings (eg, tidal volume, positive end-expiratory pressure), and titrations of vasoactive infusions. Failure to limit and control the time between physicians’ clinical estimates and TED measurements casts serious doubt on the validity of the reported data for goal 1.
The authors concluded that TED imaging can provide potentially useful estimates of cardiac output and IVV in critically ill patients. This conclusion was based solely on significant correlations found between the 2 methods among approximately 25 paired measurements. Simple linear regression is inadequate when comparing measurement methods because correlation coefficients (r values) are insensitive to systematic error, whereby high correlation coefficients can be attained even in the presence of significant measurement error.7,9,10 Correlation coefficients serve only to determine the level of association between 2 measurement methods, not the level of agreement. Hence, the data presented for goals 1 and 3 are inadequate to support the authors’ conclusions.
The level of agreement between 2 measurement methods is established by using the statistics of bias, precision, limits of agreement, confidence limits, and Bland-Altman plots, which allow for visualization of variability patterns across the measurement range.9–11 The authors elected not to incorporate these statistics into this study’s data analysis plan although they cited multiple research studies4,5,12 that employed these statistics and plots in determining agreement of cardiac output using the methods of TED imaging and PA catheters.
Because of these multiple scientific concerns, 2 conclusions of this study warrant very cautious interpretation by readers: (1) TED imaging can provide potentially useful estimates of cardiac index and IVV in critically ill patients who require mechanical ventilation (goal 3) and (2) physicians’ clinical estimates of cardiac index and IVV are often inaccurate (goal 1). This study did not provide valid or sufficient evidence to conclude that hemodynamic measurements made by nurses using TED imaging are comparable with hemodynamic measurements from PA catheters in critically ill patients. Future method comparison research should employ rigorous study designs and Bland-Altman statistics to establish whether hemodynamic measurements made by nurses using TED imaging can replace invasive measurements from PA catheters in guiding decision making in the care of critically ill patients.
Burlingame, Calif
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
