If you wish to respond to a paper or other item already published in the BJA, please go to the abstract/full text version of that item and click on the link "E-Letters: Submit a response to the article".
Electronic Letters to:
|
|
E-letters: Electronic letters published:
-
![[Read E-letter]](/icons/misc/sectionDOWN.gif)
Non-invasive estimated cerebral perfusion pressure: relatively inaccurate in brain injured patients
- Peter J Davis, David J. Grant (31 January 2005)
|
|
|||
|
Peter J Davis, Consultant Paedaitric Intensivist Bristol Royal Hospital for Children, David J. Grant
Send letter to journal:
|
In the paper of Edouard et al (1), the aim of the study was to assess the accuracy of a non-invasive assessment of cerebral perfusion pressure (CPPe), using a formula combining the phasic values of middle cerebral artery (MCA) Doppler velocities and invasively measured mean arterial pressure (AP), compared to "measured" CPP (CPPm), calculated as the arithmetic difference between AP and invasively measured intracranial pressure (ICP). Unfortunately it would appear that the statistical analysis used in this paper is neither totally correct nor complete. In the first part of the study, repeated measurements were performed on a group of 10 patients and analysed by covariance to assess linear regression. A correlation is found between the 2 measurements, but the slope value is somewhat low (0.76 on the right side and 0.65 on the left side). That this statistical method is not the best for assessing agreement between two methods of clinical measurement was demonstrated by Bland & Altman (2), nearly 20 years ago. However the bias and precision are calculated as per Bland and Altman. It is not clear from the paper whether only 1 pair of measurements per patient was used, or all the paired measurements. For statistical validity, the former should be the case, unless the calculation for measuring agreement using repeated measurements is used, which would appear not to be the case. The values are given numerically in the results but are not shown graphically, which lessens their impact. For the second part of the study, the same statistical methods are used, again producing a very low slope value of 0.55. The bias and precision were calculated correctly, both before and after the vascular reactivity test for both right and left sides. An “analysis of residuals” was performed and shown graphically. Unfortunately, only the bias figures are shown, no mention is made of the “limits of agreement”, and unlike a Bland-Altman plot, the differences are plotted against the measured CPP rather than the mean of the 2 methods of assessing CPP, which leads to a statistical artefact, in that the difference will be related to the measured CPP (3). From the numbers provided by Edouard et al, it is possible to perform the appropriate calculations for the "limits of agreement" between the two methods of measuring cerebral perfusion pressure ["Limits of agreement" = Bias +/- (1.96 x Precision)]. These calculations provide the figures that are the most clinically relevant in deciding whether one method can be used in place of the other. Using group B before the change in tidal volume as the "best" results, the non-invasive assessment using transcranial Doppler sonography (CPPe) may overestimate the cerebral perfusion pressure by approximately 13mmHg (13.3 on right, 13.9 on left) or underestimate it by up to 19mmHg (18.9 on right, 16 on left), in patients with moderate intracranial hypertension (measured ICP = 28mmHg). These differences are substantial and imply that it may not be possible to use (CPPe) and (CPPm) interchangeably. For group A, assuming that the bias and precision measurements have been performed correctly, i.e only one pair of measurements from each patient, the non-invasive assessment (CPPe) may overestimate the cerebral perfusion pressure by approximately 30mmHg (28.4 on right, 32.4 on left)or underestimate it by over 42mmHg (42.6 on right, 42.4 on left). These differences are frankly enormous. This is not obvious from either Fig 1 or Fig 2, which appear to show a good correlation between the 2 methods. More concerning still is that with a fall in tidal volume, causing a rise in PaCO2 and PECO2, a subsequent rise in the average measured ICP (28 to 38mmHg) and a fall in average measured CPP by 7mmHg, the limits of agreement for group B become even more widely spread, with the indirect method (CPPe) overestimating CPP by up to 35-40mmHg (34.8 on right, 39.6 on left) or possibly underestimating it by as much as 20mmHg (18.6 on right, 20 on left). This would be most unacceptable for clinical purposes, especially in those patients at risk of having a low CPP. As Edouard et al point out in the discussion of their paper, bilateral and continuous measurement of MCA velocities automatically linked to a measurement of AP may have potential benefit for the assessment of CPP when it is neither practical nor safe to measure ICP directly. Unfortunately on the basis of their preliminary data, it would appear to be a somewhat inaccurate method, particularly in unstable patients with a risk of severe intracranial hypertension. In contradiction to the assertion in the conclusion of the abstract, that non-invasive estimation of CPP can be used for brain monitoring of head-injured patients, even allowing for the proviso that the accuracy of the method may depend on the level of intracranial hypertension, much more research with appropriate statistical analysis will be needed before such a technique can be applied clinically in such patients, if at all. 1. Edouard AR, Vanhille E, Le Moigno S, Benhamou D, Mazoit J-X. Non- invasive assessment of cerebral perfusion presure in brain injured patients with moderate intracranial hypertension. Br J Anaes 2005; 94: 216-21 2. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; i: 307-10 3. Gill JS, Zezulka AV, Beevers DG, Davies P. Relationship between initial blood pressure and its fall with treatment. Lancet 1985; i: 567-9 Conflict of Interest:None declared |
|||