To the Editor:
I read with great interest the article by Briegel et al.,1 recently published in Anesthsiology, in which the authors present a brilliant new method for noninvasive intermittent blood pressure monitoring. For validation, the authors compared a set of blood pressure data simultaneously measured using the new method and blood pressures recorded from a femoral artery catheter.
For further analysis of a total of 1,887 measurements recorded in 110 patients, Briegel et al. presented an analysis for clustered observations according to Bland and Altman. After their cornerstone paper in 1986,2 Bland and Altman have detailed their approach in several following papers.3,4 I would like to make some comments on the statistical approach used by Briegel et al. to compare the two methods.
First, a further reduction to a data set of five measurements for each patient that resulted in only 550 paired measurements was argued with a high weight for mean arterial pressure values between 75 and 85 mmHg and a need for more homogenous distribution. I am skeptical about this statistical approach, because (1) excluded data may have larger limits of agreement than included data; (2) excluding data located in the center of a scatter plot potentially results in a better correlation as expressed by Pearson’s r; and (3) the method presented by Bland and Altman allows for different numbers of measurements per subject. Hence, excluding data is not necessary.
Furthermore, Briegel et al. refer to Bland and Altman’s approach but do not specify their analysis. It remains unclear whether the authors used both the variance for repeated differences between the two methods on the same subject (calculate as the residual mean square) and the variance for the differences between the averages of the two methods across subjects (calculated as the difference between mean squares for subjects and the residual mean square), as proposed by Bland and Altman.
Finally, a single measurement period of the new blood pressure measurement method presented in the article lasts about 60 s, which is long compared to the conventional oscillometry. During this period the tissue pressure of the upper arm is raised significantly, potentially influencing later measurements. Therefore, estimating the repeatability of the new method should be analyzed, as proposed by Bland and Altman.
Competing Interests
The author declares no competing interests.