Comparison of oscillometric blood pressure measurement by two clinical monitors.

Sir,

I read with interest the article comparing oscillometric blood pressure measurement by two different clinical devices by Hegde et al.[1] The study essentially aims to compare measurements of blood pressure taken in the same patient using two electronic devices in a single sitting, over 10 min. As we can see, the study design here is noninterventional. However, I am sceptical about the statistical method used to interpret the data and, hence, the conclusion.

The authors opted for repeated measure ANOVA for analysing the results, which is inappropriate for this kind of study. Repeated measure ANOVA, also known as a within-subject design, is considered appropriate when all the subjects in a random sample are tested under different clinical conditions, or before and after the intervention under consideration. Here, we have different measurements for each of the samples, as each sample is exposed to different conditions. Statistically, it is not appropriate to use the standard ANOVA in those cases as the data violates the assumption of independence of data and, also, it will not be able to model the correlation between the repeated measures. The kind of data appropriate for this design would be a measure repeated over time, a measure repeated across more than one condition or several related and comparable measures. By collecting data from the same participants repeatedly, individual differences are eliminated or reduced as a source of between-group differences.

In clinical studies comparing two different measurement techniques, the investigator needs to see whether the two methods agree sufficiently to be used one in place of the other. One may also be interested to know which method is more accurate and reliable when measured repeatedly, comparing with a gold standard. In such investigations, when analysis is done using correlation coefficients, the interpretation of correlation can be misleading.[2] Limits of agreement is more informative when a researcher is interested in replacing one measurement method with another. Such an analysis is also useful to assess repeatability of one measurement method.[34] Alternatively, a significant between-method correlation could be demonstrated by linear regression analysis.[25]

In a study comparing two blood pressure measurement methods, van Ittersum et al. observed that the mean (95% CI) difference between the sphygmomanometer and oscillometric device differed significantly (P<0.01) within the case and control groups (within-subject design).[6] However, the mean differences between the two methods did not show a significant relation with the level of either systolic or diastolic blood pressures in linear regression analysis, and after dividing the distribution of blood pressures into tertiles.[6]

It is clear that repeated measures ANOVA is useful in interventional study designs. As far the present study is considered, measuring the blood pressure by one method cannot affect the readings taken subsequently. In fact, the inference drawn from the study done by Hegde et al. using repeated measures ANOVA would be that measurement of blood pressure by one device might cause a significant change in subsequent blood pressure recordings, which is biologically not plausible.[5] Therefore, in the present study, agreement between the methods should have been assessed by calculating differences between paired values of blood pressure measurements. Results must have been reported according to the Bland and Altman analysis.[2] To test the hypothesis that for each parameter the bias was zero, the paired t-test would be appropriate.[45]

Misinterpretations arise when inappropriate statistical methods are applied and, therefore, it is important to carefully consider the study objectives and the type of data obtained in clinical investigations before statistical methods are applied.[5] As the main objective of the present study was to compare the results obtained by the two methods in one individual subject, measurement of agreement would have been the appropriate statistic. Based on the present study, it cannot be accepted that the difference in mean blood pressure between these two devices was medically significant. Also, as the authors pointed out rightly, there is no gold standard method to measure mean blood pressure. More and more recent studies are taking mean blood pressure into consideration; it would be even more interesting to consider invasive methods to measure mean blood pressure.[7]