Measurement of blood pressure in the office.

Measurement of Blood Pressure in the Office To the Editor: The algorithm for office blood pressure recently proposed by Myers et al1 does not allow for the discernment of isolated clinic or masked hypertension. In a recent study we measured observed automated blood pressure (nurse present), unobserved automated blood pressure (3 readings at 3-minute intervals after 5 minutes of rest), and ambulatory blood pressure in groups of subjects with normal blood pressure, isolated clinic hypertension, and untreated essential hypertension. Mean daytime awake and automated unobserved blood pressures were similar in subjects with isolated clinic hypertension, whereas in normotensive and hypertensive groups, mean daytime awake blood pressures were similar to observed automated blood pressures (Table). Applying the proposed algorithm, 29 subjects (75%) with isolated clinic hypertension would go along the “continue-to-follow” pathway. Isolated clinic hypertension is known to be associated with an increased risk of both diabetes mellitus2 and hypertension.3 Moreover, in our study, impaired glucose tolerance was a common finding in subjects with isolated clinic hypertension, even in subjects with normal fasting glucose.4 There is currently no outcome data to suggest that isolated clinic hypertension subjects who go along the proposed “continue-to-follow” pathway have any less risk of future cardiovascular events than those who go along the proposed “home or 24-hour ambulatory blood pressure monitoring” pathway. Until outcome data are available for unobserved automated blood pressure readings, all subjects with isolated clinic hypertension need home and/or 24-hour ambulatory blood pressure monitoring over time to be able to discern if and when these patients develop sustained hypertension. This is particularly important given the frequency of impaired glucose tolerance in these subjects. There is another important methodological issue in the article by Myers et al1 that needs to be considered. The Bland-Altman plots suggest that the relationship between the average of the 2 blood pressures (in particular, the automated and ambulatory awake blood pressures) and the difference of the 2 blood pressures are linear, suggesting proportional bias. Also, the scatter of the 2 blood pressures appears to increase as the average of the 2 blood pressures increases, indicating likely heteroscedacity. Hence, using normal 95% CIs on untransformed data may not be the correct technique to use, as was highlighted in a recent review of the Bland-Altman method by Ludbrook.5