Ambulatory Blood Pressure Monitoring and Diagnostic Errors in Hypertension

Abstract

Random variability of blood pressure complicates the diagnosis and subsequent treat ment of hypertension. To evaluate the importance of the number of blood pressure measurements in the correct diagnosis and control of hypertension, the authors used a Bayesian model to estimate the true average blood pressure of a group of newly diagnosed hypertensives, then calculated the diagnostic error that would result from monitoring methods using 24 daytime measurements or from using only three random monitoring measurements. The study population consisted of 129 individuals with newly diagnosed mild hypertension according to standard criteria, who were also eval uated with an ambulatory blood pressure monitor. In true normotensives (daytime di astolic blood pressure <90 mm Hg), the negative predictive value with three measure ments was 0.92, and it rose to 0.96 with monitoring methods. In mild hypertensives (90-104 mm Hg), the positive predictive value was 0.64 with three measurements and 0.84 with monitonng methods, thus reducing the rate of false mild hypertensives from 35% to 15%. Finally, in patients with moderate or severe hypertension (>104 mm Hg), the positive predictive value improved from 0.26 with three readings to 0.61 with mon itoring methods. Similar results were observed with daytime systolic pressure mea surements. As the number of measurements increased, the diagnostic error due to the random variability of blood pressure became progressively smaller. In cases of hyper tension, the large improvement in predictive values may justify using monitoring meth ods to confirm standard diagnosis. Key words: ambulatory blood pressure monitoring; Bayesian analysis; blood pressure; diagnostic error; hypertension; random variability. (Med Decis Making 1998;18:429-435)