BIAS AND VARIABILITY IN BLOOD-PRESSURE MEASUREMENT WITH AMBULATORY RECORDERS

This study sought to determine whether patient characteristics such as age, sex, blood pressure, and pulse pressure differently affect the accuracy of an oscillometric (SpaceLabs 90207) and a microphonic (TM2420 version 7) blood pressure monitor. Blood pressure recorded by two oscillometric and two microphonic ambulatory monitors was compared with simultaneous readings by two pairs of trained, blinded observers using random-zero sphygmomanometry. One hundred and eighteen subjects (53 men and 65 women, aged 17 to 94 years; systolic pressure, 89 to 211 mm Hg; diastolic, 44 to 116 mm Hg) were studied. There were no significant differences within each observer pair or between the two observer pairs as well as no correlation between interobserver differences and patient characteristics. The differences between the monitor and trained observers' readings were 2.8+/-9.9 mm Hg systolic and 3.9+/-6.8 mm Hg diastolic for the SpaceLabs and 5.0+/-5.2 mm Hg systolic and 3.4+/-6.1 mm Hg diastolic for the TM2420. Patient characteristics that predicted measurement error were defined by multiple regression. For oscillometry, systolic measurement error was highly correlated with systolic pressure, pulse pressure, and subject age. The diastolic error was significantly correlated with pulse pressure, diastolic pressure, and subject sex. For the oscillometric monitor, patient characteristics accounted for 36.6% of the variation of the systolic error and 34.7% of the variation of the diastolic error. For the microphonic monitor, only age correlated with diastolic error, and no significant correlations were seen with systolic error. Patient characteristics accounted for only 1.2% of the systolic and 8.9% of the diastolic error. We conclude that the systematic variation of accuracy with blood pressure level and pulse pressure of the oscillometric monitor would bias and reduce the spread of blood pressure values in population studies. It may also affect the accuracy in patient groups with high or low pressures and wide or narrow pulse pressures. For the microphonic method, accuracy is affected by age. These differences in accuracy necessitate specific validations in patient groups in which problems may be anticipated for both types of monitor.