Confidence limits for interpretation of home blood pressure recordings

OBJECTIVE: Accurate and precise estimation of arterial blood pressure is needed for the management of hypertension. Multiple measurements can be obtained from recorded home blood pressures devices. Averages and other statistics can be calculated. Confidence intervals provide an assessment of the precision with which average pressures are measured. This study evaluated confidence intervals for systolic pressure, diastolic pressure, and pulse pressure from recorded home blood pressure measurements. METHODS: Fifty-three patients with high normal blood pressure, suspected white coat hypertension, or refractory hypertension were assessed by recorded home blood pressure. They were instructed to take four measurements each day for 1 week using a device that stores each measurement. Measurements were downloaded from the device to a computer, stored as data files, and analyzed by software. Average pressures and confidence intervals for each participant were rapidly calculated by standard statistical methods. RESULTS: The average width of the 95% confidence intervals for systolic pressure, diastolic pressure, and pulse pressure were 8, 5, and 6 mmHg, respectively with large inter-individual differences. Significant positive correlations were found between the width of the confidence intervals and the average systolic (P<0.05) or pulse pressure (P<0.01). However, the correlation for diastolic pressure was not significant. CONCLUSION: Confidence intervals can be easily calculated for recorded home blood pressures with device storage and may provide a useful approach when cut-off points for classification as normal pressure, high normal pressure, or definite hypertension are to be excluded. This information, combined with knowledge of other risk factors, may help guide decisions for improved management.