Comparison of blood pressure measurements between an automated oscillometric device and a Hawksley random-zero sphygmomanometer in the northern Sweden MONICA study

BACKGROUND: The Hawksley random-zero sphygmomanometer (random-zero) has been used widely in epidemiological observation studies. This study compares blood pressure measurements using the random-zero with measurements using an automated oscillometric device and suggests a correction of the automated oscillometric measurements to enable comparisons of blood pressure levels over time. METHODS: The northern Sweden MONICA population survey 2009 included 1729 participants, 853 men and 876 women, 25-74 years old. Blood pressure was measured using both random-zero and an automated oscillometric device in all participants. The Omron M7 digital blood pressure monitor was used for automated oscillometric measurements. A linear mixed model was used to derive a formula to adjust the automated oscillometric readings. RESULTS: Automated oscillometric measurements of systolic blood pressure were generally lower than random-zero measurements in women [oscillometric mean 122.1 mmHg (95% confidence interval: 121.0-123.2) versus random-zero mean 124.4 mmHg (123.5-125.5)], whereas automated oscillometric measurements of systolic blood pressure were generally higher than random-zero measurements in men [oscillometric 131.1 mmHg (130.0-132.2) versus random-zero 129.0 mmHg (127.9-130.1)]. For diastolic blood pressure, automated oscillometric measurements were higher in both women [oscillometric 79.9 mmHg (79.2-80.5) versus random-zero 76.7 mmHg (76.0-77.4)] and men [oscillometric 83.1 mmHg (82.4-83.8) vs. random-zero 81.2 mmHg (80.6-81.9)]. The difference also varied with age and order of measurement. Adjustment of the automated oscillometric measurements using mixed model regression coefficients produced estimates of blood pressure that were close to the random-zero measurements. CONCLUSION: Blood pressure measurements using an automated oscillometric device differ from those with random-zero, but the oscillometric measurements can be adjusted, on the basis of sex, age and measurement order, to be similar to the random-zero measurements.