Bench testing of 10 home blood pressure monitors using an oscillometric simulator demonstrated good intra-model reproducibility but notable inter-model differences.
BACKGROUND: Patients and doctors often use home blood pressure monitoring (HBPM) to assess the control of hypertension. Despite its popularity there has always been some uncertainty with regard to its accuracy, reliability, reproducibility, and comparability. Although there are pre-market HBPM standards of accuracy, there are no standards to assure accuracy of individual HBPM units after they have been brought home. OBJECTIVE: Determination of reliability, reproducibility, and comparability of 10 models of home blood pressure monitors. METHODS: We used a Biotek BP Pump as an oscillometric simulator of systolic and diastolic blood pressures to determine reliability, reproducibility, and comparability of 10 devices. RESULTS: All of the units tested, except the Pollonex BP1000, produced reproducible readings with the pooled SD of four blood pressure settings less than 3.10 mmHg both for systolic and for diastolic measurements. The oscillometric blood pressure pump method was found to be very reproducible, with pooled differences of less than 2 mmHg and SD of less than 0.5 mmHg for a repeated series of measurements using the same monitor. Different machines of the same model were also very comparable, with pooled differences of less than 3.6 mmHg and pooled SD less than 0.7 mmHg both for systolic and for diastolic readings. There were 11-14 mmHg differences between models for all of the simulated blood pressure readings except that a 27 mmHg difference was measured at the 200 mmHg systolic blood pressure level. These differences will not necessarily be the same for measurements with humans instead of oscillometric signal generation. A system for grading the accuracy of the tested HBPM that defines accuracy of HBPM as within +/-2 SD of the average of 85 measurements is described. CONCLUSION: All models of home blood pressure monitors tested, with the exceptions of the Pollonex BP1000, produced reproducible readings and different machines of the same model were comparable.
Yarows et al. (1999) studied this question.