Accuracy of automated blood pressure measurements in the presence of atrial fibrillation: systematic review and meta-analysis

Atrial fibrillation (AF) affects ~3% of the general population and is twice as common with hypertension. Validation protocols for automated sphygmomanometers exclude people with AF, raising concerns over accuracy of hypertension diagnosis or management, using out-of-office blood pressure (BP) monitoring, in the presence of AF. Some devices include algorithms to detect AF; a feature open to misinterpretation as offering accurate BP measurement with AF. We undertook this review to explore accuracy of automated devices, with or without AF detection, for measuring BP. We searched Medline and Embase to October 2018 for studies comparing automated BP measurement devices to a standard mercury sphygmomanometer contemporaneously. Data were extracted by two reviewers. Mean BP differences between devices and mercury were calculated, where not reported and compared; meta-analyses were undertaken where possible. We included 13 studies reporting 14 devices. Mean systolic and diastolic BP differences from mercury ranged from −3.1 to + 6.1/−4.6 to +9.0 mmHg. Considerable heterogeneity existed between devices (I2: 80 to 90%). Devices with AF detection algorithms appeared no more accurate for BP measurement with AF than other devices. A previous review concluded that oscillometric devices are accurate for systolic but not diastolic BP measurement in AF. The present findings do not support that conclusion. Due to heterogeneity between devices, they should be evaluated on individual performance. We found no evidence that devices with AF detection measure BP more accurately in AF than other devices. More home or ambulatory automated BP monitors require validation in populations with AF.