Sixteen studies (4,456 patients, range 27 to 997) were included in the review. Details on sampling strategy were poorly reported but most appeared to use a convenience sample; one study used a randomised sample and two studies enrolled consecutive patients. The proportion of quality items fulfilled ranged from 45 to 95%.
Ten studies concluded that automated devices were less accurate than manual devices, particularly the mercury sphygmomanometer, but most concluded that the differences were insufficient to cause concern when used clinically.
One study found that automated devices consistently had lower readings than manual readings but after five readings the differences converged.
Two studies stated that there was sufficient agreement between automated and manual readings to warrant their use in clinical settings. However, based one of these studies, 70% of systolic measurements and 77% of diastolic measurements were within 5mmHg of the manual measurement, and 91% of systolic measurements and 95% of diastolic measurements were within 10mmHg.
Three studies showed that the Dinamap automated device was less accurate than manual readings.
Three studies reported that various types of automated devices overestimated systolic readings compared with manual readings.
Two studies found that automated devices performed less well when used to measure blood pressure in hypertensive patients.
One study reported that a wrist monitoring device was consistent with mercury sphygmomanometer readings, but this was not supported by a second study.
One study compared two automated devices with a mercury sphygmomanometer and reported differences between all three readings.