Using Hawksley random zero sphygmomanometer as a gold standard may result in misleading conclusions

We combined a database of paired blood pressure measurements taken using the Hawksley random-zero sphygmonanometer and a standard mercury sphygmomanometer and a database of paired measurements made on a SpaceLabs 90202 ambulatory blood pressure recorder and standard sphygmomanometer to determine how the SpaceLabs 90202 would have fared if it had been assessed against the Hawksley random-zero sphygmomanometer instead of a standard sphygmomanometer. The pooled database contained 255 triplicate readings. Using the standard sphygmomanometer as gold standard, the Spacelabs had a median error of 2 mm/Hg for both systolic and diastolic. Against the Hawksley random-zero sphygmomanometer, median error was -3 mm systolic and -6 mm diastolic. The proportion of errors > 10 mm rose from 11% (systolic) and 9% (diastolic) with the standard sphygmomanometer to 16% and 29% with the Hawksley random-zero sphygmomanometer. Because it underestimates systolic and diastolic pressures, the use of the Hawksley random-zero sphygmomanometer as a gold standard may have resulted in misleading conclusions about performance of some automated BP recorders.