The Hawksley random zero sphygmomanometer. Repeat experiment exonerates instrument.

EDITOR,-The two most common sources of interobserver variation in the measurement of blood pressure are measurement bias and digit preference. It is to reduce these errors that use of the Hawksley random zero sphygmomanometer is advocated. Once again, however, the accuracy of this instrument has been questioned, and it seems that, rather than reducing bias, this sphygmomanometer underestimates the true blood pressure. ' The manufacturer claims that "operator bias is eliminated from the measurement" (manufacturer's promotional literature), but even this has been questioned by the results of a small, informal study.2 In a much larger study we have confirmed that observer bias and digit preference are not eliminated by use of the Hawksley sphygmomanometer. I During a multicentre, double blind trial of a new antihypertensive agent 3621 measurements of blood pressure were made. Hawksley random zero sphygmomanometers that either were new or had been serviced immediately before the study were used. The investigators were asked to round up their measurements to the nearest 2 mmHg, use Korotkoff phase V to estimate diastolic pressure, and use a large cuff where appropriate. The study was carried out in 20 centres (seven hospitals and 13 general practices); a total of 173 patients attended a clinic on up to seven occasions. Each centre's preference for the terminal digit was examined for the three measurements random zero, systolic blood pressure, and diastolic blood pressure with the X2 test. In only three centres was no preference for the terminal digit detected. In the remaining centres five showed a preference in all three measurements, eight showed a preference in two measurements, and four centres showed a preference in one measurement. Zero was the digit chosen most commonly. These data reinforce the view of Conroy and colleagues that use of the Hawksley random zero sphygmomanometer should be abandoned. NMARY LAWSON