Observer error in blood pressure measurement.

This paper describes an experiment undertaken to determine observer error in measuring blood pressure by the auscultatory method. A microcomputer was used to display a simulated mercury manometer and play back tape-recorded Korotkoff sounds synchronized with the fall of the mercury column. Each observer's readings were entered into the computer, which displayed a histogram of all readings taken up to that point and thus showed the variation among observers. The procedure, which could easily be adapted for use in teaching, was used to test 311 observers drawn from physicians, nurses, medical students, nursing students and others at nine health care institutions in Ottawa. The results showed a strong bias for even-digit readings and standard deviations of roughly 5 to 6 mm Hg. The standard deviation for the systolic readings was somewhat smaller for the physicians as a group than for the nurses (3.5 v. 5.9 mm Hg). However, the standard deviations for the diastolic readings were roughly equal for these two groups (approximately 5.5 mm Hg).