The Probable Error of Blood‐pressure Measurements

Statistical analysis of 16,320 blood‐pressure measurements by three observers on 130 subjects has led to the following conclusions applicable to the conditions, subjects, and observers detailed in the text:— 1. When the effects of temporal differences and differences between the two arms are eliminated, systematic differences between the observers in this experiment are insignificant. 2. Significant differences in the variability of determination made by different observers were found. 3. The probable error of measurement of a single blood‐pressure observation is 1·2–1·8 mm. for systolic pressures and 1·8–2.· mm. for diastolic pressures when readings are made under conditions of rest and adequate time is permitted for the establishment of postural equilibrium. 4. The gain in precision obtained by averaging successive readings is greater for diastolic pressure than systolic pressure. 5. An average of more than 5 observations of blood‐pressure does not result in a useful gain in precision. Grateful acknowledgment is made by the authors to the Works Progress Administration (O.P. No. 465–03–3–631, Unit A‐8) for clerical assistance in the analysis of the data. For statistical advice and criticism of the manuscript we are indebted to Dr. Herbert S. Conrad and Dr. Robert C. Tryon.