Variability in Blood Pressure Measurements from Recorded Auscultation Sounds

Alan Murray1, Dingchang Zheng2, Chengyu Liu3, David Graham1, Jeff Neasham1, Adrian Cossor4, Clive Griffiths1
1Newcastle University, 2Anglia Ruskin University, 3Southeast University, 4AC Cossor


It is clinically recognised that blood pressure measurement is important, and that the standard manual auscultation method is still the current “gold standard”. Also, it is the international required method for the evaluation of automated blood pressure measurement devices. However, there is still much to assess in the variability of the manual technique. This study was designed to evaluate the variability of repeat measurements by the same operators, and between operators, when evaluating the same data. Ten young volunteer subjects with no known cardiovascular disease were studied. Korotkoff sounds were recorded from a standard stethoscope head connected to an acoustic microphone, and the audio sounds recorded to a computer during a blood pressure measurement with cuff deflation, while subjects sat quietly on a chair. All recordings were replayed blindly and independently, to two trained operators. The operators identified systolic and diastolic blood pressure (SBP and DBP) without knowledge of the subject. After all recordings had been analysed once, the analysis was repeated. In this young group SBP ranged from 95 to 112 mmHg, and DBP from 58 to 78 mmHg. The differences for repeat listening measurements for each observer was -1.6 ± 4.1 (mean ± SD) and -0.1 ± 1.5 mmHg for SBP, and 0.0 ± 1.4 and 0.1 ± 2.3 mmHg for DBP. The differences between the two operators was 0.8 ± 2.2 and 2.2 ± 3.7 mmHg for the two repeat measurements for SBP, and -0.9 ± 2.4 and -0.8 ± 1.9 for DBP. We have shown similar variability between operators as between repeat measurements on identical recordings.