Accuracy of two pulse oximeters during maximal cycling exercise

This study compared the measurement of oxygen saturation of haemoglobin (SaO2) by two pulse oximeters (Ohmeda Biox 3700e and Criticare 504 USP) with the measurement of SaO2 in arterial blood samples by CO-oximetry. Unlike many previous validation studies, arterial blood was sampled in ground glass rather than plastic syringes. Twenty men, 11 well-trained cyclists (mean +/- SE, age = 23.3 +/- 1.5 years, mass = 71.4 +/- 1.1 kg VO2max = 77 +/- 1 ml.kg1.min-1) and 9 relatively untrained subjects (age = 27.1 +/- 2.8 years, mass = 78.1 +/- 2.2 kg VO2max = 51 +/- 3 ml.kg 1.min-1) performed two maximal cycle ergometer tests each in an hypobaric chamber. The tests were at 745 mm Hg or 695 mm Hg with simultaneous measurement of SaO2 by the pulse oximeters and the CO-oximeter at rest, minute 7 of exercise and at VO2max. The best correlations, to the Co-oximeter measurement (SCO-OXO2) were found when all data from rest and exercise were combined (Criticare: r = 0.94; Ohmeda: r = 0.91). The bias measurements showed the Ohmeda underestimated SCO-OXO2 at all levels (mean = -2.5 +/- 1.9%) and the Criticare overestimated SCO-OXO2 at all levels, although to a lesser degree (mean = 0.9 +/- 1.5%). In conclusion, these results highlight the need for validation of individual pulse oximeters and that the effect of dyshaemoglobins must also be considered.