Accuracy and precision of fourteen pulse oximeters

Two sets of seven pulse oximeters (Criticare CSI-502; Nellcor-N200; Datex-Satlite; Physio-Control-Lifestat 1600; Critikon-Oxyshuttle; Ohmeda-Biox 3700; Ohmeda-Biox 3740; Radiometer-Oxi; Spectramed-Pulsat; Kontron-7840; Biochem-Ox2000; Invivo-4500; Engström-EOS; Novametrix-505) were studied in two groups of eight healthy subjects, aged 26-50 yrs. The transcutaneous oxygen saturation (SpO2) was compared with arterial oxygen saturation (SaO2) measured in simultaneously with drawn blood samples (OSM2 Radiometer) at four 20 min steady-state levels of inspired oxygen fraction (FIO2) (0.21, 0.10, 0.08 and 0.07; SaO2 99-55%) in a conditioned chamber. Both the error in accuracy (mean SpO2-SaO2 difference), and the error in precision (SD of the differences) remained less than 3% for the two highest FIO2 levels (SaO2 greater than 83%) but, during deeper hypoxia, they were increased to 8% and 5%, respectively. An instrumental systematic bias affected accuracy in particular. We concluded that a good agreement between SpO2 and SaO2, as reflected by the Bartko's intraclass coefficient, was observed in nine instruments.