1995 OPEN FORUM Abstracts
EVALUATION OF THE NONIN FINGER PHANTOM AS A STANDARD FOR CHECKING PULSE OXIMETRY PERFORMANCE
Teresa Volsko RRT Thomas J. Kallstrom RRT, Robert L. Chatburn RRT. St. Elizabeth's Hospital Medical Center, Youngstown, OH, Rainbow Babies & Childrens Hospital Cleveland, OH.
Pulse oximeters are unique among patient monitors in that they cannot be calibrated nor can calibration be truly verified. The purpose of this study was to determine if an inexpensive, commercial device simulating a human finger could produce SpO2 measurements that were within the error specifications supplied by pulse oximeter manufacturers. METHOD: Five brands of pulse oximeters were evaluated. We used the Nonin Finger Phantom to simulate SpO2 values of 95%, 90%, and 80%. Pulse rate was simulated by manual compression of the device using a metronome at rates of 120, 84, and 60 beats/min. For each saturation level, 8 measurements (different probes) were made at each pulse rate (n=24). SpO2 measurements were temperature corrected. Bias and imprecision of measurements were evaluated with t-and chi-square tests. Inaccuracy intervals (Levine, Fromm. Critical Care Monitoring. Mosby, 1995:29-33.) were constructed to include 95% of future measurements at the 99% confidence level.
Results: Observed and manufacturer's specified standard deviation (SD, Spec. SD), observed and specified mean difference between measured and true values ([Delta-bar] Spec. [Delta-bar]):
Brand/Probe SDSpec. SD p [Delta-bar]Spec. [Delta-bar] p
Novametrix Y1.42.0< 0.001 2.1 0 < 0.001
Ohmeda Oxytip 0.62.4< 0.001 1.7 0 < 0.001
Nellcor D-251.62.0< 0.001 2.2 0 < 0.001
Nonin 8000K21.22.0< 0.001 0.8 0 < 0.001
BCI Finger Probe 3024 1.12.0< 0.001 1.7 0 < 0.001
Nellcor I-201.92.0 0.4450.8 0 < 0.001
Inaccuracy intervals are shown below compared to specified intervals (dotted lines at ± 4%):
SEE ORIGINAL GRAPH
Conclusions: Simulated saturations show less imprecision but more bias than manufacturer's specifications. Inaccuracy intervals indicate that individual measurements may fall outside specified values by chance for well functioning oximeters. The Finger Phantom is a useful tool if mean values of repeated measurements are used. Spot checks with single measurements may not be adequate for some brands of pulse oximeters.