Natural speed, please.
Reliable and accurate pulse oximetry requires the design and operation of the entire system to function together. Contributions ranging from the sensor technology, cabling and electronics, signal processing and alarm technologies, to communicating results to caregivers through local and remote displays, are all critical and interrelated (Figure 1). Many years of experience in each of these individual areas has allowed us to refine and expand Nellcor™ pulse oximetry technology into a robust tool for providing critical information to caregivers. Each of these areas requires significant attention to detail and is deserving of a separate discussion regarding its contribution to accuracy within the Nellcor™ OxiMax pulse oximetry system. Accurate and reliable SpO2 readings, however, begin at the sensor as each of the downstream processes is only as good as the quality and predictability of the sensor signals that feed into them. The optical interfaces between sensor and blood-perfused tissues; the optical, electrical and mechanical design of the sensors themselves; and robust sensor calibrations serve as the foundation of pulse oximetry technology.
Demonstrating a pulse oximetry system’s SpO2 reading accuracy requires comparisons to the known true arterial oxygen saturation. We describe here the testing methodology and performance results comparing SpO2 readings from the Nellcor™ OxiMax pulse oximetry system, along with four other commercially available systems, to the gold-standard for blood oxygenation measurements: arterial SaO2 as determined by laboratory CO-oximetry.
Controlled Laboratory Study Environment Manufacturers determine the fundamental and underlying accuracy of pulse oximetry in controlled laboratory settings using consented healthy adult volunteers along with procedures designed to emphasize safety, control and reproducibility. Volunteers are instrumented with an indwelling arterial catheter for periodic blood sampling and are then exposed to progressive stepwise hypoxic air/nitrogen mixtures to attain the specified saturation range. Stable oxygen saturation levels are maintained to ensure that tissues at the pulse oximetry sensor site are at the same arterial saturation as found at the radial artery. Study methodology for validating accuracy under stable conditions is outlined in the pulse oximetry International Standard, ISO 9919.
Accuracy is stated statistically as the root-mean-square of the differences (ARMS) between the observations of SpO2 and simultaneously sampled blood SaO2.* When evaluating baseline pulse oximeter accuracy, subjects are kept relatively immobile and have normal levels of perfusion. ARMS levels observed in clinical settings are commonly greater than in laboratory studies, but do not necessarily indicate an oximeter system is inaccurate. Determining a hospitalized patient’s “true” value of SaO2 for comparison to SpO2 is itself subject to significant error – hemodynamic instabilities and varying oxygen saturation create “noise” unrelated to the pulse oximeter, since the underlying saturation can differ meaningfully at the two sampling locations at the instant of data collection2,3. Interferences with pulse oximetry, such as severe anemia or the presence of intravascular dyes or dyshemoglobins, can also affect SpO2 readings. 4,5,6 Without strictly controlling such influences during a study, results become highly dependent on the particular sampled population and often confuse comparisons among published values.
Accuracy Study In order to demonstrate the baseline SpO2 reading accuracy in the Nellcor™ OxiMax pulse oximetry system and compare it with four other commercially available systems, we conducted a prospective study of ten sensor/monitor combinations at Nellcor’s performance testing laboratory in Pleasanton, CA, following the study methodology outlined in the previous section. While only a subset of OxiMax sensors and monitors are shown here, each combination of sensor and monitor has been validated within the entire Nellcor™ OxiMax family to ensure the products’ reliable and accurate performance. The ten tested systems are shown in Table 1. Twelve adult volunteers were studied, with more than 300 blood draws collected over a 53.3% to 99.5% SaO2 span (Table 2). Bias, precision and ARMS values were calculated for each system. The z-statistic was used to determine if differences between observed ARMS values and performance levels of 2% (70%100% SaO2) or 3% (60%-80% SaO2) were significant at a level of p<0.05.
"The z-statistic"
An American speaker would say 'zee-statistic' rather than 'zed-statistic'.