SESSION TITLE: Miscellaneous Critical Care SESSION TYPE: Original Investigation Poster PRESENTED ON: Wednesday, October 26, 2016 at 01:30 PM - 02:30 PM PURPOSE: The oxygen saturation readouts from a pulse oximeter (Radical-7®; Masimo, Inc., Irvine, CA) and a Point-of-Care analyzer (i-STAT®; Abbott Laboratories, Princeton, NJ) were observed to disagree appreciably on a patient who presented with severe alkalemia. This prompted us to undertake an accuracy analysis of these two instruments. METHODS: We consulted the specification sheet of the Radical-7 to determine the accuracy of that device. Then, we generated an SaO2 figure for paO2 values that were allowed to vary between forty and one hundred torr, in one-torr increments, for this alkalemic data set using a polynomial equation originally described by G. Richard Kelman (Kelman GR. Digital computer subroutine for the conversion of oxygen tension into saturation. J Appl Physiol 1966; 21: 1375-1376). Next, we implemented the on-board algorithm incorporated within the i-STAT for this profoundly alkalemic blood gas data set, reasoning that the device and its’ algorithm would be maximally challenged by ABGs that elicit appreciable shifts in the oxyhemoglobin dissociation curve (OHDC). The OHDC for the Kelman algorithm and the i-STAT algorithm were plotted on the same grid, using Numbers® spreadsheet software (Apple, Inc., Cupertino, CA), providing us with a digital and a diagrammatic comparison of their respective results. Finally, we determined the upper and lower limits of the uncertainty intervals for both the pulse oximeter and the Point-of-Care analyzer and scribed these elements on the single composite image. RESULTS: The Radical-7 specifications indicate that this device can be expected to be within +/- 4% of the actual SaO2, to the 95% level of confidence (+/- two standard deviations). The results generated by the i-STAT were remarkably similar to those generated by Kelman’s methodology, being well within one percent of each other throughout most of the clinically relevant range. The i-STAT’s algorithm operates on the prevailing paO2, which the device determines in real-time, but that paO2 figure is itself subject to error. Consequently, the i-STAT’s uncertainty interval was altered accordingly when that datum was varied by +/- 2 SD. The digital results allow observers to determine the differences in performance of the two instruments precisely for any given paO2, while the presentation of the same data in diagrammatic form confers a global sense of the relative performance of the two devices throughout the clinically relevant range of paO2s. CONCLUSIONS: In this mathematical modeling study, the accuracy of the oxygen saturation readout from the Point-of-Care analyzer was found to compare quite favorably to the accuracy displayed by the pulse oximeter. In addition, the OHDC generated by the i-STAT was strikingly similar to the corresponding curve generated by a methodology that has long been considered “the Gold Standard”. CLINICAL IMPLICATIONS: The current-day ICU abounds with electronic instruments, some of which display the same parameters. Becaue they base their clinical decisions, at least in part, on the readouts supplied by these devices, it behooves ICU Team members to be aware of which of these instruments tend to be more accurate than others. DISCLOSURE: The following authors have nothing to disclose: Bob Demers, Joshua Cosa, Melvin Welch No Product/Research Disclosure Information
Read full abstract