Performance Analysis of Two Point-of-Care Analyzers for Generating Oxygen Saturation Values During Acid-Base Homeostasis

Abstract

Background:

The i-STAT (Abbott Laboratories, Princeton, NJ) and the RapidPoint 500 (Siemens, Norwood, MA) point-of-care analyzers each generate oxygen saturation values at the bedside by applying an on-board algorithm to the oxygen tension (PaO₂) figure generated by their internal miniaturized Clark electrodes. We sought to compare the accuracy of the digital readouts supplied by each analyzer, both to each other, and also to a polynomial expression first described in 1966, when applied to a homeostatic arterial blood gas (ABG) data set (pH = 7. 40; PaCO₂ = 40 torr; [HCO3-] = 24.0 mEq/L).

Methods:

We implemented the i-STAT's and the RapidPoint's algorithms for this data set for integer values of PaO₂ranging between 10 and 100 torr. This succeeded in generating an oxyhemoglobin dissociation curve (OHDC) for each of the analyzers. We then generated an actual OHDC using a polynomial equation originally described by Kelman1. The OHDCs for the Kelman Equation, the i-STAT's algorithm, and the RapidPoint's algorithm were plotted on the same grid, using Numbers spreadsheet software (Apple, Cupertino, CA). This provided us with a visual comparison of the respective accuracy of the analyzers over a broad range, as compared to the Kelman Equation, which has long been considered to be the accepted standard.

Results:

The accuracy of each of the analyzers was observed to be impressive, with the RapidPoint being slightly more accurate than the i-STAT. In the Figure displaying below, the actual OHDC, generated using the Kelman Equation, displays in red, while the RapidPoint-generated OHDC displays in blue, and the OHDC referable to the i-STAT displays in black.

Conclusions:

In this mathematical modeling study of oxyhemoglobin saturation, the accuracy of the oxygen saturation readouts from two point-of-care analyzers were found to be within two percent of the actual value throughout the physiologic range, in the presence of homeostasis. Furthermore, the OHDCs generated by the point-of-care analyzers were strikingly similar to the corresponding curve generated by a methodology that has long been considered to be the accepted standard, lying well within one percent of the prevailing actual saturation at PaO₂ values lying between 23 and 100 torr.

1. Kelman GR. Digital computer subroutine for the conversion of oxygen tension into saturation. J Appl Physiol 1966;21:1375-1376.