Evidence for the need of bedside accuracy of pulse oximetry in an intensive care unit

Objective: To compare pulse oximetry saturation (SpO(2)) with arterial bloo d gas saturation (SaO(2)) obtained during clinical routine to determine the optimal lowest reliable value of SpO(2) in ventilator-dependent patients b efore setting up a nurse-directed protocol of FIO2 titration. Design: Prospective clinical study. Setting: Surgical intensive care unit in a university hospital. Patients: Thirty-three patients with a pulse oximeter probe in whom arteria l blood gas was measured with a radial artery line. Interventions: SpO(2) was recorded by the nurses and compared with SaO(2) o btained by blood gas analysis with a co-oximeter. Two sensors currently use d in our surgical intensive care unit and connected to a monitor (HP OmniCa re M1165/66A; Hewett Packard, Andover, MA) were tested. In group I, the Dur ensor DS 100A (Nellcor Puritan Bennett, Pleasanton, CA), a reusable sensor, was used. In group II, the Oxisensor D25L (Nellcor Puritan Bennett), a non reusable sensor, was used. Measurements and Main Results: In group I, 64 data pairs were obtained. In this group, SaO(2) ranged from 87 to 98% and SpO(2) ranged from 92 to 100%, The bias was -1.90% and the limits of agreement ranged from -5.56 to 1.76% . In group 11, 47 data pairs were obtained. In this group, SaO(2) ranged fr om 87 to 99% and SpO(2) ranged from 92 to 100%. The bias was -2.49% and the limits of agreement ranged from -6.62 to 1.64%. Conclusions: In the range of SaO(2) tested, regardless of the sensor used, SpO(2) overestimated SaO(2). Large limits of agreement were found. Based on this result, the authors concluded that before defining a nurse-directed p rotocol of FIO2 titration with SpO(2), the material used daily must be eval uated. A minimum threshold SpO(2) value of 96% in both groups I and II is m ore reliable to ensure SaO(2) greater than or equal to 90%.