Evaluation of accuracy of pulse oximetry in dogs

The accuracy of a pulse oximeter was evaluated over a wide range of arterial oxygen and carbon dioxide tensions, using 2 probes (finger probe and ear probe) and 2 monitoring sites (tongue and tail) in anesthetized dogs. The arterial oxygen saturation of hemoglobin (SaO2) measured directly with a multiwavelength spectrophotometer was compared with saturation estimated by pulse oximetry (SpO2). Linear regression analysis of the pooled data from 399 simultaneous measurements of SpO2 and SaO2 indicated a highly significant correlation of SpO2 with SaO2 (r = 0.97; P less than or equal to 0.0001). Although the mean difference (+/- SD) between SpO2 and SaO2 for pooled data was small (-0.06 +/- 6.8%), SpO2 tended to underestimate high SaO2 values (greater than or equal to 70%) and to overestimate low SaO2 values (less than 70%). When SaO2 values were greater than or equal to 70%, the ear probe applied to the tail was less accurate (produced a significantly greater SpO2-SaO2 difference) than the ear probe on the tongue, or the finger probe at either site. When SaO2 values were less than or equal to 50%, the finger probe applied at the tail was more accurate (produced significantly smaller SpO2-SaO2 differences) than the ear probe at either site. When SaO2 values were less than or equal to 70%, high arterial carbon dioxide tension (greater than or equal to 60 mm of Hg) was associated with greater overestimation of SaO2.