Accuracy and trending capability of haemoglobin measurement by noninvasive pulse co-oximetry in anaesthetized horses

Objective: To assess the accuracy and trending capability of continuous measurement of haemoglobin concentration [Hb], haemoglobin oxygen saturation (SaO₂) and oxygen content (CaO₂) measured by the Masimo Radical-7 pulse co-oximeter in horses undergoing inhalational anaesthesia.

Study design: Prospective observational clinical study.

Animals: A group of 23 anaesthetized adult horses.

Methods: In 23 healthy adult horses undergoing elective surgical procedures, paired measurements of pulse co-oximetry-based haemoglobin concentration (SpHb), SaO₂ (SpO₂), and CaO₂ (SpOC) and simultaneous arterial blood samples were collected at multiple time points throughout anaesthesia. The arterial samples were analysed by a laboratory co-oximeter for total haemoglobin (tHb), SaO₂ and manually calculated CaO₂. Bland-Altman plots, linear regression analysis, error grid analysis, four-quadrant plot and Critchley polar plot were used to assess the accuracy and trending capability of the pulse co-oximeter. Data are presented as mean differences and 95% limits of agreement (LoA).

Results: In 101 data pairs analysed, the pulse co-oximeter slightly underestimated tHb (bias 0.06 g dL^-1; LoA -1.0 to 1.2 g dL^-1), SaO₂ (bias 1.4%; LoA -2.0% to 4.8%), and CaO₂ (bias 0.3 mL dL^-1; LoA -2.1 to 2.7 mL dL^-1). Zone A of the error grid encompassed 99% of data pairs for SpHb. Perfusion index (PI) ≥ 1% was recorded in 58/101 and PI < 1% in 43/101. The concordance rate for consecutive changes in SpHb and tHb with PI ≥ 1% and < 1% was 80% and 91% with four-quadrant plot, and 45.8% and 66.6% with Critchley polar plot.

Conclusions: Pulse co-oximetry has acceptable accuracy for the values measured, even with low PI, whereas its trending ability requires further investigation in those horses with a higher [Hb] variation during anaesthesia.