Prediction of Arterial Oxygen Saturation With Partial Pressure of Oxygen Using a Turnover Model in Volunteers

Background: The relational equation for predicting arterial oxygen saturation (SaO₂) with a partial pressure of oxygen (PaO₂) has been described and used in clinical settings. However, the equation has limitations as it was derived empirically. This study aimed to assess the relationship between PaO₂ and SaO₂ using a turnover model. Material and Methods: In a controlled desaturation study to assess the accuracy of a pulse oximeter, volunteers (n = 12) breathed hypoxic gas mixtures via a mouthpiece. Various target SaO₂ values were achieved within the range of 70%–100%. PaO₂ and SaO₂ were measured using a CO-oximeter. A turnover model was fitted to PaO₂–SaO₂ pair data. The performance of the two SaO₂ prediction methods (conventional formula and turnover model) was evaluated using additional volunteer data not used in the model-building process. Results: The pharmacodynamic parameters were as follows: k_out (1/min) = 4.45 for Asians, 0.93 for Africans, I_max = 0.837 and IC₅₀ (mmHg) = 79.2%, γ = 5.24. PaCO₂ and pH were not significant covariates. The median prediction and median absolute prediction error were 5.6% and 5.7%, respectively, for the conventional formula, and 0.1% and 1.2%, respectively, for the turnover model. Conclusion: The relationship between PaO₂ and SaO₂ was better explained by the turnover model than by the conventional formula within the 70%–100% SaO₂ range.