Accuracy of pulse oximetry in preterm and term infants is insufficient to determine arterial oxygen saturation and tension

Abstract

Aim: Oxygen saturation is frequently monitored with pulse oximetry to assess vital signs in critically ill patients. Optimally, pulse oximetry closely tracks arterial oxygen tension (PaO2), which provides guidance in oxygen titration. We investigated whether monitoring peripheral oxygen saturation (SpO2) could accurately guide oxygen titration in newborn infants. Methods: Twenty seven thousand two hundred thirty seven SpO2 readings were retrospectively paired with arterial oxygen saturation (SaO2) and PaO2 results from blood gas analyses performed in infants with arterial catheters in place. Results: SpO2 overestimated SaO2 readings by 2.9 ± 5.8%. When pulse oximetry readings were within the defined oxygen saturation target range, 7809 (20.9%) SaO2 values were below and 2830 (7.6%) exceeded the target range. In 57% of patients, PaO2 levels < 6 kPa was diagnosed while SpO2 readings were > 90%. PaO2 > 11 kPa was recorded in 19% of cases, when SpO2 readings were < 95%. Infants treated with supplemental oxygen showed a threefold increased risk of hypoxaemia compared to infants breathing room air. Sensitivity and specificity for detecting upper and lower target range limits were fair to good. For SpO2 values below 91%, ISO quality criteria were no longer fulfilled. Conclusions: Based on arterial blood gas analyses as reference, pulse oximetry readings did not fulfil the performance requirements for titrating oxygen in neonatal patients.