P1.225: Is preductal saturation helpful during neonatal resuscitation?

Background: When neonatal resuscitation is anticipated, a preductal pulse oximeter probe is placed on the neonate's right hand. This probe measures preductal saturations (SpO2) and heart rate (HR) as recommended by the Neonatal Resuscitation Program. Recently electrocardiogram (EKG) has replaced pulse oximetry to measure HR in the delivery room. Hassan et al. 2015 have shown increased bias and decreased accuracy of pulse oximetry during resuscitation using a porcine model. No studies have validated the variance in pulse oximetry using a transitioning resuscitation model with fetal arterial oxygenation (PaO2) comparable to newborn infants.

Methods: To evaluate the variance in preductal pulse oximetry during resuscitation using a transitioning ovine cardiac arrest model. After IACUC approval, near-term lambs (138 - 141d) gestation equivalent to term neonates were resuscitated post-cardiac arrest induced by asphyxia. Nonin pulse oximetry (SenSmart Model X-100) probe was used to measure preductal SpO2. SpO2 values before, during resuscitation, and 5 min post return of spontaneous circulation (ROSC), if achieved, were used. The data points were collected every 4 seconds during the prespecified observation period. Percentage of measurable values were recorded. SpO2 values were further analyzed between lambs that achieved ROSC vs. No ROSC. Post ROSC values were analyzed for 5 min. Non-parametric tests with a median, interquartile range, and F-test for variance in SpO2 values were used to analyze the data.

Results: Figure1 shows the characteristics of 10 lambs. Fifty percent of the lambs took 12 (10,18) min to achieve ROSC. The PaO2 value was 15 (10, 24) mmHg during resuscitation. We obtained 2944 data points (baseline - 160, during resuscitation (10min) - 2238, post ROSC (5min) - 546). Of all the recorded data points, we measured 97% during baseline, 48% during resuscitation (10 min), and 98% post ROSC (5 min). The variance in baseline values between lambs that achieved no ROSC vs. ROSC was F - 1.037, p=0.84. A high variance was observed during resuscitation between the no ROSC vs. ROSC lambs (10 min), F-1.759, p< 0.001, showing unreliable data that cannot be used to predict outcomes (Fig2a, 2b, 2c). During the first 5 min of resuscitation, we analyzed SpO2 between no ROSC vs. ROSC for each minute (Fig 3), which showed wide variability, confirming unreliable readings.

Conclusion: In a large pulse oximetry dataset, measurable SpO2 values were less than 50% of the recorded values during resuscitation. Significantly high variance confirms the unreliable and unpredictable nature of SpO2 during resuscitation.