Transcutaneous oxygen monitoring : effect of electrode position and gestation on arterial partial pressure of oxygen for the first three days of neonatal life

Transcutaneous measurement (TCM) yielding transcutaneous partial pressure oxygen (TcPO[2]) reading provides a non-invasive method of continuously monitoring oxygen tension in critically ill mature and premature neonates. This study evaluates the relationship between TcPO[2] and the direct measurement of partial pressure of oxygen (PaO[2]). Aim: To identify sites associated with the highest correlation and least discrepancy between the TCM and the direct PaO[2] measurement by comparing factorial combinations of TCM probe placement sites and gestational categories of neonates </=3 days old. Method: Over a period of 10 months, PaO[2] and TcPO[2] readings were collected from eligible neonates </=3 days of age who were admitted to a neonatal intensive care unit (NICU) and required an arterial line and TCM. Readings from four thorax and four abdominal probe placement sites were analysed. Neonates (n=98) were stratified into three gestational groups: <28 weeks, 28-32 weeks and >32 weeks for analysis. Results: The highest TcPO[2]-PaO[2] correlations occurred with gestation periods >32 weeks. Less mature neonates, between 28-32 weeks' gestation, recorded lower TcPO[2]-PaO[2] correlations. For very premature neonates (<28 weeks) the TcPO[2]-PaO[2] correlation for the thorax was extremely low with abdominal readings correlating more highly. Discrepancies between TcPO[2] and PaO[2] readings ranged from underestimation of PaO[2] by thorax TcPO[2] readings for neonates <28 weeks' gestation, to overestimation of PaO[2] by thorax TcPO[2] readings for neonates 28-32 weeks' gestation. Higher levels of actual PaO[2] were associated with underestimation of PaO[2] from TcPO[2] whereas lower actual PaO[2] were associated with overestimation of PaO[2] from TcPO[2]. Conclusions: TcPO[2] readings taken from the abdomen appear preferable to thorax readings. Thorax TcPO[2] readings cannot be recommended for neonates <28 weeks' gestation. Other combinations of site and gestational group provided correlations that might be considered high, but TcPO[2] readings nevertheless failed to account for substantial PaO[2] variance for gestations and sites, leading to occasional gross errors of prediction, sufficient to compromise the maintenance of safe PaO[2] levels for the relevant cases. Systematic biases in the prediction of PaO[2] from TcPO[2] could exacerbate the risk of hypoxia and hyperoxia in clinical settings if TcPO[2] readings are interpreted as almost equivalent to PaO[2]. However, systematic discrepancies between TcPO[2] readings and PaO[2] readings appear less of a threat to oxygen tension management than random errors of prediction. Predictability of PaO[2] from TcPO[2] should be considerably higher than occurred with the current findings in order to permit accurate control of oxygen tension in premature neonates.