Sl. Watkin et al., A comparison of pulse oximetry and near infrared spectroscopy (NIRS) in the detection of hypoxaemia occurring with pauses in nasal airflow in neonates, J CLIN M C, 15(7-8), 1999, pp. 441-447
Objective. The aim of this study was to compare the ability of NIRS and pul
se oximetry to detect changes in cerebral oxygenation occurring in response
to a pause in nasal airflow (PNA). Methods. Twenty-one recordings of cereb
ral oxygenation index by NIRS together with oxyhemoglobin saturation by pul
se oximetry were measured on 17 preterm infants with a history of apnoea. P
hotoplethysmography was used to confirm the accuracy of the pulse oximetry
data. PNA events were defined as pauses of greater than 4 seconds in a ther
mistor trace measuring nasal air flow. Results.Baseline variability in oxyg
enation index (Hbdiff) was found to be from -0.12 to +0.13 mu mol 100 g bra
in(-1). A fall in Hbdiff or SpO(2) was defined as a decrease of greater mag
nitude than 2 standard deviations from the baseline, i.e., -0.12 mu mol 100
g brain(-1) and 3% respectively. In 68% of 468 PNA events a fall in oxyhem
oglobin saturation (SpO(2)) was detected and in 56% a fall in Hbdiff was de
tected. In 20% of events there was no fall in cerebral oxygenation despite
a fall in SpO(2). In 8% of PNA episodes we recorded a fall in cerebral oxyg
enation but no fall in SpO(2). When a fall in cerebral oxygenation was reco
rded, the fall was greater when the event was also associated with a fall i
n SpO(2) (median (interquartile range (IQR)) 0.32 (0.21-0.69) vs. 0.25 (0.1
6-0.43) mu mol 100g brain(-1), p < 0.05). When all the PNA episodes were re
viewed no close correlation was shown between the magnitude of change in ce
rebral oxygenation and the change in SpO(2) for small changes in both indic
es. However, large falls (> 1.5 mu mol 100 g brain(-1)) in cerebral oxygena
tion were closely associated with large changes in SpO(2). Conclusions. We
conclude that both techniques are sensitive to changes in oxygenation durin
g PNA. Small changes in cerebral Hbdiff and arterial SpO(2) do not always c
orrelate for physiological reasons. A change in Hbdiff of > 0.3 mu mol 100
g brain(-1) is likely to be physiologically significant and is associated w
ith a change in SpO(2) of 12%.