Aim: Pulse oximetry is a quick, easy, noninvasive method widely used f
or monitoring oxygen saturation (SaO(2)) in the neonatal period. The g
reatest recognized problem with SaO(2) readings measured from the oxim
eter is artifact, arising from an inability to record accurately durin
g movement. The aim of this study was to determine the proportion of p
ulse oximeter readings affected by movement artifact during sleep in t
erm and preterm infants. Method: Polygraphic recordings were obtained
from 11 term acid 6 preterm infants at postconceptional ages (SD) of 3
9.8 (0.8) and 39.3 (1.5) weeks respectively. The polygraphic and compu
ter recordings were divided into epochs of 30 seconds duration and ide
ntified as active sleep (AS), quiet sleep (QS), indeterminate sleep (I
S), or wakefulness (AW), using electrophysiological parameters and beh
avioral observations. Movement artifact was identified by visual exami
nation of polygraphic computer recordings using strict criteria. Signa
l containing artifact was removed from recordings and the percentage o
f artifact time present per recording calculated. Results: Signal arti
fact was present in recordings of all infants studied, comprising an a
verage state time of 19% during quiet sleep, 49% of active sleep, 49%
of indeterminate sleep and 91% of wakefulness. A significant differenc
e in the proportion of artifact present in recordings of term and pret
erm was observed only during quiet sleep. Conclusion: Movement artifac
t during pulse oximetry recordings is dependent on behavioral state, a
nd overall affects up to 50% of recorded traces. A reliable and more a
ccurate noninvasive method of recording oxygen saturation is thus need
ed, for use in both neonatal nurseries and in sleep studies, to aid in
accurate clinical decision-making.