Actigraphy correctly predicts sleep behavior in infants who are younger than six months, when compared with polysomnography

Abstract

Actigraphy has been widely used in adults and children. In infants, validation of actigraphy has typically used a comparison with behaviorally determined sleep state classification rather than polysomnography (PSG). This study validated actigraphy against PSG for determining sleep and waking states in infants who were younger than 6 mo. Twenty-two healthy infants, 13 term and 9 preterm, were studied at three different matched postconceptional ages. Actigraph data were compared with PSG recordings in 1-min epochs. Agreement rate (AR), predictive value for sleep, predictive value for wake, sensitivity. and specificity were calculated and compared between activity thresholds and across ages with two-way ANOVA for repeated measures. Thirty-two validation studies were analyzed. Overall AR with PSG of 93.7 +/- 1.3 and 91.6 +/- 1.8 were obtained at 2-4 wk and 5-6 mo, respectively, at the low activity threshold setting, whereas the auto activity threshold gave the best agreement with PSG at 2-4 mo (AR 89.3 +/- 1.3%). Sensitivity values of 96.2 +/- 1.1% at 2-4 wk, 91.2 +/- 1.5% at 2-4 mo, and 94.0 +/- 1.9% were obtained at these same settings. There was no difference across ages in AR or sensitivity. PVW and specificity values were low in this study. We conclude that actigraphy is a valid method for monitoring sleep in infants who are younger than 6 mo.