Validation of actigraphy in hospitalised newborn infants using video polysomnography

Abstract

Actigraphy has been established as a reliable sleep assessment tool in adults; however, its utility in newborns remains unknown. Validation of actigraphy in newborns may provide a significant insight into the physiological and pathological acquisition process of mature diurnal sleep patterns and subsequent morbidities in both newborns and their mothers. Thus, the present study aimed to evaluate the accuracy of sleep-wake detection by overnight actigraphy in a cohort of newborns. Simultaneous recording of polysomnography and actigraphy data was performed in 40 newborns admitted to a tertiary neonatal intensive care unit (NICU). A mixed-effects logistic regression model to explain the sleep state identified by polysomnography was employed using the actigraphic activity score as a fixed independent variable and the individual newborn's identity as a random effect. To evaluate the usefulness of the actigraphic activity score as a surrogate marker of sleep, a receiver operating characteristic (ROC) curve analysis was performed using the variables that were used in the mixed-effects logistic regression model, and the area under the curve (AUC) was assessed. The results showed that polysomnography-determined sleep epochs were associated with a smaller activity index on actigraphy (odds ratio per 10 activity indices increase 0.81, 95% confidence interval [CI] 0.79-0.84). The AUC for the ROC curve was 0.87 (95% CI 0.87-0.88, range 0.54-0.99). An activity score of 124 showed the maximum overall accuracy (90.2%, 95% CI 87.7-92.1). Our present study suggests that sleep-wake states of NICU-hospitalised newborns can be precisely determined using actigraphy on the ankle.