Actigraphic assessment of a polysomnographic-recorded nap: a validation study

This study aimed to determine if actigraphy could differentiate sleep and wake during a daytime nap and no-nap rest period. Fifty-seven subjects participated in the study; 30 subjects were in the nap group and the remaining 27 in the no-nap comparison group. All subjects wore actigraphs while simultaneously undergoing polysomnography (PSG). Three actigraphic sensitivity levels (high, medium, low) and two interval duration minimums (15 and 40 min) were used to score the nap and no-nap data. The variables examined included total sleep time (TST), sleep latency (SL), wake after sleep onset (WASO) and sleep efficiency (SE). The Bland-Altman technique was used to determine concordance. Epoch-by-epoch analysis examined actigraphic accuracy, sensitivity and specificity. For the naps, all actigraph settings except low-40 showed significant correlations with TST. The high and medium settings predicted SE significantly and the high settings predicted SL significantly. Bland-Altman analyses demonstrated high settings overestimated TST while high and medium settings overestimated SE. Overall, for the nap condition accuracy for the actigraph was 82-86%, sensitivity was 92-96% and specificity was 40-67%. In the no-nap condition, accuracy for the actigraph was 60-84%, sensitivity was 47-78% and specificity was 60-86%. Medium-40 and low-40 were the only settings that did not misidentify sleep in the no-nap condition. These results suggest that actigraphy can predict TST, SE and SL reliably, depending upon parameter settings, and actigraphy is a highly sensitive but not specific measure for daytime naps. Different actigraphy settings may be optimal depending upon the variables of interest. Discrimination of sleep and wake during periods of waking quiescence is not as robust as during periods of mainly daytime sleep.