Comparison of Actigraphy and Polysomnography in a Sleep Centre Cohort

Abstract

ABSTRACT Polysomnography (PSG) is the gold standard for sleep assessment, but actigraphy offers a low‐cost, non‐invasive multi‐night alternative. MotionWatch 8 was compared to PSG in a tertiary sleep centre cohort ( n = 107; age 37.6 ± 14.9 years; 56% female), examining whether actigraphy‐PSG agreement varied by demographics, or clinical factors. Patients underwent inpatient PSG with actigraphy for eight nights, including the inpatient night. Actigraphy overestimated total sleep time (TST; +27.31 min), sleep efficiency (SE; +6.09%), and fragmentation index (FI; +11.04 e/h, compared with PSG arousal index, exploratory) and underestimated wake after sleep onset (WASO; −14.86 min) versus PSG ( p < 0.05 except WASO). 95% limits of agreement were wide (TST = −96.78 to +151.40 min). Correlations were strong for TST ( r = 0.728), moderate for SE ( r = 0.540) and WASO ( r = 0.581), and weak for FI ( r = 0.328; all p < 0.001). Older age was associated with greater TST/SE overestimation and WASO underestimation ( p < 0.05). Periodic limb movements in sleep (PLMS; 17%) were linked to smaller TST/SE overestimation, and WASO underestimation, but greater FI overestimation ( p < 0.05); psychiatric comorbidity, OSA status, and AHI were non‐significant. MotionWacth 8 actigraphy provided clinically meaningful group‐level estimates, particularly for TST, but wide interindividual variability, showed actigraphy and PSG not being equivalent or interchangeable. PLMS altered the magnitude and direction of actigraphy–PSG differences, reducing sleep overestimation while exaggerating movement‐derived fragmentation. Recording setting influenced actigraphy‐derived sleep continuity, so findings from a single inpatient comparison should not generalise to home sleep. These findings support cautious use of actigraphy as a PSG adjunct, with awareness of limitations in detecting quiet wakefulness and movement‐related fragmentation.