Discrepancy between wrist-actigraph and polysomnographic measures of sleep in patients with stable heart failure and a novel approach to evaluating discrepancy

Wrist-actigraphy is often used to measure sleep characteristics in a variety of populations, but discrepancies between actigraphic and polysomnographic measures have been noted in populations experiencing poor sleep quality. The purpose of this study is to examine the discrepancy between these measures and risk factors for discrepancy in people with heart failure using a novel index. We used sleep measures simultaneously recorded by actigraphy and polysomnography, and clinical data from a cross-sectional study of 155 patients with heart failure (age = 60.5 [16.1] years; 65.2% male) recruited from evidence-based heart failure disease management programmes. The discrepancy and consistency between the two measures were evaluated using Bland-Altman plots, intra-class correlations and a newly developed index that represents activity counts in wake episodes. Overall, participants had short total sleep time (327.7 [95.9] min) and poor sleep efficiency (71.3 [16.0]%) on polysomnography. The discrepancies between sleep measures were small in patients less than 60 years old, and there was excellent consistency (intra-class correlation = 0.81) compared with older patients who had poorer consistency (intra-class correlation = 0.53) on total sleep time. Higher daytime motor activity, poor sleep quality and more severe insomnia were associated with smaller discrepancies in older, but not younger, patients, and associations were more sensitively detected by the new index. These findings suggest the importance of aging, disability and co-morbidity that may influence motor activity from which sleep estimates are scored with actigraphy. The new index may be useful in identifying factors associated with the correspondence between actigraphy and polysomnography.