Introduction Total sleep time (TST) can be estimated by both actigraphy and polysomnography; however, there could be discrepancies between the two tests especially in patients with sleep disorders. The clinical implication for large differences between the two test is not fully understood.

Methods We analysed 2-weeks actigraphy and inpatient polysomnography data of 156 patients attending our tertiary sleep service as part of a service evaluation project.

Results Although there was a significant correlation in TST between the two tests (ρ=0.21, p=0.01), actigraphy tended to underestimate total sleep time (average difference 24 minutes) and there was a large variation (-165 minutes to +214 minutes, 95% confidence interval on the Bland-Altman test). When analysing patients with insomnia (n=33) separately, the correlation was non-significant (p=0.91), and the difference was larger (46 mins/-181 to +274/). In patients with OSA (n=62) and PLMS (n=66), the difference was minimal (17/-171 to 206/and 12/188 to 212/mins; OSA and PLMS, respectively). The difference between TST estimated on PSG and actigraphy significantly correlated with daytime sleepiness (ρ=0.18, p=0.03) and inversely related to the presence of depression (42±102 and 11±90; depression and no depression respectively). There was no further correlation between the difference and any clinical outcomes.

Discussion There is a potentially clinically significant discrepancy in total sleep time depending which test is used to estimate. The difference may relate to insomnia, sleepiness and depression which need to be taken into account when interpreting the data.