Résumé
To investigate sleep perception during a 32-h continuous bedrest polysomnography in patients with idiopathic hypersomnia (IH), nonspecified hypersomnia (NSH) and healthy controls (HC).
Patients with suspected IH, and HC (n = 14) underwent a regular polysomnography, a modified multiple sleep latency test (MSLT) and a 32-hour bedrest protocol, from 11PM till 7AM the day after, without circadian synchronizers. Patients were diagnosed with IH with long-sleep-time (IH-LST, n = 72) if they had a positive bedrest (sleep duration ≥ 19h), without LST (IH-nLST, n = 10, sleep duration < 19h, MSLT mean sleep latency ≤ 8min), or NSH otherwise (n = 26). Afterward participants estimated their sleep periods over the 32 hours. Relative errors were computed from perceived and objective sleep durations, compared across groups and analyzed in relation to clinical and sleep parameters.
NSH and IH-nLST showed higher relative errors compared to IH-LST and HC, during the whole bedrest (median error respectively 33.9%, 22.8%, 2.4%, 5.9%, p < .01), and specifically during daytime (7AM to 11PM, p < .01) and the second night (11PM to 7AM, p < .01). A linear negative association of relative error was found with objective sleep duration, shorter sleepers overestimating more their sleep duration, with a steeper covariance in NSH and IH-nLST (p < .01). Adjusted on objective sleep duration, sleep overestimation was associated with subjective and objective daytime sleepiness, and increased number of sleep bouts.
Overestimation of sleep duration is frequent in patients with hypersomnolence during a free-running PSG, especially in IH-nLST and particularly severe in NSH. This is a first step towards understanding the mismatch between subjective complaints and objective measures in these patients.