ABSTRACT While artificial light in urban environments was previously thought to override seasonality in humans, recent studies have challenged this assumption. We aimed to explore the relationship between seasonally varying environmental factors and changes in sleep architecture in patients with neuropsychiatric sleep disorders by comparing two consecutive years. In 770 patients, three‐night polysomnography was performed at the Clinic for Sleep &amp; Chronomedicine (St. Hedwig Hospital, Berlin, Germany) in 2018/2019. Sleep times were adjusted to patients' preferred schedules, patients slept in, and were unaware of day‐night indicators. Digital devices and clocks were not allowed. Days were spent outside the lab with work or naps not allowed. After exclusions (mostly due to psychotropic medication), analysis was conducted on the second PSG‐night in 377 patients (49.1 ± 16.8 year; 54% female). Sleep parameters were plotted as 90‐day moving‐averages (MvA) across date‐of‐record. Periodicity and seasonal windows in the MvA were identified by utilizing autocorrelations. Linear mixed‐effect models were applied to seasonal windows. Sleep parameters were correlated with same‐day photoperiod, temperature, and sunshine duration. The MvA of total sleep time (TST) and REM sleep began a 5‐month‐long decline shortly after the last occurrence of freezing 24‐h mean temperatures (correlation of TST between 2018 and 2019 at 2‐month lag: rs 361 = 0.87, p &lt; 0.001; maximum peak‐to‐nadir amplitude: ΔTST ~ 62 min, ΔREM ~ 24 min). The MvA nadirs of slow wave sleep (SWS) occurred approximately at the autumnal equinox (correlation between 2018 and 2019: rs 361 = 0.83, p &lt; 0.001). Post hoc testing following significant linear mixed‐effect model indicate that TST and REM sleep were longer around November till February than May till August ( ΔTST = 36 min; ΔREM = 14 min), while SWS was 23 min longer around February till May than August till November. Proportional seasonal variation of SWS and of REM sleep as percentages of TST differed profoundly (SWS = 31.6%; REM = 8.4%). In patients with neuropsychiatric sleep disorders living in an urban environment, data collected in 2018 show similar patterns and magnitudes in seasonal variation of sleep architecture as the 2019 data. Interestingly, whereas SWS patterns were consistent between years with possible links to photoperiod, annual variations of TST and REM sleep seem to be related to times of outside freezing temperature. For generalization, the data need to be confirmed in a healthy population. No clinical trial was registered.

Relating Photoperiod and Outdoor Temperature With Sleep Architecture in Patients With Neuropsychiatric Sleep Disorders