• presence of REM phases [ Time Frame: Up to 5 consecutive nights from 10:00 pm to 7:00 am ]
  analyzed as binary variables:0% or > 0% of Total sleep time
• Duration of SWS phases [ Time Frame: Up to 5 consecutive nights from 10:00 pm to 7:00 am ]
  analyzed as binary variables: < 9% or > 9% of Total Sleep Time

• presence of REM phases [ Time Frame: Up to 5 consecutive nights from 10:00 pm to 7:00 am ]
  We analyzed as binary variables:0% or > 0% of Total sleep time
• Duration of SWS phases [ Time Frame: Up to 5 consecutive nights from 10:00 pm to 7:00 am ]
  We analyzed as binary variables: < 9% or > 9% of Total Sleep Time

This is a prospective observational study conducted from April 2016 to October 2017 in the general-neurologic ICU In a university-affiliated hospital serving a population of 1 million habitant in the north-east of Italy. Patients were considered eligible if they were older than 18 and remained in the ICU for at least 24 hours. Patients with anamnestic severe dementia, severely decreased mental state defined as Richmond Agitation-Sedation Scale (RASS) lower than -3, and language barrier were excluded. Patients coming from other hospitals were excluded, because they might have been treated with long-lasting sleep or sedative drugs, which could deeply modify sleep quality. Patients needing continuous infusion of sedative drugs were also excluded since this could interfere with the interpretation of PSG data. Finally, patients with frontal wounds, brain lesions in the frontal lobes, external ventricular drainage and noninvasive mechanical ventilation were excluded because of the impossibility to position the PSG device.

The study was conducted in accordance with the Declaration of Helsinki and was approved by the local Ethics Committee of Brescia (NP 23332). Patients' informed consent was waived due to the lack of definition of a legal representative of temporarily incapacitated adult patients in the Italian legislation. The informed consent was obtained from the surviving patients as soon as they regained their mental competency. Family members received detailed information on the study scope and protocol. .

Sleep was monitored using a simplified PSG, Sleep Profiler (Advanced Brain Monitoring) with the device placed by the nursing staff, from 10:00 pm to 7:00 am, for a minimum of one night to a maximum of five consecutive nights . In case of occurrence of any of the exclusion criteria, such as positioning of external ventricular drainage, the PSG monitoring was interrupted. The authors collected data on age, sex, the Simplified Acute Physiology Score (SAPS II) reason for ICU admission, length of ICU stay, and presence of delirium, assessed three times daily using the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU). Any administration of drugs that could influence sleep were also recorded, including benzodiazepine as anxiolytics, opioids, dexmedetomidine, zolpidem, and melatonin.

The Sleep Profiler consists of a band with three channels for EEG, one channel for the electromyography of the masseter muscle (a touch proof connector with two electrodes), a plethysmography for heart rate, a microphone to register both ICU noise and patient snoring, and a 3D accelerometer to record head movements23. The registered parameters were then analyzed by a software that identifies sleep stages and Cortical Arousals (CA), and a report was generated for each night.

The authors analyzed the association of nocturnal noise, defined as the percentage of Total Sleep Time (TST) during which noise exceeds 40 decibel (dB), with duration of Rapid Eye Movement (REM )and Slow Wave Sleep (SWS) phases, expressed as a percentage of TST, and number of CAs.

STATISTICAL ANALYSIS Characteristics of the study population were summarized as number (percentage) for binary variables, and as mean (standard deviation, SD) or median (interquartile range, IQR), as appropriate, for continuous variables.

The study outcomes were duration of the REM and SWS stages and number of CAs, all analyzed as binary variables: REM: 0% or > 0% of the TST; SWS: ≤9% or > 9% of TST (threshold previously used in the literature24,25); CAs: ≤43 or >43 CAs per hour (threshold corresponding to the median value in our ICU, in the absence of information from the literature).

The association of noise with REM, SWS and CAs was analyzed using mixed-effect logistic regression models for repeated measures (measurements over nights for each patient), adjusting for SAPS II, age and sex as potential confounders. A sensitivity analysis were also performed in which further adjustment for the administration of drugs influencing sleep were added. Results are presented as odds ratio (OR) and 95% confidence interval (95% CI).

All statistical analyses were performed using STATA (Stata Corp, Texas, USA).

Inclusion Criteria:

• older than 18
• recovered in ICU
• remained in the ICU for at least 24 hours.

Exclusion Criteria:

• patients with anamnestic severe dementia
• patients with severely decreased mental state (defined as Richmond Agitation-Sedation Scale, RASS, lower than -3)
• presence of language barrier
• patients coming from other hospitals
• patients needing continuous infusion of sedative drugs
• patients with frontal wounds, brain lesions in the frontal lobes, external ventricular drainage and noninvasive mechanical ventilation were excluded because of the impossibility to position the PSG device