Determinants of intensive care unit light exposure: an observational cohort study of patient, temporal, and structural factors.

Introduction: Patients with critical illness experience profound circadian disruption, driven in part by atypical light exposure in the intensive care unit (ICU). While prior studies characterize ICU light, the extent to which in patient, temporal, and structural factors determine light intensity remains unclear. Methods: We conducted an observational cohort study of light intensity in the rooms of patients admitted to one of thirteen medical ICU (MICU) rooms equipped with continuous light monitoring between February 2021 and 2022. Patients were included if they remained in the monitored MICU room for at least 48 h. We collected patient demographics and clinical features. We recorded light intensity (lux) every 60 s, and then calculated hourly mean lux, hourly proportion of measures ≥250 lux, and hourly proportion of measures ≤10 lux for the first 48 h of each patient's MICU admission. Intraclass correlation coefficients (ICC) quantified variance in each light outcome attributable to patient-level factors. Multivariable multilevel models assessed adjusted associations between patient, temporal and structural factors and light outcomes. Both ICCs and multilevel models were generated separately for day (05:00-22:59) and night (23:00-04:59). Results: Among the 337 patient-room pairs, the mean (± standard deviation) age of patients was 66.1 ± 14.8 years; mean SOFA score was 6.8 ± 3.6; and daytime and nighttime mean lux were 173.8 ± 94.6 and 63.7 ± 88.2, respectively. ICCs indicated that patient-level factors explained 26.3% of daytime and 46.9% of nighttime variance in mean lux; ICC results were similar for other light outcomes. In multivariate models we found that temporal (e.g., time of day, season of ICU admission) and structural factors (e.g., room direction) strongly influenced daytime light levels, whereas nighttime variation was more heavily driven by patient-level factors such as mechanical ventilation status. Conclusion: ICU light exposure shows substantial between-patient variability, particularly at night. In multilevel models, we found that this variability is driven by clinical, temporal and structural factors. Our findings suggest that structural design alone is insufficient to normalize ICU light exposure, and that interventions targeting nocturnal care behaviors and daytime light provision may be necessary to mitigate circadian disruption in patients with critical illness.