The problem with a room that never gets dark
An intensive care unit is designed for continuous observation, and that design logic produces continuous light. Corridors stay lit, monitors glow, staff enter with task lighting, and the patient’s eye-level exposure may barely change between 14:00 and 02:00. The physiological concern is straightforward: the human circadian system depends on a robust contrast between light and dark, and the ICU removes much of that contrast.
What follows is often stated more confidently than the evidence allows. It is one thing to show that circadian disruption is common in critically ill patients and associated with delirium. It is another to show that changing the lighting changes delirium incidence. The retrieved literature supports the first claim reasonably well and the second claim only partially, with mixed and mostly non-confirmatory intervention results. This article separates those two questions, because the practical decisions in a ward depend on knowing which one is settled.
What the observational evidence shows: association, not causation
The strongest retrieved signal comes from a prospective cohort study by Li and colleagues (2023) in a surgical ICU. The unit was windowless and closed, with a standardized lighting and noise protocol: artificial light on during the day, off from 21:00 to 06:00, with the nursing station lit continuously and patient lights switched on for night interventions. Average noise was 62.50 dB(A) during the day and 58.85 dB(A) at night. Among 76 patients, 17 developed delirium, an overall rate of 22.37%. Melatonin and cortisol secretion showed biological rhythmicity in the non-delirium group but not in the delirium group. The authors concluded that disturbance of the melatonin and cortisol rhythm was associated with delirium.
This is a useful finding and a limited one. It is a single-center cohort of 76 patients, with delirium assessed by CAM-ICU twice daily, and the design cannot establish whether circadian disruption contributed to delirium, resulted from it, or shared a common cause such as illness severity. The authors themselves frame the relationship as an association. For clinical practice, that means circadian disruption is a plausible marker of risk and a plausible target, not a proven causal lever.
The window study: a real signal with real confounders
Lee and colleagues (2021) published a retrospective single-center study of 150 medical ICU patients admitted to single rooms, six with windows and six without. The cumulative incidence of delirium was 21.7% in the window group versus 43.3% in the windowless group (relative risk 1.996, 95% CI 1.220–3.265). Admission to a room with a window was independently associated with lower delirium risk (adjusted odds ratio 0.318, 95% CI 0.125–0.805). The authors concluded that exposure to natural light through windows was associated with a lower incidence of delirium.
The result is striking, and the limitations matter. Room allocation was not randomized; patients were typically assigned to the first available room. The two groups differed at baseline in antipsychotic use and alcohol abuse, and the analysis adjusted for several confounders but cannot adjust for everything that differs between a windowed and a windowless room, including staff behavior, view, and patient expectations. A retrospective association of this size is a reason to measure light properly in prospective studies, not a reason to promise families that a window prevents delirium.
The intervention pilots: mixed, and often not designed to answer the delirium question
The intervention literature is where overclaiming most often happens. The retrieved records show several relevant studies, and their designs differ enough that they should not be pooled into a single verdict.
Spies and colleagues (2024, Critical Care Medicine) conducted a proof-of-concept pilot whose primary purpose was to determine whether a multicomponent change in room design prevents ICU delirium, with a secondary assessment of lighting conditions and serum melatonin. A multicomponent design change is not a lighting trial; if delirium incidence changes, the effect cannot be attributed to light alone. The retrieved excerpt does not report the results, so no effect size can be stated here.
Lucchini and colleagues (2025, Intensive and Critical Care Nursing) compared new lighting technology with traditional fluorescent bulbs and examined sedation and delirium in the ICU. Again, the retrieved record identifies the question but not the outcome, so the direction and magnitude of any effect are not available in the evidence used for this article.
Linders and colleagues (2026, Intensive and Critical Care Nursing) published a systematic review of lighting in intensive care units, patient outcomes, and directions for future research. The existence of such a review indicates that the field has accumulated enough studies to synthesize, but the retrieved excerpt does not provide pooled estimates or quality assessments. A systematic review is only as strong as the trials it contains, and the trials described above are small, heterogeneous, and rarely blinded.
The most important statement in the retrieved literature may be the one from Wilcox and colleagues (2024, Thorax), a scoping review on ICU interventions to promote sleep and circadian biology in reducing incident delirium. It states plainly that, despite plausible pathophysiological mechanisms, research is needed to confirm the relationship between sleep, circadian rhythm, and delirium in ICU patients. That is the correct epistemic status: plausible mechanism, insufficient confirmation.
Why measurement is the missing layer
Most ICU lighting studies describe interventions in words: dynamic lighting, cycled lighting, brighter days, dimmer nights. Few report the quantities that determine circadian effects at the eye. Illuminance in lux is a start, but the circadian system responds primarily to short-wavelength light, and two rooms with the same lux can differ substantially in melanopic effectiveness depending on spectrum and viewing geometry.
CIE S 026 defines a system of metrology for ipRGC-influenced responses to light, including melanopic equivalent daylight illuminance (mEDI). The retrieved ICU studies do not report CIE S 026 melanopic values, and no retrieved source validates a melanopic target for delirium prevention. Any melanopic specification for an ICU is therefore a proposed measurement framework, not an evidence-based standard. That distinction should be stated explicitly in any protocol document, because it determines whether the protocol is generating data or claiming an established benefit.
The practical consequence is that an ICU lighting protocol should be written in measurable terms: clock times for light and dark periods, illuminance at the patient’s eye, spectral description where available, and a documented dark window. This does not guarantee a delirium benefit. It does make the intervention auditable and comparable across units, which is the precondition for the trials the scoping review calls for.
A measurable light-and-dark template for an ICU
The following is a design template for generating local data, not a validated delirium-prevention intervention. It is deliberately specified in numbers so that it can be measured, documented, and revised. The values are proposed operational targets, not research findings.
Daytime period (approximately 08:00–20:00)
- Target eye-level illuminance at the patient’s head position: measure at 10:00 and 14:00 with a calibrated meter, recording both vertical illuminance at the eye and, where a spectrometer is available, melanopic EDI.
- Open blinds where windows exist; position the bed so the patient’s eyes have a line of sight to the window where clinically possible.
- Avoid leaving the room in task-light-only conditions during daytime hours; document any period when the patient is in relative darkness for procedures.
Night period (approximately 20:00–08:00)
- Protect a dark window: target the lowest feasible eye-level illuminance, with task lighting directed away from the patient’s eyes during checks and procedures.
- Record the duration of any light exposure above the night target, including monitor brightness and door openings.
- Document noise alongside light, because the retrieved cohort used a combined lighting and noise protocol and the two are difficult to separate in practice.
Documentation
- Log eye-level illuminance at three fixed times (for example 10:00, 14:00, 02:00) for each patient day.
- Record the light and dark periods actually achieved, not only the intended schedule.
- Report delirium screening with CAM-ICU at fixed times, so that any local change can be compared against a defined outcome.
This template will not tell a unit whether lighting reduces delirium. It will tell a unit whether its lighting is what it claims to be, which is the question that must be answered before the outcome question can be tested.
What would change the conclusion
An adequately powered randomized trial with delirium as a prespecified outcome, measured light exposure including melanopic metrics, and blinding of outcome assessors where feasible would change the conclusion. So would a systematic review with pooled estimates and quality assessment showing consistent benefit across trials. Until then, the honest position is that ICU lighting is a plausible, measurable environmental lever whose delirium benefit is not established. The pilots are hypothesis-generating. They are not practice-changing on their own.
That position is not nihilism. It supports action that is low-risk and measurable: protect darkness at night, provide meaningful light exposure by day, measure what is actually delivered, and document outcomes. It also supports restraint in what clinicians tell families. A window is not a treatment. A lighting schedule is not a proven prevention. But a unit that measures its light environment is in a position to learn, and that is more than most ICUs currently do.
FAQ
Does bright light during the day prevent ICU delirium?
The retrieved evidence does not establish that. Observational studies show associations between circadian disruption and delirium, and a retrospective study found lower delirium incidence in windowed rooms, but intervention pilots are mixed and a 2024 scoping review states that research is needed to confirm the relationship. Bright light is a plausible intervention, not a proven one.
Should I ask for a window bed for a relative in intensive care?
It is reasonable to ask, and the window study provides a signal that window exposure is associated with lower delirium incidence. But room allocation is usually determined by clinical need and availability, and the study was retrospective and single-center. A window should not be presented as a guaranteed protective measure.
What is melanopic EDI, and why does it matter in an ICU?
Melanopic equivalent daylight illuminance is a metric defined in CIE S 026 for the effect of light on ipRGC-influenced responses, including circadian entrainment. It matters because lux alone does not capture the spectral quality of light reaching the eye. The retrieved ICU studies do not report melanopic values, so any melanopic target for an ICU is a proposed measurement framework rather than an established standard.
Is there a registered trial that could settle this?
The retrieved sources do not identify a specific registered trial with delirium as a prespecified outcome and measured melanopic exposure. The scoping review by Wilcox and colleagues (2024) calls for research to confirm the relationship, which implies that the definitive trial has not yet been reported in the retrieved literature.