Patient Room Lighting Design: What the Research Says, and Six Approaches Compared

The comparison above is deliberately electric-only: it is the light you can guarantee. But most patient rooms have a window, so we re-ran all six designs with the worst window regulations allow: a 4 x 3.4 ft opening, 13.6 sq ft of glazing, right at the FGI minimum of 8 percent of the floor area for this room, with a 3 ft sill, clear double glazing, and the glass a full 10 ft from the patient's pillow, under the darkest big-city daytime sky in our set, Seattle on an overcast December 21, at its best hour (noon). On its own, that window delivers 71 mel-EDI to the flat patient, 163 at 45°, and 242 sitting up, rising as the bed rises, because the raising gaze rotates toward the glass. Every "Open this model in CircadianLab" link in this article now includes exactly this window and sky, so what you open is the realistic room; the electric-only figures in the cards are what remains if you delete the window in the tool.

mel-EDI at the patient's eye, lying flat and sitting up, for each design without and with the worst-case window. Green meets 250 mel-EDI (or UGR under 22).

Three observations. Lying flat, the window barely matters: it adds 66 to 129 mel-EDI, low in the supine patient's peripheral field, and it even improves the flat glare readings slightly by brightening the background the eye adapts to. The overhead options stay firmly intolerable (UGR 30 to 32), so no verdict changes. Sitting up, the window changes everything: the raised gaze points straight at the glass, the window alone delivers 242 of the 250 target, and with any of the six designs on top, the dose now holds in every bed position ( open the headwall design sitting up with this window). But the same view makes the window the room's dominant glare source: by itself it reads UGR 25.6 sitting up, past the comfort limit, the familiar experience of facing a bright gray sky from a hospital bed. The electric layers soften it by brightening the room around it, yet every design still lands between 21 and 23.3 sitting up, and only the headwall stays clearly under the UGR 22 line. A small window under a bright sky is a harsh little patch; glare control at the glass (a light shade or film) trades away exactly the dose it just delivered.

Why not just design around the window, then? Because the rescue is conditional and the failure is not: those sitting-up numbers are the best hour of that overcast day, and at 8 am, or behind a drawn shade or privacy curtain, the same window delivers almost nothing while the electric numbers hold. Take the window as margin, not as the plan: the electric layer still has to clear the target on its own, which is what the comparison above measures and why option 6 remains the recommendation.

There is no national count of how many US patient rooms have windows, but the requirement is close to universal. Medicare's Conditions of Participation require an outside window (or outside door) in every patient sleeping room, with a sill no higher than 36 inches in new construction,^[11] and the FGI Guidelines, the design code most states adopt for hospitals, require patient rooms to be "provided with natural light by means of a window to the outside," with glazing of at least 8 percent of the floor area.^[12] Modern med-surg rooms, in short, have windows.

The gaps are in critical care and older building stock, which codes grandfather in. In a 2024 survey of 30 US pediatric ICUs, one unit in five still had windowless rooms.^[13] At a major academic neuro-ICU, 42 percent of a decade of subarachnoid-hemorrhage patients were treated in windowless rooms,^[14] and at Massachusetts General Hospital, 37 percent of 3,527 ICU admissions (2020-2023) were to windowless rooms.^[15]

Why it matters is one of the oldest findings in evidence-based design: surgical patients with a window view of trees left nearly a day sooner and took fewer strong analgesics than matched patients facing a wall,^[16] depressed inpatients in sunny rooms were discharged 15 percent sooner (and MI patients in sunny rooms fared better),^[17] spinal-surgery patients on the bright side of the ward used 22 percent less analgesic medication per hour,^[18] and two classic ICU studies found roughly twice the delirium in windowless units.^[19] Not every study agrees (the neuro-ICU cohort found no outcome difference, and the MGH study's windowed rooms actually had more delirium in a retrospective design the authors caution is confounded^[14][15]), but the weight of the evidence, and every design guideline, favors the window.

Exterior daylight is enormous (a clear sunny day is on the order of 100,000 lux, overcast 5,000 to 20,000), which is why even a fraction of it can carry a room. But every layer between sky and eye takes a share, and the shares multiply. Clear double glazing passes about 78-82 percent; the solar-control low-E units on modern hospital curtain walls pass 51-70 percent (Vitro's widely used Solarban 60/70/90 line spans exactly that range^[21]); tinted (bronze/gray/green) substrates commonly land between 10 and 45 percent.^[21] Retrofit window films, added later for heat and glare, run 15-50 percent,^[22] and roller shades with a 1-5 percent openness fabric block 95-99 percent of what reaches them. A closed hospital privacy curtain is effectively opaque.

Our model's 62 percent double glazing is therefore a friendly assumption, and because through-window light scales almost linearly with transmittance, you can re-scale every number in this section in your head:

The pattern to notice: any tint below roughly 50 percent puts even the best-case sky at risk, and the moment a shade comes down for glare (see the orientation card for why it will), daylight's circadian contribution is essentially gone. Tint and shading are chosen for heat and comfort by facade engineers who are not calculating mel-EDI; if a circadian daylight strategy is intended, the glazing schedule has to be part of the lighting design.

Under an overcast sky, orientation barely matters: the light is diffuse, and all four facades see the same gray dome. Orientation is a clear-sky variable, and on clear days the four facades behave completely differently. Peak usable mel-EDI at the eye (excluding direct-sun-in-the-eyes moments), by facade:

South is the healthcare facade: the only orientation whose best hours come in winter (the low sun reaches deep into the room), holding 250 from roughly mid-morning to mid-afternoon on clear winter days in every sunny city. East and west each own half a day, mornings or afternoons, and fail the other half; their peaks are summer numbers. North gets the least direct sun, only low-angle beams on early summer mornings and evenings and none in the middle of the day, which makes it the gentlest light in the table but also the weakest: it clears 250 only around summer midday and never in winter, anywhere, not even in Phoenix (winter best: 175).

East and west also carry the failure mode that undoes real-world daylight strategies: a low sun square in the patient's face. At 8 am on a clear June morning, an east window puts 8,000 to 14,700 mel-EDI on the supine eye at UGR 57 to 62, far beyond intolerable, with the same show on west facades before sunset and on south facades in mid-winter at low latitudes ( open Seattle's clear December noon and see the beam land on the pillow). A patient who cannot move away from it, or a nurse who walks in on it, closes the blinds, and the room drops from thousands of mel-EDI to essentially zero for the rest of the day. Glare control and circadian dose come from the same piece of fabric, and glare wins.

Putting the climate data and the room model together, we can estimate how many days a year daylight alone fails the daytime dose, defined the way WELL v2 shapes it: 250 mel-EDI at the patient's eye for at least four hours, beginning by noon.^[2] For each city we simulated the room across the solar year under clear, intermediate, and overcast skies, then weighted each sky by the city's NOAA count of clear, partly cloudy, and cloudy days.^[20]

Three lines, three lessons. The regulation-minimum window facing north fails every single day of the year in every city, Phoenix included: 13.6 sq ft of diffuse sky simply cannot push 250 mel-EDI ten feet to the pillow for four hours, ever. The same generous 8 x 5 ft window rotated from south to north gives up 100 to 200 days a year, which makes facade orientation the cheapest daylight decision a hospital planner controls. And even the best case, a big south window, still fails about a month of days in Phoenix (27, mostly overcast winter mornings), two months in Denver (63), and a third of the year in Seattle (127). There is no city and no code-compliant window for which daylight alone is a dependable circadian dose.

Method notes: supine patient, eye 10 ft from the glass, clear double glazing, blinds open all day, no exterior obstructions; partly cloudy days modeled as the CIE intermediate sky; cloudiness assumed uniform across seasons (most cities are cloudier in winter, so the true failure counts skew slightly higher); south-window clear-sky days counted as compliant even when part of the dose arrives as direct sun that would in practice force the blinds closed.

The chart above is generous by construction: clear glazing, blinds open all day, a single bed close to the glass, nothing built across the street. Every real departure from that pushes the failure counts higher, not lower.

• Behind dark glazing: below roughly 50 percent transmittance the margins vanish, and a 15-35 percent solar film turns even Phoenix's best hours marginal.^[21][22]
• Whenever the shade is down, which the orientation card shows is a matter of when, not if, on east, west, and winter-south exposures. The chart counts those clear-sky days as compliant; a patient who closes the blinds against the beam turns them into failures.
• Deeper in the room: our patient sits a friendly 10 ft from the window. Daylight's useful reach is roughly 1.5 to 2.5 times the window head height,^[25] so a second bed, a deeper room, or the minimum window shrinks the dose fast. Simulation studies of hospital rooms reach the same conclusion, needing 30-40 percent of the whole facade glazed just to cover the room with a circadian-effective dose,^[23] and field measurements in offices find the 250 mel-EDI target reliably met only near unobstructed windows.^[24]

None of this is an argument against windows: the outcome evidence above is an argument for them, they deliver view, orientation to time of day, and free dose whenever conditions line up. It is an argument about what to size the electric layer for. Size it for the worst case, the November morning, the north-facing room, the drawn shade, and the second bed, so it clears 250 mel-EDI on its own with daylight arriving as a bonus on top. That is exactly the multi-source design in option 6, and why the daylight scene in CircadianLab adds window light on top of the electric solution rather than replacing it. Model your own facade, city, and glazing before trusting the window with the dose.