How the hospital room affects recovery

Situation

Hospitals are planned by efficiency parameters. Square metres and bed count drive the investment calculation; hygiene and fire safety decide the permit. The room as a factor on patients and staff does not show up in the planning model.

An evidence base from 2005 to 2025 documents this factor anyway. Walch and colleagues compared 89 patients after spine surgery on the bright and the dark sides of the same hospital ward in a 2005 prospective cohort study (Walch et al., 2005, Psychosomatic Medicine 67(1):156-163). The bright side received 46 per cent more sunlight. Patients on the bright side needed 22 per cent fewer analgesics per hour at p=0.047, their analgesic costs ran 21 per cent lower at p=0.047 and reported stress was lower. The design is a prospective cohort, weaker than a randomised trial, and it measures the economically relevant endpoint directly.

Al Khatib, Samara and Ndiaye published a systematic review of the field in 2024, run on a PRISMA protocol with a bibliometric component (Al Khatib et al., 2024, Frontiers in Built Environment). The search covered literature from 2010 to 2023 on biophilic design in hospitals. The reported associations with shorter length of stay are directionally consistent across the included studies. Postoperative pain demand and staff stress show the same directional consistency, in both cases with methodological heterogeneity in the underlying primary studies. Where the primary studies document effect magnitudes robustly, those magnitudes do not support every marketing claim of the biophilic design industry.

Tekin, Corcoran and Urbano Gutiérrez published a meta-synthesis of the qualitative data from the Maggie’s Centres in the UK in 2023 (Tekin et al., 2023, Frontiers of Architectural Research 12(1):188-207). Maggie’s are an institutional special case of cancer-companion architecture. Maggie Keswick Jencks died in 1995. The first centre opened in Edinburgh in 1996. About two dozen houses now exist, most of them in the UK and some abroad in Hong Kong, Tokyo, Barcelona and Groningen. They are house-like buildings with gardens, daylight and domestic rooms, places that cancer patients can visit between treatments.

Youn, Kang and Lee used functional near-infrared spectroscopy in 2025 to measure what happens in the brains of nurses who spend ten minutes in a fully greened room against a white-walled control room (Youn et al., 2025, IJERPH 22(10):1571). Across 21 nurses at a hospital in Cheon-An, South Korea, oxygenation in the dorsolateral prefrontal cortex dropped significantly in the greened room at p<0.001. The sample is small but the measured mechanism is plausible: the brain had less regulation to perform in the quieter setting.

The DRG logic runs across all of this. Across the German case-payment system, comparable European systems and the US Diagnosis-Related-Groups billing, a shorter length of stay translates directly into the cost per case. Staff turnover acts as its own cost factor in the current nursing shortage. Both magnitudes appear in the biophilic literature as outcomes, but no hospital has yet run them through its own DRG calculation.

Finding

The economically translatable effects fall first on postoperative pain demand, which feeds directly into the medication and complication costs inside the case-payment envelope. The length of stay sits next to that on the revenue side of the case payment, and in a market with a chronic nursing shortage the stress load on staff becomes its own cost factor through absence and vacancy.

Walch 2005 is the hard pain anchor in this line. The 22 per cent reduction in analgesics per hour and the 21 per cent reduction in analgesic costs both reach p=0.047 in a prospective cohort. Walch measured the effect of sun exposure, not greening or material choice, and the attribution to daylight is direct.

Tekin and colleagues delivered the qualitative validation at an established special case in 2023. Daylight stood at the top for patients in the Maggie’s meta-synthesis, with fresh air and greenery close behind. For staff, privacy and quiet led instead. Patients prioritised these room factors over the strictly medical service of the Centres. This is qualitative evidence, and it still lines up with the economically measurable daylight effect in Walch.

Youn 2025 adds a neurophysiological hint. The drop in dlPFC oxygenation points to reduced cognitive load. Linking Youn’s mechanism hint to Walch’s measured outcome is a conjecture without a demonstrated pathway. Youn measured nurses in a greened setting while Walch measured patients under daylight, so the populations and the interventions differ. What stays plausible is that the brain spends less regulatory capacity in a quieter setting and leaves more room for other processes. The effect is documented in Walch’s economic endpoints and in Youn’s neurophysiological measure. The common mechanism is an open research question.

The same architectural feature that measurably eased the workload on nurses in Youn’s experiment can also bring an infection risk inside a hospital. Surial and colleagues measured a vertical green wall in a Swiss hospital in 2021 and documented elevated concentrations of bacteria and fungi in the surrounding air, on the wall surface itself and in the irrigation water that fed the plants, including Aspergillus niger and Aspergillus terreus (Surial et al., 2021, Infection Control & Hospital Epidemiology 43(2):273-275). Their own words: “the elevated concentrations of microorganisms in the air, on the plant wall, and in the water of this vertical garden led the hospital’s infection prevention committee to forego any further indoor plant installations.” The committee dropped further indoor planting after the measurement.

The two findings do not contradict each other directly. They show that a single intervention in biophilic design, the green wall, can simultaneously ease cognitive load and bring an infectious microbiome that matters in a hospital setting. The institutional consequence is a differentiated approach by patient group and location. Sehulster and Chinn set this out for the Centers for Disease Control and Prevention in 2003 in the HICPAC guideline (Sehulster and Chinn, 2003, MMWR Recommendations and Reports 52(RR-10)). Fresh and dried flowers and potted plants are forbidden there in patient areas for immunosuppressed patients as a Category II recommendation. For immunocompetent patients they remain permitted, also as a Category II recommendation. The rule grades the restriction by immune status instead of banning plants outright.

The evidence on daylight, view and outdoor access is not touched by the Surial finding. It is a different class of intervention with a different risk profile.

The evidence has other limits. Bulaj, Forero and Huntsman published a 2025 perspective paper in Frontiers in Medicine formulating biophilic design as a modulatory hypothesis on the psychopharmacological response (Bulaj et al., 2025, Frontiers in Medicine). The paper is explicitly not a systematic review and not an RCT. It provides no primary data of its own. It focuses on therapeutic home environments outside the hospital setting. In this form the modulation claim cannot be turned into an institutional calculation and remains an open research question.

The small sample in Youn 2025 is also a limit. 21 nurses are too few to generalise from. And the Al Khatib review, systematic in protocol, still draws on primary studies whose quality varies across the included papers. Correlation and causation are not consistently separable in this material.

My own view is that the evidence base already supports an institutional consequence. Anyone who plans or finances hospitals can take the documented effects as a quantifiable factor in the investment calculation, with the uncertainty marked. An operator can put the same factor on staff load and its downstream cost. Today neither group does this.

Research context

The academic translation of this material into health-economics models is still pending. Al Khatib 2024 delivers the field overview, Tekin 2023 the qualitative validation at the Maggie’s Centres. Youn 2025 adds a first neurophysiological mechanism through the measured dlPFC activity. Walch 2005 and Ulrich 1984 stand alongside them as older but hard anchors.

Roger Ulrich showed in Science in 1984 that postoperative patients in hospital rooms with a view of trees had shorter stays and required less potent analgesics than patients facing a brick wall (Ulrich, 1984, Science 224(4647):420-421). The finding has rarely been directly replicated. It sits at the start of the line in which Walch and the newer work stand.

Several gaps remain open. No published comparison of DRG case costs exists between biophilically designed and conventionally equipped hospitals. The existing economic estimates derive extrapolated savings from length-of-stay reductions without performing the calculation directly. The DRG comparison can be run with existing methods. No one has run it.

No randomised trial exists for the Bulaj modulation thesis in the hospital setting either. The nearest cohort evidence comes from Walch on analgesic demand under daylight and from Ulrich on analgesic demand under a nature view. Both support the modulation conjecture at the cohort level without replacing an RCT.

A third gap concerns the capital market. No major healthcare REIT names biophilic design as a stand-alone investment criterion across its holdings. Welltower has built the welltowerLIVING brand, but the brand targets wellness housing in the senior segment, not the hospital setting. Institutional investors have not adopted the biophilic line as a portfolio criterion.

These findings sit inside an older line. Beatley, Jones and Rainey edited a 2018 overview of the healing-environments debate, “Healthy Environments, Healing Spaces” (University of Virginia Press). Shan Jiang reviewed the Chinese literature on therapeutic landscapes and healing gardens against the Western research in 2014 (Jiang, 2014, Frontiers of Architectural Research 3(2):141-153). The newer empirical work is methodologically denser than the older reviews, but it rests on the same untested assumption that room design shifts recovery.

Implications

For healthcare REITs and private hospital operators. No major REIT makes biophilic design a portfolio criterion for its hospital line today. Welltower’s welltowerLIVING brand shows that operators already treat wellness as a selling point in the senior segment. It has not yet crossed into the hospital portfolio. Whoever spends the next two years collecting pilot data from their own holdings and benchmarking it against comparison groups has, by 2028 or 2029, an investment argument that competitors do not have. Technically the move is straightforward. What it requires is the willingness to run an outcome comparison that current REIT reporting does not yet expect.

The comparison a REIT would need already has a design template. Walch and colleagues did not run a renovation experiment. In 2005 they compared patients already assigned to the bright and dark sides of the same ward (Walch et al., Psychosomatic Medicine), which means a portfolio holding older buildings with uneven daylight distribution can run the same comparison on existing stock before committing capital to a new one. Shepley and colleagues ran a before-after comparison in 2012 that mixed the daylight change with other renovations at the same time (Shepley et al., HERD), and Youn and colleagues surveyed only 21 nurses in 2025, too few to generalise from (IJERPH). A REIT-run pilot answers both objections cheaply: hold the building the same and vary only where the daylight already differs, and run it across enough beds that the sample outgrows Youn’s.

For public hospital trusts. The Walch figures reach 22 per cent on analgesic demand and 21 per cent on analgesic cost, and the Al Khatib corpus and the Maggie’s material point the same way. Many of the individual moves, like window placement and daylight ingress, sit inside the planning envelope already. Direct outdoor access on the recovery side is structurally harder, but it is among the most highly rated factors in the Maggie’s data. The argument belongs in the DRG negotiation. Today it surfaces mainly in the architectural competition. A trust that can demonstrate a measurable effect on length of stay or on analgesic demand brings a measured number into the case-payment argument. The Walch figures allow a first approach. They do not replace a house-specific audit.

Not every source here counts equally in a negotiation, and the trust should lead with the strongest one. Walch and colleagues reported 22 and 21 per cent figures from a prospective cohort with a p-value attached (Walch et al., 2005, Psychosomatic Medicine); Al Khatib and colleagues reviewed more studies in 2024 and confirmed the same direction, but flagged methodological heterogeneity in the underlying work (Frontiers in Built Environment); Tekin and colleagues collected qualitative data at Maggie’s Centres in 2023, useful for ranking which room features patients value most rather than for pricing them (Frontiers of Architectural Research). Lead the DRG argument with the Walch number. Use Al Khatib and Tekin to explain why the planning budget should prioritise daylight and outdoor access ahead of other room features.

For nursing organisations and personnel representation. Shepley and colleagues reported in 2012 from a quasi-experimental before-after comparison of two intensive care units in New Hampshire that mean staff absence fell from 38 to 23 hours per person in the new daylight-supplied unit (Shepley et al., 2012, HERD 5(2):46-60). The design is a quasi-experiment, the daylight effect is confounded with other changes between the old and the new ward and the significance is borderline at p=0.05. Even with that caveat, Youn 2025 adds a neurophysiological mechanism in the same direction. The findings belong in pay and structural negotiations on working conditions, including where the architectural competition does not have personnel in view. Absence is a cost line that the daylight unit moved.

The honest version of this claim is modest, and it is still usable in a negotiation. Shepley and colleagues changed daylight and other building features at once, so they cannot credit the 38-to-23-hour drop to daylight alone. A second, independent line corroborates it: Youn and colleagues measured oxygenation in 2025, a different method, a different country and a different outcome than Shepley’s, and still found reduced cognitive load in a greened setting (IJERPH). A representation body that cites the absence figure as confounded but corroborated keeps the credibility to bring it back when better data arrives.

For the pharmaceutical industry, in a conditional reading. The Bulaj modulation thesis is not proven. Walch on daylight and Ulrich on a nature view support it at the cohort level without proving it. If the thesis is confirmed, the heterogeneity of the spatial setting in phase-III studies would produce an effect-dilution risk. The consequence would be to systematically record spatial variables like daylight exposure and visual contact with nature in study protocols, with the acoustic setting as a separate methodological block. Folding that into the next study generations protects against a distortion that the current study design does not control for. This recommendation stands under the proviso that the modulation thesis is confirmed in a randomised trial. Until that point it is a study-design question that should become visible in the methods sections of the next phase-III publications.

The action recommended here has a lower evidentiary bar than the thesis it responds to. Recording daylight exposure and nature-view access in a study protocol costs little and requires no proof that the modulation thesis is true. Bulaj and colleagues are explicit that their own paper is a perspective. It carries no primary data and is neither a systematic review nor a trial (Bulaj et al., 2025, Frontiers in Medicine). A study sponsor can adopt the recording practice now, as a hedge against an unmeasured confound, while treating any claim that room design changes drug response as unresolved until a randomised trial tests it directly.