Functional recovery of a resilient hospital type

Four adaptation options for ‘Nightingale’-type hospital ward buildings devised with practising clinicians are presented and evaluated. The adaptations recover functionality in an archaic ward configuration by delivering care to current UK National Health Service (NHS) models whilst preserving resilience to summer overheating. The investigation builds on recent work that demonstrates the significant resilience to heatwaves enjoyed by such traditionally constructed communal dormitories, the dominant UK hospital type between the late 1850s and 1939. Nightingale wards are potentially well-ventilated naturally, with good dilution of airborne pathogens. Although condemned as outdated by health ministers in recent years, many remain in use. As financial retrenchment suggests economical, creative refurbishment of hospitals will be required rather than new-build and replacement, the authors argue for health estates’ strategies that place value on resilience in a changing climate. Proposed adaptation options are investigated to assess resulting internal airflows and patient exposure to airborne pathogens. Options are costed and payback periods calculated to the standard public sector methodology. The proposed adaptations save time and cost over new-build equivalents. Selection of the most appropriate option is dependent on the characteristics of the patient cohort and care required.


Introduction
In 2001 the UK government demanded that the National Health Service (NHS) abandon the traditional healthcare model of a communal hospital ward and adopt the principle of single patient rooms, hitherto reserved for the very unwell and the privately insured. Although presented as a patient-facing 'consumerist' policy, part of a comprehensive NHS Modernization programme, more complex performance drivers preoccupying the UK Department of Health (DH) at the time determined the policy shift: increased space standards to facilitate inclusiveness in the implementation of the consumerist agenda, in effect destabilizing existing ward geometries (Department of Health, 2008a); poor infection control statistics at a significant number of acute hospitals damaging public confidence and adding significantly to healthcare cost (Plowman et al., 2001); patient pressure to achieve universal single-sex accommodation across the NHS Estate (NHS, 2013a); achieving patient privacy and 'Dignity on the ward' (Department of Health, 2008a); and the policy for wholesale replacement of the retained NHS Estate through publicprivate partnerships (PPPs) (Pollock, Shaoul, & Vickers, 2002). What came to be known as the Darzi Report, published in 2008(Department of Health, 2008b, further stressed the empowerment of patients and the principle of dignity and respect towards patients, although it was curiously mute on policy directions for the physical environment for healthcare. Nonetheless, the 2008 Health Building Note 04-01 (Department of Health, 2008a) specifically recommended that 50% of beds should be in single rooms, reminding readers that this merely repeated the advice given in the original Health Building Note 4 published in 1997. (Department of Health, 2013) The 2001 Labour Party manifesto, 'Renewing Public Services: NHS Reform', promised a re-elected Labour government would 'create a new type of hospitalspecially built surgical units, managed by the NHS or the private sector . . . ' (Labour Party, 2001). Pre-1949 inpatient wards built to Florence Nightingale's original mid-19th-century specifications as open-plan dormitories for 24 -30 patients (Nightingale, 1859) were specifically condemned. The DH put forward £120 million for their replacement, stating that: Older hospitals may care for older people on Nightingale wards -wards where staff can find it difficult to provide an appropriate environment for older people. £120 million will be spent over three years, making many of these wards into wards particularly for the use of older people. This will bring in more fourbedded bays, with more privacy and peace; rooms available for private conversations; and single rooms for those who are most vulnerable. the potential disadvantages of this form of accommodation (single bed wards) include: reduced social interaction with other patients leading to isolation, less surveillance by staff, an increased likelihood of rescue failure, and higher rates of slips, trips and falls. (Maben, 2009) Maben referred to Mooney (2008) and Snow (2008) for concerns about impacts on nursing staff, who argued that single rooms require greater nursing input. She also cited Roger Ulrich's highly influential arguments for single-room hospital wards delivered as key findings of his group's researches into 'evidence-based design' (EBD) in healthcare of which he was the originator. Ulrich et al., reporting on a collective literature review carried out in 2004, advised: The findings further support the importance of improving outcomes for a range of design characteristics or interventions, including single-bed rooms rather than multi-bed rooms, effective ventilation systems, a good acoustic environment, nature distractions and daylight, appropriate lighting, better ergonomic design, acuity-adaptable rooms, and improved floor layouts and work settings. (Ulrich et al., 2008) Sadler et al., citing Ulrich, were explicit: 'create 100% single-patient rooms'. In their summary of the cost of their recommended improvements they concluded: 'Single-patient rooms are now the standard for new hospital construction and, therefore, no additional costs are assumed' (Sadler, DuBose, Malone, & Zimring, 2008). Ulrich was invited to Britain to advise government; the single-room policy remains fundamentally unaltered (Department of Health, 2010). However, despite the very positive reception of this work across the NHS (NHS Scotland, 2011), the deteriorating financial situation implies that an assumed 'nil cost' is not applicable in the UK.
The intense resourcing challenge faced by all NHS Trusts militates against new buildings. Furthermore, Trusts also face the probability of an increasing frequency of extreme climate events, not least summer heatwaves, whilst being obligated to deliver energy and carbon reductions to ambitious NHS Carbon Reduction Strategy targets (NHS Sustainable Development Unit, 2010). Overheating in hospitals due to heatwaves is documented as impacting on the health and well-being of patients, staff and visitors (Carmichael et al., 2012). National Statistics report a clear correlation between mortality in hospitals and heatwaves in England and Wales (Kovats, Johnson, & Griffiths, 2006, p. 8), concluding: After accounting for the usual pattern of mortality by place of death, a larger than expected proportion of the excess deaths in the elderly occurred in hospitals and nursing homes; and In the non-elderly population, there was a large excess of mortality observed in nursing and residential homes, although the absolute numbers of deaths were small.
The report attributes more than 2,000 deaths to the August 2003 heatwave in England and Wales.
Members of the nursing profession have consistently expressed reservations about this policy. For example, nursing staff reported in interviews with the authors (at Bradford Royal Infirmary (BRI) in 2011) and in published articles their belief that the communal Nightingale ward arrangement is actually beneficial in certain care settings, for older patients, for example, a rapidly increasing in-patient constituency. The Nursing Times reported in 2011: We should not dismiss the value of Nightingale wards. [ . . . ] I would like the profession to take the positive aspects of the Nightingale wards and see them tailored to today's needs. (David, 2011) The article gained many positive comments from nurses, the majority anonymised, who deduced that the single-room agenda was driven by non-clinicians, policy-makers and the public specifically.

Characteristics of the 'Nightingale' ward
The configuration of a Nightingale ward belongs to the genus of 'pavilion plan' hospitals. The 'pavilion plan' separates hospital wards into discrete, cross-ventilated buildings, connected by a circulation route across one end only. Although the nurse Florence Nightingale is the figure popularly associated with the introduction of this idea into British hospital design at the end of the 1850s, the so-called 'Nightingale' ward resulted from advocacy on the part not only of Nightingale but also of the Scottish surgeon John Roberton and the Editor of The Builder, George Godwin (King, 1966). In a paper of 1856 given in Manchester, Roberton criticized the poor ventilation of many British hospitals, suggesting that a better approach was found in the 'pavilion plans' of many continental European examples. He was especially positive about the hospital at Bordeaux, France. Roberton's paper, which was subsequently published with illustrations, informed a critical account by Godwin in The Builder of the recently completed Netley Military Hospital. Godwin continued to deploy and promote Roberton's ideas, with several further articles appearing in The Builder on the subject of hospital ventilation; some were written by Roberton himself (King, 1966;Godwin, 1858) (Figure 1).
During this period, Nightingale was preparing answers for the Royal Commission on Barracks and Hospitals, which began taking evidence in May 1857. Her ideas were published in her book Notes on Hospitals (Nightingale, 1859), which also included three papers from The Builder likely to have been written by Roberton (King, 1966). Nightingale's interest in this subject stemmed from her experience of very high mortality rates in the two military hospitals at Scutari during the Crimean War of 1853 -56. She deduced that patient recovery was linked to the opportunities to vent out 'bad air'. 'Good' ventilation, i.e. cross-ventilation in her model, would suppress the incidence of 'cross-infection', not understood as bacterial but derived from 'miasmas' (Thompson & Goldin, 1975). These 'miasmas' developed in part from the exhaled products of the human body, especially when sick, and were 'always highly morbid and dangerous' (Nightingale 1859). William Farr (1807 -83) and fellow sanitary reformers promoted belief in miasma, attributing to it the 1849 cholera epidemic (Halliday, 2001). Nightingale insisted that 'natural ventilation, or that by open windows and open fireplaces, is the only means for procuring the life-spring of the sickfresh air', adding, 'no artificial ventilation will do this'. This comment may refer to recent attempts to ventilate hospitals mechanically, notably at the fully sealed York County Hospital of 1849, a bellowsdriven forced ventilation experiment by Queen Victoria's 'Physician Extraordinary' in which air was driven into the wards and from which it was removed by an aspirator (Burdett, 1893). The failure of the system led to the installation of opening windows in 1859. Nightingale, like Roberton, argued that each hospital ward should be located in a free-standing pavilion, connected to the rest of the hospital by a circulation route crossing one end only of the block. Her ideal was a single-storey pavilion containing just one ward, though she admitted that two-storey wards were acceptable. The pavilions were to be separated by a distance equal to twice the height of the buildings. Beds for 20 -32 patients were to be located along each side of the ward perpendicular to the walls, with one window per bed and the windows located opposite each other. A ward for 20 patients would be 80 ft long, 25 ft wide and 16 ft high (Nightingale, 1859), the preferred axis being north -south, a window for every two beds, the windows consuming at least onethird of the wall surface area, located 2-3 ft off the floor and within 1 ft of the ceiling (Figure 2).
Nightingale suggested that the windows might be made of plate or double glass, but heating was seen as a trivial exercise relative to the challenge of ventilation. She derived the required supply of fresh air by calculating the 'miasmatic' emanations from a typical sick soldier: 370 ft 3 per day from 16 soldiers and 123 ft 3 a night, generating 16 pints of water, the fatal scenario being: 'the consequent re-introduction of excrementious matter into the blood through the function of respiration . . . ' (Nightingale, 1859, p. 11). In fact, assuming the relative humidity to be 50%, to replace this quantity of vitiated air this yields a very low minimum air supply rate, some 0.167 litres/second/ person, approximately 1/60th of the contemporary standard.
Her ideas were replicated nationally. The first completed 'pavilion plan' hospital was the Herbert Military Hospital in Woolwich (begun in 1863), though Blackburn Infirmary had been planned on these lines in early 1858; its completion was delayed and Woolwich was finished first (King, 1966). In fact Nightingale's version of the continental pavilion hospital plan dominated hospital design in the UK into the 1930s (Thompson & Goldin, 1975) and also beyond, sometimes coupled to advanced ventilation strategies (Fair, 2014), though there was a parallel interest in circular ward planning during the 1880s (Taylor, 1988)

Infection control and ward design
Whilst Nightingale did not fully understand the mechanisms of infection, she advocated good ventilation and adequate bed-spacing to reduce disease risk, a principle that still applies today. Although theories of 'miasma' had been disproved by the 1880s (Ayliffe & English, 2003), ventilation strategies intended to dilute and disperse 'miasma' are now known to be effective also in venting out airborne microorganisms emanating from patients, visitors and staff. Indeed, early advocates of germ theory effectively appropriated the language of the proponents of miasmas in advocating good ventilation and this move may have aided their success (Tomes, 1998).
Evaluating the relationships between design and infection risk requires consideration of transmission routes. The transmission of infection through direct contact and poor hand hygiene is predominantly a behaviour rather than a design issue, though it is likely that good practice is promoted by physical segregation of patients and proximity of hand basins as well as maintaining manageable bed occupancy rates and demands on healthcare staff (Kibbler, Quick, & O'Neill, 1998;Beggs et al., 2006). Indeed, poor management and overcrowding in large open wards has been associated with high rates of healthcare-associated infections (Commission for Healthcare Audit and Inspection, 2007).
Understanding of airborne transmission of infection stemmed from the pioneering work of Wells (1935). True airborne transmission is cited as when pathogen-carrying particles, typically , 5 mm in diameter, are released through actions such as coughing and sneezing, travel with the air in a space and then are inhaled by susceptible occupants. Tuberculosis, measles and chicken pox are all well-known airborne infections, and there is evidence that influenza (Tellier, 2009;Milton, Fabian, Cowling, Grantham, & McDevitt, 2013) and severe acute respiratory syndrome (SARS) (Yu et al., 2004;Qian, Li, Nielsen, & Huang, 2009) may be transmitted in this way too. Droplet transmission also involves airborne dispersion, but is often regarded as a form of indirect contact transmission as the mechanism involves deposition of particles onto surfaces leading to environmental contamination. Common infections such as influenza and rhinovirus are thought to be predominantly dropletborne, however there is evidence that many hospital pathogens such as C. difficile (Roberts et al., 2008), methicillin-resistant Staphylococcus aureus (MRSA) (Kumari et al., 1998) and norovirus (Marks et al., 2003) may all be spread in this way. Regardless of the exact transmission mechanism, both airborne and droplet-borne transmission involves the release of particles into the air, and risk therefore depends on the fate of the particles. Quantitative evidence directly relating ventilation to airborne infection risk originated in operating theatre studies (Lidwell et al., 1982) and early work on tuberculosis transmission (Riley et al., 1957). Together with a raft of experimental-and modelling-based studies this has led to consensus that infection does indeed transmit via airborne routes and that ventilation is an appropriate control measure (Li et al., 2007). This body of research has led to the guidance on healthcare ventilation used today (Department of Health, 2007;WHO, 2009;ASHRAE, 2003).

Resilience of the Nightingale ward building type
The basic resilience of the Nightingale ward pavilions at BRI was established as part of the 'DeDeRHECC' ('Design and Delivery of Robust Hospital Environments in a Changing Climate') project (Lomas, Giridharan, Short, & Fair, 2012). Temperature data were collected in two Nightingale Wards at the hospital between 2009 and 2011. Figure 3 depicts the current ward building. Aluminium thermal break windows with a central top-hung light limited to 100 mm maximum opening providing less than 0.09 m 2 were installed in the late 1990s to replace the original quadruple-banked steel hopper windows, alternately topand bottom-hung, offering approximately 50% free area, some 1.35 m 2 . The data collected indicate the temperatures in all spaces monitored fell comfortably within the recommendation for hospital wards recommended by HTM 03-01 of 18 -288C , although peak external temperature was an undemanding 24.18C in this period.  report that although night-time temperatures regularly exceeded 248C, potentially affecting sleep, these occurred largely during the heating season, suggesting that a reduced set-point may reduce the incidence of higher night temperatures. A temperature of 268C was exceeded for only 3 h in Ward 8 and for 1 h in Ward 9, with an absolute maximum of 27.48C, despite the much-reduced opening window area. There was no evidence of overheating due to higher summer ambient temperatures or solar gain. The DeDeRHECC team adopted the adaptive thermal comfort standard BS EN 15251 for freerunning naturally ventilated buildings as a more reliable indicator of comfort (British Standards Institution) than the current DH guidance. For health buildings it offers bands of tolerance related to vulnerability, Category 1 being most vulnerable.  summarize the researchers' position. Fewer than 2% of the recorded temperatures exceed the Category I upper threshold, but the data yielded insufficient evidence of resilience to high ambient temperatures.
A dynamic thermal model of the ward was developed using Integrated Environmental Solutions modelling software (IES, 2011) and was calibrated against observed data to investigate the wards' thermal performance and energy demands  By comparison with other standard NHS building types, a recent paper studying the demand of a 1960s' tower building with hybrid ventilation predicted an energy demand for space conditioning alone of 101 GJ/100 m 3 (Short, Lomas, Renganathan, & Fair, 2012). Meanwhile CO 2 emissions for the BRI Nightingale wards are predicted to be about 30 kgCO 2 /m 2 for environmental control purposes, which, using the crude adjustment noted above, would uplift to about 53 kgCO 2 /m 2 , very significantly less than the Chartered Institution of Building Services Engineers (CIBSE) TM46 benchmark for 'Hospitals; clinical and research' of 129.3 kgCO 2 /m 2 (CIBSE, 2008). Here, then, is a hospital ward type that could deliver the NHS carbon reduction target, but which must be realigned with contemporary service models in order to become an operational solution to the NHS conundrum.

Adaptive options for the Nightingale building envelope
Though good, the performance of the Nightingales can be improved.  described three simple and incremental refurbishment options, summarized in Figure 4.
The first option adds 100 mm of insulation to the walls and 300 mm to the roof, opens up the triple light windows (ensuring safety with an external steel grillage), and provides a sunshade at each opening. Trickle vents are recovered behind a new perimeter heating element, for winter ventilation. The second option adds to this strategy ceiling fans operated by patients; whilst the third option introduces 100 mm diameter high-level air inlets above each bed space, between each window, with a damper and a simple convective heating device fixed to the internal face to enable supply air to be pre-heated and/or recirculated within the space. Primary heating and cooling are delivered through the installation of radiant panels. The addition of radiant cooling eliminates entirely the risk of overheating.
The present paper assumes the second option as the base treatment of the envelope but without operating fans except during summer heatwaves. Prediction of the dispersal of pathogens considers wind effects but not the action of multiple ceiling fans. Annual energy demands and CO 2 emissions of the refurbished Nightingale ward were predicted using the dynamic thermal model and the Bradford 2010 weather file for the summer period, 1 May to 30 September, as recorded in Table 1.
The performance of the Nightingale wards in a future climate has been predicted for current and future typical and extreme temperature years, the 2005 test reference year (TRY), containing monthly data typifying Bradford chaining the most typical January to the most typical February, etc., and the 2004 design summer year (DSY) depicting the third hottest year in the 22-year string based on the mean temperature recorded between April and September, the 90th percentile. Future weather years were created from the UKCP09 future climate projections by the University of Exeter assuming an A1B global emissions development scenario producing TRYs and DSYs for the 30year periods centred around 2030, 2050 and 2080 for the 5 km grid square covering Bradford. The method used has been fully described by Eames, Kershaw, and Coley (2011), and is summarized in . Higher temperatures increase gradually in the TRYs but quite rapidly in the DSYs; the difference between the temperatures in typical and extreme years becomes more pronounced so that risk-based decisions on the incorporation of mechanical cooling become more complex. The dynamic thermal model was used to predict temperatures in the Nightingale ward as currently exists and in the refurbished ward as option 2, as appropriate.
The internal heat gains, window-opening strategy and control strategies (e.g. for the cooling option) were maintained as for the 2010 analyses described above. Neither the existing nor the refurbished building will overheat in typical years, as judged by the HTM 03 and BS EN 15251 criteria, but in the 2050s warmer night-time temperatures may be experienced (1) remove stone, insulate (70 mm) and replace stone; (2) High level 100 mm air inlet ducts through solid wall; (3) radiant panel for hot and cold water; (4) opening lights in the existing windows with guards as needed externally; (5) introduce slow wide-span fans above the beds, an option not incorporated into the computational £uid dynamics (CFD) analysis; (6) shading and lightshelves of perforated white powder-coated aluminium to suppress glare and achieve a more even daylight distribution; and (7) seal vents and remove radiators/convectors as the changing climate requires some measure of radiant cooling about 2050 Source: freeze-frame image, same as Figure 3. Highlighted results show that the exceedance is deemed important in that it could not be easily corrected by re¢ning the control strategy. It is assumed that during the winter (October^April) the space will not overheat due to elevated ambient temperatures and solar gains, so the exceedance limit for BS EN 15251 is . 438 h above upper category threshold.
Functional recovery of a resilient hospital type (although these might be ameliorated easily with a refined window-opening regimen if the windows are openable to a useful degree in sufficient numbers).
However, although HTM 03 shows overheating occurs in the existing building and with refurbishment option 1 in the DSYs as early as the 2030s, the BS EN 15251 approach indicates that the refurbishment options that do not incorporate cooling will remain comfortable in both typical and extreme years right up to the 2080s. The existing building is predicted to overheat based on the Category I thresholds during the 2050s. The addition of mechanical cooling (i.e. the radiant ceiling innovation in option 3) would ensure there is no overheating in either typical or extreme temperature years right up to the 2080s.
The analysis summarized here indicates that the inherent resilience of the Nightingale wards, together with the northerly location of Bradford (and thus modest summertime temperatures even in the 2080s), would enable passive retrofit to succeed in producing a building that is comfortable until towards the end of this century. However, it would be essential also to offer a reconfiguration of the open wards to align the spaces more closely with the NHS Modernization policy objectives. The rest of the paper presents and tests options for such reconfiguration.

Infection control analysis of Nightingale wards
Studies assessing airflows and infection risk in naturally ventilated wards are few, not least because of the challenges involved in measuring airflows. Escombe et al. (2007) showed that in some cases ventilation rates in naturally ventilated spaces could be much higher than those achievable by mechanical ventilation, while Qian et al. (2010) conducted a detailed study in a naturally ventilated ward in Hong Kong and showed the dependency of the ventilation rate on the external wind conditions. Of relevance to the current paper is a recent study conducted in a Nightingale ward at St Luke's Hospital, also in Bradford (Gilkeson, Camargo-Valero, Pickin, & Noakes, 2013). This is an older ward than those considered in , but the construction, orientation and ventilation strategies are similar. Using CO 2 as a tracer gas, this study showed that with external wind speeds of 1.0 -4.0 m/s ventilation rates of 3.4 -6.5 air changes per hour (ACH) were achieved in the ward with only 60% of the windows open. Moreover, the study explored the influence of partitions between beds on the distribution of a tracer released at a representative patient location on the leeward or windward side of the building. Results considered the total exposure to the tracer over 25 min at different patient locations. In an open ward this was seen to be relatively uniform, suggesting a reasonable degree of mixing.
With partitions in place the results showed a redistribution of the tracer, with higher exposure close to the source and in the bed immediately opposite, but lower exposure in neighbouring beds. However, the overall exposure in the ward was comparable in both cases. If exposure can be considered analogous to infection risk, then these results suggest that the cross-ventilated Nightingale ward, with the right wind conditions, is capable of achieving ventilation rates comparable with the six ACHs recommended by the Department of Health (2007) and that partitioning the ward may be feasible without substantial impact on infection risk.
Adaptive recon¢guration options These basic options were initially outlined in schematic form and then developed in consultation with staff at BRI, examining patient safety, supervision, the cycle of inspection, the dispensing of medicines and treatments, the distribution and collection of food, and the night-time care model with reduced staff. It emerged that a key factor in considering the future of the Nightingale wards as functioning clinical spaces was the perceived risk of airborne cross-infection both in the open ward and in variants offering some subdivision. The relative safety of the propositions was evaluated using a computational approach to predict dwell times for air in the region of patients and their neighbours giving some insight into infection control implications.

Air£ow simulation approach
While ventilation and infection risk are clearly related, assessing the ventilation performance of a hospital ward, particularly a naturally ventilated one, is not straightforward. Tracer gas experiments offer a possible approach to assessment, however they cannot be carried out in occupied spaces and the methodology is not straightforward (Escombe et al., 2007;Gilkeson, Camargo-Valero, Pickin, & Noakes, 2013). Moreover it is not generally feasible to use such an approach to explore design changes. Computational fluid dynamics (CFD) is a numerical simulation approach that can be used quickly to explore building airflows and ventilation effectiveness under a range of different circumstances. By discretizing mass, momentum and energy equations that govern fluid flow across a mesh of elements representing the room geometry, it is possible to evaluate spatial distribution of parameters such as velocity, pressure and temperature within a building. While computational resources often limit simulations to idealized steady-state scenarios, the approach is widely used in building airflow assessment, including hospital-based studies (Tang et al., 2011). Here a series of CFD simulations were carried out to investigate the impact of the various ward internal redesigns on both ventilation characteristics and thermal comfort.

Model geometry and mesh
For consistency, each ward configuration is based on the same plan area of 10 × 8 m and the height remains constant at 3.7 m, leading to a ward air volume of approximately 300 m 3 . Figure 6 illustrates the simplified ward layouts considered in the study, derived from those proposed in Figure 5.
The traditional Nightingale ward model, Figure 6(a), is a section of the full ward consisting of six beds (three per side), each measuring 2.0 × 1.0 × 0.7 m. Figure 6(b) shows the second configuration with fullheight partitions between beds; previous studies have underlined the potential for these as an infectioncontrol measure (Noakes, Sleigh, Escombe, & Beggs, 2006;Gilkeson et al., 2013). In the Pullman-style layout, Figure 6(c), the beds are arranged into three pairs and the doors to each compartment are assumed to be open allowing for cross-ventilation via the corridor. Figure 6(d) shows the zigzag scheme comprising an island of skewed beds together with 2.1 m-high partitions segregating the immediate vicinity of each bed.
Although not shown in Figure 6(d), a slight variation of the skewed island design is also considered with both the partitions and beds raised above the ground by 0.15 m. The final configuration, Figure 6(e), is similar in layout to the traditional ward, subdivided to achieve gender separation but with the beds positioned on the end walls (as opposed to the side walls) with the addition of an external access corridor. A notable feature of this layout is the asymmetric distribution of windows; only two are present on the windward side of the ward (the central one is missing to cater for a toilet/shower room), whereas a total of six are present on the leeward side, i.e. three on the corridor and three supplementary windows positioned above it.
Computer-aided design (CAD) models were developed using Ansys Workbench, version 13.0.0 SP2 (Ansys  Figure 7 shows this grid structure for the 'Pullman' ward layout. The global cell count ranged from 0.8 to 1.5 million cells depending on the ward layout.

Boundary conditions
Cross-ventilation is modelled by treating the open windows as rectangular inlets, each measuring 0.15 × 1.00 m. On the windward side an inlet velocity, U IN , was imposed with a magnitude of 0.5 or 1.5 m/s. For the traditional ward these inlet velocities correspond to ventilation rates of two and six ACHs, respectively. Air was assumed to enter at an angle of 458 to the horizontal axis. This methodology has been shown to replicate experimentally determined flow patterns through casement windows in a Nightingale hospital ward (Gilkeson et al., 2011). All ward configurations were assumed to contain similar small radiators. With the focus of this study being on the relative performance of ventilation and thermal comfort between wards, small geometrical details (e.g. lighting and equipment) are neglected. For the same reason patient geometry is omitted, however typical thermal output (per patient) is accounted for using an appropriate temperature applied to the upper surface of each bed. Thermal boundary conditions for summer and winter cases are given in Table 2.

Solution process
Steady-state airflow simulations were run in parallel on a high performance computing cluster with 16 processors (2 × dual-quad core 2.26 GHz Nahalem processors and 24 Gb of SDD RAM) using Fluent, version 13.0 (Ansys Inc., 2013, http://ansys.com/ products/fluid-dynamics). Solutions for the governing fluid flow equations (continuity, momentum and turbulence) were computed using second-order discretization and the standard k-e turbulence model in conjunction with standard wall functions and the SIMPLE algorithm. Simulations ran for 10 000 iterations during which the absolute residual levels had dropped between four and six orders of magnitude ensuring good convergence. Although the simulations were steady-state, it is possible to exploit the resulting velocity field and determine the residence time in each region of the air volume being studied. In order to achieve this, a passive scalar equation was solved in conjunction with domain-wide source terms which enabled the distribution of the local ventilation rate, V L , to be found.

Air£ow simulation results
Altogether eight different ward-ventilation configurations are considered and they are summarized as follows: a: Traditional -windward ventilation b: Partitioned -windward ventilation c1: Pullman style -windward ventilation c2: Pullman style -leeward ventilation d1: Skewed island -windward ventilation d2: Raised skewed island -windward ventilation e1: External corridor -windward ventilation e2: External corridor -leeward ventilation The rationale for simulating leeward ventilation exclusively for the Pullman (c) and external corridor wards (e) is that these are the only asymmetric ward configurations, therefore it is important to consider both potential wind directions. In the analysis of ward airflows three ventilation parameters are considered, namely: (1) V G , the global ventilation rate through each ward, (2) V B , the ventilation rate local to each bed, calculated in a spherical volumes, 1.0 m diameter in the breathing zone of a laying patient, and (3) V L , the local ventilation rate measured in each individual cell of the grid. Figure 8 shows the observed airflow patterns in the form of path lines released from the inlet windows which are coloured by relative ventilation rate. These reveal the complexity of indoor airflow even for relatively simple cross-ventilation. The traditional (a), partitioned (b) and skewed island (d1, d2) flow fields are all very similar. They are dominated by the inlet flows which are guided cross-ward by the ceiling. A small proportion of this fresh air directly bypasses the ward and exits through the outlet windows, whereas the remainder mixes in the patient zone before exiting. The asymmetry present in the Pullman (c1, c2) and external corridor ward (e1, e2) geometries leads to differing flow fields depending on the flow direction. For both of these wards, leeward ventilation leads to substantial mixing regions in each respective corridor with significantly reduced air movement in the vicinity of the hospital beds. Therefore, the asymmetry and greater geometric complexity present in these particular designs do not lend themselves to stable and uniform ventilation characteristics. In contrast, the simpler designs appear to be very effective solutions and they are more suited to natural crossventilation.

Airflow patterns
Bed-level ventilation rates Figure 9 shows the distribution of bed-specific ventilation rates for all ward configurations under the influence of low wind speeds (i.e. 0.5 and 1.5 m/s, respectively).
For proposals (a) to (d) the two imposed wind speeds lead to global ventilation rates of two and six ACHs, respectively; the inclusion of partitions does slightly reduce the air volume in some cases but the overall effect on V G is minimal. The addition of the corridor to the extended ward (cases (e1) and (e2)) increases the overall volume from 300 to 346 m 3 , which reduces V G accordingly (constant wind speeds are assumed in all cases). Considering the results for the traditional ward (Figure 9(a)), beds A, B and C experience uniform ventilation rates, which correspond to the ventilation supplied by the adjacent inlet windows. On the opposite side of the ward the central bed (E) experiences an increase in ventilation for V G ¼ 2 ACH, which is notably higher than that experienced by beds D and F at either side. This is attributable to the airflow at either end of the ward being slowed down by a boundary layer attaching itself to the end walls. In contrast, the higher wind speed (V G ¼ 6 ACH) leads to proportionately higher ventilation rates for beds D and F, which suggest that the distribution of ventilation rates at this side of the ward is sensitive to the wind speed. In the partitioned ward case ( Figure  9(b)) there is a distinction present in the observed ventilation rates at either side of the ward but with little variation across the inlet and outlet walls. The presence of the partitions serves to stagnate the airflow as it progresses cross-ward, leading to noticeable reductions in V B on the leeward side.
For the Pullman-style hospital ward (Figure 9(c)) there is a clear variation in ventilation rate throughout the Here, the maximum local ventilation rate observed for the high wind speed is almost nine ACHs, which is 50% higher than the supply rate. This occurs above bed D which is directly opposite the central inlet window and adjacent to the door opening. It follows that as the airflow passes this central bay, it rushes through the door opening thereby increasing the ventilation rate local to bed D. In contrast, the leeward ventilation regimes show that when the air enters the opposing side of the building, the route of entry to each bay is poorly defined and this leads to reduced air movement and thus lower ventilation (Figure 9(d)).
Results from both skewed island cases (Figure 9(e, f)) show that the raised partitions have very little impact on the observed ventilation rates. The differences in V B from bed to bed are minimal which underlines the uniformity of the airflow in the centre of the ward. This characteristic can be expected because the beds are clustered centrally which is away from the regions experiencing high flow gradients, i.e. near inlets, outlets and perimeter walls. Overall, the ventilation rates are slightly lower than those observed in the traditional ward, however the uniformity of flow suggests improved stability in the ventilation conditions (assuming straight cross-ward ventilation).
The observed ventilation rates seen in the external corridor ward are fairly constant under the influence of windward ventilation (Figure 9(g)). This is explained by the fact that the beds adorn both end walls of the ward, each of which is supplied with fresh air from the same window. A point of note is that the absolute ventilation rates are significantly lower for this case because, as described above, the central window on the windward side of the building is omitted due to a toilet/shower room being present. There are six windows present on the opposing side of the building, which explains the dramatically higher ventilation rates seen for the leeward case (Figure 9(h)). The peak ventilation rate is approximately 14 ACHs occurring by bed C, which is just behind the doorway connecting the corridor to the ward. As with the Pullman layout, leeward ventilation for the external corridor case leads to a significant variation in ventilation from bed to bed, however in this case the variability stems from the interaction between the three high-mounted inlets and the inflow through the corridor opening.

Local ventilation rates
Whilst the bed-specific ventilation rates give a clear indication of the air exchange rates, which can be expected by patients, analysing the distribution of local ventilation rates throughout each ward is a convenient way of comparing them. This was achieved using a horizontal plane spanning each ward with 7700 analysis points situated at a height of 1.2 m above the ground.
Figures 10 and 11 present histograms of the distribution of V L in this plane for low and high wind speeds, respectively. For the low wind speed cases, the most striking feature is the range of V L observed, which is particularly broad for the Pullman ward (Figure 10(c) and (d)) and the extended ward with leeward ventilation (Figure 10(h)). These wider ranges are indicative of poorer mixing which leads to a greater spectrum of data. In contrast, the markedly narrower spectrum observed for the traditional ward ( Figure 10(a), skewed-island designs (Figure 10(e) and (f)) and the external corridor ward with windward ventilation (Figure 10(g)) indicate greater mixing levels with a more even distribution. It is also noticeable that the peak of the distribution is below two ACHs, the supply ventilation rate for the ward, in the skewed island design, compared with slightly above in the traditional open layout. The same overall trends are seen in the histograms for the high wind speed of 1.5 m/s in Figure 11, which shows that the distribution of ventilation rates is insensitive to wind speed.
Although the histograms provide a wealth of absolute ventilation rate data, Figure 12 shows contour plots of the relative ventilation rate, V L /V G , which is a measure of how well the inlet flows reach various parts the wards. For V L /V G ¼ 1.0 the ventilation rate at any given point is equal to the global value, V G ; when V L /V G , 1.0 the local ventilation rate is below V G ; and for V L /V G . 1.0 the ventilation is greater than V G . As would be expected, the contour plots clearly show a high relative ventilation rate near the inlets and in most cases this extends along the roof where the inlet flows are entrained. The range of relative ventilation rates is relatively small for the traditional, partitioned and skewed island designs (Figures 12(a), (b), (c) and (d)), whereas great variability is present in the remaining cases. In particular, the Pullman ward (leeward ventilation) exhibits very high air exchange rates in the corridor adjacent to the inlets and the bay walls restrict airflow, thereby slowing the flow rates through the actual bays.

Thermal characteristics
In analysing the temperatures in the ward two parameters were of interest, namely: (1) the mean average of the local bed-level temperatures (i.e. the average of all six beds), T B-AVE (8C) and (2) the difference between the minimum and maximum bed-level temperatures, DT (8C). Tables 3 and 4 show the data for summer and winter conditions respectively.
Overall, the difference in average temperatures between wards during summer is relatively small, whereas these differences are more pronounced during winter. The greater variation in winter occurs by virtue of the greater temperature ranges present; incoming air has a temperature of 128C and the radiator temperature is 458C, yet the summer conditions assume the inlet air matches the wall temperatures (228C) and the radiators are turned off.
As the ventilation rate increases, the temperatures generally drop as the faster airflow removes more heat. For summer conditions the warmest ward is the Pullman layout (windward ventilation) and the greatest variation in bed-to-bed temperatures is also highest for this case but with a leeward ventilation regime; the latter observation reinforces the view that this ward induces the greatest variability in ventilation flows (and thus the temperature distribution). For winter conditions, the traditional ward experiences the in reality variation in surface temperatures and proximity of patients to radiators and external walls will influence thermal comfort as well as the air temperatures.
Functional evaluation of the adaptation options Figure 13 records the original plan of the BRI Nightingale wards. The adaptation options were interrogated by clinical staff at BRI responsible for the care of elderly patients within Ward 3, a Nightingale ward in its original configuration but broadly divided by gender into three sections. Those consulted were the divisional general manager, nursing sister, head of nursing, a consultant in elderly care and various staff nurses. They all reported positive operational benefits in caring for the elderly from the open arrangement, echoing Maben: for the exceptionally frail and fearful, the other patients are a resource, intervisibility between patients is a positive benefit. Staff were keenly aware of coroners' reports into the ward deaths criticizing the use of side rooms, verdicts recording that patients were not viewed sufficiently often and were generally isolated. Divisional General Manager Neary believed: '"security" was equated directly with "dignity"', using 'security' in the broadest sense to include infection risk, and this was a misconception. 2 Patrolling of wards had been phased out, but was likely to be reintroduced. Relatives needed to be able to locate staff quickly. Ease of circulation within the ward was critical, nurses' stations were more effective as mobile trolleys, particularly for the patient admission process; 'crash' trolleys for use in treating cardiac arrest needed to be 'on hand', not parked in corridors; relatives should not be parked in corridors; trolleys bringing food and drink, dispensing pharmaceuticals, bearing X-ray equipment, ultrasound, all required to pass each other. Medical students progressed in groups of six at BRI. The central aisle of the Nightingale enables staff to assess patients' walking capability in judging their potential for home discharge.
Discussion of the adaptation options recorded the following observations by the staff and allowed the options to be developed further in light of their likely operational use: . Subdivision of the ward to create single-bedded bays This approach moves the ward closer to the DH ideal (and ironically is the pattern of the mediaeval hospital, e.g. the Heiligen Geist in Lubeck). Cubicles are formed in full-height partitioning, the doors likely to be maintained open. Twelve beds can be accommodated with a nurses' station rather than the 20 currently accommodated. Dimensionally the subdivision of open Nightingale wards into single rooms with dedicated bathrooms is inherently inefficient, not least because of the minimum bed spacing dimension and therefore the window spacing dictated by Nightingale. The drawing depicts the arrangement within a Nightingale pavilion at Bradford somewhat narrower than the norm so that circulation is restricted but in a full-width ward the additional 2-3 ft would enable greater functionality. Access to bathrooms  remains limited, providing external bathroom pods to each stack of rooms would remove almost all glazing, however four are shown at the each corner of the ward (Figure 14). The BRI clinical staff opposed this configuration for elderly care but were more positive about its potential for short-stay, day-case elective surgery. Nonetheless many of the positive benefits associated with the open-ward arrangement are negated.
. Subdivision of the ward to create multi-bedded bays on one side of the ward and a corridor on the other This approach is reminiscent of a traditional first class 'Pullman' railway carriage forming six compartments of two beds. Each two-bed ward is provided with a bathroom pod, a significant step towards the NHS modernization aspiration. The additional width of a standard ward would ease the space available for the ward rooms ( Figure 15). BRI clinical staff were enthusiastic. Rooms could be organized by gender but on a flexible basis, beds could be offset to break immediate intervisibility. .

Centralization of the beds
This option places a wardrobe-height spine bulkhead carrying water supply and other services required at the bedhead, along the centre of the wards, with 14 beds facing the windows. The Cabinet Office showed some enthusiasm for this configuration, which the authors do not claim to have invented. 3 It is vaguely reminiscent of an airline business class cabin arrangement. Patients no longer face each other but enjoy a view protected from direct solar gain and glare, retain some contact with patients to either side, and can receive visitors in relative privacy. Each patient has an individual washbasin, wardrobe and drawer unit. Tracked curtains could provide complete visual if not acoustic privacy for treatment. Discussion with hospital staff suggested that earlier versions with a straight central spine could develop a 'zigzag' geometry in plan, with beds offset for greater privacy, enabling beds to be pushed in and out of the ward to operating theatres, imaging suites and treatment rooms ( Figure 16). BRI staff were unconvinced that this arrangement offered benefits for elderly care because there is no direct intervisibility between beds. 'Wayfinding' may be difficult for the confused, delirious and alcoholic who would have difficulty identifying their allocated bed. However, the arrangement provides an interesting compromise for care of the less vulnerable, preserving the beneficial airflows observed within the original type. .

Externalized circulation
An external corridor is added to the eastern perimeter of each floor, steel-framed with timber stud construction clad in lightweight materials, insulated to contemporary UK Building Regulation standards, accessed through doorways formed in window openings cut down to floor level. The design attempts to recover opening window area on the corridor side but the asymmetry affects the dispersion of pathogens as reported above. The removal of circulation from the main body of the building releases spaces to take three to five beds, each ward served by a new bathroom tower (Figures 17 -19). Here the additional width gained within the ward rooms is highly beneficial, the full width available afforded adequate space between beds for It emerges, therefore, that different ward arrangements are appropriate for different care purposes, that more open arrangements may have higher functionality, if only in patient safety terms, and this safety is less likely to be imperilled by enhanced risk of airborne cross infection than hitherto (1) windowsill dropped to the £oor and the opening widened to form a door into the ward; (2) upper section of the window opening retained to vent/light directly into the ward; (3) £oor to new external circulation, 200 mm insulation; (4) steel hangers carry additional corridors and suspend steel frames of added circulation; (5) steel-framed roof construction, 300 mm insulation; (6) elevation of the added circulation element; and (7) horizontal centre pivot windows retain the former opening window areas in the Nightingale ward behind  Costs Table 5 records elemental costings of the four fundamental refurbishment options per floor. These costs were calculated by DeDeRHECC project partners Davis Langdon AECOM. Services renewals account for 40 -50% of overall construction cost, weighting the Skewed island option in particular.
Table 6 reveals however that the Skewed island and Pullman options are the less cost-intensive in costs per m 2 and 60 -70% of the equivalent new-build cost. The benefit of reduced contract time is only indicated indirectly by the allowances for main contractor preliminaries. Six months to one year to recover lost bed spaces may be saved with additional value-formoney (VfM) benefits. Table 7 records discounted lifecycle energy costs for the base case 'do nothing' option against option 1. Little lifecycle cost difference between the various refurbishment options emerged. The only major difference affecting energy performance between the 'do nothing' case and the four other options is the addition of insulation resulting in a reduction in heating energy.
The four options, then, by reconfiguring patient environments to offer more privacy and dignity, recover inherently resilient buildings. Savings in energy use are available but trivial in comparison with other operational costs. More significant are the potential savings in the avoidance of mechanical cooling installation later in the century. Current UK Treasury VfM models are unable to include such future savings, let alone intangibles such as business continuity or reduced mortality as a consequence of more effective observation. Note: Refurbishment costs include external insulation installed behind the existing outer stone skin, allowing for removal and reinstatement, at »150 000. Source: Derived from the Davis Langdon AECOM cost database. Notes: (1) Costs represent energy costs for one £oor only to allow a comparison to be made with the cost model. Simplistically these energy costs could be prorated by £oor area for multiple £oors.
(2) Energy costs only include energy demands included in the Loughborough model . It is believed that these energy demands only include space heating, some small power and lighting. They therefore do not include various other energy demands commonly found in hospitals, such as medical equipment.
(3) No maintenance or replacement costs are included since these are thought to be similar between the base case and the various architectural/building services options.
(4) Retail energy costs are taken from Department for Energy and Climate Change projections^IAG guidance.
(5) The reduction in costs is mainly a result of the addition of insulation, which should be seen in the light of the estimated »150 000 costs for insulation rather than the full costs of refurbishment which result in functional improvements (improved privacy, for example) but do not have a bene¢t that could be monetized easily.

Conclusions
This paper has amplified previous work on the resilience of Nightingale wards by analysing new internal configurations, devised in close consultation with clinicians. These arrangements all introduce higher levels of privacy and dignity (in line with current policy) into the resilient envelopes of Nightingale ward buildings without compromising that innate resilience. Insofar as the desire to achieve greater control of airborne pathogens drove recent disenchantment with Nightingale wards, the reconfiguration options are modelled to predict patterns of airflow and likely concentrations of pathogens in air with a higher dwell time. Model results indicate that good cross-ventilation is still possible with adjustment to the internal layout, provided care is taken to ensure there are appropriate ventilation openings to enable through flow of air. This is particularly important in the two cases with a corridor at the side of the patient rooms (options c and e) where the presence of an internal wall can act to block effective airflows. The construction detail of the adaptation schemes is developed and costed in detail. It is significantly lower than new-build alternatives and quicker to deliver. Net present value calculations using the UK government discount rate are undertaken to predict payback periods, the fundamental data required by HM Treasury in determining viable policy.
The authors argue for a review of policy as the deep financial retrenchment required of the NHS redirects emphasis towards refurbishment. Current economic circumstances place a particular premium on lighttouch refurbishment of NHS sites; similarly, a wholly reasonable concern to improve the patient experience also means that cosmetic changes can be favoured over more substantial interventions. In addition, Trusts are reluctant to lose capacity and have concerns about construction noise and dust transmission. The options presented here address these concerns whilst delivering fundamental improvements in privacy and dignity and demonstrating that, with sensible reconfiguration, Florence Nightingale's original approach might yet be viable for the 21st-century NHS. Indeed, this work implies that designers might productively re-evaluate techniques and strategies for good ventilation first deployed in the pre-modern era in order to answer the challenges of contemporary low-energy architecture.