Understanding Thermal Conditions, Thermal Comfort and Adaptive Behaviours in Naturally Ventilated Multi-patient Wards in Connaught Hospital, Freetown, Sierra Leone

How can the limited cooling capacity of natural ventilation in multi-patient wards in hot-humid settings with limited resources be extended by a better understanding of the links between thermal conditions, thermal comfort, and occupant adaptive behaviours? Natural ventilation remains the primary environmental mechanism for cooling and airborne infection control in hospitals among the poorest countries with the weakest public health systems across the equatorial zone. In these hospitals, a future rise in energy demand for cooling, made necessary by climate change and increased medical care expectations in inadequate buildings is expected to exacerbate existing infrastructural problems. Their occupants will be exposed more frequently and over prolonged periods to thermally uncomfortable indoor environments. Recommendations about the thermal conditions in naturally ventilated inpatient hospital facilities in hot-humid settings have been overlooked by all existing international standards while overheating criteria for naturally ventilated hospital spaces exist only for those located across the temperate zones. The thermal environmental performance assessment in naturally ventilated inpatient facilities in hot-humid settings becomes more challenging by the lack of any previous overheating assessment of naturally ventilated inpatients facilities in both new and historical buildings where on-site environmental monitoring was applied. Although knowledge about occupant adaptive behaviours, especially among those who are most vulnerable to thermal discomfort, is fundamental for the efficient mitigation of indoor overheating, there is a lack of empirical evidence regarding thermal performance and adaptability in inpatient facilities with hot and humid conditions. In naturally ventilated hospital wards, where the physiological and behavioural capacity for thermal adaptability is determined by each occupant’s type role in a strictly regulated environment while the dispersion of the indoor thermal conditions remains unstable, experienced thermal heat stress might be disproportionate to human thermal vulnerability. In this project, rather than understanding thermal comfort perceptions, practices, and expectations in hospital wards as passive stimuli to transient thermal conditions, the aim is to investigate the dynamic links between the ambient environment and occupants’ thermal comfort perceptions and adaptive behaviours while considering the impact of relative humidity, indoor airflows, personal factors, and spatial, temporal, and seasonal conditions. A mixed-methods longitudinal field survey was conducted over nine weeks during the rainy (September 2016) and dry seasons (March-April 2017) in eight naturally ventilated wards at the main tertiary government-run hospital with equatorial-monsoonal climate at one of the epicentres of the 2014-16 Ebola outbreak with the following main research objectives: a) to identify the spatial attributes linked with hospital design in the tropics before the 1940s; b) to define the associations between thermal conditions and spatial attributes, operational schedules and occupant-controlled window opening behaviours during contemporary hospital operation; c) to determine the ranges of neutral, comfortable and preferred temperatures, relative humidity values and airflows and the thermal adaptive capacity among nurses, patients and visitors; d) to assess the impact of seasonal, temporal, spatial and environmental conditions and personal factors on thermal comfort and adaptive behaviours. The case-study hospital is in Africa's west coastal zone, at a historical site in a central urban location and consists of a complex of buildings built consecutively between the 1920s and the 2000s, including a pavilion-plan building, which is composed of eight Nightingale wards. Infection control practises were integrated with scientifically standardised protocols and nursing routines following one-week piloting and co-designing processes with doctors and nurses. Context-specific infrastructural challenges and safety concerns hindered the installation of a network of sensors and the monitoring of the existing ceiling fans, which was intermittent due to regular electricity power cuts. A multidisciplinary dataset collected according to the ASHRAE 55: 2013 was composed from environmental and behavioural data. Twenty-one semi-structured interviews with twelve doctors and nine head nurses, 750 Thermal Comfort Interviews (T.C.Is.) (45,000 data), indoor and outdoor environmental monitoring (7,933 hours), window-opening behaviours (1,914 photos) and movement mapping (17 hours) comprised the collected dataset. In total, twenty participants were excluded from the analysis of the T.C.Is. due to their exposure to high airflows coming from personal fans. The final sample consisted of 50.68% (370) nurses, 25.62% (187) patients and 23.70% (173) visitors, who were interviewed across four surgical (43.70%), two medical (14.50%) and two mixed (42.60%) wards. The history and the general model of the hospital complex, which was digitally reconstructed, was informed by archival evidence from Freetown State Library, the National Archives in London, and the British Library, and by a thorough building survey. Empirical and experimental findings were produced through descriptive and non-parametric inferential statistics, predictive correlation (Spearman coefficients, Kendall’s W test coefficients and Cramer’s V effect size), predictive regression (simple linear and ordinal logistic and probit regression), time-series regression, content analysis and thermodynamic modelling. In 1864, British colonial officers building on Florence Nightingale's extensive work published a best practice framework for the design of barrack hospitals in British India. Although thermal comfort was not their primary focus, by conceptualising the ward as an instrument for efficient infection control through natural ventilation, they created a system of spatial components to maximise climate sensitivity, airflow rates and nurses' control over the thermal conditions in the ward. Despite the fact that the case-study Nightingale wards embodied climate-responsive characteristics influenced by these ideas; during contemporary operation, the drivers of their environmental performance were similar to those in the rest of the case-study wards. In all selected wards the windows lacked adequate shading devices and double-glazed windows, while internal window curtains trapped solar radiation and, by convection, induced higher adjacent air temperatures. At the same time, heat gains by conduction through the heavyweight external and internal walls and by convection through the uninsulated ceilings and floors reduced the potential of nocturnal cooling, contributing to higher night-time overheating. In naturally ventilated multi-patient wards in hot-humid settings with limited resources, the impact of the spatial attributes on ventilative cooling is likely to be different between diverse building typologies and seasons, with cooler indoor temperatures being associated with higher openable window coverage during night-time in the pavilion plan typology during the rainy season (Spearman coefficient=-0.34, p-value<0.001) and in other contemporary typologies during the dry season (Spearman coefficient=-0.32, p-value<0.001), while deeper plan layouts could have a protective impact against indoor overheating especially during night-time over the dry season (Spearman coefficient=-0.63, p-value<0.001) only in the modern building typologies. Despite the statistically insignificant correlation with outdoor temperature and relative humidity levels, occupant-controlled widow operation in the case-study wards displayed weak correlations with rising indoor temperature during the rainy season (0.13<Spearman coefficient<0.19, p-value<0.01) and falling indoor relative humidity values (Spearman coefficient=-0.14, p-value<0.01). The reported high levels of awareness for adaptive behaviours among all occupant types were not reflected in their realised adaptive actions. Comparisons between reported and observed individual adaptive behaviours showed that nurses drank water, visitors moved to cooler places, and patients asked for help. Nurses’ responses revealed that actions for the restoration of patients’ thermal comfort were an integral part of nursing care and in line with doctors’ advice and patients’ needs. Reported thermal discomfort was driven by perceptions of high levels of indoor relative humidity during the rainy season, elevated indoor temperature during the dry season and low levels of indoor airflows during both seasons. Through probit regression of the preference votes, the range of acceptable indoor thermal conditions was defined by operative temperatures varying between 29.00 and 30.00oC during the rainy season and between 28.00 and 29.00 oC during the dry season, relative humidity levels from 66.00 to 69.00% during the rainy season and around 71.00% during the dry season while acceptable airflows stand at 0.9m/s during both seasons. Outdoor environmental conditions were weak predictors of experienced indoor thermal discomfort. Patients expressed the highest levels of sensitivity to thermal discomfort to rising operative temperatures. At the same time, their votes displayed the strongest influence (Cramer’s V effect size: 0.50-0.69, p-value<0.001) by diverse building and ward typologies, gender, water and food consumptions and operation of building controls. Furthermore, allocation of patients in buildings with architectural and engineering characteristics like the case-study Nightingale typology can have a strong alleviating impact of thermal comfort during the rainy season (Cramer’s V effect size: 0.50-0.66, p-value) while acceptability of the indoor airflows among patients might increase if they can control window operation (Cramer’s V effect size: 0.54, p-value