Climatic and psychosocial risks of heat illness incidents on construction site

Applied Ergonomics 53 (2016) 25e35

Contents lists available at ScienceDirect

Applied Ergonomics journal homepage: www.elsevier.com/locate/apergo

Climatic and psychosocial risks of heat illness incidents on construction site Yunyan Andrea Jia a, *, Steve Rowlinson b, c, Marina Ciccarelli d a

School of Built Environment, Faculty of Humanity, Curtin University, Perth, Australia Department of Real Estate and Construction, The University of Hong Kong, Pokfulam, Hong Kong c Key Laboratory of Eco-environment in Three Gorges Reservoir Region under Ministry of Education, Chongqing University, Chongqing, China d School of Occupational Therapy and Social Work, Faculty of Health Sciences, Curtin University, Perth, Australia b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 21 April 2015 Received in revised form 17 July 2015 Accepted 19 August 2015 Available online xxx

The study presented in this paper aims to identify prominent risks leading to heat illness in summer among construction workers that can be prioritised for developing effective interventions. Samples are 216 construction workers' cases at the individual level and 26 construction projects cases at the organisation level. A grounded theory is generated to define the climatic heat and psychosocial risks and the relationships between risks, timing and effectiveness of interventions. The theoretical framework is then used to guide content analysis of 36 individual onsite heat illness cases to identify prominent risks. The results suggest that heat stress risks on construction site are socially constructed and can be effectively managed through elimination at supply chain level, effective engineering control, proactive control of the risks through individual interventions and reactive control through mindful recognition and response to early symptoms. The role of management infrastructure as a base for effective interventions is discussed. © 2015 Elsevier Ltd and The Ergonomics Society. All rights reserved.

Keywords: Occupational heat stress Climatic heat risk Construction workers Effective interventions Socio-ergonomic model

1. Introduction The effect of global warming is leading to growing numbers of heat related diseases (Lin et al., 2009) and a general increase in mortality (Vaneckova et al., 2008). Continuous increase in ambient temperature amplifies the risk of occupational heat stress to outdoor workers, with construction workers being a vulnerable population (Xiang et al., 2014). This creates a challenge to the adaptability of existing international standards that provide benchmarks for heat stress management in the workplace (Parsons, 2013). Over decades, interventions on occupational heat stress have been developed using predominantly a “hot room” experimental method (Wyndham et al., 1967) e a setting that can hold control variables constant, yet isolated from the complexity of real life working environment. As a result, existing threshold-based intervention strategies (e.g. ACGIH, 2013; ISO 7243, 1989) are developed to manage the risks among homogeneous work crews in highly structured environments such as military training or steel mill works where decision making can be centralised and work-rest

* Corresponding author. School of Built Environment, Faculty of Humanity, Curtin University, GPO Box U1987, Perth, WA 6845, Australia. E-mail address: [email protected] (Y.A. Jia). http://dx.doi.org/10.1016/j.apergo.2015.08.008 0003-6870/© 2015 Elsevier Ltd and The Ergonomics Society. All rights reserved.

regimens can be routinely exercised. This assumed context is however a far cry from the nature of a dynamic workplace such as a construction site, where moving-around laborious tasks are performed under rapidly changing weather conditions. Unlike the manufacturing industry, the construction industry is organised in project-based organisations and operated through numerous instant decisions made by frontline personnel. These decisions are results of prioritisation among conflicting and short-term goals, in which progress pressure and financial incentives often prevail over safety concerns. In such a context, the experimentally defined environmental thresholds are found to be “security biased” (Budd, 2008) for being over-conservative and counterproductive, and thus never progress beyond the policy stage. The laboratory-validated interventions are implemented as competing tasks with the production tasks, leading to the elimination of one risk while introducing another, and therefore are rarely effective. To bridge this gap, a comprehensive, contextualised theoretical model is needed to explain and predict the effectiveness of interventions of heat stress management in a complex work context. Meanwhile, primary risks for heat illness incidents need to be identified to inform management decisions for prioritisation of resources and attentions. The objective of this study is to develop a theoretical model of heat stress management in construction project organisations and

26

Y.A. Jia et al. / Applied Ergonomics 53 (2016) 25e35

to identify primary heat risk factors contributing to heat illness incidents happened on construction sites. 2. Methodology A dilemma in the dominant positivist epistemology underpinning existing heat stress studies is that the subject must be taken out of the natural work context in order to test the effect of a set of predetermined, ‘objectively’ existing heat stress factors or interventions (e.g. Fujii et al., 2008). Yet once taken out of the context, both the subject and the factors are no longer the same. This, to a large extent, accounts for why scientifically validated interventions need to be customised in order to be useful in real world environments. We based our research on a social constructivist stance (Berger and Luckmann, 1966). Starting from the well-established rational model which defines six factors for heat stress, i.e., air temperature, humidity, radiant heat, air velocity, clothing effect and metabolic heat (Malchaire, 1995; Parsons, 2003), we explore how stakeholders in the construction industry make sense of these factors and henceforth how the heat risk factors as social constructs are linked to effectiveness of the interventions being practiced. On this stance, we treated participants of the study as both subjects and informants, and self-reported data with equal importance as the objectively measured environmental and physiological data during the triangulation analysis. We decided to use a naturalistic, non-intrusive and triangulation approach (Lincoln and Guba, 1985) for this study on three concerns. First, the study must be non-intrusive to the on-going work process so to preserve the natural working environment. Second, all stakeholders who have a role in heat stress interventions on site needed to be involved in data collection in order for triangulation to work. Third, while the variables suggested by the rational heat stress model are measured quantitatively and continuously, we did not assume we had known the risks and the interventions needed. Rather, we assumed that little was known and new factors were to be elicited, therefore a grounded theory approach was necessary (Strauss and Corbin, 1990). 2.1. Instrumentation Continuous data including worker heart rates and environmental heat stress (temperature, humidity, solar radiant heat, and Wet Bulb Globe Temperature (WBGT)) were recorded at 1-min intervals. Data on risks and interventions in the specific sociomanagerial environment of each construction site were collected with three instruments. They included a data collection sheet for the construction workers sample, a questionnaire and interview guide for the managers sample and a site observation checklist. The three instruments share two sets of core questions, i.e., perceptions of risks and effectiveness of interventions. The items in the questionnaire for measuring perception of heat risks and perceived effectiveness of interventions were designed based on a review of the literature [key references include (HSE, 2002; Leithead and Lind, 1964; Parsons, 1995, 2006), existing guidelines on heat stress management (Abu Dhabi EHS Center, 2012a, 2012b; ACGIH, 2009; AIOH, 2003; CIC, 2008; CSAO, 2000; OSHA, 1999) and findings from a pilot study by the research team in 2010]. The items were designed using a five-point Likert scale. Equipment used for collecting environmental and physiological data are presented in Table 1. A section of critical incident report was included in the worker data collection sheet to obtain workers' personal heat illness experiences. This included an exploratory question of whether the respondent had personal experience of heat illness on site, and if so, a brief description of the incident scenario, time, the work

environment, perceived symptoms, treatment, perceived causes of the incident, and any other background information the respondent regarded as relevant to the incident. The critical incident technique (CIT) is a qualitative data collection method that guides participant to focus on personally experienced event and describes factual details around the event. It was first developed in the Aviation Psychology Program of the US Air Forces (Flanagan, 1954) and has been widely used in various research fields since (Butterfield et al., 2005; Tuuli and Rowlinson, 2010). Compared with the commonly used qualitative methods in which respondents are asked to provide general opinions about certain topics, CIT gives the respondents a focus and directs their attention to factual observations; therefore it elicits more valid data. 2.2. Sampling Sampling of data collection was stratified by three criteria. First, by the types of projects, i.e., building work, civil engineering work, and RMAA (repair, maintenance, minor alteration and addition) work, which were having different organisational forms, workforce cultures and site environment characteristics. A second criterion for stratification of sampling was indoor and outdoor work; and the third criterion was to include major trades vulnerable to heat stress, including bar bender, bar fixer, carpenter, concreter, bricklayer, plasterer, welder, scaffolder, HAVC (heat, air ventilation and cooling) fitter, MEP (mechanical electrical plumbing) worker, demolition worker, tunnelling worker, and an open list to be suggested by site managers. 2.3. Data collection protocol As any ethnographic study that needs to be informed of insiders' view of the target group, this study needed to identify ‘gatekeepers’ as initial contacts to lead the researchers to get access to other informants in the field (Lincoln and Guba, 1985). However, unlike a cultural group as a typical ‘field’ in an ethnographic study, target groups in this study were construction project organisations structured in hierarchies. For both safety and commercial reasons, access to construction sites had to be endorsed top-down by senior management. Therefore the research team first contacted senior management of clients or contractors, through the second author's network, to identify available sites that met the criteria of the research brief. Site managers or safety managers were then assigned to lead the researchers to the selected construction sites. However, these ‘gatekeepers’ who provided access to the site workers, were not part of the workers' community per se. Approaching the workers through their supervisors has a potential risk of introducing defensive attribution bias into the interview data (Hofmann and Stetzer, 1998), and therefore the researchers needed to build direct rapport and reciprocal trust with the workers to eliminate the ‘top-down’ effect such that workers can provide valid information (Bailey, 2007). Argyris (1952) identifies four factors that lead to bias in data collected by formal interviews: (i) interviews are new psychological situations that are associated with tension and uncertainty; (ii) a formal interview setting represents an authoritarian relationship that triggers a defensive mechanism within the respondent; (iii) the formal interview represents intrusion by an outsider into the respondents' personal relationship with their leaders; and (iv) disparities in language and mannerisms between the researchers and the workers and practitioners can trigger defence mechanism among the respondents. Trust building between the researcher and the respondents is thus of vital importance for validity of field research and is more than keeping confidentiality of their personal information. Trust is to be developed through prolonged

Y.A. Jia et al. / Applied Ergonomics 53 (2016) 25e35

27

Table 1 Instruments for collecting environmental and physiological data.

1 2 3 4 5 6 7 8 9 9 10

Measured variable

Equipment/Instrument

Manufacturer

Further reference

Dry bulb temperature Wet bulb temperature Globe temperature Relative humidity WBGT indoor WBGT outdoor Air velocity Tympanic temperature Resting heart rate Blood pressure Blood glucose

QuesTemp™ 48N

3M, USA

(Bernard and Barrow, 2013; Rowlinson and Jia, 2014)

11 Heart rate 12 Skin temperature (chest) 13 Perceived exertion 12 Fitness (measured by YMCA 3-min step test) 13 14 15 16 17 18

Body height Body weight Sleep quality Smoking Alcohol drinking Current medication

Beaufort Wind Force Scale e estimated wind speed ranges from zero to 3 m/s. Braun Thermoscan IRT 4020 Kaz Europe SA, Switzerland Omron HEMe7211 Accu-Chek Active

(NOAA, 2008) (Dzarr et al., 2009; Rowlinson and Jia, 2014)

Omron Healthcare, Kyoto, Japan Roche Diagnostics Ltd, Mannheim, Germany Polar Electro Oy, Finland

Polar RS800CX N NA1/APAC1 heart rate monitor HOBO TidbiT v2 Temperature Data Logger, HOBO Data Loggers by Onset, USA Also used in collecting temperature inside UTBI-001 safety helmet, see Table 9. Rating of perceived exertion scale (RPE) (Borg, 2005) Polar RS800CX used as timer (YMCA of the USA, 2000) A 12-inch high bench Self-made Korg MA1 Solo metronome KORG INC., Japan Measured by a metre stick Means of four measurements before and after work A 3-point Likert scale: how was your sleep quality last night (1 e bad; 2 e OK; 3 egood) Self report (1-No; 2 e A little; 3-Yes) Self report (1-No; 2 e A little; 3-Yes) Self-report (0- none; 1 e hypertension; 2 e heart disease; 3 e chest disease; 4 e other disease)

engagement (Lincoln and Guba, 1985) and through reciprocity and reactivity (Creswell, 1998). On the other hand, Dean and Whyte (1958) observe that respondent's desire to please the researcher during the interview can also introduce distorted information into the data. Therefore the data collection protocol for this study was carefully devised to minimise the aforementioned risks to bias. This includes careful identification of the researchers' boundaries, role and extent of influence in the organisation throughout the process of field study. With a socio-political awareness, researchers need to address the interest of different stakeholders, establish rapport and nurture the relationship. To the worker participants, researchers have a responsibility to be trustworthy, demonstrating respect and goodwill. To the management team, researchers need to engage them in a joint mission of improving the site condition and thus, their job performance. That is, researchers should be diligently creating winewin situations whenever a new group joined the scene. One more constraint specific to climatic heat stress research is that the available period for data collection is restricted to days of hot weather within one season of a year e in Hong Kong it is 3e4 months excluding days of heavy rain or typhoon. Bearing in mind the aforementioned issues, a two-day protocol was developed to

guide the data collection (see Table 2 for details). A crew of six to 12 workers from each construction site were recruited to participate in the study. Over two consecutive working days, researchers met the workers in two sessions, before and after work. Between the two sessions, the workers went on with their regular work tasks during which heart rate, skin temperature, and environmental heat stress were continuous recorded at 1-min intervals. Food and drinks were provided to workers during the four sessions. Semi-structured interviews with workers were conducted in the last session. No audio recording was used during the interviews to keep the process informal and to minimise possible defensiveness triggered by formality; instead the researchers kept notes during the interviews. 2.4. Data analysis Data were analysed in four steps. First, independent samples ttests were conducted to analyse questionnaire data to identify significant differences in perception between the managers and the workers, between two major ethnic groups (Chinese and Nepalese), and between age groups (below and over 50 years old) in the workers sample. In addition, demographic and physiological variables, including surface-to-mass ratio, BMI, fitness, blood pressure,

Table 2 Field data collection protocol. Prior-work session

Day time

After-work session

Informal interviews (workers): activities, hydration, critical incidents; Personal baseline data: weight, BP, fitness; Administer questionnaire to workers Day1 night Workers fill in questionnaires at home, fasting for 8 h prior to blood glucose test. Researchers download continuous data, maintain and clean equipment. Semi-structured interviews Day 2 Collect completed questionnaires from workers; Continuous recording at workplace: ibid.; (workers): risks, intervention, and Personal baseline data measurement: ibid., blood Site observation: ibid; critical incidents Interview (managers): risks, interventions, innovations glucose test Day 1 Introduction; Demographic data: trade, tenure, age, gender; Personal baseline data: height, weight, tympanic temperature, blood pressure (BP), resting heart rate (RHR), clothing

Continuous recording at workplace: environmental heat stress, including air temperature, relative humidity, solar radiant heat, WBGT; heart rate, skin temperature; Site observation: activities, risks and interventions

28

Y.A. Jia et al. / Applied Ergonomics 53 (2016) 25e35

blood glucose, ageing, smoking, sleep quality and effective heat risk, were analysed with deviation analysis (Hale et al., 1997; n, 1984) to identify factors that were out of the normal Kjelle range. These included body surface-to-mass ratio, body mass index (BMI), fitness, blood pressure, blood glucose, ageing (above the age of 50 years old), smoking habit, and sleep quality (more details of the criteria can be found in Rowlinson and Jia, 2015, p. 181). Effective heat risks were analysed by computing the maximum allowable exposure time (Dlim) (ISO 7933, 2004; Malchaire et al., 2001) and compared the result with the actual continuous work time (CWT) of the situation (Rowlinson and Jia, 2014). These initial analysis results were used as building blocks for triangulation with the qualitative analysis. Second, the multiple sources of data were triangulated to construct a case for each individual worker, including crossvalidated information from the individual's physiological and selfreported data, co-workers’ reports, relevant managers' interview data, environmental data and on-site observations. Second, the triangulated cases were analysed using a grounded theory approach (Strauss and Corbin, 1998) to generate a theoretical model for understanding the risks, consequences, intervention strategies and their interrelationships in the construction management context. Three research assistants conducted open coding, the results of which were discussed in a meeting led by the first author, who afterwards continued with axial coding, memo writing and theory generation in frequent discussion with the second author. Third, critical incident analysis was conducted to identify prominent risk factors for heat illness. For this analysis, the data were re-organised into 36 individual heat illness cases, and were constructed with narratives of the cases and their background information including individual health condition, co-workers, supervisors and researchers' observations of work characteristics and workplace conditions. The constructed heat illness cases were then analysed against the generated theoretical model to identify prominent heat risks on construction sites using three methods: epidemiological analysis of temporal and demographic patterns; deviation analysis of physiological and lifestyle variables; and supplemented by content analysis of the qualitative data (Graneheim and Lundman, 2004). 3. Results 3.1. Obtained sample Field data were obtained from 26 construction sites in Hong Kong, including: six new building projects, three foundation works, three tunnelling projects, six civil engineering projects, five interior renovation projects, one building demolition project, and two roadside maintenance projects. There were six more projects lined up to participate the study but were not included, because sample saturation had been achieved. In the managers sample, questionnaire data were obtained from a sample of 96 managers from both client and contractor organisations under 37 different job titles, including Managing Director, Quality Manager, Site Agent, Engineer, Safety Manager, Site Nurse, Inspector, etc. Among them, 38 site-based personnel participated in semi-structured interviews. In the workers sample, physiological data were obtained from 216 individual construction workers, among which 207 were coupled with interview and questionnaire data. The sample includes 37 trades. Characteristics of environmental and metabolic heat stress data and demographic and personal health data for the workers sample have been published in Rowlinson and Jia (2014). Independent samples t-tests were conducted to compare the risk perceptions between managers and workers. Significant perception differences included solar radiation (t ¼ 5.03, p < 0.001), humidity

(t ¼ 9.85, p < 0.0001), and alcohol drinking (t ¼ 3.47, p < 0.005). Descriptive statistics of workers sample can be seen in Table 3. 3.2. Generated theoretical model Drawing from the whole database, the generated theoretical model is shown in Fig. 1. Central to the model is a four-stage process from heat-hazard-as-exists on the construction site; to effective heat stress that impacts on human body (CP1); to the onset of heat illness (CP2); then to the serious consequences of heat illness including heat exhaustion and heat stroke (CP3); and finally to fatality (CP4). A clear pattern was identified that most preventive actions were triggered in response to the perception of early symptoms of a heat disorder. Therefore, early symptoms, including heat rash, heat cramp, dizziness and fatigue are encapsulated as a node in the process of developing heat illness to separate proactive and reactive prevention strategies. The four stages can be managed at the subsequent stages with engineering controls (MP1), control of individual factors (MP2), reactive preventions (MP3) or effective emergency response (MP4). More proactively, the heat hazards can be eliminated by institutional factors such as project scheduling or daily work planning (EP1), by individual's mindful adjustment of their work pace and clothing (EP2), or by supply chain factors, e.g., prefabrication that relieves workers from site-based work thus eliminate the exposure (EP3). Interventions at all stages are enabled or constrained by organisational factors of the construction project. These include resources provision, systemic practices at the project level, procedures at the team level and established cultures and norms. Effective engineering controls need to be enabled by the effectiveness of the facilities (ECP7) and institutional support including their provision, procedure of using it and training (ECP6), and individual action of actually using them (ECP8). Individual factors can be managed, intervened or selected by project level systemic practices such as health screening, financial incentive system, training, or regulations (ECP5). Reactive preventions, including individual response to perceived early symptoms, informing supervisors or co-workers to get support, organisational initiatives to identify early symptoms among workers and stop the process, need to be enabled by individual factors including mindfulness, knowledge, and correct risk perception (ECP1) and institutional enablers such as buddy system and healthcare staffing, etc. (ECP4). Effective emergency responses need to be enabled by first-aid protocol and regular drills in the organisational procedures (ECP3), while individual factors such as latent illness can make a significant contribution to the result of death (ECP9). Moreover, factors in the upper stream of the supply chain and outside of the project organisation, such as permanent work design, client requirements, prefabrication, PPE and machinery design and manufacture, to some extent determine the practice on construction site (ECP10). 3.3. Critical incident analysis Guided by the generated model, individual cases that reported experiences of heat illness incidents were analysed to identify prominent risks. Of the 207 workers (21 female and 186 male) who participated in interviews, 40 workers reported heat illness incidents on site. Excluding four cases that reported incidents of other people, 36 valid heat illness cases (17.4%) were obtained. The 36 cases involved 35 male workers and one female worker. These cases were distributed across 17 of the 37 trades represented in this survey. Reported symptoms of heat disorders include heat rash (1 case), fatigue (10 cases), feeling thirsty (3 cases), feeling discomfort (5 cases), dizziness (18 cases), difficulty in breathing (4 cases), heat cramp (3 cases), dehydration (3 cases), over sweating (4 cases), dry

Y.A. Jia et al. / Applied Ergonomics 53 (2016) 25e35

29

Table 3 Characteristics of workers sample. Variables

N

Mean

Age Body height (m) Body weight (kg)  Tympanic temperature ( C) RPE (personal average)

216 216 216 216 204

43.0 1.67 67.1 36.4 13.6

Fitness (by YMCA test)

200

Blood Glucose Ethnicity Fitness (by BMI) Blood pressure Gender Smoking

204 216 216 198 216 212

95% CI

Median

SD

41.5 to 44.4 44.5 10.9 1.66 to 1.68 1.67 0.07 65.5 to 68.7 66.7 11.6 36.3 to 36.4 36.4 0.45 13.3 to 13.9 13.2 2.0 Ranges and percentages Very poor, 7 (3.5%); Poor, 29 (14.5%); Below average, 23 (11.5%); Average, 39 (19.5%); Above (19.5%); Good, 34 (17.0%); Excellent, 29 (14.5%) Hypoglycaemia, 4 (2.0%); Normal, 152 (74.5%); Pre-diabetes, 39 (19.1%); Diabetes, 9 (4.4%) Chinese, 196 (90.7%); Nepalese, 17 (7.9%); Pakistani, 2 (0.9%); Vietnamese, 1 (0.5%) Underweight, 10 (4.6%); Normal, 121 (56.0%); Overweight 71 (32.9%); Obesity, 14 (6.5%) Normal, 154 (77.8%); Hypotension, 2 (1.0%); Hypertension, 42 (21.2%) Female, 21 (9.7%); Male, 195 (90.3%) Smoker, 107 (50.5%); Non-smoker, 105 (49.5%)

Range 19e64 1.49e1.85 37.7e106.9 34.7e37.5 8.6e20.0 average, 39

Fig. 1. The generated socio-ergonomic model.

and hot skin (1 case), fever (8 cases), headache (6 cases), vomiting (8 cases), loss of control (2 cases), fainting (7 cases), non-sweating (1 case), and heat stroke (2 cases). 3.3.1. Results of epidemiological analysis 3.3.1.1. Temporal patterns of incidents. Fig. 2 shows temporal pattern of the reported heat disorders cases. The earliest time slot reported in the critical incident was 8 am. There was a peak right before lunch, and another at 2 pm. lasting until 3 pm., and then dramatically decreased due to the tea break at most sites. No incident was reported after 4 pm. 3.3.1.2. Age. The sample was divided into five age cohorts for analysis. No heat disorder cases were found in the 18e25 year age group. Whilst average working metabolic rate decreased consistently with age, the highest percentage (23.7%) was found in the

26e35 year old group and then decreased with age. (Table 4). 3.3.1.3. Gender. The sample of workers included 21 females and 195 males. Only one heat disorder case (4.8%) was reported among the females. Thirty-four (17.4%) heat disorder cases occurred in the male sample; however, this result is more of a reflection of the trade distribution between male and female workers in the construction industry, where females often do lighter work such as cleaning or site reception. Female workers in this sample include 15 general workers, one demolition worker, two pipe layers, one plasterer, and two tilers. The one female worker who experienced a heat disorder did so while she was under financial pressure (i.e. running her own sub-contracting company) and while conducting heavy physical work (casting concrete). At the time of data collection she had closed her business and was working as a cleaner on site.

30

Y.A. Jia et al. / Applied Ergonomics 53 (2016) 25e35

Fig. 2. Temporal pattern of the frequency of heat disorder incidents on site.

Table 4 Heat illness cases in five age groups and their average working metabolic rates. Age groups (years)

N (number of participants)

Number of HI cases [n (rate ratio)]

Mean metabolic rate (W/m2)

95% CI

SD

Median

Range

n (number of datasets)

18e25 26e35 36e45 46e55 56e65

17 37 64 71 25

0

213.6 198.0 185.4 171.0 163.4

209.9e217.3 196.1e200.0 184.0e186.8 169.7e172.3 161.5e165.3

81.5 63.8 63.8 62.3 53.8

200.4 190.9 177.5 164.0 157.6

55.0e506.6 58.5e509.9 57.5e665.6 57.2e592.8 56.6e334.0

1874 4210 8016 8761 3052

9 11 11 3

(24.3%) (17.2%) (15.3%) (12%)

3.3.1.4. Smoking habit. Divided by smoking habit, the smokers group demonstrated a slightly higher rate of heat disorder cases (17.8%) than non-smokers group (15.2%). However, when the 21 female workers, all of whom were non-smokers, were excluded from analysis, the percentage of heat disorder cases was slightly reversed (Table 5). 3.3.1.5. Fitness, blood pressure and blood glucose. Rate ratio of heat illness cases categorised by blood pressure (BP), blood glucose (BG), and fitness levels measured by body mass index (BMI) and YMCA scores respectively, are shown in Table 6. In terms of BMI, the underweight and obesity groups exhibited higher rates of heat illness cases, with the former reflecting effects of small body surface area for heat dissipation and the latter indicating the consequences of excessive metabolic heat. Contradictory to common beliefs, the groups within the normal ranges for BP and BG were found to have the highest rates of heat illness cases. 3.3.2. Prominent heat risk factors on construction site Table 7 lists elaboration of the prominent risk factors identified from 36 individual cases of heat illness on site through content

Table 5 Percentage of heat disorder cases in smoking and non-smoking groups.

N Heat disorder cases Percentage of cases

Whole sample

Controlled gender (male sample)

Smoker

Non-smoker

Smoker

Non-smoker

107 19 17.8%

105 16 15.2%

107 19 17.8%

84 15 17.9%

analysis. Four of the cases were captured on the study day. Data in such cases were triangulated with the record of climatic heat stress and metabolic rate record (see Table 8 for details). The data were analysed with the Predicted Heat Strain (PHS) model to compute the maximum allowable work time (Dlim), which was further compared with the actual continuous work time in the work situation. It is noteworthy that all the reported heat illness incidents were stopped by timely responses from individual initiatives, including stopping work or washing to cool down the body (Table 9). 4. Discussion In Case 13, the worker states that he experienced heat illness at work due to a hot weather and heavy workload. However, analysis results of the workplace heat stress data only indicate a mild warm environment, a moderate workload (in the range of metabolic rate classified by ISO 7243, 1989) with unlimited allowable continuous work time as predicted by the PHS model. Meanwhile his personal health data indicates he is aged 63 with hypertension and very poor fitness; his general description of himself indicated he was under financial pressure. In such a case the causality is interpreted as personal health condition and psychological stress instead of the worker's perceived reasons. The environmental heat stress only played a trigger for the heat illness incident. The prominent risk identified for this case is thus identified as personal health factors, suggesting a periodical health-screening would be necessary for heat illness prevention. The prominent risk for Case 31, as reported by the subject, was the heat accumulated inside the safety helmet. “During 11.30 am today when I wore safety helmet, I felt dizzy. I had to take off the helmet lest I fainted. And I was well. But when I put it on

Y.A. Jia et al. / Applied Ergonomics 53 (2016) 25e35

31

Table 6 Percentage of heat disorder cases by fitness levels. Personal health factors Fitness by BMI

Underweight Normal Overweight Obesity Excellent Good Above average Average Below average Poor Very poor Hypotension Normal Hypertension Hypoglycaemia Normal Pre-diabetes Diabetes

Fitness by YMCA test

Blood pressure

Blood glucose

N

Heat illness cases

Rate ratio of cases

10 121 71 14 29 34 39 37 24 29 7 2 153 61 4 152 39 9

3 18 10 3 2 2 11 7 6 3 2 0 28 8 0 31 4 1

30.0% 14.9% 14.1% 21.4% 6.9% 5.9% 28.2% 18.9% 25.0% 10.3% 28.6% 0% 18.3% 13.1% 0% 20.4% 10.3% 11.1%

Table 7 Prominent risk factors identified. Category

Identified risks

Elaboration

Climatic heat

Lack of wind

 Confined or semi-confined spaces, e.g., tunnels  Still-wind weather  Building lifecycle stages: a building project during the glazed phase; renovation work conducted within occupied residential buildings; demolition work during which a building is wrapped in nets  Locations at the seaside, on the mountain slope of an island, and in tunnels  Locations at large open sites of civil engineering projects, formwork and concreting on roof tops  The foundation stage of building projects for which no natural shade is available or shading facilities are not provided  The time period from 11 am to 3 pm on a hot day,  Vulnerable trade: leveller, scaffolder, concreter, shot-firer and rebar worker  Welding work  Concreting work for which the concrete generates extra heat  Maintenance work conducted roadside during which climatic heat is aggravated by the extra heat generated by passing vehicles  Safety helmet, boots and reflective vest which are impermeable to water vapour

Humidity Solar radiant heat

Workplace heat

Clothing effect Metabolic heat

Personal factors

PPE

    Continuous work time  Workload Work pace

Ethnicity Ageing Acclimatisation

Fatigue

Personal health and life style Dehydration Risk perception Job skills Mindfulness Psychological Stress

         

Heavy physical work generates more metabolic heat Work pace imposed by supervisor Certain work are paced by material, machine or team coordination Self-paced work: individual worker fails to control within a safety limit Long continuous work without break. In some cases breaks are dismissed either by supervisors or production-focused financial incentive system Ethnic minority experiences communication problems, leading to inadequate safety knowledge and job skills. Ageing in combination with latent illness and poor fitness can result in heat illness. Acclimatisation has a physical dimension and a mental dimension No formal acclimatisation procedure is being practiced. Discontinuous job of construction workers as a result of pyramid sub-contracting system also leads to lack of acclimatisation Unusual hot say in a non-summer season Sleepiness due to long working hours, long traveling time from home to site, works on night shift, offework activities, e.g. alcohol drinking, or individual health problem leading to poor sleep quality Lack of energy before or after lunch Fatigue appears to be a precursor of heat illness in some cases and a syndrome in other cases A combination of latent illness factors can cause heat illness

 Dehydration is due to lack of provision of worker, or no breaks are given for drinking water, or individual negligence  Underestimation of risk leads to unrealistic work pace and ignorance of early signs of heat illness  Poor job skills cost more metabolic heat to complete the task. Ethnic minority is often in such a situation due to communication problem and ineffective training. This corroborates with results of previous research (Gun and Budd, 1995; HSE, 2002).  ‘The most effective prevention is to ask the worker himself from time to time: “Are you safe now?”’ (quote from a rebar worker's interview)  Emotional stress due to conflict with supervisor or cynicism to the management system (licensing or reward).

again, the dizziness came back. It happened once more at around 2.30 pm.” This is cross validated by data recorded on another occasion. To verify this report, researchers further investigated the temperature within the safety helmet in subsequent site studies. Data loggers were attached in two workers' safety helmets to record temperature data during work. The recorded peak

temperatures and the compared environmental temperatures are reported in Table 9. The recorded highest temperature within the safety helmet was 43.7  C, which was 11.5  C higher than the environmental temperature. Mitigation strategy of the ‘PPE-heat hazard’ was observed in another construction site where management allocated a “PPE-free” area at a safe location near the

32

Y.A. Jia et al. / Applied Ergonomics 53 (2016) 25e35

Table 8 Details of heat illness cases that occurred on data collection days. Cases

Case 13 (PTER07)

Trade

Plasterer (indoor renovation work) Fender

Case 31 (HKUC04)

Date of incidents

19 Aug 2011

8 Jun, 2011 8 Jun 2011

7 Jul 2011 8 Jul 2011 8 Jul 2011

Time of incidentsc

3e4 pm

2e3 pm

2e3 pm

1e2 pm

2e3 pm

30.5 29.3 2.1 29.9e31.0 28.7e34.5 31.2 31.0 0.82 31.0e31.4 30.3e32.9 37.6 34.3 5.8 36.1e39.1 32.8e48.8 74.4 75.6 3.5 73.5e75.3 67.7e78.6 111.7 111.5 7.77 109.7e113.7

35.7 36.6 1.98 35.2e36.2 31.2e37.4 37.4 37.8 1.32 37.1e37.8 34.5e39.3 52.0 54.6 5.49 50.6e53.4 39.8e56.4 50.1 49.2 4.44 49.0e51.2 43.7e59.7 106.9 109.0 11.24 104.0 e109.8 79e125 211.9 223.9 63.38 195.7 e228.2 55.0 e313.9 0.1 0.6 clo

36.0 36.0 0.54 35.8e36.1 35.0e37.3 38.3 38.3 0.49 38.2e38.5 37.4e39.3 53.4 53.0 53.8 53.0e53.8 50.6e56.0 45.3 45.1 1.70 44.9e45.8 41.8e48.9 106.9 106.1 18.0 103.3e110.6

0.5 0.6 clo

36.7 37.1 0.99 36.5e37.0 33.8e38.1 39.2 39.2 0.87 39.0e39.4 37.6e41.5 55.8 56.6 2.91 55.1e56.6 46.7e59.5 42.3 42.3 2.37 41.7e42.9 37.0e47.2 104.2 104.0 7.61 102.3 e106.2 93e120 197.1 195.7 42.86 186.1 e208.0 133.8 e285.8 0.1 0.6 clo

Rigger, plant operator

Case 33 (CSP03)

Case 34 (CSP12)

Rigger

Rigger

Range

101.9e223.3

(m/s) Static thermal insulation Water vapour permeability (m) (kg) (Min)

0.1 0.6 clo

11 am e12 pm 34.4 34.4 0.24 34.3e34.4 33.9e34.8 32.8 32.8 0.27 32.8e32.9 32e33 48.2 48.2 0.55 48.0e48.3 47.1e49.3 68.1 68.1 0.84 67.9e68.3 66.1e69.7 115.8 117 9.0 113.5 e118.1 94e143 252.2 256.5 33.1 243.7 e260.1 172.2 e351.8 0.5 0.6 clo

0.24

0.24

0.24

0.24

0.24

0.24

Body height Body weight Dlimd CWTd (Min) Identified deviations in personal health conditions Reported symptoms Self-reported causes of heat illness

1.65 63.3 Unlimited 90e150 Age 63; hypertension; very poor fitness Feeling not well; vomiting Hot weather, heavy workload

1.74 55.2 48 120e180 e

1.74 55.2 Unlimited 60e120

1.75 82.7 61 90e150 e

1.75 82.7 57 60

1.79 71.5 46 60e120 Smoker

Interview/observation

Has HI history; under financial pressure



WBGT ( C)a

Mean Median SD 95% CI Range Mean Median SD 95% CI Range Mean Median SD 95% CI Range Mean Median SD 95% CI Range Mean Median SD 95% CI

27.3 27.3 0.11 27.2e27.3 27.0e27.5 30.8 30.8 0.09 30.80e30.85 30.7e31.0 30.9 30.9 0.08 30.85e30.90 30.7e31.0 64.6 64.3 1.18 64.3e64.9 62.2e67.5 103.9 103.8 5.6 102.8e104.9

Range Metabolic ratea (W/m2) Mean Median SD 95% CI

89.2e115.6 169.4 (E) 169.1 25.9 164.5e174.2

Dry bulb Temperaturea  ( C)

Globe Temperaturea  ( C)

Relative humiditya (%)

Heart ratea(W/m2)

Wind speed Estimated clothing effectb

100e130 237.0 236.4 28.4 229.6e244.3 194.2e304.1

Dizzy Safety helmet

Fainting Hot weather, strong sunlight, fatigue Temperature in safety helmet is Long working hours of compared in Table 9. the construction site, slept only 4 h (fatigue)

71.2e173.0 226.9 (E) 224.4 83.9 209.9e243.9 60.6e534.3 0.1 0.6 clo

Fainting Hot weather, strong sunlight Did not have lunch (fatigue)

a

Data were recorded at 1-min intervals. Clothing effect was estimated according to ISO 9920, combined effect of safety helmet, no-sleeve reflective vest, cotton t-shirt and safety boots (ISO 9920, 2009). Time range of the incidents was estimated as 30 min before and after the reported time point of the incidents. Means of heat stress parameters were calculated as means of the 60 min range. d CWT is the actual continuous working time from start of the shift to the reported time of heat illness case. Dlim is the maximum continuous working time as predicted by the PHS model under the condition of the individual condition of body weight, height, clothing effect, metabolic rate and environmental heat. b c

workplace. There, the workers were allowed to take off their PPE and take a break from both work and the heat. The heat accumulation effect cannot be adequately estimated with the PHS model. Improvement suggested by Wang et al. (2011) provides a crossvalidation of this case. In this study of Hong Kong construction workers, we found that demographic and personal health factors did not have a strong

consistent influence on cases of heat disorder except for age, which, however, exhibits a pattern contrary to common beliefs that old people are more vulnerable to heat. The rate of heat disorders was highest among 26e35 year olds and then decreased with increasing age. Although at individual level increasing age is generally associated with despairing heat dissipation capacity through the lower levels of fitness, the demography of the construction workforce

Y.A. Jia et al. / Applied Ergonomics 53 (2016) 25e35

33

Table 9 Temperature recorded inside the safety helmet. 



Subject (trade)

Date: 23 August 2011

Environmental temperature (dry bulb temperature,  C)

Temperature recorded in safety helmet ( C)

Difference ( C)

MTRO03 welder

9.56 am 11.10 am 13.53 pm 14.26 pm 16.00 pm 10.31 am 11.17 am 13.49 pm 14.44 pm 15.58 pm

31.4 33.0 33.9 33.3 32.8 32.2 33.2 34.0 34.2 32.7

40.0 40.9 43.2 37.9 39.0 43.7 39.1 41.2 40.8 38.6

8.6 7.9 9.3 4.6 6.2 11.5 5.9 7.2 6.6 5.9

MTRO06 carpenter

however changes with the ageing process whereby unfit people gradually drop out of the industry, as selected and configured by the heavy physical job demands (Marchant, 2013). The trend can also be explained by the gap between people's actual physiological capacity of dissipating heat and their perception of the capacity. Individual adaptation behaviours to modify the situation are triggered by the latter, while heat illness develops based on the former. In the dominant masculine culture of the construction industry, workers tend to underestimate risks and overestimate one's own physical capacity (Hofstede, 2001; Marchant, 2013). Underestimation of the risk and overestimation of one's own capacity were reported by both the workers and the managers samples in our study as being the major causal factors of heat illness on construction sites. It would be reasonable to hypothesize that the perception-capacity gap is most significant in the 26e35 age group when the body's physiological capacity starts to deteriorate while the mind's recognition of this deterioration lags behind. Through the ageing process, people become more realistic and experienced in their estimation of, and response to, the risks of heat illness. This hypothesis warrants further study. Another explanation to the inconsistent ageing effect on heat tolerance is the level of job skills (HSE, 2002). Those who can survive the job throughout a life span accumulate experience, skills, and expertise in their particular trade, which assist them to cope with the extra workload caused by environmental stressor (Parsons, 2014, p. 405). Earlier studies have concluded that females are at an advantage in humid hot weather due to having a higher surface-to-mass ratio; while males are advantaged in dry hot weather due to a higher sweating capacity (Kenney, 1985; Shapiro et al., 1980). More recent studies found gender differences tend to disappear when fitness, surface area-to-mass ratio and state of acclimation are equalised (Havenith, 2001). Results of our study show a lower rate of heat illness cases among female workers in humid hot weather, which seems to concur with the earlier conclusions. However, other than the biological difference, our results should firstly be attributed to the differences in the trade selection between the two genders in the construction industry. It is noteworthy that results of the critical incident analysis did not identify temperature as one of the prominent risk factors contributing to the occurrence of heat illness. This can be explained by the context of the organisational context of construction site as a loosely coupled system (Orton and Weick, 1990; Weick, 1976), in which individuals are engaged in spontaneous adaptations triggered by sense, perception and comprehension of local hazards, and enabled by a personal repertoire of coping strategies. Thus, it is important to recognise that the human being is an active agent that responds and reacts to the environment through both physiological adaptations, e.g., acclimatisation, and behavioural adaptations, in which feeling of discomfort is coupled with actions to reduce the exposure. Here, sense and knowledge intimately interact to enable an individual to respond to what he or she perceives about the

environment, and make effective adaptations by keeping hydrated, adjusting the work pace, and making timely response to early signs of heat disorder. Parsons (1995) noted that behavioural responses such as “taking off clothing, reduced work rate and movement away from heat source” (p.6) can be seen as part of human thermoregulatory responses and taken as indicators of heat strain. Taking this forward, findings of this study show that these responses are associated with immediate change in the level of heat stress and thus to certain extent constitute heat stress itself. From an organisational perspective, the risks are also dependent on project organisations' responsibility to inform frontline staff of the climatic heat risks, provide knowledge of safe work in heat, and remove the constraints or incentives that drive the construction workers to ignore early warning signs of heat illness. Thus, we argue that heat stress is not only a physiological and psychological phenomenon (Parsons, 2014), but also a social construct. Donoghue et al. (2000) examined 106 hospitalised heat exhaustion cases among mining workers and concluded that few hospitalised heat exhaustion cases occur in a situation below 28.5  C-WBGT, a threshold prescribed by ACGIH. In our study, early symptoms of heat illness were reported in situations below this threshold, and the majority did not get heat illness in a much hotter environment. This finding suggests three fundamental issues in the existing experimental methods to determine the WBGT thresholds. First, the thresholds are set up assuming a standard work force that are young and fit, which, in the real life, are always characterised by wide variety of individual differences. Second, without comparing to an external standardized threshold, individual can well perceive and respond to early symptoms of heat disorders, thus preventing heat exhaustion from happening at all. Third, in the context of a self-paced work environment such as a construction project organisation, it is the psychosocial environment (e.g. financial incentives or lack of knowledge) that drives workers to ignore their early symptoms and develop more serious consequences of heat illness. Therefore, as noted by Parsons (2014), “an underlying simple deterministic model of human interaction with the thermal environment will be inadequate and laboratory experiments will be of limited value” (Parsons, 2014, p. 95). A systemic approach to addressing these organisational factors would be more effective in controlling heat stress than would a set of prescriptive thresholds. The results remind us that the population suffering from ‘occupational’ heat stress is different from the general population, insofar as the demography of the occupational group has been configured by the nature of their specific work. Contrary to the common belief that heat disorder cases occur among trades characterised by heavy physical outdoor work, our results indicate that it occurs within all trades in construction work. A most physiologically vulnerable worker can be safe in hot weather over the warning threshold, while a most physically fit worker can experience heat illness in seemingly safe weather. What make the difference are mindfulness and the effect of autonomous adaptation

34

Y.A. Jia et al. / Applied Ergonomics 53 (2016) 25e35

behaviour, triggered by a risk perception that ‘fits’ the individual person. Effective and cost-efficient interventions are those that fit the specific population and integrate with their specific job characteristics, organisational mechanism and production process. 5. Conclusion Our study demonstrates that heat illness on construction site is a consequence of systemic issue encompassing environmental hazards, personal physiological conditions and organizationally enabled individual autonomous adaptation. Its occurrence reflects more of the socio-managerial effects than are predicted by a rational heat balance model. Heat risks are socially constructed and mitigated. In that sense, the risks of heat stress need to be reconceptualised as social constructs that direct to effective interventions. The naturalistic, multi-sourced data collection protocol developed in this study provides a methodology to enable incident causations to be interpreted in the context of weather conditions, personal health, the work environment, and autonomous adaptation. Whilst the rational ergonomics model provides a basis for explaining the physiological mechanism of heat exchange between the human body and its immediate thermal environment, our study connects it to the socio-managerial context of construction project management to generate a socio-ergonomic model. The socio-ergonomic model developed in this study contributes to the existing body of knowledge in three aspects. First, it differentiates the concept of heat hazards as existence from the concept of heat stress risks having impact on human body, thus explicates a timing dimension as a key factor of implementing effective interventions. Second, the concepts of heat risks are redefined to include psychosocial dimensions to be considered holistically in the development of a systemic intervention. Third, the moderating, enabling and constraining role of institutional environments within a project organisation in heat stress management on the construction site is identified and elaborated. As demonstrated this study, the generated model is a useful framework in guiding the analysis of heat illness cases to identify prominent risk factors in specific working conditions, and more generally, to guide accident investigation to identify causalities among institutional issues at organisational level or in the supply chain. Practically it can be used to guide project planning in the construction industry to ensure effective interventions. The fact that some of the climatic heat factors were identified as prominent risk factors in the critical incident analysis of our study highlighted the inadequacy of effective engineering controls on humidity, solar radiant heat and ventilation on construction sites in Hong Kong at the time of the field study. As an outcome of our study these shortfalls have been adequately addressed by the updated guidelines for the Hong Kong construction industry (CIC, 2013). The region-specific sample is a limitation of this study. The identified temporal and spatial characteristics of heat hazards should be understood within the context of the oceanic climate of Hong Kong. While a full spectrum of identified institutional factors is identified in a separate paper (Rowlinson and Jia, 2015), further study is needed for cross-regional comparison to generalise the model and test its robustness. Acknowledgements The authors would like to acknowledge Mark SHARP, Gigi TSANG Wing Chi, Tas Yong KOH, Martin TUULI, YIN Xianting Eva, Sam C. M. HUI, ZHANG Yu, SHEN Yuzhong John, Chi Wah LAU, Glen JOE, Michael TSE, LU Wei Tony, MA Shichao Athenda, JU Chuangjing Carrie, ZHANG Dan Dora, LIANG Yanhong, LI Jingkai Jack, BI Mo, ZHOU Zhipeng Roger, LI Xiaoyang, CHIU Wai Yee Betty, GOH Ching

Siew, and CHEN Nazhen for their help during the research process. Organizational and individual participants of the study are also gratefully acknowledged. The research was financially supported by a number of funding sources, including Hong Kong Construction Industry Council; Seed Funding Grant for Applied Research (Project No. 201109160027) and Leung Kau Kui Research and Teaching Endowment Fund (Project No. 201011165012) at the University of Hong Kong; Key laboratory of Eco-environment in Three Gorges Reservoir Region under Ministry of Education, Chongqing University, China; and Eminent Visiting Fellowship from Faculty of Humanity, Curtin University, Australia.

References Abu Dhabi EHS Center, 2012a. Code of Practice EHSRI e Safety in the Heat (Abu Dhabi EHSMS Regulatory Framework CoP 11.0). Abu Dhabi Environment, Health and Safety Center, UAE. Abu Dhabi EHS Center, 2012b. Technical Guideline: Safety in the Heat (Abu Dhabi EHSMS Regulatory Framework). Abu Dhabi Environment, Health and Safety Center, UAE. ACGIH, 2009. Heat Stress and Heat Strain: TLV® Physical Agents 7th Edition Documentation. American Conference of Governmental Industrial Hygienists; Cincinnati. ACGIH, 2013. 2013 TLVs® and BEIs® e Based on the Documentation of the Threshold Limit Values for Chemical Substances and Physical Agents & Biological Exposure Indices. American Conference of Governmental Industrial Hygienists, Cincinnati. AIOH, 2003. Heat Stress Standard & Documentation Developed for Use in the Australian Environment (Developed by Ross Di Corleto, Gerry Coles and Ian Firth). The Australian Institute of Occupational Hygienists Inc, Tullamarine. Argyris, C., 1952. Diagnosing defenses against the outsider. J. Soc. Issues 8 (3), 24e34. Bailey, C.A., 2007. A Guide to Qualitative Field Research, second ed. Pine Forge Press, Thousand Oaks, California. Berger, P.L., Luckmann, T., 1966. The Social Construction of Reality: a Treatise its the Sociology of Knowledge. Anchor Books, New York. Bernard, T.E., Barrow, C.A., 2013. Empirical approach to outdoor WBGT from meteorological data and performance of two different instrument designs. Ind. Health 51, 79e85. Borg, G., 2005. Scaling experiences during work: perceived exertion and difficulty. In: Stanton, N., Hedge, A., Brookhuis, K., Salas, E., Hendrick, H. (Eds.), Handbook of Human Factors and Ergonomics Methods. CRC Press, Boca Raton, 11e11~11e17. Budd, G.M., 2008. Wet-bulb globe temperature (WBGT) e its history and its limitations. J. Sci. Med. Sport 11, 20e32. Butterfield, L.D., Borgen, W.A., Amundson, N.E., Maglio, A.-S.T., 2005. Fifty years of the critical incident technique: 1954e2004 and beyond. Qual. Res. 5 (4), 475e497. CIC, 2008. Guidelines on Site Safety Measures for Working in Hot Weather. Construction Industry Council, Hong Kong. CIC, 2013. Guidelines on Site Safety Measures for Working in Hot Weather (Version 2). Construction Industry Council, Hong Kong. Creswell, J.W., 1998. Qualitative Inquiry and Research Design: Choosing Among Five Traditions. Sage, Thousand Oaks, CA. CSAO, 2000. Heat Stress: Guidelines for Recognition, Assessment, and Control in Construction. Construction Safety Association of Ontario, Etobicoke. Dean, J.P., Whyte, W.F., 1958. How do you know if the informant is telling the truth? Hum. Organ. 17 (2), 34e38. Donoghue, A.M., Sinclair, M.J., Bates, G.P., 2000. Heat exhaustion in a deep underground metalliferous mine. Occup. Environ. Med. 57, 165e174. Dzarr, A.A., Kamal, M., Baba, A.A., 2009. A comparison between infrared tympanic thermometry, oral and axilla with rectal thermometry in neutropenic adults. Eur. J. Oncol. Nurs. 13 (4), 250e254. Flanagan, J.C., 1954. The critical incident technique. Psychol. Bull. 51 (4), 327e358. Fujii, R.K., Horie, S., Tsutsui, T., Nagano, C., 2008. Effectiveness of a head wash cooling protocol using non-refrigerated water in reducing heat stress. J. Occup. Health 50, 251e261. Graneheim, U.H., Lundman, B., 2004. Qualitative content analysis in nursing research: concepts, procedures and measures to achieve trustworthiness. Nurse Educ. Today 24, 105e112. Gun, R.T., Budd, G.M., 1995. Effects of thermal, personal and behavioural factors on the physiological strain, thermal comfort and productivity of Australian shearers in hot weather. Ergonomics 38 (7), 1368e1384. Hale, A.R., Heming, B.H.J., Carthey, J., Kirwan, B., 1997. Modelling of safety management systems. Saf. Sci. 26 (1/2), 121e140. Havenith, G., 2001. Human surface to mass ratio and body core temperature in exercise heat stressda concept revisited. J. Therm. Biol. 26 (4e5), 387e393. Hofmann, D.A., Stetzer, A., 1998. The role of safety climate and communication in accident interpretation: implications for learning from negative events. Acad. Manag. J. 41 (6), 644e657.

Y.A. Jia et al. / Applied Ergonomics 53 (2016) 25e35 Hofstede, G., 2001. Culture's Consequences: Comparing Values, Behaviors, Institutions, and Organizations Across Nations, second ed. Sage Publications, Thousand Oaks, Calif. HSE, 2002. The Development of a Practical Heat Stress Assessment Methodology for Use in UK Industry (HSE Research Report 008, Prepared by Damian Bethea and Ken Parsons). Health & Safety Executive, UK. ISO 7243, 1989. Hot Environments e Estimation of the Heat Stress on Working Man, Based on the WBGT-index (Wet Bulb Globe Temperature). International Standard Organisation, Geneva. ISO 7933, 2004. Ergonomics of the Thermal Environment Analytical Determination and Interpretation of Heat Stress Using Calculation of the Predicted Heat Strain. International Standard Organisation, Geneva. ISO 9920, 2009. Ergonomics of the Thermal Environment e Estimation of Thermal Insulation and Water Vapour Resistance of a Clothing Ensemble. International Standard Organisation, Geneva. Kenney, W.L., 1985. A review of comparative responses of men and women to heat stress. Environ. Res. 37 (1), 1e11. n, U., 1984. The deviation concept in occupational accident controldI: defiKjelle nition and classification. Accid. Analysis Prev. 16 (4), 289e306. Leithead, C.S., Lind, A.R., 1964. Heat Stress and Heat Disorders. Cassell, London. Lin, S., Luo, M., Walker, R.J., Liu, X., Hwang, S.-A., Chinery, R., 2009. Extreme high temperatures and hospital admissions for respiratory and cardiovascular diseases. Epidemiology 20 (5), 738e746. Lincoln, Y.S., Guba, E.G., 1985. Naturalistic Inquiry. Sage Pubns, Beverly Hills. Malchaire, J., 1995. Methodology of investigation of hot working conditions in the field. Ergonomics 38 (1), 73e85. Malchaire, J., Piette, A., Kampmann, B., Mehnert, P., Gebhardt, H., Havenith, G., , et al.Griefahn, B., 2001. Development and validation of the predicted heat strain model. Ann. Occup. Hyg. 45 (2), 123e135. Marchant, T., 2013. Keep going: career perspectives on ageing and masculinity of self-employed tradesmen in Australia. Constr. Manag. Econ. 31 (8), 845e860. NOAA, 2008. Beaufort Wind Scale. National Oceanic and Atmospheric Administration, USA. http://www.spc.noaa.gov/faq/tornado/beaufort.html (accessed 28.01.15). Orton, J.D., Weick, K.E., 1990. Loosely coupled systems: a reconceptualization. Acad. Manag. Rev. 15 (2), 203e223. OSHA, 1999. OSHA Technical Manual. Chapter 4, Heat Stress OSHA Technical Manual. Occupational Safety & Health Administration, Washington, DC. http:// www.osha.gov/dts/osta/otm/otm_iii/otm_iii_4.html (accessed 28.01.15). Parsons, K., 1995. International heat stress standards: a review. Ergonomics 38 (1), 6e22.

35

Parsons, K., 2003. Human Thermal Environments, second ed. Taylor and Francis, London. Parsons, K., 2006. Heat stress standard ISO 7243 and its global application. Ind. Health 44, 368e379. Parsons, K., 2013. Occupational health impact of climate change: current and future ISO standards for the assessment of heat stress. Ind. Health 51, 86e100. Parsons, K., 2014. Human Thermal Environments: the Effects of Hot, Moderate, and Cold Environments on Human Health, Comfort, and Performance, third ed. CRC Press, London. Rowlinson, S., Jia, A.Y., 2014. Application of the predicted heat strain model in development of localized, threshold-based heat stress management guidelines for the construction industry. Ann. Occup. Hyg. 58 (3), 326e339. Rowlinson, S., Jia, A.Y., 2015. Construction accident causality: an institutional analysis of heat illness incidents on site. Saf. Sci. 78, 179e189. Shapiro, Y., Pandolf, K.B., Avellini, B.A., Pimental, N.A., Goldman, R.F., 1980. Physiological responses of men and women to humid and dry heat. J. Appl. Physiol. 49, 1e8. Strauss, A., Corbin, J., 1998. Basics of Qualitative Research: Grounded Theory Procedures and Techniques, (second ed.). Sage, Thousand Oaks, CA. Strauss, A., Corbin, J., 1990. Basics of Qualitative Research: Grounded Theory Procedures and Techniques. Sage Publications, Newbury Park, Calif. Tuuli, M.M., Rowlinson, S., 2010. What empowers individuals and teams in project settings? A critical incident analysis. Eng. Constr. Archit. Manag. 17 (1), 9e20. Vaneckova, P., Beggs, P.J., Dear, R. J. d, McCracken, K.W.J., 2008. Effect of temperature on mortality during the six warmer months in Sydney, Australia, between1993 and 2004. Environ. Res. 108, 361e369. Wang, F., Kuklane, K., Gao, C., Holmer, I., 2011. Can the PHS model (ISO7933) predict reasonable thermophysiological responses while wearing protective clothing in hot environments? Physiol. Meas. 32 (2), 239e249. Weick, K.E., 1976. Educational organizations as loosely coupled systems. Adm. Sci. Q. 21 (1), 1e19. Wyndham, C.H., Allan, A.M., Bredell, G.A.G., Andrew, R., 1967. Assessing the heat stress and establishing the limits for work in a hot mine. Br. J. Ind. Med. 24 (4), 255e271. Xiang, J., Bi, P., Pisaniello, D., Hansen, A., Sullivan, T., 2014. Association between high temperature and work-related injuries in Adelaide, South Australia, 2001e2010. Occup. Environ. Med. 71, 246e252. YMCA of the USA, 2000. YMCA Fitness Testing and Assessment Manual, fourth ed. Human Kinetics, Champaign, Illinois.