Int J Biometeorol
DOI 10.1007/s00484-017-1346-9
SPECIAL ISSUE ON TRANS-DISCIPLINARY APPROACHES TO CLIMATE CHANGE
Estimated work ability in warm outdoor environments depends
on the chosen heat stress assessment metric
Peter Bröde 1 & Dusan Fiala 2 & Bruno Lemke 3 & Tord Kjellstrom 4,5
Received: 26 July 2016 / Revised: 20 March 2017 / Accepted: 28 March 2017
# ISB 2017
Abstract With a view to occupational effects of climate
change, we performed a simulation study on the influence of
different heat stress assessment metrics on estimated work-
ability (WA) of labour in warm outdoor environments.
Whole-day shifts with varying workloads were simulated
using as input meteorological records for the hottest month
from four cities with prevailing hot (Dallas, New Delhi) or
warm-humid conditions (Managua, Osaka), respectively. In
addition, we considered the effects of adaptive strategies like
shielding against solar radiation and different work-rest sched-
ules assuming an acclimated person wearing light work
Electronic supplementary material The online version of this article
(doi:10.1007/s00484-017-1346-9) contains supplementary material,
which is available to authorized users.
* Peter Bröde
broede@ifado.de
Dusan Fiala
dfiala@ergonsim.de
Bruno Lemke
bruno.lemke@nmit.ac.nz
Tord Kjellstrom
kjellstromt@yahoo.com
1
Leibniz Research Centre for Working Environment and Human
Factors at TU Dortmund (IfADo), Ardeystrasse 67,
44139 Dortmund, Germany
2
Ergonsim—Human Thermal Modelling, Marxzell, Germany
3
School of Health, Nelson Marlborough Institute of Technology,
Nelson, New Zealand
4
Centre for Technology Research and Innovation (CETRI),
Lemesos, Cyprus
5
School of Demography, Australian National University,
Canberra, Australia
clothes (0.6 clo). We assessed WA according to Wet Bulb
Globe Temperature (WBGT) by means of an empirical rela-
tion of worker performance from field studies (Hothaps), and
as allowed work hours using safety threshold limits proposed
by the corresponding standards. Using the physiological
models Predicted Heat Strain (PHS) and Universal Thermal
Climate Index (UTCI)-Fiala, we calculated WA as the percent-
age of working hours with body core temperature and cumu-
lated sweat loss below standard limits (38 °C and 7.5% of
body weight, respectively) recommended by ISO 7933 and
below conservative (38 °C; 3%) and liberal (38.2 °C; 7.5%)
limits in comparison. ANOVA results showed that the differ-
ent metrics, workload, time of day and climate type deter-
mined the largest part of WA variance. WBGT-based metrics
were highly correlated and indicated slightly more constrained
WA for moderate workload, but were less restrictive with high
workload and for afternoon work hours compared to PHS and
UTCI-Fiala. Though PHS showed unrealistic dynamic re-
sponses to rest from work compared to UTCI-Fiala, differ-
ences in WA assessed by the physiological models largely
depended on the applied limit criteria. In conclusion, our study
showed that the choice of the heat stress assessment metric
impacts notably on the estimated WA. Whereas PHS and
UTCI-Fiala can account for cumulative physiological strain
imposed by extended work hours when working heavily un-
der high heat stress, the current WBGT standards do not in-
clude this. Advanced thermophysiological models might help
developing alternatives, where not only modelling details but
also the choice of physiological limit criteria will require at-
tention. There is also an urgent need for suitable empirical data
relating workplace heat exposure to workability.
Keywords Climate change . Heat stress . Productivity . Work
capacity . Model . Index

Introduction
Workplace heat exposures are an important occupational
health concern in tropical and subtropical countries (Spector
and Sheffield 2014) where for several months each year, tem-
peratures exceed 30–35 °C. Humidity is another major heat
stress factor, and air movement and direct heat radiation play a
role in many workplaces (Parsons 2014). The ongoing and
future climate change makes the heat situation worse for many
working people (Hanna et al. 2011; Kjellstrom et al. 2016;
Kjellstrom and Crowe 2011; Kjellstrom et al. 2009a;
Kjellstrom et al. 2009b; Kjellstrom et al. 2013), and cooling
systems (Gao et al. 2011; Barwood et al. 2009; Chinevere
et al. 2008) are often not available or are difficult to provide.
As heat stress can also be associated with impaired muscular
performance or decreased work capacity (Hettinga et al. 2007;
Cheuvront et al. 2010), there is a growing interest to assess the
potential effects of climate change in terms of work efficiency
or productivity (Dunne et al. 2013; Blazejczyk et al. 2014;
Kjellstrom et al. 2016; Kjellstrom et al. 2009c).
In order to assess the current and future potential workplace
heat exposures and to develop policies for prevention of heat
related health threats and losses of workability, which is an alter-
native term to work capacity and an input into labour productivity,
it is important to have scientifically based heat stress indicators
available (Spector and Sheffield 2014). One approach to quantify
the effect of heat stress on workability is to consider a working
environment as workable as long as limiting criteria from ergo-
nomics standards are not exceeded (Kjellstrom and Crowe 2011;
Kjellstrom et al. 2013; Lundgren et al. 2014; Blazejczyk et al.
2014). Popular metrics applied in this area have recently been
reviewed (Havenith and Fiala 2016; Gao et al. this special issue)
and are used to produce corresponding global and local heat maps
(Kjellstrom et al. this special issue). In this respect, common indi-
ces like the Wet Bulb Globe Temperature (WBGT) (ISO 7243
1989), the Predicted Heat Strain (PHS) (ISO 7933 2004) and the
recently developed Universal Thermal Climate Index (UTCI)
(Bröde et al. 2012) are of particular interest. The latter is based
on the advanced physiological UTCI-Fiala model (Fiala et al.
2012) that was in good agreement with other standards and exper-
imental data in occupational settings (Bröde et al. 2013; Bröde
et al. 2016; Kampmann et al. 2012).
The effect of heat stress on human physiology is assessed
by the increase in body core temperature and by the amount of
sweating indicating thermoregulatory strain (Parsons 2014),
but various criteria are in use. Based on an early WHO report
(WHO Scientific Group 1969), the standard ISO 7933 (2004)
suggests that considering the inter-individual variability in
core temperature a limit of 38 °C will protect most workers
from Bexcessive, often disturbing, physiological changes^ oc-
curring at about 39.2 °C (Malchaire et al. 2000). However,
during development of this standard, a limit of 38.2 °C was
reported to be also compatible with these safety goals
Int J Biometeorol
(Malchaire et al. 2000) and this is indeed used by some met-
rics (Miller and Bates 2007). A laboratory study suggested
core temperature limits for the occurrence of heat exhaustion
depending on the hydration status with 38.1 °C for
hypohydrated and 38.3 °C for euhydrated persons (Sawka
et al. 1992), and a recent simulation study on the effects of
personal protective equipment even employed a limit of 39 °C
(Xu et al. 2016). Similarly, for sweat loss, dehydration levels
above 3% of body weight should be avoided (Parsons 2014);
however, the standard applies a limit of 7.5% for an average
person, because usual drinking habits over a shift are taken
into account (Malchaire et al. 2000; ISO 7933 2004).
This diversity in the applied heat stress metrics and limit
criteria makes comparisons between different studies difficult.
So, the aim of this contribution is to present and discuss dif-
ferent approaches to quantifying occupational heat exposures
in relation to workability and to assess which methods may be
suitable for inter-study comparisons and for assessments of
future potential heat exposures as an outcome of local climate
change. Our approach was to simulate whole-day work shifts
under extreme climatic conditions with varying workloads
and work-rest schedules applying different heat stress assess-
ment metrics.
Material and methods
For this simulation study, we assumed an acclimated person
wearing light work clothes with thermal insulation of 0.6 clo
(1 clo = 0.155 m2 K/W) and performing light, moderate or
heavy work with typical values of metabolic heat production
(M) chosen to M = 180, 297 and 414 W, respectively, in
accordance with the international standard ISO 7243 (1989).
Climatic conditions
Meteorological records representative for the hottest month in
1999 from four cities with prevailing hot-dry [Dallas (32° 46′
33″ N 96° 47′ 48″ W, humid subtropical climate with dry winds
in summer and Köppen-Geiger classification Cfa), New Delhi
(28° 36′ 50″ N 77° 12′ 32″ E, humid subtropical climate with
Köppen-Geiger classification Cwa)] or warm-humid climate
[Managua (12° 8′ 11″ N 86° 15′ 5″ W, tropical wet climate in
warm season with Köppen-Geiger classification Aw), Osaka
(34° 41′ 38″ N 135° 30′ 8″ E, humid subtropical climate with
Köppen-Geiger classification Cfa)], respectively, were available
from the US National Climatic Data Centre at National Oceanic
and Atmospheric Administration (NOAA, USA). We obtained
hourly averages of air temperature (Ta), wind speed (va), air hu-
midity expressed as water vapour pressure (pa) and global solar
radiation (Kglob) at the person level (1.1 m) as described by
Lemke and Kjellstrom (2012). We used this information to cal-
culate mean radiant temperature (Tmrt) applying the BioKlima
Int J Biometeorol
software version 2.6 (Blazejczyk 2011; Blazejczyk et al. 2014).
In order to increase the variability in the climatic conditions under
study, we also calculated these values assuming identical values
of Ta and pa as before, but using Tmrt = Ta and a reduced wind
speed va = 1 m/s. Such settings had been applied in earlier studies
(Lundgren et al. 2014; Kjellstrom and Crowe 2011; Kjellstrom
et al. 2009c, 2013) of situations representing a workplace
shielded against solar radiation, as e.g. found in shaded semi-
open spaces or indoors with low wind and relative air move-
ments resulting from the worker’s activities. Minute-by-minute
values as required input to the physiological models (cf. the
BPHS and UTCI-Fiala model of thermoregulation^ section be-
low) were calculated from the hourly data by spline interpolation
(de Boor 1978) and are illustrated in Fig. 1.
Work profile scenarios
We simulated an extended 12-h shift with continuous work from
06:00 to 18:00 interrupted by a 1-h midday break from 12:00–
13:00. This reference scenario was complemented by profiles with
varying metabolic rates and work-rest schedules. For the
intermittent profile, we assumed for each hour 50 min work
followed by 10 min rest; for the profile with a long break, the
resting periods from the intermittent profile were gathered to a
prolonged midday break from 11:00 to 13:50. In order to study
the effect of self-pacing (Miller et al. 2011), we also considered
work profiles with reduced metabolic rates. In one profile
(Reduced Met 1), metabolic rates were actually chosen to
5/6 × M + 1/6 × Mrest with resting metabolic heat production
Mrest defined below, so that hourly averaged metabolic rates
matched those of the intermittent profile with 10 min rest per hour.
Net muscular efficiency (Enet) relates M to mechanical output (P)
by Enet = P/(M − Mrest) (Whipp and Wasserman 1969), thus im-
plying that for Reduced Met 1, mechanical output P was 5/6 of
that from the continuous profile. In another profile (Reduced Met
2), metabolic rate was set to 5/6 of the metabolic rate from the
continuous profile, meaning here that P was again 5/6 of the
Fig. 1 Climatic conditions from
06:00 to 18:00 as simulated for
four cities. Dashed lines represent
the assumed wind speed
(va = 1 m/s) and mean radiant
temperature (Tmrt = Ta) for
shielded conditions
continuous profile when estimated according to gross efficiency
Egross = P/M (Whipp and Wasserman 1969; Hettinga et al. 2007).
For break periods, we assumed that they took place under
the actual climatic conditions with an associated metabolic
rate of Mrest = 117 W (ISO 7243 1989). The time course of
metabolic rates for the different work profiles and workloads
is included in the Electronic Supplemental Material (ESM)
accompanying this text and is shown in ESM-Fig. 1.
Heat stress assessment metrics and workability
WBGT
Hourly WBGT values were estimated from the meteorological
data as described earlier (Lemke and Kjellstrom 2012). The heat
stress assessment by WBGT relies on the definition of limit
reference values (WBGTlim) to which the actual WBGT values
are compared. WBGTlim decreases with increasing rate of meta-
bolic heat production (M) given in Watts. There are differing
approaches for its calculation using the Eqs. 1 and 2 proposed
by ISO/DIS 7243 (2015) and the National Institute for
Occupational Safety and Health (NIOSH) (Jacklitsch et al.
2016), respectively.
WBGTlim;ISO=DIS ¼ 34:9−M =46 ð1Þ
WBGTlim;NIOSH ¼ 56:7−11:5  log10 M ð2Þ
When the actual WBGT value exceeds WBGTlim, the stan-
dard ISO 7243 (1989) suggests for each hour to limit the
allowed working time to a fraction of an hour, which can be
used as an estimate of workability and is calculated by Eq. 3
(Malchaire 1979; d’Ambrosio Alfano et al. 2014)
Workability
   
WBGTlim;rest −WBGT
¼ max 0; min 1; ð3Þ
WBGTlim;rest −WBGTlim
 Int J Biometeorol

In Eq. 3, WBGTlim,rest denotes the reference value obtained

by inserting resting metabolic rate Mrest = 117 W (ISO 7243
1989) in Eq. 1 or 2, respectively. The application of the min
and max functions ensures that workability estimates stay be-
tween 0 and 1. Applying Eq. 1 for computing WBGTlim and
WBGTlim,rest in Eq. 3 yielded workability estimates according
to ISO/DIS 7243 (2015), which were mostly below those
obtained for the NIOSH criteria (Jacklitsch et al. 2016) from
Eq. 2, especially for high workload (Fig. 2). This
corresponded to the differences in WBGTlim occurring at high
workload with ISO/DIS using lower limit values, i.e. being
more protective than NIOSH (Fig. 3).
Another metric based on WBGT stems from the ^high
occupational temperature health and productivity
suppression^ programme (Hothaps) (Kjellstrom et al. 2009a)
relating reduced workability as observed in field studies with
gold mine workers in South Africa (Wyndham 1969) and rice
harvesting workers in India (Sahu et al. 2013) to simulta-
neously recorded WBGT values (Kjellstrom et al. 2014).
These data are given for low, moderate and high workloads,
respectively. Contrary to the workability estimates based on
WBGTlim with a lower limit of zero (Eq. 3), they propose a
lower limit of 10%, i.e. working is possible for 6 min within
each hour even under extreme heat (Fig. 2). For simplified
calculation purposes, Hothaps workability values were ap-
Fig. 3 Hourly WBGT values for the continuous work profile under non-
proximated in this study by a two-parameter logistic function shielded (a) and shielded (b) climatic conditions in four cities with a hot
given by Eq. 4. (Dallas, New Delhi) or warm-humid (Managua, Osaka) climate com-
pared to limiting WBGT reference values (WBGTlim) for low, moderate
WorkabilityHothaps ¼ 0:1 þ 0:9=ð1 þ ðWBGT=α1 Þα2 Þ ð4Þ and high workload calculated according to NIOSH (solid light grey lines)
or ISO/DIS 7243 (dotted light grey lines), respectively
In Eq. 4, the sets of parameters (α1, α2) were fitted to the
data by non-linear least squares (Bates and Watts 1988) and
PHS and UTCI-Fiala model of thermoregulation
amounted to (34.64, 22.72) for low, (32.93, 17.81) for mod-
erate and (30.94, 16.64) for high workload, respectively,
Simulations with PHS were performed using the software pro-
yielding a very close fit with R2 > 0.99 (Fig. 2).
vided by the standard (ISO 7933 2004; Malchaire and Piette
2004), modelling an acclimated person with 75 kg body weight
wearing clothing with clothing insulation of 0.6 clo.
The UTCI-Fiala model (Fiala et al. 2012) is an adapted
version of the BFiala thermal Physiology and Comfort^
(FPC) model (Fiala and Havenith 2016; Fiala et al. 2001) used
to derive UTCI, the BUniversal Thermal Climate Index^
(Bröde et al. 2012). It simulates a person with 73.4 kg body
weight and was extended for the purpose of this study
allowing simulations with different metabolic rates and expo-
sure durations (Bröde et al. 2016) and considering the effect of
hidromeiosis (Candas et al. 1980), i.e. reduced sweating
caused by increased local skin wettedness in humid conditions
(Fiala and Havenith 2016). It also implemented a detailed
model of the clothing characteristics (Havenith et al. 2012),
which were chosen here to correspond to the assumed light
Fig. 2 Percentage workability determined by WBGT using limit values
work clothes (cotton long-sleeved shirt, long trousers, under-
calculated according to ISO/DIS 7243 (2015) or NIOSH (Jacklitsch et al.
2016) or using the empirical relationship from Hothaps (Kjellstrom et al. pants, socks) plus shoes with an estimated insulation of 0.6 clo
2014) (Fiala 1998).
Int J Biometeorol
With both models, time courses of rectal temperatures and
sweat rates were predicted for 12-h shifts from the minute-by-
minute input of Ta, va, pa, Tmrt and metabolic rate for the
different climates, shielding conditions, workloads and work
profiles. Sweat rates were integrated over time and expressed
as cumulated sweat loss in percentage of body weight. Rectal
temperature and cumulated sweat loss were compared to three
sets of limit criteria: standard limits (ISO 7933 2004;
Malchaire et al. 2000) with 38 °C rectal temperature and sweat
loss of 7.5% of body weight; conservative, i.e. more protective
limits with the rectal temperature boundary identical to the
standard, but a reduced sweat loss limit of 3% body weight;
and a liberal limit with sweat loss criterion identical to stan-
dard, but higher rectal temperature limit of 38.2 °C (Malchaire
et al. 2000; Miller and Bates 2007). For each minute, work-
ability was then assumed if both rectal temperature and sweat
loss stayed below the limits. Workability was set to zero for
the rest of the shift as soon as the sweat loss limit was
exceeded.
Data analysis and statistics
Workability was also set to zero for all resting periods im-
posed by the intermittent and long break scenarios, as well
as for the 1-h midday break, which was excluded from the
summarising analyses described below. Hourly workability
values were expressed as percentages computed from the
minute-by-minute values, and means with standard deviations
(SD) were calculated related to the factors climate type (hot
vs. warm-humid), assumed shielding (non-shielded vs.
shielded), time of day (before vs. after midday), work profile
(continuous, intermittent, long break, Reduced Met 1,
Reduced Met 2), workload (low, moderate, high) and the nine
heat stress assessment metrics.
In order to determine the parts of total variance in hourly
workability attributable to those factors, we calculated vari-
ance components using ANOVA (Littell et al. 2006).
Comparisons between the nine heat stress assessment metrics
were performed by computing Spearman rank correlation co-
efficients (rs) of hourly workability. All calculations were car-
ried out using SAS® version 9.2 (Littell et al. 2006).
Results
As we simulated 120 12-h scenarios in total, the descriptions in
the BWBGT, physiological heat strain and workability^ section
will focus on the continuous work profile, while detailed presen-
tations of all conditions are available in the Electronic
Supplemental Material (ESM) accompanying this text. The over-
all pattern of workability related to different determinants will be
analysed in the BDeterminants of estimated workability^ section.
WBGT, physiological heat strain and workability
Hourly WBGT (Fig. 3) followed the course of air temperature
and mean radiant temperature (Fig. 1) with values increasing
from morning to midday and decreasing again towards the
evening. WBGT was higher for the non-shielded (Fig. 3a)
compared to the shielded condition (Fig. 3b) with approxi-
mately 2 °C difference during the hottest hours of the day
and for hot (Dallas, New Delhi) compared to warm-humid
climates (Managua, Osaka). In New Delhi, WBGT in the
morning was higher than at the other locations, and conse-
quently the limit values (WBGTlim) for continuously working
with high load were reached 1 to 2 h earlier. The correspond-
ing results for the other work profiles are illustrated by ESM-
Fig. 2, showing identical hourly WBGT, but modified
WBGTlim, because only the latter depends on hourly metabol-
ic rate, which is calculated as time-weighted average of resting
and work metabolic rates for the different workload levels and
work-rest schedules.
Figure 4 exemplifies the physiological heat strain indicated
by rectal temperature and sweat rate expressed as cumulated
percentage loss of body weight, which were predicted by PHS
and UTCI-Fiala for continuous work. The corresponding il-
lustrations for the other work profiles are presented in ESM-
Fig. 3a–e. Thermal strain increased considerably with grow-
ing workload and, similarly to WBGT, was higher for hot
compared to warm-humid locations, especially considering
sweat loss. PHS and UTCI-Fiala predicted similar rectal tem-
peratures for hot climates (Dallas, New Delhi) accompanied
by slightly higher sweat rates for PHS, whereas in warm-
humid locations (Managua, Osaka), sweat loss predictions
by the two models were more similar, but PHS predicted low-
er rectal temperatures than UTCI-Fiala.
The sets of standard, conservative and liberal limit values,
respectively, showed tremendously different outcomes when
comparing them to the model responses. Whereas the standard
limits were not reached with low and moderate workload, the
conservative dehydration limit was exceeded around noon
with moderate workload and even with low workload in the
afternoon in the hot climates of Dallas and New Delhi, indi-
cating an immediate prohibition of continuing work under
those conditions. Surpassing the more liberal core temperature
limit of 38.2 °C was notably delayed with high workload in
hot climates and did hardly ever occur under warm-humid
conditions in Managua and Osaka (Fig. 4).
Figure 5 illustrates the hourly workability profiles with
different workloads estimated by the nine metrics based on
WBGT, PHS and UTCI-Fiala when working continuously
under non-shielded climatic conditions in New Delhi. The
corresponding profiles for the other climatic conditions in
the different cities are given in ESM-Fig. 4a–j.
Workability estimated by WBGT-based assessment metrics
was slightly reduced during the hottest period of the day

Fig. 4 Rectal temperatures and cumulated sweat loss predicted by PHS
and UTCI-Fiala for the continuous work profile with midday break from
12:00 to 13:00 under non-shielded (a) and shielded (b) climatic
(13:00–15:00) already for low workload, and more pro-
nounced for moderate workload. Contrary, for low workload,
PHS and UTCI-Fiala indicated reduced workability only with
conservative limits, but in connection with standard and liber-
al physiological limit values showed unrestricted workability
(100%) over the whole day with low workload (Fig. 5a) and
till the late afternoon (16:00 for PHS, 17:00 for UTCI-Fiala;
Fig. 5b) with moderate workload. With high workload, all
metrics exhibited severe restrictions on workability, which
appeared less pronounced for WBGT compared to PHS and
UTCI-Fiala, especially in the afternoon (Fig. 5c).
Int J Biometeorol
conditions for different workloads in four cities. Horizontal reference
lines indicate the standard, liberal and conservative limit values for rectal
temperature and sweat loss, respectively
Determinants of estimated workability
Figure 6 summarises the influence of climate, shielding,
time of day, work profile, workload and heat stress metric
on estimated hourly workability. Looking at the means
and SD related to those factors in Fig. 6a showed reduced
workability in hot compared to warm-humid climates, un-
der non-shielded vs. shielded conditions and in the hours
after the noon break compared to the morning. As expect-
ed, the workability was lower for continuous work com-
pared to scenarios with more breaks or reduced metabolic
Int J Biometeorol

Fig. 5 Hourly percentage workability estimated by different metrics for continuous work at low (a), moderate (b) and high (c) workload under non-
shielded climatic conditions in New Delhi

rates, but the reduction was less than the decrease in workloads. All those mean values were accompanied by
workability observed for high vs. moderate and low considerable standard deviations.

Fig. 6 Variability of percentage workability related to climate type, way interactions calculated by ANOVA. The residual term in b also
assumed shielding, time of day, work profile, workload and heat stress subsumes the forth and higher order interactions as well as those two-
assessment metric illustrated as means ± SD (a) and by partitioning the and three-way interaction terms which explained less than 2% of total
total variance in relative components (b) for main effects and up to three- variance
 Int J Biometeorol

When partitioning this variability into its components by
ANOVA (Fig. 6b), we noticed that almost three quarters of
total workability variance were attributable to workload, time
of day, the different heat stress assessment metrics and their
interactions, as indicated by the light-grey columns in Fig. 6b.
Climate type and its interaction with metrics, workload and
time of day accounted for another 10% of variance, whereas
shielding against sun radiation and the different work profiles
together only explained less than 2%. Similarly, all other two-
and three-way interactions accounted for less than 2% and
were added to the higher order interactions in the residual term
amounting to 16% of total variance.
Tables 1 and 2 cross-tabulate means and SD of percentage
hourly workability related to the heat stress assessment met-
rics and workload interacting with time of day (Table 1) and
climate type (Table 2), respectively, as those factors explained
more than 80% of workability variance (Fig. 6b). Whereas
restrictions in workability with low workload were only pre-
dicted by the physiological models using the conservative
limits for the afternoon (Table 1) and under the hot climates
in New Delhi and Dallas (Table 2), WBGT-based metrics
indicated reductions in averaged workability for moderate
workload that were greater after break and in hot climates
amounting to 8–9% for Hothaps and to 12–13% for NIOSH
and ISO/DIS. In contrast, PHS and UTCI-Fiala hardly im-
posed any restrictions for moderate workload when using
standard and liberal limits. With the liberal limits, only minor
reductions below 5% emerged even for high workload under
warm-humid conditions in Managua and Osaka and before
midday break. Considerably reduced workability was estimat-
ed for high workload especially in hot climates and after the
midday break with peak reductions above 60% with the stan-
dard limits. For these conditions, WBGT-based assessment
metrics predicted less severe constraints than PHS and
UTCI-Fiala applying standard or conservative physiological
limits.
Concerning the different sets of limit values for PHS and
UTCI-Fiala, considerable differences in percentage workabil-
ity of more than 50% were observed between the conservative
and liberal limits especially after the midday break (Table 1)
and in hot climates for high workload (Table 2). Differences
between WBGT-based metrics were lower amounting to max-
imally 3% for moderate work in warm-humid climates and
inclining to 13–14% with high workload after break and in
hot conditions. A general pattern with less severe restrictions
for the Hothaps compared to the NIOSH and ISO/DIS metrics
emerged in accordance with Fig. 2.
Using Spearman rank correlation coefficients (rs), Table 3
compares the nine heat stress assessment metrics concerning
estimated hourly workability. All correlations were positive
and statistically significant (p < 0.001) due to the high number
of 1272 simulated work hours. WBGT with ISO/DIS and
NIOSH limit values showed the highest inter-correlation
(rs = 0.95), and were also highly correlated with the field data
metric Hothaps (rs = 0.82). Correlations of the WBGT-based
metrics with PHS and UTCI-Fiala were about 0.6 for the stan-
dard and conservative physiological limits and dropped fur-
ther to about 0.4 for the liberal limits. Comparing PHS with
UTCI-Fiala, the highest correlations between 0.7–0.8 were
found when they shared the same set of limit values, whereas
rs ranged only between 0.3 and 0.4 when contrasting the con-
servative with the liberal limits for PHS and UTCI-Fiala,
respectively.
Discussion
Aiming at the impact of climate change on labour productivity,
this study dealt with workability estimated by different heat stress
assessment metrics ranging from the widely used index WBGT,
over the heat balance index PHS with the ability to predict body
core temperature and sweat loss, Bwhich puts it closer to a

Table 1 Mean (SD) of hourly

percentage workability in relation Workload Low Moderate High
to workload and time of day for
different metrics, which were Time of day Before After Before After Before After
based on WBGT using the break break break break break break
Hothaps approach and limit Metric Mean (SD) of hourly workability (%)
values, calculated according to
ISO/DIS 7243 and NIOSH, and WBGT Hothaps 99 (1) 98 (2) 96 (2) 91 (5) 90 (5) 78 (10)
on the PHS and UTCI-Fiala ISO/DIS 100 (0) 100 (0) 97 (5) 88 (17) 86 (11) 64 (19)
models using conservative (cons),
NIOSH 100 (0) 100 (1) 97 (5) 88 (16) 90 (9) 72 (18)
standard (std) and liberal (lib)
criteria for core temperature and PHS cons 100 (0) 64 (39) 98 (5) 13 (14) 78 (21) 0 (0)
dehydration level, respectively std 100 (0) 100 (0) 100 (0) 97 (8) 84 (22) 35 (37)
lib 100 (0) 100 (0) 100 (0) 97 (7) 99 (4) 64 (34)
UTCI-Fiala cons 100 (0) 84 (22) 100 (1) 24 (20) 56 (29) 0 (0)
std 100 (0) 100 (0) 100 (0) 100 (1) 57 (29) 31 (26)
lib 100 (0) 100 (0) 100 (0) 100 (1) 96 (10) 72 (30)
Int J Biometeorol

Table 2 Mean (SD) of hourly

percentage workability in relation Workload Low Moderate High
to workload and climate type for
different metrics, which were Climate type Hot Warm-humid Hot Warm-humid Hot Warm-humid
based on WBGT using the
Hothaps approach and limit Metric Mean (SD) of hourly workability (%)
values, calculated according to
ISO/DIS 7243 and NIOSH, and WBGT Hothaps 98 (2) 99 (0) 92 (6) 96 (2) 80 (11) 88 (5)
on the PHS and UTCI-Fiala ISO/DIS 100 (0) 100 (0) 87 (17) 98 (4) 67 (21) 83 (12)
models using conservative (cons),
NIOSH 100 (1) 100 (0) 87 (16) 97 (4) 73 (19) 88 (10)
standard (std) and liberal (lib)
criteria for core temperature and PHS cons 64 (39) 100 (0) 48 (48) 63 (38) 32 (35) 46 (47)
dehydration level, respectively std 100 (0) 100 (0) 97 (8) 100 (0) 41 (36) 78 (33)
lib 100 (0) 100 (0) 97 (7) 100 (0) 66 (36) 96 (9)
UTCI-Fiala cons 84 (22) 100 (0) 53 (47) 71 (31) 26 (31) 30 (38)
std 100 (0) 100 (0) 100 (1) 100 (0) 35 (25) 52 (33)
lib 100 (0) 100 (0) 100 (1) 100 (0) 70 (29) 98 (7)

physiological model" (Havenith and Fiala 2016), to the advanced
UTCI-Fiala model of human thermoregulation. We simulated
whole-day shifts with varying workloads and work profiles un-
der extreme heat stress conditions recorded in hot (Dallas, New
Delhi) and warm-humid regions (Managua, Osaka), as those will
probably affect the working population more frequently under
climate change scenarios (Beggs 2014; Gosling et al. 2011; IPCC
2007; Nilsson and Kjellstrom 2010).
Shielding against sun irradiation is an obvious adaptive strat-
egy to attenuate heat stress frequently used in well insulated
places (Blazejczyk et al. 2014; Gao et al. this special issue;
Konarska et al. 2014; Matzarakis et al. 2007; Shashua-Bar and
Hoffman 2000). Accordingly, workability was higher when the
workplace was shielded against solar radiation. At the same time,
the effect of shielding was smaller than the impact of workload or
time of day (Fig. 6), perhaps because in our simulations, the
benefit of reduced mean radiant temperature was partially
counteracted by the simultaneously assumed lower air velocity
reducing convective cooling. Although the settings used here for
the shielded condition (with Tmrt = Ta and a reduced relative air
velocity va = 1 m/s mainly resulting from the worker’s activities)
had been applied in earlier studies (Lundgren et al. 2014;
Kjellstrom and Crowe 2011; Kjellstrom et al. 2009c, 2013),
and though sun shading in urban built environments is often
accompanied by wind sheltering (Oke 1987, Allegrini and
Carmeliet 2017) possibly reducing wind speed to still air for
the meteorological conditions from Fig. 1 (Ghiaus et al. 2006),
the major focus of our paper was on comparing the different heat
stress metrics, and the shielded situation was introduced to in-
crease the variation in the climatic conditions under study (cf. the
BClimatic conditions^ section). It is thus important to note that a
systematic investigation on the effects of varying solar radiation-
wind combinations, e.g. including high wind speeds with
va > 10 m/s as employed in a recent study on the performance
of elite cyclists (Junge et al. 2016), is certainly out of the scope of
this paper and will require future dedicated studies.
As expected, workability was hardly an issue with light
work in the assumed climate conditions, but was reduced with
moderate or high workload, which are the prevailing condi-
tions reported from field studies (Lundgren et al. 2014;
Venugopal et al. 2016). Air and mean radiant temperatures
reached their maximum values during the afternoon and were

Table 3 Spearman rank

correlation coefficients between WBGT PHS UTCI-Fiala
hourly percentage workability
from 1272 simulated work hours Hothaps ISO/DIS NIOSH cons std lib cons std lib
determined by different metrics
based on WBGT and on PHS and WBGT Hothaps – 0.82 0.82 0.64 0.56 0.39 0.66 0.57 0.36
UTCI-Fiala models with standard ISO/DIS 0.82 – 0.95 0.56 0.65 0.44 0.63 0.65 0.43
(std), conservative (cons) and
NIOSH 0.82 0.95 – 0.57 0.62 0.44 0.60 0.57 0.42
liberal (lib) physiological
assessment criteria PHS cons 0.64 0.56 0.57 – 0.53 0.38 0.84 0.42 0.34
std 0.56 0.65 0.62 0.53 – 0.69 0.56 0.74 0.63
lib 0.39 0.44 0.44 0.38 0.69 – 0.40 0.51 0.78
UTCI-Fiala cons 0.66 0.63 0.60 0.84 0.56 0.40 – 0.64 0.35
std 0.57 0.65 0.57 0.42 0.74 0.51 0.64 – 0.52
lib 0.36 0.43 0.42 0.34 0.63 0.78 0.35 0.52 –

higher in hot compared to warm-humid regions (Fig. 1), thus
explaining at least partly the more severe reduction in work-
ability after the midday break and for hot climates. Together
with the different metrics, workload, time of day and climate
type determined the largest part of workability variance and
also modified the influence of the heat stress metrics on esti-
mated workability as outlined in the BWorkability and heat
stress metrics^ section below.
Workability and work output
Alterations of the work profile by inserting additional breaks or
by lowering metabolic rate caused a modest increase in work-
ability compared to working continuously with the prescribed
workload. However, discussing the influence of these modifica-
tions on workability in comparison to the continuous work pro-
file with the view on labour productivity requires consideration
of the muscular power output (P). If workability for both the
continuous profile (WAcontinuous) and the modified conditions
(WAmod) are 100% of the whole-day shift, the corresponding
ratio in work output (Pmod/Pcontinuous) will be 5/6 for the long
break and intermittent scenario, because working time will be
5/6 of the continuous work profile. Similarly, the work profiles
with reduced metabolic rate will also yield a ratio Pmod/
Pcontinuous = 5/6 according to either net muscular efficiency for
the Reduced Met 1 or to gross efficiency for the Reduced Met 2
profile as shown in the BWork profile scenarios^ section. If
WAcontinuous or WAmod does not equal 100%, the output ratio
can be estimated for our settings by Eq. 5
Pmod =Pcontinuous ¼ WAmod =WAcontinuous  ð5=6Þ ð5Þ
Fig. 7 Work output from profiles with different work-rest schedules
compared as ratio to the continuous scenario in relation to the continuous
work’s percentage workability estimated for whole-day shifts by different
Int J Biometeorol
This was used to compare the work output of the modified
profiles in relation to the continuous profile. As shown in
Fig. 7, these ratios increased with decreasing workability for
the continuous profile estimated by the different metrics. They
even exceeded unity for very low workability predicted by
PHS or UTCI-Fiala indicating that continuously performing
at a prescribed workload may yield inferior output than adap-
tively adjusting workload or working duration in the heat.
This is in line with the benefits ascribed to self-pacing
employed by experienced workers as an adaptive strategy to
ensure safety under heat stress (Miller et al. 2011; Kampmann
2000; Schlader et al. 2011).
Notably these results were not obtained with the Hothaps
metric, which predicted a constant 5/6 power output ratio.
This was presumably due to a too coarse categorisation of
workload (low, moderate, high) which did not capture the
subtle changes in hourly metabolic rate introduced by the
modified work profiles. This issue might be resolved in future
adaptations of the Hothaps metric by relating the reduction in
workability to the difference of actual WBGT to the limit
reference value depending on metabolic rate, as exemplified
in ESM-Fig. 5 for the ISO/DIS limits from Eq. 1.
Workability and heat stress metrics
Several occasions with work output ratios below the expected
value of 5/6 were another unexpected feature of Fig. 7, which
almost exclusively concerned estimates for the intermittent
scenario by PHS and implied that workability was lower for
the intermittent profile compared to continuous work. Indeed
heat stress metrics. Locally estimated smoothing splines (LOESS) are
added for the different metrics, where for PHS and UTCI-Fiala the results
with the standard, conservative and liberal limits were merged in one plot
Int J Biometeorol
we found that maximum rectal temperatures from PHS for the
intermittent work profile (ESM-Fig. 3b) were always higher
than the maximum predicted for continuous work (ESM-
Fig. 3a) with differences up to 0.31 °C. The UTCI-Fiala model
showed the opposite with maximum rectal temperatures al-
ways being lower for intermittent work as expected. This
was related to different physiological responses to a rest from
work as illustrated by Fig. 8 for the continuous and long break
scenarios, respectively. We found that UTCI-Fiala predicted a
less steep decrease in body core temperature during break
compared to PHS, which showed an unrealistic instantaneous
decline followed by a steep increase when work started again.
These limitations in the dynamic behaviour of PHS predic-
tions had been reported before (Gebhardt et al. 2009) and are
also an issue for the ongoing revision of the standard by ISO
TC159/SC5/WG1 (Olesen et al. 2016).
Despite these dissimilarities between PHS and UTCI-Fiala
in thermoregulatory modelling, the main distinguishing fea-
tures between the physiological modelling approaches to
workability were the applied limit criteria. Whereas workabil-
ity was almost unrestricted for light and moderate work with
the standard limits, the conservative limits lead to severe con-
straints in hot climates. The standard and conservative limits
apply the same criterion of 38 °C for core temperature, but use
different limit values for cumulated sweat production with
7.5% (standard) vs. 3% (conservative) of body weight. So
the limits for sweat rate are a key feature of the observed
differences in estimated workability, which corroborate the
importance of providing measures enabling workers to com-
pensate for excessive sweat loss by liquid intake during heat
stress (Lundgren et al. 2014; Gao et al. this special issue), as it
is implicated by the standard limits (Malchaire et al. 2000).
Standard and liberal limits shared the sweat loss criterion, but
core temperature was allowed to increase to 38.2 °C in the
liberal case. This slightly higher core temperature limit, which
was also considered as an option when developing the stan-
dard criteria (Malchaire et al. 2000) and is applied by other
Fig. 8 Decline in rectal
temperature around the midday
break for the profiles with
continuous work and with a long
break, respectively, predicted by
PHS and UTCI-Fiala model with
low, moderate and high
workloads under non-shielded
climatic conditions in New Delhi.
Horizontal reference lines
indicate the standard and liberal
limit values for rectal temperature
heat stress indices (Miller and Bates 2007), substantially in-
creased workability estimated for high workload, e.g.
predicting almost unconstrained work in the warm-humid
conditions simulated here.
Differences between the WBGT-based metrics were due to
the Hothaps workability analysis representing actual averages
in field studies and the NIOSH and ISO/DIS estimates incor-
porating a Bsafety margin^ to protect heat sensitive workers.
The resulting deviations were generally limited, with Hothaps
restricting workability less compared to NIOSH or ISO/DIS.
This was more pronounced under high workload in hot cli-
mates (Table 2) with WBGT values in the range where
NIOSH and ISO/DIS are more protective (Fig. 2). For mod-
erate workload, WBGT indicated more constrained workabil-
ity in comparison to PHS and UTCI-Fiala with standard
limits. This is in line with the intended preventive features of
NIOSH and ISO/DIS as screening limits based on WBGT
(Parsons 2014; Olesen et al. 2016), which had been associated
with an overestimation of actual heat strain for moderate work
(Kampmann et al. 2012). That study (Kampmann et al. 2012)
had also shown that PHS tended to underestimate actual rectal
temperatures in the critical range of 38 °C, whereas UTCI-
Fiala provided unbiased predictions. This might explain the
large differences between PHS and UTCI-Fiala with standard
limits under high workload conditions (Table 1).
On the other hand, WBGT was less restrictive with high
workload and for afternoon work hours compared to PHS and
UTCI-Fiala. The latter might be partly related to the insuffi-
cient regard WBGT pays to constrained evaporation in humid
conditions under high workload (Claassen and Kok 2007;
Budd 2008) related to hidromeiosis (Candas et al. 1980), as
included here in the UTCI-Fiala model. It might also be ex-
plainable because WBGT assesses the actual thermal environ-
ment on an hourly basis independently of previous heat stress,
and thus neglects the cumulated physiological strain induced
by extended work hours (Venugopal et al. 2016), which is
captured by the physiological modelling approach.

Interestingly, this corresponds to lower productivity observed
during the fifth than during the first work hour in an agricul-
tural field study (Sahu et al. 2013).
Limitations and outlook
Our study utilised a simulation approach to compare the dif-
ferent heat stress indices and limit criteria. At least, the sce-
narios were based on meteorological records representing ex-
treme summer conditions, which are expected to occur more
frequently under climate change. We assessed workability in
relation to different heat stress tolerance limits resulting for
super-threshold strain levels in restricted working times,
which will directly influence the predicted productivity of
occupations with predominant muscular work. Hence, our
findings will probably be representative for the type of agri-
cultural or industrial work prevailing in developing and
emerging economies (Kjellstrom et al. 2009b). While this
study was based on simulations of theoretical work situations,
it would be interesting to compare our results with those ob-
tained in experiments carried out in real work places.
However, we could not identify any relevant publicly avail-
able datasets.
Estimating the practical consequences of climate change
settings on companies or economies in financial terms would
require linking the heat stress metrics with whole-year weather
scenarios and economic data characterising the agriculture or
industry sector and region of interest, which certainly is be-
yond the scope of this paper. But the strong dependency of
workability on workload found in this study underpins the
potential of reducing muscular demands by supporting tech-
nologies or modified work processes as a heat stress mitiga-
tion measure (Gao et al. this special issue). Furthermore, the
decrease in workability during the afternoon hours might sug-
gest avoiding extended working durations, e.g. by modified
shift schedules. One should note that currently, only PHS and
UTCI-Fiala would allow for assessing such duration related
measures, as WBGT does only consider the actual climate and
workload, but no cumulative time effects. Spending the break
periods at shaded, ventilated or cooled places would be an
additional option for heat stress mitigation. Its benefits could
in principle be assessed by all metrics introduced here, though
PHS might be considered with caution in this regard because
of its limitations in modelling dynamically changing condi-
tions mentioned above.
In addition, it is important to note that work re-
organisations and technical developments reducing muscular
workload often increase the requirements on sensorimotor and
cognitive skills which might be already impaired at body heat
storage and dehydration levels below the heat tolerance limits
considered here (Hancock and Vasmatzidis 2003; Lieberman
2007). As the effect sizes are divergent between different
types of cognitive tasks (Hancock and Vasmatzidis 2003;
Int J Biometeorol
Secher and Ritz 2012) and because there is a lack of studies
empirically relating cognitive work performance to heat stress,
our models could not cover such effects quantitatively.
Additional developments will emerge from the ongoing
European Heat-Shield project (www.Heat-Shield.eu), with
new estimates of workability risk functions still to be
published (Kjellstrom et al. this special issue), potentially in-
cluding the beneficial effects of heat stress mitigation mea-
sures (Gao et al. this special issue).
Though workability estimates from the nine metrics were
all positively correlated (Table 3), there were obvious differ-
ences between the heat stress assessment metrics, but their
evaluation was not possible in this study due to the lack of
independent validation data. For future inter-disciplinary cli-
mate change impact research, there is a need to develop robust
heat stress assessment metrics, which need to be easily
interpreted and meet the practical needs at workplaces. It
would be worthwhile if, in addition to projected climatic sce-
narios, predictive models also considered the consequences of
modified types or organisation of work, for them to become
also applicable to occupations with greater portions of mental
or knowledge work. We only identified two studies from the
mining (Wyndham 1969) and agricultural sector (Sahu et al.
2013) that quantified the impact of heat exposure levels on
labour productivity, and those data were the basis for the
Hothaps assessment metric presented in this study. Thus, a
key requirement will remain the availability of corre-
sponding empirical data for both model development
and validation purposes supported by rising open data
initiatives (Viola 2016).
Conclusions
In conclusion, our study showed that the choice of the heat
stress assessment metric impacts heavily on the estimated
work ability. Assessments based on WBGT gave lower work-
ability when applying heat safety criteria (NIOSH and ISO/
DIS) compared to using the average performance in field stud-
ies (Hothaps). The WBGT assessments forecast slightly stron-
ger effects in situations with moderate workload especially in
hot conditions compared to PHS and UTCI-Fiala. The latter,
however, were more sensitive in situations when working
heavily under high heat stress, because they could account
for cumulative physiological strain imposed by extended
work hours. Advanced thermophysiological models might
thus help developing alternative heat stress assessment met-
rics, where not only modelling details but also the choice of
physiological limit criteria will require attention. In addition,
more empirical research on work ability and productivity re-
lated to actual heat exposure data is urgently needed for model
validation and to provide improved evidence for climate
change impact assessments.
Int J Biometeorol
Acknowledgements Initial ideas to this study were developed during
the BInternational Workshop On Occupational Heat Exposure Indicators
For Use In Climate Change Impact Assessments^ held at Lund
University from 13 to 15 August 2012 with support from the Swedish
Council for Working Life and Social Research (FAS), the Centre for
Medicine and Technology for Working Life and Society (Metalund),
the Strategic Research Area Epidemiology for Health (Epihealth) and
the Swedish International Development Cooperation Agency (Sida).
Bruno Lemke’s input was supported by the internal research funds of
the Nelson-Marlborough Institute of Technology. Tord Kjellstrom’s con-
tribution was supported by the HEAT-SHIELD project under EU frame-
work programme Horizon 2020 (grant no. 668786). The authors are par-
ticularly grateful to Ms. Mina Sandusky for thorough and excellent lan-
guage editing of this manuscript.
Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of
interest.
References
Allegrini J, Carmeliet J (2017) Simulations of local heat islands in Zürich
with coupled CFD and building energy models. Urban Climate. doi:
10.1016/j.uclim.2017.02.003
Barwood M, Davey S, House J, Tipton M (2009) Post-exercise cooling
techniques in hot, humid conditions. Eur J Appl Physiol 107(4):
385–396. doi:10.1007/s00421-009-1135-1
Bates DM, Watts DG (1988) Nonlinear regression analysis and its appli-
cations. John Wiley & Sons, Inc., New York
Beggs PJ (2014) Climate change and biometeorology, the International
Society of Biometeorology and its journal: a perspective on the past
and a framework for the future. Int J Biometeorol 58(1):1–6. doi:10.
1007/s00484-013-0696-1
Blazejczyk K (2011) BioKlima v. 2.6—universal tool for bioclimatic and
thermophysiological studies. http://www.igipz.pan.pl/Bioklima-
zgik.html. Accessed 2016–02-12
Blazejczyk K, Baranowski J, Blazejczyk A (2014) Heat stress and occu-
pational health and safety—spatial and temporal differentiation.
Miscellanea Geographica—Regional Studies on Development
18(1):61–67. doi:10.2478/mgrsd-2014-0011
Bröde P, Fiala D, Blazejczyk K, Holmér I, Jendritzky G, Kampmann B,
Tinz B, Havenith G (2012) Deriving the operational procedure for
the Universal Thermal Climate Index (UTCI). Int J Biometeorol
56(3):481–494. doi:10.1007/s00484-011-0454-1
Bröde P, Blazejczyk K, Fiala D, Havenith G, Holmér I, Jendritzky G,
Kuklane K, Kampmann B (2013) The Universal Thermal Climate
Index UTCI compared to ergonomics standards for assessing the
thermal environment. Ind Health 51(1):16–24. doi:10.2486/
indhealth.2012-0098
Bröde P, Kampmann B, Fiala D (2016) Extending the universal thermal
climate index UTCI towards varying activity levels and exposure
times. In: Brotas L, Roaf S, Nicol F, Humphreys MA (eds) 9th
International Windsor Conference, Cumberland Lodge, Windsor,
UK, 7–10 April 2016. Network for Comfort and Energy Use in
Buildings, http://nceub.org.uk, London, pp 73–79
Budd GM (2008) Wet-bulb globe temperature (WBGT)—its history and
its limitations. J Sci Med Sport 11(1):20–32. doi:10.1016/j.jsams.
2007.07.003
Candas V, Libert J, Vogt J (1980) Effect of hidromeiosis on sweat
drippage during acclimation to humid heat. Eur J Appl Physiol
Occup Physiol 44(2):123–133. doi:10.1007/BF00421090
Cheuvront SN, Kenefick RW, Montain SJ, Sawka MN (2010)
Mechanisms of aerobic performance impairment with heat stress
and dehydration. J Appl Physiol 109(6):1989–1995. doi:10.1152/
japplphysiol.00367.2010
Chinevere T, Cadarette B, Goodman D, Ely B, Cheuvront S, Sawka M
(2008) Efficacy of body ventilation system for reducing strain in
warm and hot climates. Eur J Appl Physiol 103(3):307–314. doi:
10.1007/s00421-008-0707-9
Claassen N, Kok R (2007) The accuracy of the WBGT heat stress index at
low and high humidity levels. Occupational Health Southern Africa
13(2):12–18
d’Ambrosio Alfano FR, Malchaire J, Palella BI, Riccio G (2014) WBGT
index revisited after 60 years of use. Ann Occup Hyg 58(8):955–
970. doi:10.1093/annhyg/meu050
de Boor C (1978) A practical guide to splines. Springer-Verlag, New York
Dunne JP, Stouffer RJ, John JG (2013) Reductions in labour capacity
from heat stress under climate warming. Nature Clim Change 3(6):
563–566. doi:10.1038/nclimate1827
Fiala D (1998) Dynamic simulation of human heat transfer and thermal
comfort. PhD Thesis, De Montfort University, Leicester, UK
Fiala D, Havenith G (2016) Modelling human heat transfer and temperature
regulation. In: Gefen A, Epstein Y (eds) The mechanobiology and
mechanophysiology of military-related injuries. Springer International
Publishing, Cham, pp 265–302. doi:10.1007/8415_2015_183
Fiala D, Lomas KJ, Stohrer M (2001) Computer prediction of
human thermoregulatory and temperature responses to a wide
range of environmental conditions. Int J Biometeorol 45(3):
143–159. doi:10.1007/s004840100099
Fiala D, Havenith G, Bröde P, Kampmann B, Jendritzky G (2012) UTCI-
Fiala multi-node model of human heat transfer and temperature regula-
tion. Int J Biometeorol 56(3):429–441. doi:10.1007/s00484-011-0424-7
Gao C, Kuklane K, Holmér I (2011) Cooling vests with phase change
materials: the effects of melting temperature on heat strain allevia-
tion in an extremely hot environment. Eur J Appl Physiol 111(6):
1207–1216. doi:10.1007/s00421-010-1748-4
Gao C, Kuklane K, Östergren P-O, Kjellstrom T (in press) Occupational
heat stress assessment and protective strategies in the context of cli-
mate change. Int J Biometeorol. doi:10.1007/s00484-017-1352-y (this
special issue)
Gebhardt H, Kampmann B, Müller BH, Bux K (2009) Calculation of
cooling phases in warm and hot environments using the PHS-model.
Occup Ergon 8(4):195–204. doi:10.3233/OER-2009-0167
Ghiaus C, Allard F, Santamouris M, Georgakis C, Nicol F (2006) Urban
environment influence on natural ventilation potential. Build
Environ 41(4):395–406. doi:10.1016/j.buildenv.2005.02.003
Gosling SN, Warren R, Arnell NW, Good P, Caesar J, Bernie D, Lowe JA,
van der Linden P, O’Hanley JR, Smith SM (2011) A review of
recent developments in climate change science. Part II: the global-
scale impacts of climate change. Prog Phys Geogr 35(4):443–464.
doi:10.1177/0309133311407650
Hancock PA, Vasmatzidis I (2003) Effects of heat stress on cognitive
performance: the current state of knowledge. Int J Hyperth 19(3):
355–372. doi:10.1080/0265673021000054630
Hanna EG, Kjellstrom T, Bennett C, Dear K (2011) Climate change
and rising heat: population health implications for working
people in Australia. Asia Pac J Public Health 23(2 suppl):
14S–26S. doi:10.1177/1010539510391457
Havenith G, Fiala D (2016) Thermal indices and thermophysiological
modeling for heat stress. Comprehensive Physiology 6(1):255–
302. doi:10.1002/cphy.c140051
Havenith G, Fiala D, Blazejczyk K, Richards M, Bröde P, Holmér I,
Rintamaki H, Ben Shabat Y, Jendritzky G (2012) The UTCI-
clothing model. Int J Biometeorol 56(3):461–470. doi:10.1007/
s00484-011-0451-4
Hettinga F, De Koning J, de Vrijer A, Wüst R, Daanen H, Foster C (2007)
The effect of ambient temperature on gross-efficiency in cycling.

Eur J Appl Physiol 101(4):465–471. doi:10.1007/s00421-007-
0519-3
IPCC (2007) Summary for policymakers. In: Solomon S, Qin D,
Manning M et al (eds) Climate change 2007: the physical science
basis. Contribution of Working Group I to the Fourth Assessment
Report of the Intergovernmental Panel on Climate Change.
Cambridge University Press, Cambridge
ISO 7243 (1989) Hot environments; estimation of the heat stress on
working man, based on the WBGT-index (wet bulb globe tempera-
ture). International Organisation for Standardisation, Geneva
ISO 7933 (2004) Ergonomics of the thermal environment—analytical
determination and interpretation of heat stress using calculation of
the predicted heat strain. International Organisation for
Standardisation, Geneva
ISO/DIS 7243 (2015) Ergonomics of the thermal environment—assess-
ment of heat stress using the WBGT (wet bulb globe temperature)
index (ISO/DIS 7243:2015); German and English version prEN ISO
7243. Beuth Verlag, Berlin
Jacklitsch B, Williams WJ, Musolin K, Coca A, Kim J-H, Turner N (2016)
NIOSH criteria for a recommended standard: occupational exposure to
heat and hot environments. Cincinnati, http://www.cdc.gov/niosh/docs/
2016-106/pdfs/2016-106.pdf. Accessed 2016–03-17
Junge N, Jørgensen R, Flouris AD, Nybo L (2016) Prolonged self-paced
exercise in the heat—environmental factors affecting performance.
Temperature 3(4):539–548. doi:10.1080/23328940.2016.1216257
Kampmann B (2000) Zur Physiologie der Arbeit in warmem Klima.
Ergebnisse aus Laboruntersuchungen und aus Feldstudien im
Steinkohlenbergbau. Habilitation Thesis, Bergische Universität
Wuppertal, Wuppertal
Kampmann B, Bröde P, Fiala D (2012) Physiological responses to tem-
perature and humidity compared to the assessment by UTCI,
WGBT and PHS. Int J Biometeorol 56(3):505–513. doi:10.1007/
s00484-011-0410-0
Kjellstrom T, Crowe J (2011) Climate change, workplace heat exposure,
and occupational health and productivity in Central America. Int J
Occup Environ Health 17(3):270–281
Kjellstrom T, Gabrysch S, Lemke B, Dear K (2009a) The ‘Hothaps’ pro-
gramme for assessing climate change impacts on occupational health
and productivity: an invitation to carry out field studies. Glob Health
Action 2:10.3402/gha.v2i0.2082. doi:10.3402/gha.v2i0.2082
Kjellstrom T, Holmér I, Lemke B (2009b) Workplace heat stress, health
and productivity—an increasing challenge for low and middle-
income countries during climate change. Glob Health Action 2:10.
3402/gha.v2i0.2047. doi:10.3402/gha.v2i0.2047
Kjellstrom T, Kovats RS, Lloyd SJ, Holt T, Tol RSJ (2009c) The direct
impact of climate change on regional labor productivity. Archives of
Environmental & Occupational Health 64(4):217–227. doi:10.1080/
19338240903352776
Kjellstrom T, Lemke B, Otto M (2013) Mapping occupational heat ex-
posure and effects in South-East Asia: ongoing time trends 1980-
2011 and future estimates to 2050. Ind Health 51(1):56–67
Kjellstrom T, Lemke B, Otto M, Hyatt O, Dear K (2014) Occupational
heat stress: contribution to WHO project on BGlobal assessment of
the health impacts of climate change^. Climate Change Health
Impact & Prevention (ClimateCHIP, http://www.climatechip.org).
Mapua, New Zealand, http://climatechip.org/sites/default/files/
publications/TP2014_4_Occupational_Heat_Stress_WHO.pdf.
Accessed 2016–03-15
Kjellstrom T, Briggs D, Freyberg C, Lemke B, Otto M, Hyatt O (2016) Heat,
human performance, and occupational health: a key issue for the assess-
ment of global climate change impacts. Annu Rev Public Health 37(1):
97–112. doi:10.1146/annurev-publhealth-032315-021740
Kjellstrom T, Freyberg C, Lemke B, Otto M, Briggs D (in press)
Estimating population heat exposure and impacts on working people
in conjunction with climate change. Int J Biometeorol (this special
issue)
Int J Biometeorol
Konarska J, Lindberg F, Larsson A, Thorsson S, Holmer B (2014)
Transmissivity of solar radiation through crowns of single urban
trees—application for outdoor thermal comfort modelling. Theor
Appl Climatol 117(3):363–376. doi:10.1007/s00704-013-1000-3
Lemke B, Kjellstrom T (2012) Calculating workplace WBGT from me-
teorological data: a tool for climate change assessment. Ind Health
50(4):267–278. doi:10.2486/indhealth.MS1352
Lieberman HR (2007) Hydration and cognition: a critical review and
recommendations for future research. J Am Coll Nutr 26(5 Suppl):
555S–561S
Littell RC, Milliken GA, Stroup WW, Wolfinger RD, Schabenberger O
(2006) SAS® system for mixed models, Second edn. SAS Institute
Inc., Cary
Lundgren K, Kuklane K, Venugopal V (2014) Occupational heat stress and
associated productivity loss estimation using the PHS model (ISO
7933): a case study from workplaces in Chennai, India. Glob Health
Action 7:10.3402/gha.v3407.25283. doi:10.3402/gha.v7.25283
Malchaire J (1979) The TLV work-rest regimens for occupational expo-
sure to heat: a review of their development. Ann Occup Hyg 22(1):
55–62. doi:10.1093/annhyg/22.1.55
Malchaire J, Piette A (2004) Predicted Heat Strain (PHS) model compu-
tation programme written in Quick Basic: exact copy of the pro-
gramme of annex D of the ISO 7933. http://www.deparisnet.be/
chaleur/Normes/iso7933n.pdf. Accessed 2015–08-15
Malchaire J, Kampmann B, Havenith G, Mehnert P, Gebhardt JH (2000)
Criteria for estimating acceptable exposure times in hot working
environments: a review. Int Arch Occup Environ Health 73(4):
215–220. doi:10.1007/s004200050420
Matzarakis A, Rutz F, Mayer H (2007) Modelling radiation fluxes in simple
and complex environments—application of the RayMan model. Int J
Biometeorol 51(4):323–334. doi:10.1007/s00484-006-0061-8
Miller VS, Bates GP (2007) The thermal work limit is a simple reliable
heat index for the protection of workers in thermally stressful envi-
ronments. Ann Occup Hyg 51(6):553–561. doi:10.1093/annhyg/
mem035
Miller V, Bates G, Schneider JD, Thomsen J (2011) Self-pacing as a
protective mechanism against the effects of heat stress. Ann Occup
Hyg 55(5):548–555. doi:10.1093/annhyg/mer012
Nilsson M, Kjellstrom T (2010) Invited editorial: climate change impacts
on working people: how to develop prevention policies. Glob Health
Action 3:5774. doi:10.3402/gha.v3i0.5774
Oke TR (1987) Boundary layer climates, 2nd edn. Routledge, London
Olesen B, D’Ambrosio-Alfano FR, Parsons KC, Palella BI (2016) The
history of international standardization for the ergonomics of the
thermal environment. In: Brotas L, Roaf S, Nicol F, Humphreys
MA (eds) 9th International Windsor Conference, Cumberland
Lodge, Windsor, UK, 7–10 April 2016. Network for Comfort and
Energy Use in Buildings, http://nceub.org.uk, London, pp 15–38
Parsons KC (2014) Human thermal environments: the effects of hot,
moderate and cold environments on human health, comfort and
performance, 3rd edn. CRC Press, London
Sahu S, Sett M, Kjellstrom T (2013) Heat exposure, cardiovascular stress
and work productivity in rice harvesters in India: implications for a
climate change future. Ind Health 51(4):424–431. doi:10.2486/
indhealth.2013-0006
Sawka MN, Young AJ, Latzka WA, Neufer PD, Quigley MD, Pandolf
KB (1992) Human tolerance to heat strain during exercise: influence
of hydration. J Appl Physiol 73(1):368–375
Schlader ZJ, Stannard SR, Mündel T (2011) Evidence for thermoregula-
tory behavior during self-paced exercise in the heat. J Therm Biol
36(7):390–396. doi:10.1016/j.jtherbio.2011.07.002
Secher M, Ritz P (2012) Hydration and cognitive performance. J Nutr
Health Aging 16(4):325–329. doi:10.1007/s12603-012-0033-0
Shashua-Bar L, Hoffman ME (2000) Vegetation as a climatic component
in the design of an urban street: an empirical model for predicting the
Int J Biometeorol
cooling effect of urban green areas with trees. Energy and Buildings
31(3):221–235. doi:10.1016/S0378-7788(99)00018-3
Spector JT, Sheffield PE (2014) Re-evaluating occupational heat stress in
a changing climate. Ann Occup Hyg 58(8):936–942. doi:10.1093/
annhyg/meu073
Venugopal V, Chinnadurai J, Lucas RAI, Kjellstrom T (2016)
Occupational heat stress profiles in selected workplaces in India.
Int J Environ Res Public Health 13(1):89. doi:10.3390/
ijerph13010089
Viola R (2016) EU-US cooperation for a better usability of open data. The
digital single market blog. DG CONNECT, EU Commission,
Brussels https://ec.europa.eu/digital-single-market/en/blog/eu-us-
cooperation-better-usability-open-data. Accessed 2016-10-14
Whipp BJ, Wasserman K (1969) Efficiency of muscular work. J Appl
Physiol 26(5):644–648
WHO Scientific Group (1969) Health factors involved in working under
conditions of heat stress. Report of a WHO Scientific Group. World
Health Organ Tech Rep 412, http://apps.who.int/iris/bitstream/
10665/40716/1/WHO_TRS_412.pdf. Accessed 2016–05-25
Wyndham CH (1969) Adaptation to heat and cold. Environ Res 2(5–6):
442–469. doi:10.1016/0013-9351(69)90015-2
Xu X, Gonzalez JA, Santee WR, Blanchard LA, Hoyt RW (2016) Heat
strain imposed by personal protective ensembles: quantitative anal-
ysis using a thermoregulation model. Int J Biometeorol 60(7):1065–
1074. doi:10.1007/s00484-015-1100-0