Time-motion analysis as a novel approach for evaluating the impact of environmental heat exposure on labor loss in agriculture workers

RESEARCH PAPER

Time-motion analysis as a novel approach for evaluating the impact

of environmental heat exposure on labor loss in agriculture workers

Leonidas G. Ioannou

, Lydia Tsoutsoubi

, George Samoutis

, Lucka Kajfez Bogataj

, Glen P. Kenny

, Lars Nybo

Tord Kjellstrom

, and Andreas D. Flouris

FAME Laboratory, School of Exercise Science, University of Thessaly, Thessaly, Greece;

Medical School, University of Nicosia, Nicosia, Cyprus;

Biotehnical Faculty, University of Ljubljana, Ljubljana, Slovenia;

Human and Environmental Physiology Research Unit, School of Human

Kinetics, University of Ottawa, Ottawa, ON, Canada;

Department of Nutrition, Exercise and Sports, University of Copenhagen, Copenhagen,

Denmark;

Centre for Technology Research and Innovation (CETRI), Limassol, Cyprus

ARTICLE HISTORY

Received 4 April 2017

Revised 30 May 2017

Accepted 30 May 2017

ABSTRACT

Introduction: In this study we (i) introduced time-motion analysis for assessing the impact of

workplace heat on the work shift time spent doing labor (WTL) of grape-picking workers,

(ii) examined whether seasonal environmental differences can inﬂuence their WTL, and

(iii) investigated whether their WTL can be assessed by monitoring productivity or the vineyard

manager’s estimate of WTL. Methods : Seven grape-picking workers were assessed during the

summer and/or autumn via video throughout four work shifts. Results: Air temperature

(26.8 § 4.8



C), wet bulb globe temperature (WBGT; 25.2 § 4.1



C), universal thermal climate index

(UTCI; 35.2 § 6.7



C), and solar radiation (719.1 § 187.5 W/m

) were associated with changes in

mean skin temperature (1.7 § 1.8



C) (p < 0.05). Time-motion analysis showed that 12.4% (summer

15.3% vs. autumn 10.0%; p < 0.001) of total work shift time was spent on irregular breaks (WTB).

There was a 0.8%, 0.8%, 0.6%, and 2.1% increase in hourly WTB for every degree Celsius increase in

temperature, WBGT, UTCI, and mean skin temperature, respectively (p < 0.01). Seasonal changes in

UTCI explained 64.0% of the seasonal changes in WTL (p D 0.017). Productivity explained 36.6% of

the variance in WTL (p < 0.001), while the vineyard manager’s WTL estimate was too optimistic

(p < 0.001) and explained only 2.8% of the variance in the true WTL (p D 0.456). Conclusion: Time-

motion analysis accurately assesses WTL, evaluating every second spent by each worker during

every work shift. The studied grape-picking workers experienced increased workplace heat, leading

to signiﬁcant labor loss. Monitoring productivity or the vineyard manager ’s estimate of each

worker’s WTL did not completely reﬂect the true WTL in these grape-picking workers.

KEYWORDS

Europe; heat strain; heat

stress; irregular work break;

productivity; WBGT; UTCI

Introduction

Occupational heat stress is a parameter that inﬂuences

several industries worldwide. In the agriculture sector,

occupational heat stress is very relevant because many

tasks rely on manual work as the prevailing and,

sometimes, only feasible method for gentle handling

of vulnerable plants. Workplace heat is difﬁcult to

mitigate in agriculture, as artiﬁcial cooling or shadow-

ing toward the sun would be detrimental to growth or

directly harmful for the plants. The global wine indus-

try is large, comprising »0.2% of world gross domestic

product (GDP),

and climate change has increased the

areas suitable for growing vines. However, climate

change may have also increased the heat exposure in

the subtropical areas, where the majority of the wine

is produced.

This is noteworthy, because wine pro-

duction is still dominated by manual work. Some tasks

(e.g., transport of the grapes) have been motorized,

but the manual tasks by the grape-picking workers

remain similar to those developed 5,500 y ago when

the vine cultivation tradition emerged.

However,

given the increase in environmental temperature

during the past ﬁve millennia in regions such as

the Mediterranean,

the workers who currently pick

the grapes carry out their jobs under adverse environ-

mental conditions. In Cyprus, for instance, the mean

CONTACT Andreas D. Flouris andreasﬂ[email protected] FAME Laboratory, Department of Exercise Science, University of Thessaly, Karyes, Trikala

42100, Greece.

& Francis.

This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/by-nc-nd/

4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transfor med, or built upon in

any way.

TEMPERATURE

2017, VOL. 4, NO. 3, 330–340

https://doi.org/10.1080/23328940.2017.1338210

maximum temperature in August (the main part of

the harvest season) is around 36



These working

conditions would be considered a heat-wave in

Central European countries such as Germany.

Laboratory studies have shown that the above-

mentioned amount of thermal strain during grape

picking can impair the human physiologic function

and capacity to perform prolonged exercise and

work.

6-8

In addition, ﬁeld studies in real-life

occupational settings have shown that workplace heat

exposure impairs productivity.

9-13

These effects trans-

late to signiﬁcant ﬁnancial implications. Speciﬁcally,

the ﬁnancial costs due to heat-related work absentee-

ism and reduced productivity in Germany during

2004 were estimated between 0.7 and 3.1 billion EUR

(0.8–3.4 billion USD),

while similar calculations for

Australia during 2013/2014 indicated total losses of

5.6 billion EUR (6.2 billion USD).

Therefore, it is

vital to identify means that will allow employees to be

more productive despite using traditional work

practices in their natural habitat and – at the same

time – will protect their health and safety as well as

improve job satisfaction.

To date, there is very limited detailed and evidence-

based knowledge on the effects of workplace heat on

agriculture workers. Notably, the literature presents

no relevant studies conducted in Europe. Previous

studies in American

and Indian

agriculture workers

assessed the effects of workplace heat on productivity

by recording the amount of hourly individual-

harvested crops. The former study reported a non-

signiﬁcant trend of productivity reduction in fruit

harvesters due to increasing wet bulb globe tempera-

ture (WBGT), while the latter study found that the

productivity of rice harvesters was signiﬁcantly

decreased at WBGT above 26



C. It is important to

consider, however, that productivity may not neces-

sarily reﬂect the true work time spent on labor (WTL,

that is, the time that a worker spends toward complet-

ing a task, not including breaks or other work-related

behaviors) since it does not take into account differen-

ces in the yield of farm crops due to geological and/or

environmental effects. For instance, factors such as

land leveling and soil moisture can locally affect grape

production by up to 50% within the same vineyard.

Thus, productivity at a certain time of the day (e.g.,

when heat levels are at maximum) also reﬂects the

amount of crop that is available for picking at that

time which, as mentioned above, can vary by up to

50% within the same piece of land. This, in turn, can

distort the ﬁnal productivity assessment of the work-

ers, and the conclusion on the effect of workplace heat

on productivity. It is important, therefore, to

discriminate between productivity and labor.

Time-motion analysis combined with physiologic

data can be valuable in assessing the physiologic strain

of workers during periods of increased occupational

heat stress.

Time-motion analysis includes analyzing

movement and the time spent on each movement

through video analysis. This method is very powerful,

since the investigator can analyze every second spent

by each worker during every work shift, thus accu-

rately assessing WTL without bias. The primary aim

of this study was to introduce time-motion analysis

for assessing the impact of workplace heat on the

WTL during grape-picking. The latter is a moderate-

intensity work activity, which is self-paced and, hence,

allows for exploring both behavioral responses and

changes in work efﬁciency (time spent on work versus

breaks). The secondary aim was to examine if seasonal

differences in environmental conditions can inﬂuence

the WTL of these grape-picking workers. Finally, the

third aim was to investigate whether WTL of grape-

picking workers can be assessed by the previously-

adopted approaches of: (i) monitoring productivity, or

(ii) the vineyard manager’s estimate of each worker’s

WTL at the end of the work shift.

Materials and methods

The experimental protocol was approved by the

University of Thessaly, School of Exercise Science

Ethics Review Board in accordance with the Declara-

tion of Helsinki. The study involved monitoring

grape-picking workers on four separate days during

the 77-day harvest of 2016 (August 16 to October 31).

Speciﬁcally, the grape harvest season was divided into

two similar time periods as follows: summer (38 days;

August 16 to September 22) and autumn (39 days;

September 23 to October 31). The 4 days that were

selected for monitoring (summer: August 17–18;

autumn: October 11–12) were chosen as the prevailing

environmental conditions were representative for each

time period. Indeed, the average daily WBGT of the

two monitored summer days (24.5 § 0.8



C) and the

average daily WBGT of the entire summer period

(23.8 § 1.3



C) were similar (p > 0.05). Similarly, the

average daily WBGT of the two monitored autumn

TEMPERATURE 331

days (19.8 § 0.3



C) was similar (p > 0.05) to the

average daily WBGT of the entire autumn period

(19.5 § 0.8



C). The WBGT data used for these com-

parisons were recorded at the Paphos International

Airport, which is located approximately 7 km from

the monitored vineyard, and were provided by www.

wunderground.com.

Given the above-mentioned difference of 4



WBGT between the summer a nd autumn periods,

we calculated the minimum required sample based

on the productivity of rice harvesters at WBGT of

27–28



C (76.6 § 2.1 bundles/h) and 31–32



(56.4 § 2.9 bundles/h) from the aforementioned

study in Indian agriculture workers.

Using these

data, an effect size of 7.67 for the differences

between our summer and autumn seasons was

expected. Assuming an a of 0.05 and b of 0.95,

three participants would provide enough power to

detect a s tatistical difference of a similar magnitude



Power Version 3.1.9.2).

Based on these calcula-

tions, a total of seven healthy and heat-acclimatized

(i.e., continuously living in the area for the  90

previous days and performing other agric ulture jobs

on a daily basis) workers with experience in grape-

picking (8–16 y of work experience) volunteered

and were recruited for this study. The workers

worked in a new vineyard each day. For all recorded

shifts during autumn, two workers (on e male and

one female) were unavailable to participate because

they were working in a different vineyard and were

replaced by the vineyard manager by one m ale

worker. Therefore, the monitored group comprised

of six workers (four males and two females) during

the summer and ﬁve worke rs (four males and one

female) during the autumn. Given this change, all

season-related comparisons included data from

workers who were monitored d uring both summer

and autumn. Written informed consent was

obtained from all voluntee rs before their participa-

tion in the study.

Two days before the start of data collection, volun-

teers underwent a familiarization session that included

information regarding all data collection procedures.

Anthropometric characteristics were also recorded at

that time. Moreover, to minimize participant bias (i.e.,

WTL being affected because the workers were being

monitored), sham measurements were performed one

day before each seasonal assessment (i.e., on August

16 and October 10). The aim was for the workers to

become familiar with the presence of the investigators

at the workplace and the video recording. During

these sham measurements, all procedures were per-

formed in the same manner as during the actual

recording days.

During each recording and sham day, the workers

were monitored from the beginning (06:00 during the

summer; 07:00 during the autumn) until the end of

the work shift, about 8 h later. Speciﬁ cally, we cap-

tured workers’ movement via video and we recorded

skin temperature and environmental data throughout

the work shift. At the end of each work shift, we

recorded the vineyard manager’s estimation regarding

the WTL of each worker. No restrictions were placed

on water/food consumption or any other kind of

work- or non-work-related behavior. To ensure that

we did not inﬂuence the workers’ normal work rou-

tine, the temperature sensors used were miniature,

wireless, and were placed on the workers ’ body during

the previous night. Also, to minimize participant bias

(i.e., WTL being affected because the workers were

being monitored), the true reason for the video

recording was hidden from the volunteers. Instead,

they were informed that the investigators were

creating a video that would describe the process of

wine making. Of course, once the data collection was

completed, all volunteers were informed about the

true purpose of the video recording and gave their

permission to analyze and publish these data.

Anthropometric measurements included height,

mass, waist-to-hip ratio, and skinfold (abdominal, tri-

ceps, thigh, suprailiac) thickness. The latter was used

to calculate body fat percentage according to the

Jackson and Pollock equation.

Body surface area was

calculated using the Du Bois formula.

Maximal

oxygen uptake was estimated using a non-exercise

prediction model.

Workers’ movement was recorded using a video

camera (Hero 4 black, GoPro, California, USA). The

video camera was positioned in close proximity to the

workers (10–40 m) without interrupting their work-

ﬂow. Temperature at the skin surface was recorded

every second at four sites using iBUTTON sensors

(type DS1921 H, Maxim/Dallas Semiconductor Corp.,

USA) to calculate the mean skin temperature [T

; 0.3

(chest C arm) C 0.2 (thigh C leg)].

The difference

(DT

) between the baseline T

(i.e., T

at time 0) and

the current T

(i.e., T

at time t) was also calculated.

Environmental data including air temperature

332 L. G. IOANNOU ET AL.

air

), relative humidity, and wind speed were mea-

sured continuously using a portable weather station

(WMR200, Oregon, USA). The weather station was

positioned within the vineyard, 20–100 m away from

the workers and was installed according to the manu-

facturer’s instructions. Solar radiation (

sol

R) data were

collected from the ofﬁcial website of the European

Union’s Joint Research Center.

The obtained

environmental data were used to calculate WBGT,

and the universal thermal climate index (UTCI) using

the Excel Heat Stress Calculator downloaded from

www.ClimateCHIP.org.

This calculator uses the

recommended method of Liljegren to calculate the

outdoor WBGT from meteorological data,

25,26

as well

as information from the ofﬁcial UTCI webpage

27,28

estimate the UTCI.

Time-motion analysis

The video recordings were used to identify work-

related behaviors. Work time spent on irregular work

breaks (WTB) was deﬁned as any unprescribed work

cessation determined by workers’ own judgment, and

not based on speciﬁc time intervals or instructions.

Lunch time was not considered as WTB because it

was prescribed by management. We also recorded the

duration of uninterrupted WTL (i.e., continuous work

without break) to delineate the impact of workplace

heat on the frequency of breaks, which is different

from the duration of WTB. It should be noted that the

workers had constant access to shade either from the

vine trees themselves or from other trees that, in some

areas, surrounded the vineyard. Thus, the WTB was

divided into two categories: the WTB during which

the workers decided to rest in the shade (WTB

shade

)

and the WTB during which the workers chose to stay

under the sun (WTB

sun

). Based on these deﬁnitions,

the following ﬁve work-related behaviors were identi-

ﬁed in the time-motion analysis: (i) WTL, (ii) uninter-

rupted WTL, (iii) WTB

shade

, (iv) WTB

sun

, and (v)

lunch (always taken outdoors). In addition, the video

recordings were used to calculate productivity using

previous methodology [i.e., the number of boxes full

of grapes picked by the entire group in each hour

divided by the number of workers (six workers during

the summer and ﬁve workers during the autumn)].

Work-related behaviors were determined for each

worker individually through time-motion analysis

that was conducted off site by two trained

analysts. The video recordings (60 min £ 8-h work

shift £ 4 days D 1,920 min total recorded over the

four monitored days) were analyzed within a 2-week

period. To minimize the risk of errors due to fatigue,

each analyst analyzed 6 h of video per day, while

taking a 1-h break every 2 h of time-motion analysis.

Due to the necessary pauses to record work-related

behavior, the overall ratio of video recording to time-

motion analyzing was 1:1.33 (i.e., for every 1 h of

video recording, 80 min were required for analysis).

Experimenter bias was minimized via training the

analysts to rate using the same short video to ensure

adequate agreement. For the same reason, the two

analysts worked in the same room and they were

instructed to seek each other’s advice in cases where

they could not make a ﬁrm decision on their own.

They were, thus, encouraged to give consensus group

ratings of work-related behaviors.

Data analysis

The T

air

, WBGT, and UTCI were grouped into

5-degree categories (< 15



C, 15–20



C, 20–25



25–30



C, 30–35



C, 35–40



C, 40–45



C, and >45



C),

while the DT

was grouped into ten 1-degree catego-

ries (<¡2



C, ¡2to¡1



C, ¡1to0



C, 0 to 1



C, 1 to



C, 2 to 3



C, 3 to 4



C, 4 to 5



C, 5 to 6



C, and >6



C).

Finally,

sol

R was divided into groups of 200 W/m

(0–

200 W/m

, 200–400 W/m

, 400–600 W/m

, 600–

800 W/m

, and >800 W/m

To address the primary aim of the study, Pearson’s

correlations were used to detect whether the hourly

WTB and productivity were associated with the hourly

air

, WBGT, UTCI,

sol

R, DT

, and T

. One-way x

was conducted to detect prevalence differences among

the four (i.e., excluding lunch) work-related behaviors.

Moreover, two-way x

analysis was used to detect

prevalence differences of the four work-related behav-

iors among the afore-mentioned categories of T

air

WBGT, UTCI, and

sol

To address the secondary aim of the study, inde-

pendent samples t-tests were used to assess differences

in hourly T

air

, WBGT, UTCI, D T

, T

, and WTB

between summer and autumn. A two-way x

was used

to detect prevalence differences between the types of

WTB (WTB

shade

and WTB

sun

) across the two seasons.

Furthermore, two stepwise multiple linear regression

analyses incorporating backward elimination at the

0.051 level were used to examine if seasonal

TEMPERATURE 333

differences in environmental conditions (T

air

sol

R),

heat stress indices (WBGT, UTCI), T

, and DT

(all

independent variables) can predict seasonal changes

in WTL or WTB (dependent variables). This stepwise

procedure allows for careful assessment of each inde-

pendent variable through the change in the coefﬁcient

of determination (R

To address the third aim of the study, simultaneous

linear regression analysis was used to predict WTL

during the previous work hour (dependent variable)

based on hourly productivity or the vineyard manag-

er’s estimate of each worker’s WTL (independent vari-

ables). Finally, a paired samples t-test was used to

investigate whether the vineyard manager’s estimate

of each worker’s WTL was similar to the actual WTL

(as assessed via time-motion analysis). The level of

signiﬁcance for all analyses was set at p < 0.05. All sta-

tistical analyses were conducted using SPSS 22.0 for

Windows (IBM, Armonk, NY, USA). All results are

presented as mean § sd, unless otherwise stated.

Results

The physical characteristics of all volunteers are

shown in Table 1. The slight differences between the

workers monitored in the summer and in autumn

resulted in no signiﬁcant differences in any of the

physical characteristics recorded (all p’s > 0.05).

There were no relationships between the physical

characteristics of the workers and the hourly WTB (all

p’s > 0.05). During the studied work shifts, the work-

ers harvested a total of 9,600 kg of grapes. The crop

yield per m

of land was 281 § 4 g of grapes. Produc-

tivity (2.2 § 1.0 boxes) ﬂuctuated from 0.0 to 4.0

boxes of grapes (each box weighed 25 kg), depending

on the type of work done at a given time.

The T

air

(mean: 26.8 § 4.8



C; range: 15.2–33.9



C),

WBGT (mean: 25.2 § 4.1



C; range: 14.0–32.8



C),

UTCI (mean: 35.2 § 6.7



C; range: 10.6–45.7



C), and

sol

R (mean: 719.1 § 187.5 W/m

; range: 0–906 W/m

)

ﬂuctuated considerably throughout the study period.

Large ﬂuctuations were also observed in T

(mean:

33.7 § 1.6



C; range: 28.2–37.5



C) and DT

(mean:

1.7 § 1.8



C; range: ¡2.6 to 6.9



C). The percentage of

work shift time spent in each category of environmen-

tal factors, heat stress indices, and DT

is shown in

Table 2. As suspected, the majority of the work shift

time during the summer period was spent under con-

ditions of high heat stress. Speciﬁcally, 56.4% of work

shift time during the summer was spent at T

air

>30



87.3% of time was spent at WBGT >25



C, 80.4% of

time was spent at UTCI >35



C, and 63.5% was spent

sol

R >800 W/m

. These results may explain the fact

that the workers spent 57.7% of the work shift time

during the summer having DT



C. During

autumn, the environment was much more temperate.

Only 1.1% of the work shift time was spent at

air

>30



C, 8.2% of time was spent at WBGT >25



19.2% of time was spent at UTCI >35



C, and no time

was spent at

sol

R >800 W/m

. During both seasons,

air

: r D 0.90; WBGT: r D 0.76; UTCI: r D 0.83;

sol

R: r D 0.81) and DT

air

: r D 0.93; WBGT:

r D 0.76; UTCI: r D 0.84;

sol

R: r D 0.87) were posi-

tively correlated with the recorded environmental fac-

tors and heat stress indices (all p’s < 0.05).

Analyses conducted to address the ﬁrst aim of the

study demonstrated that, across the entire study

period, 12.4% of the total work shift time was lost on

WTB (WTB

sun

: 8.1%; WTB

shade

: 4.3%). The hourly

percentage of work shift time spent on WTB ranged

from 0% to 76.1%. The majority of WTB was observed

in the middle 4-h period of the work shift (p < 0.001;

Fig. 1). The mean duration of WTB [01:10 § 05:11

(min:sec)] signiﬁcantly increased, while the mean

duration of uninterrupted labor [05:23 § 09:43 (min:

sec)] signiﬁcantly decreased at higher levels of WBGT,

UTCI,

sol

R, and DT

(all p’s <0.05; Figs. 2 and 3), sug-

gesting that workplace heat stress leads to increased

frequency and duration of work breaks. This was also

conﬁrmed via Pearson’s correlation where the hourly

WTB was positively associated with the recorded envi-

ronmental factors and the DT

air

: r D 0.38; WBGT:

r D 0.39; UTCI: r D 0.40;

sol

R: r D 0.32; DT

: r D 0.29;

all p’s < 0.01). Furthermore, the total work shift time

lost due to WTB was signiﬁcantly increased at higher

levels of T

air

, WBGT, UTCI,

sol

R, and DT

(all p’s <

0.01; Fig. 4). Simultaneous linear regression analyses

Table 1. Physical characteristics (mean § sd) of the studied

grape-picking workers.

Males (n D 5) Females (n D 2)

Age (years) 39.0 § 10.8 39.5 § 13.4

Height (cm) 171.9 § 2.9 170.0 § 0.0

Mass (kg) 79.1 § 14.1 61.7 § 12.2

Body mass index (kg/m

) 25.7 § 4.8 24.1 § 4.8

Body surface area (m

) 1.9 § 0.2 1.7 § 0.1

Waist to hip ratio (cm) 0.94 § 0.08 0.83 § 0.10

max (ml/kg/min) 37.6 § 7.0 36.6 § 1.5

Body fat (%) 15.6 § 6.4 20.3 § 2.5

Note: No signiﬁcant group differences were observed between summer and

autumn (p > 0.05).

334 L. G. IOANNOU ET AL.

demonstrated that there is a 0.8%, 0.8%, 0.6%, and

2.1% increase in WTB for every 1



C increase in T

air

D 0.47; F

(1,18)

D 14.97, p D 0.001), WBGT (R

0.37; F

(1,18)

D 10.01, p D 0.006), UTCI (R

D 0.69;

(1,34)

D 74.18, p < 0.001), and D T

D 0.91;

(1,9)

D 83.36, p < 0.001), respectively.

Analyses conducted to address the second aim con-

ﬁrmed the existence of signiﬁcant seasonal differences

in environmental conditions (Table 2). Indeed, all

environmental parameters and heat stress indices

were signiﬁcantly increased during the summer com-

pared with the autumn (T

air

: 28.9 § 4.2



C versus

24.3 § 4.2



C; WBGT: 27.9 § 2.5



C versus 21.9 §

3.0



C; UTCI: 38.8 § 4.9



C versus 30.8 § 5.8



sol

Table 2. Percentage of work shift time spent in each category of environmental factors, heat stress indices, and DT



C <15 15–20 20–25 25–30 30–35 35–40 40–45 >45

air

Summer 0% 22.3% 21.3% 56.4%

Autumn 21.1% 20.2% 57.6%

1.1%

WBGT Summer 0%

1.3% 11.4% 68.0% 19.3%

Autumn 4.7% 18.1%

69.0%

8.2%

UTCI Summer 0% 0.5% 0.8% 6.2% 12.1% 29.5%

50.2% 0.7%

Autumn 3.4% 5.9%

3.6%

17.1%

50.8%

18.7%

0.5%

W/m

0–200 200–400 400–600 600–800 >800

sol

R Summer 1.3% 3.2%

12.8%

19.2%

63.5%

Autumn 5.3% 6.4%

12.8%

75.5%



C < ¡ 2 ¡2to¡1 ¡1to0 0–11–22– 33–44–55–6 6 to 7

Summer 0% 1.6% 11.2%

14.0%

15.5%

22.6%

17.6% 12.6% 3.4%

1.5%

Autumn 1.6% 10.6%

17.1%

22.4%

26.3%

12.9%

5.0%

3.7%

0.3%

0.1%



Difference from category to the left within the same season statistically signiﬁcant at p < 0.05.

Difference from autumn within the same category statistically signiﬁcant at p < 0.05.

Note: T

air

D air temperature (



C); WBGT D wet bulb globe temperature (



C); UTCI D universal thermal climate index (



C);

sol

R D solar radiation (W/m

);

D difference between the baseline mean skin temperature (i.e., at time 0) and the current mean skin temperature (



C).

Figure 1. Hourly percentage (mean § sd) of work time spent on

irregular work breaks (WTB) during each hour (local Cyprus time)

of the recorded work shifts. Blue bars show results from all the

studied work shifts. Red and green bars show results from all the

summer and autumn work shifts, respectively. Asterisks indicate

signiﬁcant (p < 0.05) differences from the work hour to the left.

The lunch break (which is not included in the WTB) was always

taken at 11:00 and lasted for 30 min.

Figure 2. Mean duration of uninterrupted labor (dotted lines cor-

responding to the left vertical axis) and mean work time spent

on irregular breaks (WTB; bars corresponding to the right vertical

axis) based on the UTCI (blue color) and WBGT (green color) cate-

gories. Asterisks indicate signiﬁcant (p <0.05) differences from

the UTCI or WBGT category to the left. The reference values for

UTCI are as follows: 9–26



C: no thermal stress; 26–32



C: moder-

ate heat stress; 32–38



C: strong heat stress; 38–46



C: very strong

heat stress; > 46



C: extreme heat stress. The reference values for

WBGT are as follows:  25.6–27.7



C: no heat stress; 27.8–29.4



low heat stress; 29.4 –31.0



C: moderate heat stress; 31.0– 32.1



high heat stress;  32.2



C: extreme heat stress.

TEMPERATURE 335

773.7 § 177.0 W/m

versus 653 § 178.6 W/m

; all

p’s<0.05). It is logical to assume that these differences

generated the observed signiﬁcantly increased summer

mean hourly T

(34.3 § 1.4



C versus 32.7 § 1.5



and DT

(2.3 § 1.7



C versus 0.7 § 1.4



C) compared

with the autumn (all p’s < 0.01; Fig. 5). During the

summer, 15.3% of the total work shift time was lost on

WTB, while only 10% of work shift time was lost on

WTB during the autumn (p D 0.004). This was despite

the fact that the crop yield per m

of land was similar

in the two seasons (summer: 279 g; autumn: 285 g).

The majority of WTB was observed in the middle 4-h

period of the work shift during both summer

(08:00–12:00) and autumn (09:00–13:00) (p > 0.05;

Fig. 1). Interestingly, while the majority of WTB was

spent under the sun during both the summer (56.2%)

and the autumn (75.3%) (both p’s <0.001), the time

spent in WTB

shade

was nearly twice as large during the

summer (43.8%) compared with the autumn (24.7%)

(p < 0.001). Also, we found no signiﬁcant differences

in productivity between the summer (2.3 § 1.2 boxes)

and the autumn (2.0 § 0.8 boxes) (p > 0.05). Back-

ward linear regression analysis demonstrated that sea-

sonal changes in UTCI account for 64% of the

seasonal changes in WTL [F

(1,7)

D 10.63, p D 0.017].

Our analyses addressing the third aim showed no

signiﬁcant association between productivity and either

the recorded environmental factors, or the heat stress

indices, or the DT

(both p’s > 0.05). Productivity

was positively correlated with the WTL during the

previous work hour (r D 0.61, p < 0.001; Fig. 6). A

simultaneous linear regression analysis showed that

productivity accounts for 36.6% of the variance in

WTL during the previous work hour [F

(1,31)

D 17.34,

p < 0.001]. Finally, the vineyard manager felt that the

workers performed their jobs correctly, at the usual

pace, and at a level that was acceptable throughout the

monitored work shifts. This is indicated by his esti-

mate for each worker’s WTL which was 95.0 § 5.1%.

However, this estimate was signiﬁcantly higher (p <

0.001) than the actual WTL assessed via time-motion

analysis (87.4 § 4.5%). A simultaneous linear regres-

sion analysis demonstrated that the manager’s esti-

mate of WTL predicted 2.8% of the variance in the

actual WTL [between workers or between work shifts

for the same worker; F

(1,21)

D 0.577, p D 0.456].

Discussion

Toourknowledge,thisistheﬁrst study to use

time-motion analysis to discriminate between pro-

ductivity and labor when assessing the impacts of

workplace heat. Time -motion analysis is a qualita-

tive method that has been used for some time for

the assessment of physical performance and work

intensity in e lite team sports.

29,30

In occupational

settings, it is one of the recommended methods by

the American Conference o f Governmental Indus-

trial Hygienists to assess hand activity level

Figure 3. Mean duration of uninterrupted labor (dotted lines cor-

responding to the left vertical axis) and mean work time spent

on irregular breaks (WTB; bars corresponding to the right vertical

axis) based on the DT

(beige color) and the solar radiation (red

color) categories. Asterisks indicate signiﬁcant (p < 0.05) differen-

ces from the DT

or solar radiation category to the left. Note:

D difference between the baseline mean skin temperature

and the current mean skin temperature.

Figure 4. Loss of labor for each category of environmental fac-

tors, heat stress indices, and DT

. Each full gray body ﬁgure rep-

resents one work shift lost per ten work shifts due to work time

spent on irregular breaks (WTB). Note: T

air

D air temperature;

WBGT D wet bulb globe temperature; UTCI D universal thermal

climate index;

sol

R D solar radiation; DT

D difference between

the baseline mean skin temperature and the current mean skin

temperature.

336 L. G. IOANNOU ET AL.

threshold limit values,

and has been adopted to

determine the d egree of correlation between hand

activity/force ratings in various jobs.

Moreover,

we recently used this technique to assess the phys i-

ologic strain of electrical utilities workers in hot

regions of North America.

In the pr esent study,

we extended this approach to labor assessment,

and we strengthened the methodology by applying

stricter rules and minimizing participant and exper-

imenter bias. Introducing time-motion analysis as a

tool to assess the effects of workplace heat on pro-

ductivity and labor effort in agriculture workers

was the main aspect of this study. As such, this

paper should be considered as a method developing

paper rather than an exhaustive (large scale) study

of agriculture workers which would require a

broader study (more worker s and, possibly , differ -

ent loca tions). Though time consuming, time-

motion analysis is very powerful, since the investi-

gator can analyze every second spent by each

worker during every work shift, thus accurately

assessing labor and productivity wit hout bias. This

is supported by the fact that the present WTL and

WTB results were sensitive enough to detect differ-

ences betwee n workers, days, as well as seasons

despite the relatively small number of worker s

assessed.

To the best of our knowledge, this is the ﬁrst study to

investigate the impact of workplace heat on a group of

European agriculture workers. We chose to study

grape-picking workers because wine is one of the most

important European agricultural products

and

because these workers carry out manual labor

under

adverse environmental conditions. Indeed, our study

shows that most of the work shift time during the sum-

mer—the main part of the grape-picking period—was

spent under conditions of increased heat (T

air

>30



or WBGT >25



C, or UTCI >35



C, or

sol

R >800 W/

) and was accompanied by signiﬁcant changes in

. Our study also showed that WTB comprised

12.4% of the total work time during the recorded

Figure 5. Fluctuation in mean skin temperature (mean § sd) of the studied workers across the 8-h work shifts (local Cyprus time) during

the summer (left pane) and the autumn (right pane) study periods. The background images illustrate the same worker picking grapes

during the summer (left) and the autumn (right). Seasonal comparisons demonstrated signiﬁcant differences (p < 0.05) during the

majority of the work shift time.

Figure 6. Fluctuation (mean § sd) in the true work time spent on

labor (WTL) during the previous hour (left vertical axis; purple

bars) and the hourly productivity (right vertical axis; orange bars)

of the studied workers across the 8-h work shift (local Cyprus

time). Asterisks indicate signiﬁcant differences (p < 0.05) from

work hour to the left.

TEMPERATURE 337

grape-picking shifts. This loss of labor time was

increased (i.e., longer and more frequent work breaks)

at higher levels of WBGT, UTCI,

sol

R, and DT

, with as

much as 26.7% of the work shift time spent on breaks

at the highest recorded levels of heat stress (i.e., DT

6–7



C). Overall, we found that there is a 0.8%, 0.8%,

0.6%, and 2.1% increase in WTB for every one degree

Celsius increase in T

air

, WBGT, UTCI, and DT

respectively. It is important to also note that, following

recent suggestions,

we evaluated heat stress using sev-

eral parameters and different indices to ensure a com-

prehensive and systematic approach.

We found signiﬁcant seasonal differences in envi-

ronmental conditions, heat stress indices, T

,and

. Importantly, we also detected notable labor dif-

ferences between seasons, as more than 15% of the

total work shift time was lost on WTB during the sum-

mer, while this loss was restricted to 10% during the

autumn. Overall, seasonal changes in UTCI accounted

for 64% of the seasonal changes in WTL. Also, we

observed that the WTB

shade

was nearly twice as large

during the summer compared with the autumn, indi-

cating that workers were behaviorally thermoregulating.

This was anticipated, given the signiﬁcant changes seen

in T

, which is the main driver of behavioral thermo-

regulation.

6,35

Indeed, environmental and metabolic

heat gain lead to increased T

, which augments ther-

mal perception and/or cardiovascular strain. In turn,

this leads to increased perceived exertion that, eventu-

ally, causes reductions in work rate.

Our results demonstrating that workplace heat is

accompanied by signiﬁcant labor loss in the studied

agriculture workers are in line with previous larger

occupational studies in the literature reporting similar

effects on the productivity of workers in the agricul-

ture industry.

9,10

Agricultural workers, and especially

grape-picking workers, cannot avoid workplace heat

exposure. Based on our calculations, our workers per-

formed moderate-intensity work with an average

work rate of »330 W for men and »260 W for

women (calculation based on known energy expendi-

ture for “farming, ﬁshing, and forestry”).

As such,

work-rest ratios should be 3:1, 1:1, and 1:3 at WBGT

 29



C,  30.5



C, and  32



C, respectively.

This

would have important implications for the studied

grape-picking workers who spent 35.7%, 7.9%, and

1.6% of their work time at WBGT  29



C,  30.5



and  32



C, respectively, but were ofﬁcially not pro-

vided with any breaks other than the lunch break.

The results of the present study suggest that moni-

toring productivity or the vineyard manager’s estimate

of each worker’s WTL at the end of the work shift does

not completely reﬂect the workers’ true labor effort

during grape-picking. Indeed, productivity explained

»37% of the variance in WTL during the previous

work hour (or vice versa), while the vineyard manag-

er’s estimate of WTL was too optimistic and explained

only »3% of the variance in the true WTL assessed

via time-motion analysis. The differences between

productivity, the manager’s estimate of WTL, and the

true WTL assessed via time-motion analysis may

reﬂect the impact of geological factors that are known

to inﬂuence the grape yield by up to 50%, within the

same vineyard.

It is possible that these factors may

be less important in other agricultural jobs such as

rice

or tree fruit

harvesting. Nevertheless, we rec-

ommend that future studies should discriminate

between productivity and labor, possibly by using the

time-motion analysis technique used in the present

study or other similar methods.

The present results would be strengthened by assess-

ing a larger sample of workers and/or separate groups of

workers in different European countries, and by record-

ing also core temperature and heart rate. Studies in this

direction should be planned in the coming years aiming

to protect workers’ health and to preserve their labor

capacity and productivity despite the projected increased

frequency of heat waves due to climate change. More-

over, it is important to note that we did not control work-

ers’ clothing to ensure that we did not inﬂuence their

normal work routine and to minimize participant bias.

Based on the present results, it is concluded that

time-motion analysis accurately assesses labor effort,

evaluating every second spent by each worker during

every work shift. Moreover, the studied grape-picking

workers experienced high levels of workplace heat

which was associated with signiﬁcant loss of labor

effort reaching as high as 27%. Finally, monitoring

productivity or the vineyard manager’s estimate of

each worker’s labor effort at the end of the work shift

did not completely reﬂect the studied workers’ true

labor effort during grape-picking.

Abbreviations

difference between the baseline mean skin

temperature (i.e., at time 0) and the cur-

rent mean skin temperature (



338 L. G. IOANNOU ET AL.

sol

R solar radiation (W/m

)

air

air temperature (



mean skin temperature (



UTCI universal thermal climate index (



WBGT wet bulb globe temperature (



WTL work time spent on labor (time, min:sec)

WTB work time spent on irregular work breaks

(time, min:sec)

WTB

shade

work time spent on irregular breaks in the

shade (time, min:sec)

WTB

sun

work time spent on irregular work breaks

under the sun (time, min:sec)

Disclosure of potential conﬂicts of interest

There are no relevant ﬁnancial or other relationships that

might be perceived as leading to a conﬂict of interest in rela-

tion to this work. The funders had no role in study design,

data collection and analysis, decision to publish, or preparation

of the manuscript.

Acknowledgments

The authors are grateful to the volunteers and the vineyard

management for their participation in this study.

Funding

The study has received funding from the European Union’s

Horizon 2020 research and innovation programme under the

Grant agreement no. 668786.

ORCID

Lars Nybo http://orcid.org/0000-0002-9090-1958

Andreas D. Flouris

http://orcid.org/0000-0002-9823-3915

References

[1] Anderson K, Nelgen S. Global wine markets, 1961 to

2009: a statistical compendium. Adelaide, Australia: The

University of Adelaide Press, 2011.

[2] Falcone G, De Luca AI, Stillitano T, Strano A, Romeo G,

Gulisano G. Assessment of environmental and economic

impacts of vine-growing combining life cycle assessment,

life cycle costing and multicriterial analysis. Sustainability.

2016;8(8):793. PMID:28008371; doi:10.3390/Su8080793

[3] Finne M, Holmgren K, Sundqvist HS, Weiberg E, Lindblom

M. Climate in the eastern Mediterranean, and adjacent

regions, during the past 6000 years – a review. J Archaeol

Sci. 2011;38(12):3153-3173. doi:10.1016/j.jas.2011.05.007

[4] Price C, Michaelides S, Pashiardis S, Alpert P. Long term

changes in diurnal temperature range in Cyprus. Atmos

Res. 1999;51(2):85-98. doi:10.1016/S0169-8095(99)00022-8

[5] Heudorf U, Schade M. Heat waves and mortality in

Frankfurt am Main, Germany, 2003–2013: what effect do

heat-health action plans and the heat warning system

have? Z Gerontol Geriatr. 2014;47(6):475-482. PMID:

25119702; doi:10.1007/s00391-014-0673-2

[6] Flouris AD, Schlader ZJ. Human behavioral thermoregu-

lation during exercise in the heat. Scand J Med Sci Sports.

2015;25(Suppl 1):52-64. PMID:25943656; doi:10.1111/

sms.12349

[7] Nybo L, Rasmussen P, Sawka MN. Performance in the

heat-physiological factors of importance for hyperther-

mia-induced fatigue. Compr Physiol. 2014;4(2):657-689.

PMID:24715563; doi:10.1002/cphy.c130012

[8] Junge N, Jorgensen R, Flouris AD, Nybo L. Prolonged

self-paced exercise in the heat – environmental factors

affecting performance. Temperature. 2016;3(4):539-548.

PMID:28090557; doi:10.1080/23328940.2016.1216257

[9] Quiller G, Krenz J, Ebi K, Hess JJ, Fenske RA, Sampson

PD, Pan M, Spector JT. Heat exposure and productivity

in orchards: Implications for climate change research.

Arch Environ Occup Health. 2017:1–4. PMID:28139172;

doi:10.1080/19338244.2017.1288077

[10] Sahu S, Sett M, Kjellstro m T. Heat exposure, cardio-

vascular stress and work productivity in rice harvest-

ers in India: i mplic ations for a climate change future.

Ind Health. 2013;51(4):424-431. PMID:23685851;

doi:10.2486/indhealth.2013-0006

[11] Lundgren K, Kuklane K, Gao CS, Holmer I. Effects of

heat stress on working populations when facing climate

change. Ind Health. 2013;51(1):3-15. PMID:23411752;

doi:10.2486/indhealth.2012-0089

[12] Kjellstrom T, Holmer I, Lemke B . Workplac e heat

stress, health and productivity – an inc reasing chal-

lenge for low and middle-income countries during cli-

mate change. Glob Health Action. 2009; 2. PMID:

20052422; doi:10.3402/gha.v2i0.2047

[13] Jay O, Brotherhood JR. Occupational heat stress in Aus-

tralian workplaces. Temperature. 2016;3(3):394-411.

PMID:28349081; doi:10.1080/23328940.2016.1216256

[14] Hubler M, Klepper G, Peterson S. Costs of climate

change: the effects of rising temperatures on health and

productivity in Germany. Ecol Econ. 2008;68(1–2):381-

393. doi:10.1016/j.ecolecon.2008.04.010

[15] Zander KK, Botzen WJW, Oppermann E, Kjellstrom T,

Garnett ST. Heat stress causes substantial labour produc-

tivity loss in Australia. Nat Clim Change. 2015;5(7):647.

doi:10.1038/Nclimate2623

[16] Ramos MC, Martınez-Casasnovas JA. Impact of land

levelling on soil moisture and runoff variability in vine-

yards under different rainfall distributions in a Mediter-

ranean climate and its inﬂuence on crop productivity. J

Hydrol. 2006;321:131-146. doi:10.1016/j.jhydrol.2005.07.

055

[17] Meade RD, Lauzon M, Poirier MP, Flouris AD, Kenny

GP. The physical demands of electrical utilities work in

North America. J Occup Environ Hyg. 2016;13(1):60-70.

PMID:26317802; doi:10.1080/15459624.2015.1077966

TEMPERATURE 339

[18] Faul F, Er dfelder E, Lang AG, Buchner A. G



Power 3: a

ﬂexible statist ical power analysis program for the social,

behavioral, and biomedical sciences. Behav Res Meth-

ods. 2007;39(2):175-191. PMID:17695343; doi:10.3758/

BF03193146

[19] Jackson AS, Pollock ML. Practical assessment of body

composition. Phys Sportsmed. 1985;13(5):76-90. PMID:

27463295; doi:10.1080/00913847.1985.11708790

[20] Du Bois D, Du Bois EF. A formula to estimate the

approximate surface area if height and weight be known.

Arch Int Med. 1916;17(6):863-871. doi:10.1001/archi

nte.1916.00080130010002

[21] Jackson AS, Blair S N, Ma har MT, Wier LT, Ross RM,

Stuteville JE. Prediction of functional aerobic capacity

without exercise testing. Med Sci Sports Exerc.

1990;22(6):863-870. PMID:2287267; doi:10.1249/

00005768-199012000-00021

[22] Ramanathan NL. A New Weighting system for mean sur-

face temperature of the human body. J Appl Physiol.

1964;19:531-533. PMID:14173555

[23] European Commission. Solar radiation. European Com-

mission, 2017. [accessed 2017 Feb 02]. http://re.jrc.

europa.eu/pvgis/apps4/pvest.php.

[24] Ruby Coast Research Centre Team. Excel heat stress cal-

culator. www.ClimateCHIP.org (accessed 2017 Feb 4)

[25] Lemke B, Kjellstrom T. Calculating workplace WBGT

from meteorological data: a tool for climate change assess-

ment. Ind Health. 2012;50(4):267-278. PMID:22673363;

doi:10.2486/indhealth.MS1352

[26] Liljegren JC, Carhart RA, Lawday P, Tschopp S, Sharp R.

Modeling the wet bulb globe temperature using standard

meteorological measurements. J Occup Environ Hyg.

2008;5(10):645-655. PMID:18668404; doi:10.1080/15459

620802310770

[27] Br

€

ode P, Fiala D, B»a

zejczyk K, Holm



er I, Jendritzky G,

Kampmann B, Tinz B, Havenith G. Deriving the oper-

ational procedure for the Universal Thermal Climate

Index (UTCI). Int J Biometeorol 2012; 56(3):481–94;

PMID:21626294; https://doi.org/10.1007/s00484-011-

0454-1

[28] Universal Thermal Climate Index calculator. [accessed

2017 Feb 04]. Available at: http://www.utci.org/

[29] da Silva AI, Fernandes LC, Fernandez R. Energy expendi-

ture and intensity of physical activity in soccer referees

during match-play. J Sports Sci Med. 2008;7(3):327-334.

PMID:24149899

[30] Osgnach C, Poser S, Bernardini R, Rinaldo R, di

Prampero PE. Energy cost and metabolic power in elite

soccer: a new match analysis approach. Med Sci Sports

Exerc. 2010;42(1):170-178. PMID:20010116; doi:10.1249/

MSS.0b013e3181ae5cfd

[31] ACGIH. Hand activity level. In: 2001 TLVs and BEIs.

Cincinnati (OH): ACGIH; 2001.

[32] Wurzelbacher S, Burt S, Crombie K, Ramsey J, Luo L,

Allee S, Jin Y. A comparison of assessment methods of

hand activity and force for use in calculating the ACGIH

(R) hand activity level (HAL) TLV(R). J Occup Environ

Hyg. 2010;7(7):407-416. PMID:20446152; doi:10.1080/

15459624.2010.481171

[33] Food and Agriculture Organization of the United

Nations. FAO Statistical Yearbook 2014. Europe and

Central Asia. Budapest, Hungary: FAO Regional Ofﬁce

for Europe and Central Asia; 2014.

[34] Budd GM How should we measure occupational heat

stress? Temperature. 2016;3(3):369-370. PMID:28352143;

doi:10.1080/23328940.2016.1218992

[35] Flouris AD, Piantoni C. Links between thermoregulation

and aging in endotherms and ectotherms. Temperature.

2015;2(1):73-85. PMID:27226994; doi:10.4161/23328940.

2014.989793

[36] Tudor-Locke C, Ainsworth BE, Washington TL, Troiano

R. Assigning metabolic equivalent values to the 2002 cen-

sus occupational classiﬁcation system. J Phys Act Health.

2011;8(4):581-586. PMID:21597131. https://doi.org/10.11

23/jpah.8.4.581

[37] International Organization for Standardization. ISO stan-

dard 7243: hot environments – estimation of heat stress

on working man based on the WBGT index (wet bulb

globe temperature). Geneva: International Organization

for Standardization; 1989.

340 L. G. IOANNOU ET AL.