Appendix Table 7.

Papers selected using the fourth search strategy

Paper	Year	Country/Region	Population	Heat exposure index	Study design	Main results
Altinsoy et al (1)	2015	Turkey	Working population	WBGT	Simulation based on climate driven model	Between 2071 and 2100, deficiency in labour productivity may reach up to 52 % during the summer across so Central Anatolia, Cyprus, and parts of the Aegean and eastern Mediterranean coastal areas, particularly in agriculture and construction.
Bodin et al (2)	2016	El Salvador	Agriculture	WBGT	Physiological parameters monitoring	WBGT was >26°C from 9:00 (average max.: 29.3±1.7°C). There was a reduction of HR symptoms and dehydration before and after the intervention. Individual daily production increased from 5.1 to a high of 7.3 tons/person/day.
Delgado_Cortez (3)	2009	Nicaragua	Agriculture	WBGT	Physiological parameters monitoring and productivity	Output production increased significantly (p=0.005) among those best hydrated, from 5.5 to 8 tons of cut sugarcane per worker per day.
Dell et al (4)	2014	Global	Working population	Not specified	Narrative review	Many of the papers discussed seek to address the impact of "harvesting". Temperatures can also affect income through labour productivity, which can in turn affect health.
Esteban et al (5)	2009	Taiwan	Working population	Not specified	Simulation based on climate driven model	Annual downtime from tropical cyclones could increase from 1.5% nowadays to up to 2.2% by 2085, an increase of almost 50%. This decrease in productivity could result in a loss of up to 0.7% of the annual Taiwanese GDP by 2085.
Gubernot et al (6)	2014	USA	Working population	Not specified	Scoping review	Around 40% of HRI cases occur among workers. The length of service appears inversely related to the occurrence of HRI. Working in direct sunshine can add up to 15°F to the perceived temperature. Increased heat decreases workers’ productivity.
Hanna et al (7)	2011	Australia	Working population	Ambient temperature	Narrative review	Workers carried out on average 1 hour less work per day when temperatures exceeded 37°C (compared with days <30°C), as workers self-paced to maintain thermal comfort. Heat-related health risks increase when work is "externally paced".
Heal et al (8)	2016	Global	Working population	Not specified	Systematic review	A day with temperatures exceeding 32°C can increase local monthly mortality rates by more than 1% and reduces daily labour supply in exposed sectors by up to 14%.
Kenefick et al (9)	2007	USA	Working population	Ambient temperature	Narrative review	A study showed that when subjects were dehydrated, productivity of stacking and debarking pulpwood was reduced by 12%. Decision-making and cognitive performance are also adversely influenced by dehydration
Kershaw et al (10)	2013	United Kingdom	Other	Different levels of climate change	Simulation based on climate driven model	As we move further into the future there is an increase in summertime overheating and hence a decrease in productivity but also a decrease in cold related winter discomfort and an associated increase in productivity.
Kjellstrom (11)	2016	Global	Working population	WBGT	Simulation based on climate driven model	In South-East Asia as much as 15% to 20% of annual work hours may already be lost in heat-exposed jobs, and this may double by 2050. The annual cost of reduced labour productivity at country level already in 2030 can be several percent of GDP.
Kjellstrom et al (12)	2009	Global, with a special focus on LMIC	Working population	WBGT	Narrative review	The resulting work capacity during different hours for a person who works at a heavy work intensity of 500 W is very low: on average only 20% of work capacity remains at 12 noon.
Kjellstrom et al (13)	2009	Global	Working population	WBGT	Simulation based on climate driven model	In terms of absolute change in labour productivity by the 2080s, the greatest losses (11.4% to 26.9%) are seen under A2 in Southeast Asia, Andean and Central America, Eastern Sub-Saharan Africa, and the Caribbean.
Kjellstrom et al (14)	2011	Global, with a special focus on Central America	Working population	WBGT	Narrative review	24% of HR deaths in the USA were in the Ag/For/Fis/ Hun sector and of these 67% were in workers employed in crop production. Above 26°C, the increasing need for rest periods can be approximated by straight lines that differ depending on the work intensity
Kjellstrom et al (15)	2013	South-East Asia	Working population	WBGT	Simulation based on climate driven model	Heavy work in the shade is now affected in the hottest areas so that 50-60% of afternoon work time is lost due to heat. Heavy work in the sun has losses up to 80% and above, and even moderate work is affected: >50% of afternoon work time is lost.
Kjellstrom et al (16)	2016	Global	Working population	WBGT	Narrative review	Reduced work capacity, labour productivity, and economic loss, as well as heat impact on gross domestic product were reported in India, USA and South Africa. When hourly WBGT exceeds 26◦C, work capacity is reduced in heavy-labour jobs, and above 32◦C (90◦F) any work activity is made difficult. GDP losses will be greater than 20% by 2100.
Langkulsen et al (17)	2010	Thailand	Agriculture - Construction	WBGT	Cross-sectional	Productivity as perceived by the workers revealed that for more than half (60%) the workers, productivity loss varied from 10 to 66.7%, whereas vegetable field workers displayed no loss of productivity.
Lao et al (18)	2016	Australia	Working population	Not specified	Cross-sectional	Common heat-related symptoms including headaches, sweating, dizziness, and tiredness. Most workers highlighted that heat impedes their work efficiency and causes a substantial slowing of work rate.
Li et al (19)	2016	China	Construction	WBGT	Within-group comparison	Direct work time decreased by 0.57% and idle time increased by 0.74% when the WBGT increased by 1 °C; direct work time increased by 0.33% when the workers' experience increased by 1 year and decreased by 0.72% when the workers' age increased by 1 year.
Lundgren et al (20)	2013	Global	Working population	Not specified	Narrative review	The years 1995-2006 have rank among the warmest. Effects of increasing temperatures include a high prevalence of CKD, in Central America, especially among sugarcane workers, and a global reduction of work productivity.
Lundgren et al (21)	2014	India	Working population	WBGT	Within-group comparison	Avg WBGT was 29.7. Significant impacts on productivity in all workplaces, apart from the laundry facility, were shown, e.g. in the canteen, the core temperature limit of 38°C predicted by the model was reached in only 64 min for women.
Marchetti et al (22)	2016	World	Working population	Not specified	Narrative review	Productivity is affected after 1 hour of moderate physical work above 32 °C. In SE Asia up to 20% of annual work hours may already be lost in heat-exposed jobs. By 2080, the greatest losses are foreseen in SE Asia, Central America, Sub-Saharan Africa.
Mathee et al (23)	2010	South Africa	Outdoor workers	Ambient tremparture	Cross-sectional	Hourly max. temp. ranged from 23-36°C in Upington and 21-27°C in Johannesburg. Heat-related effects reported included sunburn, sleeplessness, irritability, and exhaustion leading to difficulty in maintaining work levels and output during very hot weather.
Mirabelli et al (24)	2010	USA (North Carolina)	Agriculture	Not specified	Cross-sectional	Working in extreme heat was reported by 281 respondents (94%), among whom 112 (40%) reported symptoms of heat illness. 37% reported changes in their work hours and 34% in their work activities.
Oyekale (25)	2015	Nigeria	Agriculture	Not specified	Cross-sectional	Missing regular times scheduled for spraying cocoa pods (45.7% in Ondo state) was one of the forms of reported climate change induced occupational stresses
Patz et al (26)	2014	Global	General population, including working population	Not specified	Systematic review	Heat stress has reduced labour capacity by 10% in summer’s peak over the past few decades. Projected reduction may double by 2050.
Pradhan et al (27)	2013	Nepal	Manufacturing industry	Heat index, humidity index, and WBGT, based on the HOTHAPS approach	Cross-sectional	The average temperature in the summer reached to over 39°C: the environmental conditions were inadequate for workers to work continuously during the day. HRI included fainting, mental irritation, laziness, sleepless nights, dehydration and giddiness.
Sahu et al (28)	2013	India	Agriculture	WBGT	Cross-sectional	At WBGT>26°C the hourly N of rice bundles collected was reduced approximately 5% per °C of increased WBGT.
Sett et al (29)	2014	India	Manufacturing industry	WBGT	Cross-sectional	There is a linear decline in productivity with an increase in maximum air temperature above 34.9°C, and the lost productivity for every degree rise in temperature is about 2%.
Singh et al (30)	2015	Australia	Working population	Not specified	Cross-sectional	Productivity is significantly reduced at 35°C, and approximately one-third of baseline work productivity can be lost in certain physically demanding jobs when working at 40°C.
Suzuki-Parker et al (31)	2016	Japan	Working population	WBGT	Simulation based on climate driven model	"Light labour safe" hours are projected to decrease by 30-40 % by the end of the 21st century. The number of "heavy labour restricted days" is projected to increase from~5 days in the 2000s to nearly 2/3 of the days in August in the 2090s.
Venugopal et al (32)	2015	India	Working population	WBGT	Cross-sectional	Workers with heavy workloads reported more heat-related health issues (chi square = 23.67, p ≤ 0.001) and reduced productivity (chi square =15.82, p ≤ 0.001), especially outdoor workers. Of the workers’ assessed in both hotter and cooler seasons, 65% vs. 32% were exposed to higher than recommended WBGT levels in hotter and cooler seasons respectively. This corresponded to a significant increase in self-reported productivity losses (p ¤ 0.016).
Xiang et al (33)	2014	Global	Working population	WGBT, heat stress and air temperature	Systematic review	Manual workers be at risk of heat stress, especially those in LMIC. 79% of identified studies indicated that participants were suffering from heat strain, with outdoor workplaces (90%) being much higher than indoor workplaces (65%).
Zander et al (34)	2015	Australia	Working population	Not specified	Cross-sectional	70% said heat made them less productive on at least 1 day in the previous year: on average, they were 35% less productive on days on which they had suffered from heat, were less productive on 10 days, and worked for 27.1 h less.
Zhao et al (35)	2016	China	Outdoor workers	Daily Tmax	Simulation based on climate driven model	The total HTS is estimated at 38.6 billion yuan/y over the 1979-2005 period (0.2% of GDP), the share of GDP devoted to HTS could become as high as 3% at the end of 21st century.
Zivin et al (36)	2014	USA	Working population	Historical and forecasted temperature distribution-daily maximum and minimum temperature, precipitation, snowfall, and relative humidity.	Econometric model	At daily max. temp. >85°F, workers in industries with high exposure to climate reduce daily time allocated to labour by as much as 1 h. Almost all of the decrease in time allocated to labour happens at the end of the day.