Occupational heat stress, heat-related effects and the related social and economic loss: a scoping literature review

Introduction While there is consistent evidence on the effects of heat on workers' health and safety, the evidence on the resulting social and economic impacts is still limited. A scoping literature review was carried out to update the knowledge about social and economic impacts related to workplace heat exposure. Methods The literature search was conducted in two bibliographic databases (Web of Science and PubMed), to select publications from 2010 to April 2022. Results A total of 89 studies were included in the qualitative synthesis (32 field studies, 8 studies estimating healthcare-related costs, and 49 economic studies). Overall, consistent evidence of the socioeconomic impacts of heat exposure in the workplace emerges. Actual productivity losses at the global level are nearly 10% and are expected to increase up to 30–40% under the worst climate change scenario by the end of the century. Vulnerable regions are mainly low-latitude and low- and middle-income countries with a greater proportion of outdoor workers but include also areas from developed countries such as southern Europe. The most affected sectors are agriculture and construction. There is limited evidence regarding the role of cooling measures and changes in the work/rest schedule in mitigating heat-related productivity loss. Conclusion The available evidence highlights the need for strengthening prevention efforts to enhance workers' awareness and resilience toward occupational heat exposure, particularly in low- and middle-income countries but also in some areas of developed countries where an increase in frequency and intensity of heat waves is expected under future climate change scenarios.

/fpubh. . of the body is disrupted, and physiological pathways resulting in heat-related illness, acute outcomes (e.g., myocardial infarction), or exacerbations of pre-existing diseases (e.g., cardiovascular and respiratory outcomes) are activated (2). Individuals working in the heat are also prone to physical strength losses and cognitive function impairments, leading to work-related injuries (3), missed workdays, and productivity reductions (4) and, in the long term, may develop chronic kidney impairment (5, 6). Health and productivity outcomes related to heat strain have a huge impact in terms of social and economic costs on the different actors involved (7): the workers themselves due to the temporary or permanent health and quality of life impairments and missed wages, the farm or factory due to necessity of maintaining production despite employees absences or output reductions, the healthcare system due to the healthcare expenditures due to workers seeking care, the social security or insurance system due to reimbursements to laborers for injuries, permanent disability, or occupational diseases, and the whole country or region in terms of reductions of the gross domestic product due to production losses in specific economic sectors. Moreover, climate change is expected to worsen heat exposure in some regions exceeding work-related productivity thresholds (8).
Heat exposure in the workplace is a growing hazard throughout the world, considering climate change scenarios showing a universal increase in heat extremes virtually in every region, but larger in Central and South America, the Mediterranean region, north Africa, the Arabian Peninsula, India, and Southeast Asia (9). Most of the affected regions are low-income economies mostly relying on manual labor and manufacturing work with agriculture and construction being the economic sectors at higher risk of heat exposure and at higher workload intensity than others. The quantification of economic impacts of heat exposure in the workplace is of worth for individual companies, labor policymakers, insurance companies, but also for occupational safety and healthcare systems and should be taken into account when analyzing markets and economies at both the local and global scale. The knowledge of economic losses related to heat may serve as a basis to plan prevention measures at company level with a view on cost-benefit analysis, to set up specific heat adaptation policies, or to strengthen social security systems by enclosing climate risk concerns, especially toward poorer population and countries (10).
Differently from the strong evidence available on the effects of heat on workers' health and safety, there is still limited but growing evidence on the resulting social (7,11) and economic impacts (12). The lack of standardized methodologies for evaluation (epidemiological vs. econometric studies), as well as in the operational definitions of productivity loss (lost worktime, reported physical and cognitive performance reductions, and work output reductions in case of manual workers), heat-related productivity losses, and economic costs (i.e., lost salaries and wages due to fatigue/sickness, cost per compensable claim, and healthcare costs related to treatment and rehabilitation) make it difficult to have consistent findings and clear trends. Despite the close connection between indicators of social and economic heat-related impacts and the common underlying causal pathways, no previous literature review considered both social and economic losses related to heat at the same time.
As part of the Italian WORKLIMATE Project (https://www. worklimate.it) funded by the Italian Workers' Compensation Authority (INAIL), a literature review was carried out to update the evidence on both social and economic impacts related to workplace heat exposure. To address such a comprehensive research question, a scoping review was considered to be more suitable, as suggested also by other authors (13), to address the whole body of evidence deriving from different type of studies (i.e., epidemiological and economic modeling studies).

. Methods
The scoping literature search was conducted in two bibliographic databases (Web of Science and PubMed), using both free terms and controlled vocabulary (Supplementary Table 1) to select studies published since 2010 to April 2022. Since previous reviews considered impacts of occupational heat stress on workers' productivity (7,11,14) and on economic losses (12) separately, and in consideration of the interconnections between work performance, workers' health and safety, and monetary costs, the outcomes of interest were both social impacts related to workers (i.e., work hours losses and work absences), and economic impacts for a specific group of workers or economic sector (i.e., monetary costs associated to production losses) or for social security systems (i.e., compensation for work-related injuries and diseases). Both indoor and outdoor occupational heat exposure and all potentially exposed economic sectors and tasks (e.g., manual workers) were considered. The first group of relevant studies were from epidemiological studies (both qualitative and quantitative) on workers estimating productivity losses in the field or estimating costs related to occupational heat-related illnesses and injuries. The second category of suitable studies was represented by recently conducted economic studies adopting several approaches such as structural economic models and econometric models, to estimate the impacts of climate change on labor productivity and related economic costs, using occupational health and safety recommendations in an entire economic sector and for regional or global economies. Experimental studies (e.g., on physiological responses), epidemiological studies on occupational heat-related illnesses not estimating economic implications, studies focusing only on the impact of heat exposure on workers' cognitive functions, and studies on other occupational exposures (e.g., cold, air pollution) were excluded. Only original studies were retrieved, while literature reviews (7,11,12,(14)(15)(16) were excluded but used to screen for additional relevant studies, as well as the 6th assessment report of the Intergovernmental Panel on Climate Change (https:// www.ipcc.ch/report/ar6/wg2/) (8). The selection of studies and data extraction were conducted according to PRISMA guidelines (17). The outcomes considered were as follows: -Lost productivity estimated or perceived by the worker associated with the heat exposure. -Economic costs associated with heat-related injuries or hospitalizations in workers. -Projections of productivity losses due to heat and related economic costs for current climate or under climate change scenarios.

. Results
A total of 8,151 potentially relevant records were identified after duplicates were removed, of which 104 were identified from previous reviews on the topic (7,11,12,(14)(15)(16) or other sources (Supplementary Figure 1). Out of these, 137 were assessed as full texts because potentially relevant, and, finally, 89 studies were included in the qualitative synthesis. The largest number of studies was carried out in Asia (49 studies) and the lowest in Central and South America (24 studies) (all totals include the 21 global economic modeling studies) ( Figure 1). Field studies accounted for a larger proportion of studies in Asia, Oceania, and Europe, healthcare-related studies were more prevalent in North America and Oceania, and regional modeling studies were mostly conducted in North America, Europe, and Asia. Table 1 describes the results of the epidemiological studies (n = 40), including 32 field studies and 8 studies estimating healthcarerelated costs.

. . Field studies
Most field studies (20 out of 32) were conducted in low-or middle-income countries (18, 20-22, 24, 27, 29-31, 33, 37, 38, 40-42, 45-47, 49), with only 11 studies from Europe, the USA, and Australia/New Zealand (19,23,25,26,28,34,35,39,43,44,48) and 1 multicenter study (36). Three studies were qualitative, based on interviews or focus groups (28, 33,43), while the other studies were quantitative with 27 cross-sectional and 2 longitudinal studies (26,42) and provided an estimation of the association between heat and labor productivity measured in the field or perceived by workers. The study size overall ranging from 16 (30) to 4,095 workers (21) in different occupational sectors (9 on agriculture, 4 on construction, 1 on mining, and 18 from several sectors) also includes indoor workers (13 studies). Two main approaches were followed by studies in estimating heat impacts on productivity. The first approach provides an estimation of productivity and then evaluates the losses due to heat by comparing productivity data collected at different WBGT levels. In these studies, productivity was estimated in different ways: by worktime (direct, indirect, and non-productive) (22,30), cognitive and physical performance (e.g., time to complete task/work extra hour absenteeism/taken sick leave) self-reported from questionnaires or interviews (26, 30,32,38), clinical examinations (e.g., walking speed) (42), daily output reported or measured by field instruments (e.g., tally counters) (21,25), and premature worker attrition (21). In the second approach, productivity was not measured by itself, but in terms of productivity losses due to heat in different ways: published physiological models (31), self-reported by workers (29,49), and prediction models of economic losses due to heat based on number of laborers and the given laborers salary when exceeding WBGT thresholds (44). All studies used individual productivity measures except Amini et al. (18), which evaluated productivity at area level.
Despite the great heterogeneity in the work sectors and study size, nearly all field studies consistently showed a reduction in productivity due to occupational heat exposure. The only exception was one study on office workers not providing evidence of influence of thermal stress on work performance, possibly due to the fact that the thermal stress variable evaluated included both heat and cold temperature; therefore, their single contributions on work performance could not be disentangled (26). The estimated productivity losses ranged between 0.3% and 10% reduction for an increase of 1 • C in WBGT (30,(40)(41)(42)47). Considering the whole summer season, the prevalence of workers reporting heat-related productivity loss varied among studies from 11% (46) to 81% (35). Some studies also found an association of heat with an increase in indirect non-productive time at work (30), an increase in idle time at work (30), or in personal household time needed to rest to adapt to heat stress (22). Four studies (36,44,48,49) also provided an estimate of the related economic costs by applying the productivity losses to the gross wages or income of workers with an estimated cost of 6-8 euros per hour worked in Italy (36), 1100 Canadian dollars annually per worker in Ontario (44), 655 USD annually per worker in Australia (48), and 257 euros annually per worker in Malaysia (49). In one study (45), 25% of the workers self-referred a loss in their wages due to fatigue or sickness related to heat. The study by Langkulsen et al. (27) showed a reduction in productivity only in two of the occupational sectors considered (pottery and construction) but not in the others. Only Li et al. (30) and Yi and Chan (47) adjusted for individual worker characteristics such as age and BMI. Given the cross-sectional approach adopted in most studies, the results do not allow causal inference on the association between occupational heat exposure and work productivity.

. . Studies evaluating healthcare-related costs
In contrast to the field-based studies, the eight studies estimating healthcare-related costs due to occupational heat exposure used data from administrative databases; therefore, they were mostly conducted in western countries such as Europe, Australia, the US, and Canada (50-52, 54, 56-58), with only one study from China (53). Six studies considered all occupational sectors, while three studies only considered specific sectors, such as agriculture and construction. Five studies were descriptive analyses of occupational injuries or diseases identified as heat-related and consequent compensation costs in specific occupational sectors De Sario et al. .

FIGURE
Geographical distribution of included studies based on the study design by continents.
(50-52, 56, 57), while the other three were etiological studies estimating the occupational injuries attributable to heat exposure through time-series or case-crossover analysis and then quantifying the related costs (53,54,58). The national Spanish study from Martinez-Solanas et al. (54) was the only one to estimate heat-related injuries corresponding costs including not only the direct costs attributable to social or private insurance refund to the workers (for long-term losses) or to the healthcare system but also the indirect costs associated with maintaining production, and costs of pain and suffering. The total economic impact of heat-related injuries in the study period was 320 million euros, with the costs associated with pain and suffering higher than other types of costs. The study conducted by Ma et al. (53) in China evaluated the attributable fraction of insurance pay-out related to occupational heat exposure (temperatures above the limit of the Wet Bulb Globe Temperature (WBGT) in accordance with international standards) of 4.1% (95% CI 0.2%-7.7%). In an Australian study, an increase of 1 • C in maximum temperature above 33.8 • C was associated with an increase of 41.6% in healthcare costs and 74.8% in working days lost due to heat-related injuries (58).
Two descriptive studies conducted in Washington State, US (50, 52), reported an increase in both the median cost per heatrelated injury over time (from 537 USD in 1995-2005 to 909 USD in 2006-2017) and the number of working days lost due to injury (from 46 to 93 days per claim on average). The studies also reported an increase in temperatures over time associated with the injuries. In the same study area, Spector et al. (57) estimated a median cost per claim of 654 USD specifically for the agricultural and forestry sectors similar to the previous study conducted in the area (50). These studies suggest a higher median costs related with non-compensable claims for heat-related than for total injuries suggesting a possible under-reporting of work-related accidents in this sector (50, 52, 57). Another South Australian study on construction industry (56) calculated average cost during heat waves higher than in control periods (26,381 vs. 12,747 Australian dollars), with higher costs in the urban area than in the suburbs and for specific agents of injury (i.e., work platform, electricity, and equipment). Finally, a Canadian study in Ontario (51) estimated the rate of injuries related to loss of productive worktime (injury loss time), which is equal to 1.7 cases per million full-time equivalent months in the period 2004-2010.
Studies estimating healthcare-related costs identified some worker subgroups are related to higher costs or worktime losses such as manual workers (51), Black or Latinos workers (52), new workers (56, 58), workers aged 15-24 years (51), men (53), and workers of small-(56) or medium-sized companies (53,58). Table 2 describes the results of the 49 economic studies at the global (59-70, 72, 74-80) and regional level (73,. Most studies focused on impacts of current and projected heat on workers' productivity, with the exception of some studies estimating production output reductions due to heat (59,72,83)      There was a trend of decreasing productivity with increasing WBGT, but this was not statistically significant (significant only in unadjusted model).     Descriptive analysis of heat-related illness compensation claims and risk factors (outdoor/indoor, comorbidity, hours of the day, acclimatization) and related costs.

. . Results from economic studies
Median cost for all compensable and not compensable claims for heat-related illness was 537 USD (mean 1,864 USD), higher than for total claims (not only for heat-related injuries). Also median cost for non-compensable claim was higher for heat-related illness than for total claims (513 vs. 251 USD). Median cost per compensable claim for heat-related illness was 1,916 USD, lower than for total claims (4,771 USD). For time loss claims, the median number of working days lost was 6 (46 days on average).       Analysis focused on the loss of labor productivity as a function of WBGT levels during the hottest months in reference period and under scenarios.
Reductions in work capacity during the hottest months already occur at the global level (10% reduction). By 2050 under both scenarios, work capacity loss is 2-fold higher than in the historical period (20% reduction). By 2100, the reductions in the hottest month may reach 37% based on RCP8.5 and 20% based on RCP4.5. By 2200, very significant further changes in work capacity are projected for the hottest month based on RCP8.5 (61% reduction), and 12% of population is exposed to work capacity losses. To offset these reductions a substantial increase in unskilled farm workers will be required.  Analysis based on statistical model of work output (or GDP) and thermal stress (controlling for institutions, capital stock, and education). Linear regression between GDP is produced using a combination of capital and effective labor input. Effective labor input is defined as a composite of labor hours, labor effort, and labor performance as a function of the ambient temperature. We allow for the possibility that temperature may affect GDP with a time lag, by allowing for 1, 5, and 10 lags. For the US, household data on air conditioning and heating expenditures.
Very hot countries such as Thailand, India, and Nigeria suffer negative output shocks on the order of 3-4% per capita GDP per 1 • C. Very cold countries such as the UK, Canada, Norway, and Sweden have significantly higher output in warmer years (and lower output in colder years). In the US, a household with an average age of 20 spends roughly 15% (28 USD) more per year on air conditioning and 12% (54 USD) less on heating than an otherwise equivalent household with an average age of 60 and expenditure on both air conditioning and heating are higher for households with someone at home who is working than for those with someone at home but not working.    Direct output production losses by region (billion USD). Also indirect production and total losses (in terms of value of goods and services) are calculated.
Absolute and relative heat stress-induced direct output losses based on risk function between temperature and productivity from literature (perturbed productivity) (73). Absolute output losses are then determined by multiplying the perturbed productivity with the baseline production of that region Globally, between 2000 and 2039 direct output losses increase by 47% if no further adaptation measures are taken. Regional increase in direct losses in the billions USD (e.g., in India, Saudi Arabia, or Mexico) or nearly double the direct output losses (e.g., in Northern America or Europe) within the next decades.
Matsumoto et al. Coupled socioeconomic (CGE) and climate models. Changes in labor productivity affect the labor input necessary to produce goods/services in the production functions. Climate change impact on labor productivity based on dose-response function between WBGT and work capacity estimated in literature.
The impacts were the largest for the agricultural (36.8-100% labor productivity reduction by 2100), and the lowest for the service sectors (83.0-100% productivity reduction by 2100). Labor productivity reached its minimum levels during the warmest and wettest parts of the year in already hot and humid regions (similar trends were observed for both of the mitigation scenarios as well). Such declines in labor productivity reduced production and, consequently, affected the macroeconomy. The global-level negative impact on GDP grew with temperature increases, which was about 2% per 1 • C  Interdisciplinary approach that combines climate projections, epidemiological findings, and economic analyses. Work capacity loss (a physiological variable) estimated using the dose-response function between WBGT and work capacity estimated in literature (Hothaps and ISO). The spatiotemporal data of relative worker productivity losses are matched with the gridded data on the population count to obtain population-weighted impacts on worker productivity at a regional level. and the associated economic costs are assessed by using a dynamic multi-region, multi-sector computable general equilibrium model. Autonomous mechanization of outdoor work in agriculture and construction and presence of air conditioning for indoor work is implemented in the model.
Heat stress leads to substantial reductions in worker productivity. For RCP8.5, using the Hothaps function with constant work intensity results in an average reduction of 0.7% (1.8%) in global GDP by 2050 (2100) relative to the reference period. Impacts are higher for high-intensity work in low-latitude countries of Africa, South America, and Asia. Given the assumption of absence of air conditioning and constant work intensity, reductions in worker productivity in some regions under RCP8.5 could even exceed 40% by 2100 compared to the reference. Approximately 42% of the global mitigation cost could be offset by avoiding the adverse impacts of heat on worker productivity. Agriculture and construction are the most adversely affected by heat because these sectors require many work-intensive activities in the outdoor environment. While many low-latitude regions experience considerable reductions in worker productivity, less vulnerable regions such as Oceania, North America, Former Soviet Union, and Europe, receive a comparative advantage in production of agricultural goods, which explains those moderate increases in their production. Due to the penetration of air conditioning and a lower work intensity, manufacturing is less adversely affected by heat compared to agriculture and construction. The service sector exhibits a low risk of exposure to heat.   Relative percentage change in (annual) productivity with respect to the baseline, for all countries and sectors The spatiotemporal data of work capacity loss estimated based on WBGT using the dose-response function between WBGT and work capacity found in the literature projection of loss in labor productivity from relationships between average temperature and labor productivity under scenarios of 1, 2, 3, 4 and 5 • C increases in average temperature (study period not specified).
The estimated percentage variation of labor productivity for 140 regions and for a +1 • C increase in temperature is−0.27%. The mean productivity losses range from−2.52% to−17.48%. Agriculture is the sector most significantly affected by higher heat stress. Some effects are felt by about half of the countries already at +1 • C. Work capacity (work hours loss) estimated based on WBGT using the dose-response function between WBGT and work capacity found in the literature and on safety recommendation of work/rest ratio. Daily total worktime was calculated by the hourly work capacity and summed the hourly work capacity from 9:00 AM to 5:00 PM. In order to express the labor productivity loss due to reduced worktime in economic costs, the labor input was multiplied by the ratio of the worktime reduction, and their product was used as the effective labor input to the production function. The direct cost is calculated as the additional wages required to compensate the worktime loss associated with the additional labor requirements. Presence of air conditioning for indoor work is implemented in the model.   Empirical models of the relationship between the production of value in individual industries and interannual variations in climate. The production of goods and services is measured by per capita value added. Regression models of production with temperature, rainfall, and cyclones evaluated in non-linear models.
Heat impact on total production of−2.5% per 1 • C increase. Wholesale, retail, restaurants and hotels (-6.1% per 1 • C increase), and other services (-2.2% per 1 • C increase) exhibit significant production losses. Output losses occurring in non-agricultural production (−2.4% per 1 • C increase) substantially exceed losses occurring in agricultural production (−0.1% per 1 • C increase).  Relative productivity losses (%) Cost of lost productivity per square meter as a result of thermal discomfort The loss of productivity due to thermal stress for each hour of occupancy is derived from physiological model of productivity and PMV. The cost of lost productivity per square meter as a result of thermal discomfort over the year is estimated based on the productivity per worker within a given sector. This is calculated by dividing the Gross Value Added (GVA) for that sector by the number of people employed in that sector measured as full-time equivalent (FTE). The change in relative productivity as a function of user thermal comfort is then applied to the economic output of a worker. A typical office building is used.
As the climate warms then the cost of lost productivity increases from 134 pounds per square meter in 1970s (3.2% lost productivity) to 148, 164, and 181 pounds (and 3.5%, 3.9%, and 4.3% lost productivity) per square meter in 2030, 2050, and 2080, respectively. Projections of future labor productivity losses (in terms of lost labor days) (% loss at specific WBGT level) from dose-response function between WBGT and work capacity for moderate and heavy work estimated in literature and based on safety standard (ISO) function (work hours lost due to rest and slower work due to heat) In 1975 in the hottest locations 30-40% of afternoon worktime is lost in the shade and 60-70% lost in the sun. In 2050 in hottest areas, afternoon worktime is lost due to heat up to 80% for heavy work and up to 50% for moderate work.  Labor productivity (days with reduced labor productivity as percentage of total working days) Relative productivity loss in future scenarios compared to baseline.
Projections of future labor productivity losses (in terms of lost labor days) compared to baseline climate, applying dose-response function between WBGT and work capacity (work-rest ratio) estimated in literature for moderate and heavy labor. Population-weighted future labor productivity estimated from the number of days with reduced labor productivity multiplied for reduction ratio expressed as percentage change of the total number of working days. The relative productivity loss was calculated as difference between future and current labor productivity by period.  Calculation of the number of hours would be unsafe to work (in terms of lost labor days) based on dose-response function between heat index and work capacity (work-rest ratio) estimated by NIOSH for light and moderate workloads. These findings were coupled with the future annual average number of days projected to exceed heat index thresholds by occupational category and scenario and multiplied by the number of people in each occupational category (e.g., protective service) and refer to this exposure metric as "person-days" per year. Economic losses in terms of earnings at risk (assuming that workers are not paid for the hours during which it is too hot to work) calculated based on unsafe workdays, annual median earnings, and total workdays per year. Two potential adaptation options-using an adjusted work schedule that shifts work hours to cooler times of day and lightening workloads-were also assessed.
Assuming normal work schedules and moderate workloads, nationwide nearly 3 million outdoor workers already experience 7 or more unsafe workdays per year-primarily across Southwest, Southern Great Plains, Midwest, and Southeast. This number will grow by late century to 17.1 million workers (RCP4.5) and 27.7 million workers (RCP8.5). In terms of earning loss 4.7% of earnings (or a total of $49.2 billion) would be at risk under RCP4.5 and 10.2% (or a total of $107.5 billion) under RCP8.5 by the end of the century. Both adaptation scenarios are able to reduce the number of workers at risk, especially the second measure reducing workloads to light levels. By late century, universal implementation of both adaptation measures combined with emissions reductions consistent with the RCP4.5 pathway would reduce the number of workers experiencing 7 or more unsafe workdays per year to virtually none compared with 27.7 million workers who would experience such losses with the higher emissions RCP8.5 scenario and no adaptation measures implemented.     Labor productivity losses (work hours) estimated using the dose-response functions between temperature and work performance by occupational category. The regional employment in the four occupational groups, is used to aggregate the occupational losses into a single metric representing regional labor productivity loss by period. A macroeconometric model of the European economy is then used to assess implications of change in productivity in monetary terms. In addition to the direct effect of the labor productivity shock, the model also captures the dynamic, long-term cumulative effects that operate through the capital investment processes. Economic impacts are presented as changes in annual GDP in 2013 Euros. Adaptation was also considered: diffusion of space cooling and increase in the use of robotic exoskeletons.
Productivity of labor can be 1.6% lower in the worst-case scenario (RCP8.5), with the largest reductions in southern European regions. Adaptation can reduce the productivity losses by around 40%, with higher rates of reduction for the lower warming levels. The annual economic losses in Europe could reach 563 billion euros or 1.15% of GDP by the 2080s in the worst-case scenario.  Labor productivity losses (work hours) estimated using the dose-response functions between WGBT and remaining productivity (%) by different work intensities from Hothaps models and safety standards. Adaptation solutions were also considered (i) a decrease in the supply of labor (total hours worked), (ii) a reduction in the effort applied per hour worked, (iii) a reduction in productivity, per hour worked, for a given level of effort.
These losses are 1-5% of productivity for a 1.5 • C temperature. In all countries except Jordan, the first and second largest absolute reductions in labor productivity loss are in the agriculture and construction sectors, respectively. In India the reduction is 20% of total workforce hours lost due to heat stress, the other countries losses are lower. Changing working hour patterns will be most effective in countries where temperatures are high during 'normal' working hours, and lower at other times. The split shift reduces productivity losses by between 0.9 and 8 percentage points, equivalent to reductions in lost productivity between 40% and 70% across all five countries.
. . . Regional studies Regional studies (n = 28) were also considered both from peerreviewed journals (73, 81, 82, 87- a heterogeneous impact of heat on work productivity not only among countries but also within the same country (82,95,99,102). As seen in the global studies, low-latitude, high-intensity labor settings were the most affected such as West Africa, Southeast Asia, and Central and South America. Moreover, specific local studies suggest an impact also in other regions such as southern European countries (92,97,103), some parts of the US (especially agricultural areas in Southeast and Southwest) (94,96,106), and Australia (86). For example, under medium-high emission scenario by the end of the century, a 0.4-0.9% loss in productive days was shown for Southern Europe (92), 10.2% of wages lost were estimated in the US (94), and a 16-17% labor capacity loss was predicted in China (95). Agriculture was the sector most affected by heat stress, both considering the current climate and future scenarios and among non-agricultural sectors, construction, manufacturing, transportation, service, and mining (73,83,86,98,104,105). Agriculture (97) and manufacturing sector are also expected to be impacted in terms of farm production output losses (100,101).
The evaluation of adaptation measures was marginally evaluated: Air conditioning was effective in reducing labor productivity losses in indoor settings in two European studies (84,103), with one study suggesting also a potential role for technological measures such as robotic exoskeletons (103), while measures affecting the work/rest schedule have been shown to reduce productivity loss in outdoor workers in one study in Ethiopia, Ghana, India, Jordan, and Tanzania (104) and in one US study (94) while another US study provided uncertain results (91).

. Discussion
This literature review provides an updated summary of the evidence on socioeconomic impacts of occupational heat exposure and confirms the results of previous reviews (7,11,12,(14)(15)(16) and of the latest IPCC report (8). The review also provides further evidence on the association between indoor and outdoor heat exposure and socioeconomic impacts in terms of productivity loss or costs. Throughout the different study types, a coherent picture of the social and economic impacts of heat exposure in the workplace emerges, highlighting the main pathways for heatrelated productivity losses. One pathway is in common with the general population and is related to the increased risk of acute heat-related illnesses and deaths (1) and the emergence of chronic illnesses consequences such as renal impairment (5, 6). Underlying biological mechanisms include thermoregulatory failure with cardiovascular fatigue and respiratory distress, dehydration with progressive kidney dysfunction in case of sustained chronic exposure. Another pathway is related to changes in vigilance and cognitive performance that may enhance the risk of distraction, impairment in risk perception, and reaction time leading to improper operation and injury (3). The third pathway directly related to work productivity and physical performance reductions and to the physiological need to rest during heat exposure, leading to a reduction in work hours and work output (16). All these pathways are strongly interconnected, and it is difficult to identify which plays a major role in productivity loss.
The most robust evidence in the present review derives from time-series or case-crossover studies (53,54,58). Such methods are the "gold-standard" study design to evaluate the short-term .
/fpubh. . effects of environmental exposures at the population level while controlling for time-varying confounders. Field studies represent an important piece of evidence about heat-related productivity loss, but they have the limitation of providing evidence on a small sample and related to given setting at a specific time interval (110) and only a limited number of studies adjust for potential confounders (22,39,40,47). Studies are consistent in reporting labor productivity loss perceived by the workers (19, 20, 23, 29, 32-35, 43, 45, 46, 48, 49) and have negative impacts in terms of physical performance (42) and work output (21,25). The largest body of evidence from the present review comes from economic modeling studies. These are mostly global or regional studies which apply modeled spatially resolved temperature data for the current or future climate change scenarios to risk functions from physiological studies (65,71,89,91,108) to obtain an estimation of loss in working hours which is then converted to economic costs via workers' wages or as portion of gross domestic product (GDP). Studies combining economic and climate modeling have the added value of providing current and future impact estimates which are useful for the definition of adaptation and mitigation actions. However, these models are dependent on the scenarios selected and assumptions made; thus, it is important that the uncertainty is adequately reported (71). More complex economic models, i.e., the general equilibrium models (109), are able to account for the interdependencies among sectors but also have a number of methodological challenges in particular in accounting for societal welfare changes (different by GDP), nonlinear damages, and micro-and macro-adaptation processes (8). Although methodological differences limit comparability, actual productivity losses at the global level are nearly 10% (62,65) and under the worst-case scenario (high emissions) by 2100 are expected to increase up to 30-40% (62,65,67,75). GDP losses for the same period and scenario varied between 1.8% compared to baseline (75) and 23% (59). Scenarios suggest that in regions like sub-Saharan Africa, India, Southeast Asia, and South America, productivity losses may be even greater as they will experience significant warming and a high share of the economy entails laborintensive occupations (59-64, 66-68, 75, 78, 80), experiencing over a 10 times increase in work hours lost under the worst emission scenario (62). Some studies also report substantial reductions in work capacity in the United States, Europe, and Australia (86,92,94,96,97,103,106).
Vulnerability factors increasing the risk of heat-related productivity loss may differ according to the underlying causal pathway, with potential differences among factors increasing vulnerability for heat-related diseases, heat-related injuries, and heat-related productivity loss. However, the link with socioeconomic impacts is less clear. Individual factors such as age (53,56), gender (31,37,45,49), race (52), education level (37, 53), immigration status (34), and comorbidities such as kidney failure or other conditions (21,22) have been related to higher reduction in work productivity in some studies, but the evidence is limited. The work environment may also affect worker susceptibility to productivity losses related to heat, as consistently shown in the literature. Some occupational sectors, primarily agriculture and construction, appear more affected than others, suggesting a higher impact on productivity loss due to more intense physical activities.
The agricultural sector alone accounts for two-thirds of all labor hours lost globally in 2021 at the global level (78). Other sectors or workers affected include transportation and utilities (83,98), miners (37, 58,83,86,105), and indoor workers with no air conditioning (19,100,101). Furthermore, performing heavy tasks (45,48,68,70,75,89,93), direct sunlight exposure (36,63,89), and use of personal protective equipment (PPE) (23,29,35) have been associated with productivity loss. In some cases, the work sector and task may be a multiplier of existing individual vulnerabilities, as in the case of migrant agricultural workers (34) or young manual workers (51).
Awareness of heat-related risks, health and safety actions and training, as well as workers behaviors play a key role productivity loss due to heat among workers (37). The heterogeneous perception of heat-related occupational risks and causes of productivity loss (35,37,45,46) suggests that more efforts are needed to enhance risk perception and heat-protective behaviors. Work management policies need to have a holistic approach by addressing all potential pathways linking heat exposure to workers' health, safety, and productivity (37). Specific information tools aimed to increase adaptive capacity and protective behavior especially in the most vulnerable workers can reduce impacts on productivity, as suggested by the work carried out in Italy within the Worklimate project (https://www.worklimate.it/en/home-english/). Some strengths and limitations are worth mentioning: the quality of studies was not formally evaluated, and the search was restricted to only two bibliographic databases (PubMed and Web of Science) and only to English language studies that may have restricted the geographical coverage of some areas of the world such as Central and South America and Africa. To partially counterbalance this, the inclusion of a significant number of studies (14 out of 89) from the gray literature (from academia, NGOs, or economic or policy organizations) (16, 46, 61, 62, 66, 83-86, 90, 92, 98, 100, 104) retrieved from reviews in the field (7,11,12,(14)(15)(16) ensures to include a greater number of studies from low-and middle-income countries where the issue is particularly relevant. Moreover, the scoping review was limited to studied published since 2010, but this was also the publication horizon from previous reviews (11,12,14).
Due to the heterogeneity of studies in terms of methodologies used, heat exposure indicators, and economic cost measures, a quantitative synthesis was not possible. However, the present literature review provides a clear and consistent indication of the effects of heat on productivity and costs for employers and employees, economic sectors, social security systems, and national economies. The impacts are coherent across a range of study designs and study areas although we cannot exclude that some relevant papers are missing, the possibility that publication bias could distort these results is low thanks to the inclusion of a relevant piece of gray literature as specified above. This large body of evidence can support decision-making process in terms of improving and protecting worker safety, health, and wellbeing following the Total Worker Health approach (111) also in the context of climate change resilience and response by involving all relevant stakeholders both at the policy level and at the workplace level (i.e., nurses or other healthcare practitioners and workers' compensation professionals) (112) initiatives in this field have been taken, but more efforts are needed in terms of prevention, employer and employee information, and training to raise awareness and increase resilience and behavioral adaptation. The evidence suggests that the expected impacts of climate change may be even greater and that investing resources in prevention actions in occupational settings has both social and economic benefits. Despite the consistent evidence on productivity impacts, some knowledge gaps emerge. Future research needs to address them such as the role of individual and work-related factors in increasing worker's vulnerability to productivity losses, and the evaluation of adaptation measures such as work schedule adjustments and work-level reductions only little evaluated in terms of productivity improvements (91,94,104).

. Conclusion
In conclusion, much knowledge has been accumulated about heat-related reduction in work capacity in recent years. There is an urgent need for holistic work management policies such as the Total Worker Health approach and for climate change adaptation and mitigation efforts to protect workers' health from future warming and climate extremes, especially in most vulnerable agriculture, manufacturing, and construction sectors and in very hot countries with highintensity work.