INTRODUCTION

Heat stress refers to heat received in excess of that which the body can tolerate, without physiological impairment. It is an important facet of the impact of climate change on human health (75). Heat stress stems from three factors: intra-body heat production from muscular physical activity, external (i.e., ambient) heat, and clothing that affects heat convection and sweat evaporation (64). Consequential adverse effects are generally known as heat strain and include clinical diseases, health impairments, and reduced human performance and work capacity (64, 66). Metabolism, which provides the energy source for muscular movements, is a major internal source of heat and increases with activity level. External heat influences the rate of transfer of intrabody heat away from the body. In hot ambient conditions, therefore, when heat transfer is limited, the core body temperature increases, resulting in serious health risks. Adverse health effects due to exposures to excessive ambient heat (termed here heat exposure) already occur in many parts of the world, not only during heat waves but also due to the need for intensive manual work in hot daily conditions (41). Climate change (10) will increase both the incidence and the severity of these effects. Heat exposure also affects workers' capability to undertake physical activities without harm; in hot conditions, work capacity falls, leading to a decrease of labor productivity. These effects highlight the need for better analysis of the heat/health interface.

This article draws on the large body of physiological evidence that provides indicative estimates of reduced human performance and work capacity because of increased environmental heat (4). The most vulnerable people are those in low-income tropical countries where heavy physical work is common and the hot season is long (62). Our analysis here shows that even in hot parts of some high-income countries, such as in the United States, there are likely to be increased occupational health problems as climate change progresses. In addition, excessive heat exposure affects exercise and sporting activities (6, 8), and it reduces individuals' ability to carry out daily household tasks.

HEAT STRESS AS A HEALTH AND SOCIAL HAZARD

The direct health impacts of heat exposure are usually assessed in terms of mortality (e.g., 24, 48) or hospital admissions (30, 46). Elderly people and individuals with impaired health are especially vulnerable, but heat stroke also occurs among workers who perform heavy labor in hot conditions with potentially wider social and economic implications (64). Reports on heat stroke deaths among agricultural workers in the United States [∼30 deaths each year (9)] show that symptoms of serious heat strain are often ignored as individuals continue to work beyond a safe heat exposure limit. Hundreds of cardiovascular deaths each year among construction workers in Qatar may likewise be due to workplace heat (25). In addition, more than 1,000 additional deaths (as compared to the same weeks in preceding years) in the age range 20–70 years occurred during the two weeks of extreme heat in France in August 2003 (32). The fact that most of the deaths in this age range were among men suggests that the impacts may have been at least partly due to excess heat exposures during work (though no analysis has been done). In all of Europe, as many as 70,000 heat-related deaths may have occurred during this heat wave (69), but, again, no analysis of the occupational health component was carried out. In recent years, the epidemic of fatal chronic kidney disease in sugar cane harvesters in Central America (84) has been linked to daily dehydration due to excessive sweating in the hot work environment. However, the scarcity of quantitative field studies on these occupational health issues has created gaps in the evidence needed for climate change impact assessments.

Clinical effects are not the only consequences of excessive heat exposure (Figure 1). Physiological effects also act to reduce human performance and work capacity (4, 41). These outcomes have been overlooked in several international reviews of climate change effects (e.g., 12, 83, 86) but were highlighted in the Human Health chapter of the recent Intergovernmental Panel on Climate Change (IPCC) assessment of impacts of climate change (75). Excessive heat exposure affects not only individuals but also, through impacts on performance and productivity, the local community and economy (Figure 1).

Farming communities are particularly affected by increased heat exposures due to climate change. Studies of traditional agriculture in low- and middle-income countries (11, 49, 68) have shown that up to 80% of total farming energy input (megajoules/ha/year) comes from the physical work carried out by farmers. Mechanization clearly offers a substitute for human labor in many agricultural activities (74), but this solution requires economic resources and access to a suitable energy supply and is often beyond the financial capacity of agricultural communities in most developing countries.

Other outdoor occupations, such as construction work, open cast mining, transportation, and community services may also suffer from particular heat-related problems (72). Factory and workshop buildings in hot low-income countries often have no air-conditioning or other effective cooling systems (38). Millions of workers in these countries, who often produce low-price clothes, shoes, furniture, and other consumer products for sale in high-income countries, may experience extreme heat exposures on a regular basis.

Excessive workplace heat exposure is already a problem for many of the tropical and subtropical areas of the world. Figure 2 shows 30-year monthly averages of the in shade afternoon WBGT (wet bulb globe temperature) heat index levels for the hottest month in each geographic grid cell (0.5×0.5° areas; approximately 50×50 km at the equator) for the period 1980–2009. WBGT (89) is a measure that combines, within a single index, temperature, humidity, wind speed, and heat radiation, all of which affect rates of heat transfer from the body. When hourly WBGT exceeds 26°C (79°F), hourly work capacity is reduced in heavy-labor jobs, and above 32°C (90°F), any work activity is made difficult (34).

As Figure 2 indicates, heat exposures over this recent period were sufficient to affect work activities in most of the tropical and subtropical zones. The hot areas in Africa, Asia, and Latin America are very highly populated, and more than four billion people may be affected. The southeast and southwest United States are also at risk on the basis of these values unless effects can be reduced by technologies such as air-conditioning. Historical data corroborate the importance of such cooling measures, at least in the context of factory and office work, as productivity improved when excessive heat exposure was reduced (17).

EFFECTS OF INTENSE HEAT ON HEALTH AND WELL-BEING

As little as 20% of metabolic energy used by muscles contributes to the muscular external work output (64); the remainder of energy used becomes waste heat and needs to be released from the body into the external environment if a rise in the core body temperature from the normal 37°C is to be avoided. Muscular movement is the major energy consumer in the body (4). At high air temperatures (above 34–37°C) (93–99°F), the only method of heat loss is via evaporation of sweat. In high humidity conditions, however, sweat evaporation is insufficient, and other physiological changes cannot prevent the core body temperature from rising to a dangerous level, beyond 39°C (102°F) (4). Heavy labor in hot humid environments is therefore a particularly serious health risk (41). Although it is natural to self-pace and reduce work intensity when the heat stress rises (56), some people keep working beyond the safe limit either because they need to complete work tasks or to maintain work output to get paid (9). Self-pacing will generally reduce individuals' labor productivity (e.g., 82).

Increased core body temperature and excessive sweating, leading to dehydration, can have several direct or indirect implications for health and well-being (Table 1). Exposure–response relationships are available for some of these effects, but the evidence is limited and not always suitable for quantitative analysis. This lack of information inevitably impedes attempts to estimate the global health impacts of climate change (55).

Table 1

Climate change–related health impacts of heat according to the Intergovernmental Panel on Climate Change (IPCC), the World Health Organization (WHO), and other sources [modified from De Blois et al. (15)]

Table 1 indicates that the physiological strain of redistributed blood flow, increased heart rate, and prominent sweating due to excessive heat (possibly resulting in dehydration) affect primarily the cardiovascular system. When core body temperature increases beyond 39°C (102°F), effects on the function of the CNS, such as confusion or unconsciousness, mean that people may not take preventive actions in time. Table 1 also includes indirect effects, such as forced migration, due to the environment becoming uninhabitable (57, 63, 73). This effect—and its associated health risks—may become a major international problem on a scale beyond the ongoing migrant crisis in Europe (54, 92). Table 1 also lists concerns that were not analyzed in detail by either the IPCC or the World Health Organization (WHO), including fatal chronic kidney disease, increases in violent crimes and suicides, teratogenic effects of high core body temperature, interactions with prescription drugs, and poor clinical status [increased burden of disability (58)]. For all these reasons, actual health impacts of increased heat during climate change are likely to exceed those indicated by previous estimates (e.g., 55, 27).

Table 2 focuses on the specific health effects related to work. Physical work capacity is taken into account in international and national standards for heat protection at work (64), but cognitive or psychological performance effects are considered primarily in terms of human comfort (29). Determining the effects of excessive heat on the learning ability of schoolchildren is also important, although to date this has been the subject of very little research (13). Given the extreme heat conditions that already occur in many hot, low-income countries, the resulting reduction in educational performance and social and economic development should be a greater concern for international and national agencies than it currently seems to be (no international assessment of the effects of climate change mentions it).

Table 2

Health effects and negative impacts of excessive heat exposure at work

As indicated in Table 2, the direct effects on health are not the only consequences of concern from heat stress. Reduced work performance and resultant losses of income and nutrition are also important and may lead to further adverse health effects caused by the loss of income. Estimating the impacts on work performance and productivity are difficult because of the need to make assumptions about the capacity for adaptation (e.g., by shifting work to cooler times of the day, applying cooling systems, or supplying sufficient fresh water). There has also been relatively little research from which to derive the equivalent of exposure–response functions, relating performance loss to heat stress. Only three studies have explored these functions in studies of people carrying out their usual daily work, all showing increasing levels of productivity loss with increased heat stress (59, 70, 88). The studies are not directly comparable, however; whereas the last two were based on moderate metabolic rates (300W), the study by Nag & Nag (59) related to low rates (light labor, 200W), though even so, the increasing trend with increasing heat exposure was still apparent (44).

An important factor in the effects of heat stress during work is the frequency and duration of rest breaks. These breaks provide workers with the opportunity to cool down and recover optimum function (and thus work capacity), but these breaks do reduce work time and possibly daily labor productivity. Productivity also depends on the work environment (e.g., cool location with plenty of water for rest breaks), as well as the physiological condition of the workers. For all these reasons, the impacts of climate change on work capacity and on economic productivity are sensitive to adaptation measures and are thus difficult to estimate with confidence. In already hot locations in low-income communities, the opportunities for other adaptations than reduction in work intensity are very slim.

METHODS TO QUANTIFY HEALTH IMPACTS OF HEAT

Assessments of the health impacts of exposures to excessive heat are usually based on a heat index. To be effective, this method needs to be designed to account for all the key climate factors that affect both exposures to heat and human physiological mechanisms (Table 3), including air temperature, humidity, air movement (wind speed), and heat radiation (outdoors mainly from the sun) (64). The formula used to calculate the index should also reflect the physical transfer of heat into and away from the human body and the physiological responses (e.g., sweating and vasodilation in the skin). The impact on the health and performance of an average worker performing an intense work activity can then be assessed from the index.

Table 3

Variables of importance to assessing workplace heat impacts on health and well-being

In any climatic context, the level of heat exposure depends on whether the individual is indoors, in shade, or in full sun. The duration of exposure is also critical in determining the effects of exposure, as are characteristics that influence workers' physiological tolerance to heat—e.g., the level and type of activity, acclimatization, and clothing. Preexisting health, obesity, age, and sex are also important for estimating impacts (Table 3).

More than 30 heat indexes have been developed (20), differing both in the input variables they use and in the way their effects are parameterized. Most indexes estimate comfort levels, and few have analyzed the situation in very hot tropical environments. Some indexes require specific measuring equipment, whereas others rely on graphic nomograms to estimate values from measured climate variables. For use in climate change impact assessment, an index should ideally be quantified with a mathematical formula based on climate variables for which reliable modeled estimates are available.

Some indexes do not include solar heat radiation or air movement [e.g., the Heat Stress Index, HSI (64)], and some calculations incorporate so-called standard conditions. For example, the UTCI (Universal Thermal Climate Index) assumes certain types of clothing at different heat levels and a physical activity of walking 4 km per hour (5). In analyses of impacts on occupational health and work capacity, however, the WBGT (89) has advantages. The WBGT was designed to represent heat effects on working people and has been used as a heat exposure index in the few occupational epidemiology studies available (59, 70, 88) and more generally in studies of occupational health. It has also been widely proposed in both national (e.g., 1) and international (34) guidelines and standards and has been extensively tested in laboratories and in the field by the US army (26). The in-shade or indoor values can also be calculated from routine weather station data using mathematical formulas (2, 50, 51).

Whichever index is used, the specific environment of exposure (outdoors in-sun or in shade, indoors with or without air-conditioning) also needs to be known (or inferred) to allow the risk of heat effects to be estimated (Table 3). These conditions vary both within and between countries. In low-income tropical countries, for example, indoor heat exposures would be similar to those experienced outdoors in shade owing to the lack of air-conditioning. Agricultural workers in these countries may be assumed to be working in full sun during hotter periods of the day (e.g., afternoon), though they may seek shelter in the shade when possible. However, many agricultural and construction workers cannot protect themselves in this way except, in some situations, via large sun hats (84).

Weather stations routinely measure temperature and humidity in shade [“in a place sheltered from direct solar radiation” (85)]. The contribution of heat radiation from the sun is not always routinely measured. Full-sun values can be converted from estimates of heat in full shade using available hourly solar heat radiation values by latitude and longitude (see http://power.larc.nasa.gov/cgi-bin/timeseries.cgi?&p=&email%20=dailylarc.nasa.gov). WBGT levels in full sun during hot afternoons in tropical areas are ∼2–3°C (36–37°F) hotter than the in-shade values (43).

Wind speed ideally also needs to be incorporated into the index. Wind speeds vary considerably over the short term (e.g., hourly) and also between indoors and outdoors. Indoors, air movement can be increased by fans; however, when air temperature is above 37°C (99°F) and relative humidity is above 60%, fans may actually increase the heat stress (35) because of the intensified flow of hot air over the skin. When relative humidity is very low (10%, very dry air), fans help reduce heat stress up to a temperature of 48°C (118°F) (35). Additionally, working people usually move their arms and legs, generating their own air movement over their skin (∼1 meter per second). Allowing for these various effects is clearly difficult, though fortunately for calculations, the impact of air movements on heat stress varies little at speeds above ∼1 m/s (50).

Given the complexity of the factors affecting heat exposure and stress, availability of relevant climatic and other data is clearly an important constraint in estimating impacts (Table 3). Depending on the type of assessment (e.g., retrospective or prospective) data on future as well as past conditions will be needed. The spatial representation of these data is also important because ultimately investigators usually need to apply the values to populations, which are unevenly distributed geographically. Point data (e.g., for weather stations) thus have to be extrapolated to provide estimates of population exposures. Many different methods may be used for this purpose; however, unless information on local conditions is incorporated, important variations in heat exposure—for example, the additional heating in urban areas due to urban heat island effects—tend to be ignored (61). Lack of information on intra-urban climate may mean that risks in urban areas, where the majority of people live, are underestimated.

The exposure periods and durations of relevance vary greatly depending on health outcomes and levels of stress (Table 1). Heat exhaustion, for example, may set in after less than an hour of exposure in extreme conditions (high temperature and humidity and intense work). Cardiac arrest may also occur in a short time in people with preexisting heart conditions. In contrast, effects of undernutrition and mental stresses (Table 1) may take years to manifest. Thus there is a need for different averaging times for the heat index for different circumstances. Information on the metabolic rate associated with work or other high-intensity activities (e.g., sports) is of great importance (Table 3). In addition, working people who have already suffered from health effects of heat are more vulnerable than the people without such experience (64).

Except for small—and not necessarily representative—groups of people (e.g., athletes), little of the information needed to estimate the social and behavioral aspects of heat stress is usually available at an individual level. For practical reasons, therefore, assessments of health effects generally rely on area or population averages, derived from census data, or demographic or health surveys. Nevertheless, the physiological relationships between the environment, behavior, and modifiers of heat stress are well known (4) and were taken into account when international standards were developed for heat stress (e.g., 34, 64). Likewise, much detailed research has been done on the design of clothing or clothing materials that serve as barriers against heat or that reduce heat stress (31, 33).

ASSESSMENTS OF CLIMATE CHANGE–RELATED HEAT STRESS IMPACTS IN WORKPLACES

As the preceding discussion has indicated, the potential climate change–related population-level health impacts of occupational heat stress on working people are now being recognized (18, 37, 38, 41, 53). The recent IPCC assessment of human health effects (75) highlighted this issue, but no quantitative analysis was included. To date, only two studies have attempted to assess the global health impact of heat due to climate change in any comprehensive way (55, 27). Both studies make assumptions about adaptations to climate change and the effects of susceptibility to heat stress, and both rely on epidemiological data from studies of prolonged heat waves, mainly in North America and Europe. Such heat wave effects are unlikely to represent the impacts felt in low-income, developing countries in tropical areas.

Studies on the impacts of heat stress on work capacity are also rare. An analysis from Perth, Australia (53), based on projections of Australia's climate through 2070, explored the likely physiological effects of heat exposure in association with intense physical activity (including work) on human performance. Results indicated that whereas an average person, acclimatized to heat, could safely carry out physical activity or manual labor outdoors during all but one day per year in the 1990s, climate change would increase the number of days with dangerous heat exposure to 15–26 days per year by the 2070s (53).

Using the WBGT index and the international standard (34) for safe work, Kjellstrom et al. (42) made quantitative, worldwide estimates of the impacts of increasing workplace heat on global and regional populations as a result of climate change. Analysis was based on the assumptions that the mixture of jobs outdoors/indoors and at different work intensities stayed the same and that gross domestic product (GDP) increased over the study period (1975–2050). Results showed that climate change between 1975 (as the median year of the period 1961–1990 for which climate data were used) and 2050 would reduce the available work hours in all regions. The estimated reductions at the population level varied between 0.2% for Australasia and 18.6% for Central America. Other highly affected regions were Southeast Asia (18.2%), West Africa (15.8%), Central Africa (15.4%), Oceania (15.2%), the Caribbean (11.7%), and South Asia (11.5%). A shift in the workforce distribution away from physically demanding jobs to less ardent service jobs will make these reductions smaller, and use of heat-protection methods (adaptation) will reduce them further; however, in some countries, significant reductions in labor productivity will likely remain.

A more recent study examined likely changes in work capacity for an average worker in Southeast Asia (43) using the Wyndham (88) productivity loss data. March is the hottest month in much of this region, and Figure 3 shows the large potential decrease in work capacity from 1975 to 2050 and the differences between work in the sun and work in the shade. Heavy labor in the sun is most affected and, in 2050, would face a 29% loss of annual work hours. Moderate labor in the sun faces smaller losses (about 15%), and labor in the shade or indoors face even lower levels.

Another study of the worldwide impacts of workplace heat on working people was published by Dunne et al. (18). Their analysis focused on the loss of labor productivity during the hottest months in each part of the world over the period 1975–2200, using estimated WBGT levels together with the US national (1) and international (34) standards for safe work intensities at various WBGT levels. Reductions in work capacity, defined in this way, during the hottest months already occur at the global level [6–10% reduction (18)]. By 2100, the reductions in the hottest month may reach 37% based on RCP8.5 and 20% based on RCP4.5 (18). During the coolest months, the reductions may be ∼5% and 1% for these RCPs, respectively. By 2200, very significant further changes in work capacity are projected for the hottest month based on RCP8.5 (as high as 61% reduction during the hottest month), and the annual work capacity loss is 12% using the industrial threshold limit (safety standard) for light labor [further details are shown in graphics by Dunne et al. (18)].

Another analysis of labor productivity loss for the United States (47) was based on a single study of US time-use patterns in relation to daily heat conditions. The analysis reported a reduction in the proportion of time applied to work during the hottest months in most of the United States and presented the results in terms of economic losses. Together, these studies show that reductions in human capacity for physical activity (including work, active transport, leisure, sport and performance of routine daily tasks) are likely to induce substantial work time losses, with potentially serious economic consequences in geographic regions that will experience increased duration and intensity of the hot seasons.

The first global and national estimates of the economic consequences of reduced labor productivity resulting from increasing heat exposure due to climate change were made by DARA (14), on the basis of physiological impacts and labor productivity loss data from Kjellstrom et al. (42). These results projected a total global GDP loss of US$2.1 trillion (PPP $) for 2030. As a percentage of the national GDP, losses varied markedly and were greatest in tropical low- or middle-income countries (e.g., 0.0% in the United Kingdom and Japan, 0.2% in the United States, 0.8% in China, 3.2% in India, 6.0% in Indonesia and Thailand, and 6.4% in Nigeria and Ghana) (14, 39). A second, more recent global analysis using annual average temperatures and national GDP for different countries (7) concluded that as much as 23% of global GDP in 2100 will be lost owing to climate change. Reduced labor productivity due to increasing heat was just one of a number of causes considered.

A third methodological approach was used in the study by Zander et al. (90). Self-reported estimates of absenteeism and reductions in work performance (presenteeism) caused by heat in Australia were based on information from 1,726 employed people. The individual economic losses due to heat were US$655 per person, which translates to an economic burden totaling US$6.2 billion in Australia. Future projections were not presented, but these results indicate the importance of heat exposure for economic output. Further studies would clearly improve our understanding of potentially significant social and economic effects and perhaps inspire further political motivation to mitigate the consequences of climate change.

Figure 4 shows results from a new analysis, conducted by the authors of this article (B. Lemke, unpublished observations), using climate data derived from the HadGEM2-ES model and the RCP (Representative Concentration Pathways) 2.6 and 8.5. These represent the most extreme of the potential future climate scenarios used by the IPCC in 2013 (10). Analysis was done using climate data presented on a 0.5×0.5° grid (81) for 1995 and 2085 (surrounding 30-year periods, 1980–2009 and 2070–2099). WBGT was calculated using modeled daily measurements of temperature and humidity (ISI-MIP data) for the 67,000 grid cells over land, assuming an average air movement at 1 m/s and no solar heat radiation, as would be experienced for a physical laborer working indoors (without cooling) or outdoors in the shade (50). Population and employment data for the two target years were converted to the same grids. An exposure–response function for work capacity loss was derived on the basis of two available epidemiological studies (70, 88) to estimate the percentage of daylight hours that would be so hot in 2085 that work capacity would necessarily be reduced.

The substantial reduction in work capacity (and related labor productivity) between 1995 and 2085 is clearly seen in the maps (Figure 4). The areas with the greatest risk in 2085 remain the same (Amazon region, West Africa, Arab Gulf area, Pakistan, North India, Indonesia, and parts of China), but substantial reductions in work capacity are apparent in the southeast United States, parts of Europe, Africa, and the rest of India and China. Although some of these areas have low populations (e.g., parts of the Amazon and Australia), the majority have a high population density, implying that large numbers of people will be affected. The southern coast of the Arab Gulf area has a number of large cities and few people in between. The urban heat island effect may have special importance in this region. A recent paper (63) highlights the extreme daily heat levels of the local atmosphere that will occur later this century.

Expressed as the number of person-hours lost due to heat in whole regions (i.e., the work capacity loss multiplied by the working population in each grid cell and then summed up for all grid cells in a region), the impact in the hottest regions varies between 1% and 10%, assuming that no change in heat adaptation takes place (B. Lemke, unpublished observations). These numbers are averages at the regional level for a mixed workforce (average metabolic rate=300W; in shade or indoor noncooled work). The amount of work capacity loss varies both by job intensity (activity level) and by environment (shade or sun). Locally, and nationally, therefore, impacts may be more severe, depending on the working conditions and type of employment (as illustrated in Figure 3).

The calculations we present are based on the human performance and work capacity loss for people acclimatized to the local heat situation. Physiological acclimatization takes less than two weeks (64); even once individuals are fully adjusted, they will have to utilize behavior change (work during cooler hours, take more rest periods, improve hydration, etc.) and/or technical solutions to further reduce their heat vulnerability. Air-conditioning or other cooling systems can reduce the heat effects of indoor work, but the cost and sustainability of any cooling system will influence the extent of its application (52). Many millions of people will likely be unable to protect their daily activities and work from the increasing heat due to climate change. It should also be noted that the daily use of air-conditioning makes the person more sensitive to heat exposure, as the acclimatization may be incomplete.

CONCLUSIONS

Climate change during the twenty-first century will result in increased exposures to intense heat in many parts of the world. Without effective adaptation measures, workers engaged in heavy labor or working in humid and poorly ventilated conditions face increased risks of heat stress and are likely to suffer reduced performance and work capacity as consequences. In situations where full adaptation to the increased heat levels cannot be implemented, labor productivity will diminish, with potentially large economic consequences. The physiological basis for the reduction in human performance and work capacity is well known, but the application in assessments of climate change impacts has, so far, been piecemeal.

The local loss of work capacity (or labor productivity) during daylight working hours is currently up to 10% in particularly hot areas of the world and could be as high as 30–40% by 2085. Estimates of the related global reductions in GDP project a loss of US$2 trillion per year by 2030 and another loss of 23% of global GDP in 2100. These costs will be distributed unevenly across the world, with tropical regions experiencing the greatest impacts and poor communities bearing the brunt of the effects. To reduce these impacts, a range of adaptation measures will be needed, geared toward local conditions and communities. These include changes in working practices, mechanization of heavy labor, and the installation and use of air cooling technologies. Because these adaptations are often costly and themselves take time to implement, the initial steps to introduce them are already urgently needed. On the other hand, the economic losses without climate change mitigation and heat adaptation are likely to be substantial in many affected countries and, at the global level, may be counted in trillions of USDs.

disclosure statement

The authors are not aware of any affiliations, memberships, funding, or financial holdings that might be perceived as affecting the objectivity of this review.