Abstract
Background: In response to the Paris Agreement under the United Nations Framework Convention on Climate Change, the research community was asked to estimate differences in sectoral-specific risks at 1.5 °C and 2 °C increases in global mean surface air temperature (SAT) above pre-industrial temperatures. Projections of the health risks of climate change typically focus on time periods and not on the magnitude of temperature change.
Objective: Summarize projections of health risks associated with temperature extremes and occupational heat stress, air quality, undernutrition, and vector-borne diseases to estimate how these risks would differ at increases in warming of 1.5 °C, 2 °C, and higher.
Methods: A comprehensive search strategy included English language publications since 2008 projecting health risks of climate change identified through established databases. Of 109 relevant publications, nearly all were for future time periods (e.g. in 2030 and 2050) rather than future SAT thresholds. Time periods were therefore converted to temperature changes based on the models and scenarios used.
Results: Warming of 1.5 °C is reached in about the 2030s for all multi-model means under all scenarios and warming of 2 °C is reached in about the 2050s under most scenarios. Of the 40 studies projecting risks at 1.5 and 2 °C increases of SAT, risks were higher at 2 °C for adverse health consequences associated with exposures to high ambient temperatures, ground-level ozone, and undernutrition, with regional variations. Risks for vector-borne diseases could increase or decrease with higher global mean temperatures, depending on regional climate responses and disease ecology.
Conclusions: The burden of many climate-sensitive health risks are projected to be greater at an increase of 2 °C SAT above pre-industrial temperatures than at 1.5 °C. Future projection studies should report results based on changes in global and regional mean SATs and time, to facilitate quantitative analyses of health risks and to inform the level of ambition and timing of adaptation interventions.
Export citation and abstract BibTeX RIS
Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence.
Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Introduction
Climate change is increasing global and regional temperatures and precipitation in some regions, the frequency and intensity of extreme weather and climate events, and sea levels and ocean acidification (IPCC 2013). Exposure to these changes have adverse consequences for human health and well-being (Cramer et al 2014, Ebi et al 2017, Smith et al 2014). Smith et al (2014) concluded that if climate change continues as projected under higher emission scenarios, major changes in ill health would include:
- Greater risks of injuries, diseases, and death due to more intense heatwaves and fires (very high confidence);
- Increased risk of undernutrition resulting from diminished food production and water supply in poor regions (high confidence);
- Consequences for health from lost work capacity and reduced labor productivity (high confidence);
- Increased risks of food- and waterborne diseases (very high confidence) and vector-borne diseases (medium confidence);
- Modest reductions in cold-related morbidity and mortality in some areas due to fewer cold extremes (low confidence), geographic shifts in food production, and reduced capacity of disease-carrying vectors due to exceedance of thermal thresholds (medium confidence). These positive effects will be increasingly outweighed, worldwide, by the magnitude and severity of the negative effects of climate change (high confidence).
Few projections supporting these summary statements quantified how risks could differ at specific increases in global mean surface air temperature (SAT). To determine the extent of additional health risks of climate change at increases in global mean SAT of 1.5 °C, 2 °C, and higher, we reviewed recent projections of health risks associated with temperature extremes, heat stress as it affects occupational health, air quality, undernutrition, and vector-borne diseases. This information can be used to inform implementation of the Paris Agreement under the United Nations Framework Convention on Climate Change (UNFCCC 2015).
Methods
The search strategy included projections of identified climate-sensitive health outcomes over this century published in or since 2008 through November 2017. We chose 2008 to include projections of health risks assessed in Smith et al (2014). We conducted a comprehensive literature search of the peer-reviewed, global literature, with iterative searches of PubMed, Web of Science, Ovid MEDLINE, and Embase to identify English publications using specific health outcomes and climate or climate change as search terms; search terms were applied to publication titles and abstracts (Annex 1 in supplemental material available at stacks.iop.org/ERL/13/063007/mmedia). Publications were identified for temperature extremes, occupational health, ground-level ozone and particulate matter, undernutrition, and vector-borne diseases such as malaria, dengue, West Nile virus, Lyme and other tick-borne diseases, as well as Chagas disease and leishmaniasis. We did not include food- and water-borne diseases because of the paucity of literature.
The initial database search returned 2649 publications. After removing duplicates and publications whose title or abstract indicated the analyses did not include projections of the risks of climate-sensitive health outcomes (e.g. morbidity or mortality and potential health impacts of temperature extremes) under a changing climate, 1891 publications remained. The reference lists of all publications including projections of health risks were searched for additional possible citations. The literature search was updated during peer review to identify and include additional submissions.
Abstracts were reviewed to identify papers that quantified climate change projections of health risks, leaving 109 studies. These publications were critically reviewed and summarized in terms of the region of interest; the health outcome metric used; baseline health information; climate model(s) and scenario(s) used; the time periods of interest; impacts at the baseline reported in the study; projected risks at 1.5 °C, 2 °C, and higher degrees of warming; and other factors considered, such as population change. Information on warming above 2 °C was included to put the risks of the lower thresholds of warming into perspective. Supplemental materials includes summary tables organized by health outcome, with two tables for each health outcome. One table summarizes publications projecting risks for that health outcome at 1.5 °C and 2 °C increase in SAT (e.g. the study projected for both degrees of temperature change) and another table that summarizes other publications projecting health risks for that health outcome. The first table for each health outcome directly addresses the request from the UNFCCC to understand the risks of warming of 1.5 °C and 2 °C, while the second summarizes other projections, to provide further quantifications of the health risks of climate change. There is limited information on the health risks of warming of more than 2 °C in the supplementary tables; these projections are not covered in depth because of greater uncertainties regarding health outcomes associated with higher degrees of warming and also because of inconsistencies in the time periods projected.
The diversity of baselines, scenarios, climate models, and health models used in the projections precludes quantitative comparisons of the studies. Qualitative comparisons of the results were made using expert judgement.
Table 1. Decades when 1.5 °C, 2 °C, and higher degrees of warming are projected to be reached for multi- model means.
| Generation | Scenario | Decade 1.5 °C reached | Decade 2 °C reached | dT 2080–2099 | dT 2090–2099 |
|---|---|---|---|---|---|
| SRES | B1 | 2039–2048 | 2065–2074 | 2.18 | 2.27 |
| SRES | A1b | 2029–2038 | 2045–2054 | 3.00 | 3.21 |
| SRES | A2 | 2032–2041 | 2048–2057 | 3.39 | 3.83 |
| RCP | 2.6 | 2047–2056 | a | 1.48 | 1.49 |
| RCP | 4.5 | 2031–2040 | 2055–2064 | 2.32 | 2.37 |
| RCP | 6.0 | 2036–2045 | 2058–2067 | 2.63 | 2.86 |
| RCP | 8.5 | 2026–2035 | 2040–2049 | 3.90 | 4.39 |
Nearly all projections of the health risks of climate change were for future time periods (e.g. risks in 2030 and 2050), not for specific thresholds of global mean SAT changes. Analysis therefore required a conversion from time to temperature. The year for which global mean SAT is projected to reach 1.5 °C and 2 °C above preindustrial levels was estimated using global climate model (GCM) projections employed within each health study. For each study, the following were characterized: (1) the model generation (Coupled Model Intercomparison Project (CMIP) 3 or CMIP5); (2) the emissions scenario(s) (SRES A1B/A2/B1 for CMIP3, or Representative Concentration Pathway (RCP) 2.6/4.5/6.0/8.5 for CMIP5); and (3) the number of GCMs used. If only one GCM was used in a study, the temperature projection(s) was used from that GCM, model generation, and emissions scenario(s). If more than one GCM was used, the multi-model mean temperature projection for that particular model generation and emissions scenario(s) was used. CMIP3 global mean SAT projections were obtained from the IPCC Data Distribution Centre (IPCC 2007). CMIP5 projections were obtained from the Royal Netherlands Meteorological Institute Climate Explorer (KNMI 2013).
Next, a baseline period was defined to estimate 1.5 °C and 2 °C global mean SAT increases. Rather than define a specific pre-industrial baseline period (which is challenging due to the different historical starting points among GCMs), within each GCM projection, the decade 2010–2019 was defined as the baseline. The rationale for this decision was that the center year of this decade—2015—is considered the first year for which observed global mean SAT reached 1.0 °C above pre-industrial conditions as defined by the 1850–1900 average (UK Met Office 2015). To determine when 1.5 °C, 2 °C, and higher degrees of warming above preindustrial conditions were reached for a given temperature projection, we created a 10 year projection window, that was then moved forward one year at a time, starting with 2011–2020, until the projected global mean SAT in the window was 0.5 °C, 1.0 °C, and higher above the 2010–2019 baseline. For perspective, we estimated, using the same approach, the magnitude by which global mean SAT is projected to exceed pre-industrial levels by the last decades of the 21st century, 2080–2089 and 2090–2099. We did not attempt to scale individual projections to 1.5 and 2 °C because doing so would introduce large uncertainties.
Results
Table 1 shows the decades when 1.5 °C, 2 °C, and higher degrees of warming are reached for multi- model means under the SRES and RCP scenarios used in the projections summarized. Warming of 1.5 °C was projected to be reached in about the 2030s for all multi-model means under all scenarios. Warming of 2 °C was projected to be reached in about the 2050s under most scenarios. Note that RCP2.6 does not reach 2 °C by the end of the century.
Temperature-related morbidity or mortality: Of the thirty-four studies that projected increased exposure to extreme temperatures or temperature-related morbidity or mortality over this century (see supplemental material tables S1 and S2), 15 studies projected morbidity or mortality at various degrees of warming, but did not specifically compare 1.5 and 2 °C (Anderson et al 2016, Astrom et al 2013, Benmarhnia et al 2014, Dong et al 2015, Gasparrini et al 2017, Kingsley et al 2016, Marsha et al 2016, Martinez et al 2016, Oleson et al 2015, Petkova et al 2017, Vicedo-Cabrera et al 2017, Voorhees et al 2011, Wang et al 2016, Weinberger et al 2017, Wu et al 2014).
Nineteen studies projected mortality at both 1.5 °C and 2 °C (Arnell et al 2018, Chung et al 2017, Doyon et al 2008, Garland et al 2015, Guo et al 2016, Hanna et al 2011, Hajat et al 2014, Honda et al 2014, Huang et al 2012, Huynen and Martens 2015, Jackson et al 2010, Kendrovski et al 2017, Li et al 2015, Li et al 2016, Petkova et al 2013, Schwartz et al 2015, Vardoulakis et al 2014, Wang et al 2015, WHO 2014), concluding that the projected magnitude of heat-related mortality at 2 °C was greater than for 1.5 °C. While higher risks were associated with greater degrees of projected warming, the magnitude of risks at different degrees of warming varied by region, presumably because of differences in average temperatures (e.g. risks are higher in regions with cooler average temperatures), population acclimatization, population vulnerability, the built environment, access to air conditioning and other factors.
In some regions (e.g. the UK under projected warming of 2 °C), cold-related mortality was projected to decrease with warmer temperatures; greater reductions in mortality were generally observed with higher degrees of warming (see SM tables S1 and S2). However, increases in heat-related mortality were projected to outweigh any reductions, with the heat-related risks increasing with greater degrees of warming (Hajat et al 2014, Huang et al 2012, Huynen and Martens 2015, Gasparrini et al 2017, Schwartz et al 2015, Vardoulakis et al 2014, Oleson et al 2015, Weinberger et al 2017).
Evidence suggests adaptation has reduced the impacts of heatwaves. Recent observations showed smaller health burdens in many countries during such events than during earlier time periods characterized by less adaptation (Arbuthnott et al 2016, Astrom et al 2013, Chung et al 2017, Sheridan and Dixon 2016). Assumptions of additional adaptation reduced the projected magnitude of temperature-related mortality under different warming scenarios (Anderson et al 2016, Huynen and Martens 2015, Li et al 2016, Petkova et al 2017, WHO 2014).
Impacts of heat stress on occupational health: Four projections supported the conclusion of Smith et al (2014) that increasing heat stress associated with additional climate change could further compromise safe work activity and worker productivity during the hottest months of the year (Kjellstrom et al 2013, Kjellstrom et al 2017, Sheffield et al 2013, Dunne et al 2013). However, these projections did not compare risks at 1.5 °C and 2 °C.
Without considering the complex drivers of heat stress or the potential for acclimatization and adaptation, three studies projected that large areas of the world may become inhospitable for human health and well-being as temperatures continue to increase (Pal and Eltahir 2015, Matthews et al 2017, Sherwood and Huber 2010).
Studies projected other measures of occupational health risks from higher temperatures. Worldwide projections of the costs of preventing workplace heat-related illnesses through worker breaks suggested that total Gross Domestic Product (GDP) losses in 2100 could range from 2.6%–4.0% under high greenhouse gas emission scenarios compared to current climate conditions (Takakura et al 2017). Because the relationship between the costs of heat-related illness prevention and temperature is approximately linear, the difference in economic losses was projected to be ∼0.3% less for 1.5 °C compared to 2 °C in 2100 in terms of global GDP. In China, taking into account population growth and employment structure, high temperature subsidies for employees working on extremely hot days were projected to increase from about 39 billion yuan year−1 in 1979–2005 to 250 billion yuan year−1 in the 2030s and 1000 billion yuan year−1 in 2100 (Zhao et al 2016), with higher costs under RCP8.5 than under RCPs 4.5 and 2.6.
Air quality: Climate change could alter the dispersion of primary air pollutants, particularly particulate matter, and intensify the formation of secondary pollutants, such as ground-level ozone, whose formation is temperature dependent (Orru et al 2017). There is high uncertainty of projected changes in the atmospheric concentrations of ground-level ozone and particulate matter, with large regional variations in projected changes. Of the 18 studies that projected the health risks of changes in air quality (see SM tables S3 and S4), 12 projected morbidity or mortality at various degrees of warming, but did not specifically compare 1.5 °C and 2 °C (Alexeeff et al 2016, Chang et al 2014, Fang et al 2013, Fann et al 2015, Garcia-Menendez et al 2015, Geels et al 2015, Goto et al 2016, Liu et al 2016, Orru et al 2013, Physick et al 2014, Sun et al 2015, Wilson et al 2017).
The six studies projecting risks at both 1.5 °C and 2 °C (Dionisio et al 2017, Heal et al 2013, Lee et al 2017, Likhvar et al 2015, Silva et al 2016, Tainio et al 2013) concluded that ozone-related mortality will increase with additional warming, with the risks higher at 2 °C (Dionisio et al 2017, Heal et al 2013, Lee et al 2017, Likhvar et al 2015, Silva et al 2016, Tainio et al 2013). Reductions in precursor emissions would reduce future ozone concentrations and associated mortality. Because of uncertainties in future precursor emissions that lead to the formation of ozone, most studies held emissions constant, focusing instead on projecting the risks associated with climate change impacts.
Changes in projected particulate matter-related mortality could increase or decrease, depending on climate projections (Fang et al 2013, Garcia-Mendez et al 2015, Geels et al 2015, Goto et al 2016, Likhvar et al 2015, Liu et al 2016, Silva et al 2017, Sun et al 2015, Tainio et al 2013). Emission assumptions also influence projected changes in particulate matter, affecting the magnitude and pattern of future mortality.
Undernutrition: Four studies of the risks of undernutrition with climate change supported the conclusions of Smith et al (2014) that climate change will negatively affect childhood undernutrition and stunting, through reduced food availability, and will negatively affect undernutrition-related childhood mortality and increase disability-adjusted life years lost, with the largest risks in Asia and Africa (see SM tables S5 and S6). Three studies compared health risks associated with food insecurity at 1.5 °C and 2 °C, concluding that risks are higher at 2 °C (Hasegawa et al 2016, Ishida et al 2014, WHO 2014). Warming of 1.5 °C is associated with an increase in the global undernourished population to 530–550 million; and to 540–590 million at 2 °C (Hasegawa et al 2016). Climate change impacts on dietary and weight-related risk factors were projected to increase mortality due to global reductions in food availability, fruit and vegetable consumption, and red meat consumption (Springmann et al 2016). Further, temperature increases are reducing the protein and micronutrient content of major cereal crops, which is expected to further affect food security (Myers et al 2017).
Vector-borne diseases
Malaria: Ten projections of the potential impacts of climate change on malaria globally and for China, Asia, Africa, and South America (see SM tables S7 and S8) (Caminade et al 2014, Khormi and Kumar 2016, Kwak et al 2014, Laporta et al 2015, Ren et al 2016, Semakula et al 2017, Song et al 2016, Tompkins and Caporaso 2016, Yamana et al 2016), confirmed the conclusions of Smith et al (2014) that weather and climate were among the drivers of the geographic range, intensity of transmission, and seasonality of malaria, and that the influences of temperature and precipitation are nonlinear. Within the context of an observed 62% decline in malaria mortality since the year 2000 (WHO 2016), the three studies projecting risks at 1.5 and 2 °C generally concluded the burden of malaria could increase with climate change because of a greater geographic range of the Anopheles vector and/or a longer season of disease transmission (Ren et al 2016, Semakula et al 2017, Song et al 2016). Relationships between temperature and disease incidence are not necessarily linear, resulting in complex patterns of changes in risk with additional warming. Some regions are projected to become too hot and/or dry for the Anopheles mosquito, such as in northern China and parts of south and southeast Asia (Khormi and Kumar 2016, Semakula et al 2017, Tompkins and Caporaso 2016, Yamana et al 2016). Vector populations are projected to shift in some regions with climate change, with expansions and reductions depending on the degree of local warming, the ecology of the vector, and other factors (Ren et al 2016).
Aedes and dengue: The Aedes spp. mosquito is the vector for dengue, chikungunya, yellow fever, and Zika viruses. Recent projections focused on the geographic distribution of Aedes aegypti and Ae. albopictus (principal vectors for these diseases) or on the prevalence of dengue fever, generally concluding the abundance of mosquitoes will increase by the 2030s and beyond compared to present, as will their geographic range, and suggesting more individuals at risk of dengue fever, with regional differences (see SM tables S7 and S8). (Banu et al 2014, Bouzid et al 2014, Butterworth et al 2017, Campbell et al 2015, Colon-Gonzalez et al 2013, Fischer et al 2011, Fischer et al 2013, Jia et al 2017, Khormi and Kumar 2014, Liu-Helmersson et al 2016, Mweya et al 2016, Ogden et al 2014a, Proestos et al 2015, Ryan et al 2017, Tagaris et al 2017, Teurlai et al (2015), Tjaden et al 2017, Williams et al 2014, Williams et al 2016) Projections at global and regional levels include North America, Southern Europe, Australia, China, Asia, New Caledonia, and Tanzania. As noted for Anopheles mosquitoes, some regions may become too hot and/or dry for Aedes spp. (Khormi and Kumar 2014). Six studies projected that, all else equal, exposure to Aedes mosquitoes and Aedes-transmitted viruses is projected to increase with greater warming (i.e. 2.0 °C vs 1.5 °C) (Bouzid et al 2014, Colon-Gonzalez et al 2013, Fischer et al 2011, Fischer et al 2013, Ogden et al 2014a, Mweya et al 2016). Climate change is projected to expand the geographic range of chikungunya, with greater expansion with higher degrees of warming (Tjaden et al 2017).
West Nile virus: Projections for North America and Europe under climate change suggested a latitudinal and altitudinal expansion of regions climatically suitable for West Nile virus transmission, particularly along the current edges of its transmission. They also suggested extension of the transmission season and an increase in the number of human cases, with the magnitude and pattern of changes varying by location and degree of warming (see tables S7 and S8 Belova et al 2017, Brown et al 2015, Chen et al 2013, Harrigan et al 2014, Morin and Comrie 2013, Semenza et al 2016). One study projected greater risks at 2.0 C than 1.5 C (Semenza et al 2016).
Lyme disease and other tick-borne diseases: Most projections concluded that climate change may expand the geographic range or shift the seasonality of Lyme and other tick-borne diseases in parts of North America and Europe (see SM tables S7 and S8) (Dhingra et al 2013, Feria-Arroyo et al 2014, Levi et al 2015, Monaghan et al 2015, Ogden et al 2014b, Porretta et al 2013, Simon et al 2014, Williams et al 2015). Two studies projected risks at 1.5 °C and 2 °C (Levi et al 2015, Ogden et al 2014b). If increased temperatures result in greater abundance of ticks and increased contact rates with humans, an earlier onset of the disease season may result in more cases with greater degrees of warming.
Other vector-borne diseases: Among studies projecting the risks of other vector-borne diseases (Carvalho et al 2015, Ceccarelli and Rabinovich 2015, Domsa et al 2016, Garza et al 2014, Gonzalez et al 2014, Kartashev et al 2014, McIntyre et al 2017, Medone et al 2015, Ochieng et al 2016), two projected risks for Chagas disease and leishmaniasis at 1.5 °C and 2 °C (see SM tables S7 and S8) (Ceccarelli and Rabinovich 2015, Gonzalez et al 2013). Overall, projections of other vector-borne diseases suggest climate change could increase or decrease future health burdens, with greater risks at higher degrees of warming.
Discussion
Detection and attribution studies indicate climate change is already adversely affecting human health (e.g. Ebi et al 2017), indicating that dangerous anthropogenic interference with the climate system is occurring. Identifying the magnitude and pattern of risks under different transient and stabilization increases in SAT can help inform the level of ambition and timing of adaptation and mitigation strategies and policies.
This comprehensive review summarizes the growing number of projections of the health risks of climate change, showing that higher global and regional SATs are generally detrimental to a wide array of climate-sensitive health outcomes. The evidence suggests high agreement among most studies, with broadly similar estimated risks for a particular exposure. It also highlights that the diversity of baselines, scenarios, and climate and health models used in the studies preclude the possibility of quantifying health risks across many of them (e.g. conduct a meta-analysis) but that there would be significant benefits in doing so.
Higher ambient temperatures and humidity levels can place additional stress on individuals engaging in physical activity. Measures of heat stress, particularly the wet bulb globe temperature, were developed to monitor environmental conditions during work and exercise, to determine when heat exposure could be hazardous (NIOSH 2016). With continued exposure to high ambient temperatures, and without interventions to lower core body temperature, heat stress can progress through heat stroke to death (Hanna and Tait 2015). Characteristics of the individual (e.g. age, health status, and level of physical fitness), type of activity (e.g. degree of exertion), and other factors determine disease progression. Heat stress can be reduced through adaptation by modifying metabolic heat production or heat exchange associated with convection, radiation, or evaporation. Projections of the risks of heat stress and heat mortality in warming climates do not take into account these and other critical factors, leading to low confidence in estimates of how health burdens could change with climate change.
As temperatures and other weather variables continue to change, health models need to consider how to most appropriately represent temperature-related risks in what are now the tails of the exposure distribution (e.g. extreme temperature events). Assumptions about the shape of associations in the upper tails of what is now current exposure(s) were not always stated in the studies reviewed. Particularly for high temperatures, recent projections were often based on mathematical functions where the shape of the exposure-response curve is highly non-linear (Gasparrini et al 2015, WHO 2014). But some functions, such as natural cubic splines, are likely to become linear beyond the range of the observations, which means they may not provide accurate estimates of future risks (Rocklov and Ebi 2012). Linear assumptions can significantly affect the magnitude of projected heat-related mortality risks (Rocklov and Ebi 2012). Assumptions of linearity in earlier projections were common, although it is unlikely that heat-related mortality will increase linearly with higher temperature increases because of acclimatization and adaptation, including changes in the built environment (Astrom et al 2013, Dong et al 2015, Hajat et al 2014, Huynen and Martens 2015, Kingsley et al 2016, Li et al 2015, Marsha et al 2016, Martinez et al 2016, Vardoulakis et al 2014, Wang et al 2016, WHO 2014). Assumptions about the shape of the relationships in vector-borne disease projections were often unclear. Not accounting for non-linear responses to changing hazards means that projections could over- or underestimate risks, and may not account for surprises.
Therefore, as good practice, we recommend that studies projecting health risks from climate change state assumptions about the shape of associations assumed between exposure(s) and health outcomes at higher degrees of temperature change. This is important because the effectiveness of public health interventions to adapt to hotter temperatures will depend on the accuracy of estimates of health risks associated with future warming.
We also recommend that future projections report global and regional mean SAT changes to increase comparability across studies to understand the magnitude of the challenges that will likely need to be addressed, and to estimate when those temperature changes are likely to arise; the latter can inform the timing of when adaptation interventions will likely be necessary. Developing a set of common scenarios, combining climate projections under a range of emission pathways and multiple socioeconomic development pathways, would facilitate comparisons across studies.
Without reporting temperature change associated with modeling choices in studies, it is not readily apparent when projections cross important policy-relevant temperature thresholds that could increase potential harm to population health. Reporting temperature change would make the information from health projections much more useful to decision makers planning adaptations to manage health risks. By comparing different scenarios at each degree of temperature change, it would be useful to compare the outcomes under the same temperature change, but with different levels of economic development as expressed through the Shared Socioeconomic Pathways (Ebi 2014). For example, the degree of projected temperature change is similar under RCP4.5 in the year 2100 to under RCP8.5 in 2050; however, socioeconomic conditions (e.g. economic growth, population, technology, policies and institutions) will change. Therefore, exposure to climate-related hazards and adaptive capacity will differ in 2050 and 2100.
Reporting of time slices is also needed to provide insights into the urgency associated with developing adaptation interventions to protect health and into how quickly mitigation policies can reduce the magnitude of climate change to which individuals, communities, and health systems will need to adapt. In health systems, apart from planning infrastructure investments with long lifetimes, most adaptations focus on relatively short time scales, such as implementing early warning and response systems. Long-term adaptation constitutes a series of sequential short-term decisions within an iterative risk management framework (Ebi 2011, Hess et al 2012). Therefore, projections of the magnitude and pattern of health risks in, for example, 2030, can inform adaptation planning over the next decade, while projections of risks in 2050 can inform adaptation over the subsequent decades.
To inform studies projecting future health risks, it would be helpful for health researchers to develop scenarios of population health and health systems development over this century that can extend the climate change and development scenarios (e.g. Shared Socioeconomic Pathways) (Ebi 2014, Sellers and Ebi 2017). Narratives and quantifications are needed at regional and global scales of how critical parameters affecting health could evolve under different development pathways, to improve the ability to quantify morbidity and mortality among different groups such as vulnerable people. These factors can include the extent to which health systems will be prepared to manage changing health burdens; inequities in health and income; and drivers traditionally outside the health sector, such as travel and tourism. These scenarios would support more robust projections of the magnitude and pattern of health risks associated with different degrees of regional and global changes in SAT, precipitation, sea level rise, and other variables, under different trajectories of population exposure and vulnerabilities. With this information, it would be possible to project the range of possible health benefits and risks associated with different policy choices to address climate change and its associated risks.
Scenarios also need to explicitly incorporate adaptation assumptions. For example, while planned adaptation to reduce impacts of heat on health (e.g. heat warning systems, air conditioning, monitoring and surveillance) can be effective (Anderson et al 2016, Toloo et al 2013, White et al 2017), the magnitude of risk reduction associated with specific measures is largely unknown (Deschenes 2014). Without incorporating estimates of the effectiveness of adaptation at various time slices and degrees of temperature change in studies, projected health risks are unlikely to accurately estimate the magnitude of the challenges to be managed. For example, assumptions of a constant increase in successful heat adaptation in projections of future heat health risks from climate change do not capture the complexity of regionally specific vulnerability factors and the non-linearity of climate responses, thereby significantly underestimating risks to health (Ebi et al 2016). Scenarios also need to consider limits to adaptation, such as physiological limits to acclimatization to higher temperatures.
Climate change and health vulnerability and adaptation assessments can provide a rich source of quantitative and qualitative data for planning appropriate adaptive responses to rapid climatic shifts (WHO 2013). Integration of information about non-linear relationships associated with climate and health responses into future research and the application of iterative risk management approaches to prepare for impacts are needed to reduce risks of very severe impacts (Ebi et al 2016, Hess and Ebi 2016).
There are multiple sources of uncertainty in the analyses, including uncertainties in the climate models and greenhouse gas emission pathways, assumptions underlying health models, accuracy of the health models, extent of robust inclusion of adaptation, and others. We introduced another source of uncertainty by using the decade 2010–2019 as the baseline; we did this because of the challenges of different historic starting points for each GCM. This was not likely the largest source of uncertainty.
Conclusions
Overall, the health risks of a global mean SAT increase of 2 °C above pre-industrial temperatures are projected to be greater than the risks for an increase of 1.5 °C, with generally even higher risks at greater increases in SAT. The risks may be particularly elevated for heat-related morbidity and mortality, heat stress, ground-level ozone, and undernutrition. For vector-borne diseases, the risks are more variable because warmer temperatures may result in some regions becoming too hot and/or too dry for a vector. Future concentrations of particulate matter could increase or decrease, depending on emission assumptions and projected changes in precipitation. Despite the limitations, this review supports the ambition of rapidly reducing greenhouse gas emissions to increase the probability that health risks will stay within manageable boundaries.
The Paris Agreement is an important and possibly unique opportunity for the climate and health research enterprise to inform effective decisions to prepare for and manage the health risks of additional climate change, from local to international levels. Providing policy-relevant projections of the health risks of climate change will increase the possibilities of protecting and promoting population health, today and in the future.
Acknowledgments
Part of this research was supported by the National Aeronautics and Space Administration (NNX16AO98G). The National Center for Atmospheric Research is sponsored by the National Science Foundation.