Adverse heat-health outcomes and critical environmental limits (PSU HEAT Project)

Abstract

Background:

The earth’s climate is warming and the frequency, duration, and severity of heat waves are increasing. Meanwhile, the world’s population is rapidly aging. Epidemiological data demonstrate exponentially greater increases in morbidity and mortality during heat waves in adults ≥65 years. Laboratory data substantiate the mechanistic underpinnings of age-associated differences in thermoregulatory function. However, the specific combinations of environmental conditions (i.e., ambient temperature and absolute/relative humidity) above which older adults are at increased risk of heat-related morbidity and mortality are less clear.

Methods:

This review was conducted to (1) examine the recent (past 3 years) literature regarding heat-related morbidity and mortality in the elderly and discuss projections of future heat-related morbidity and mortality based on climate model data, and (2) detail the background and unique methodology of our ongoing laboratory-based projects aimed toward identifying the specific environmental conditions that result in elevated risk of heat illness in older adults, and the implications of using the data toward the development of evidence-based safety interventions in a continually-warming climate (PSU HEAT; Human Environmental Age Thresholds).

Results:

The recent literature demonstrates that extreme heat continues to be increasingly detrimental to the health of the elderly and that this is apparent across the world, although the specific environmental conditions above which older adults are at increased risk of heat-related morbidity and mortality remain unclear.

Conclusion:

Characterizing the environmental conditions above which risk of heat-related illnesses increase remains critical to enact policy decisions and mitigation efforts to protect vulnerable people during extreme heat events.

INTRODUCTION

The earth’s climate is warming well above historical averages. Accompanying the rise in average global temperatures is an increased frequency, duration, and severity of heat waves (Hansen, Sato, & Ruedy, 2012; Hayhoe, Sheridan, Kalkstein, & Greene, 2010; Meehl, Tebaldi, Walton, Easterling, & McDaniel, 2009; Nakano, Matsueda, & Sugi, 2013; Perkins, Alexander, & Nairn, 2012; Semenza et al., 1996). At the same time, the world’s elderly population is expanding rapidly, with the number of people ≥65 years in the U.S. alone projected to increase to nearly 80 million by the year 2040 (“A profile of older Americans: 2012,” 2012). By 2050, there will be more people over age 60 than under age 15 in the world for the first time (Division). People ≥65 years exhibit disproportionately larger increases in morbidity and mortality during heat waves than younger individuals, with the large majority of excess deaths during heat waves occurring in adults over the age of 65 years (Conti et al., 2005; Kovats & Hajat, 2008). In one California study, excess mortality was found to increase by 4.3% for every 5.6°C increase in apparent temperature (Basu & Malig, 2011) (see Appendix for definition of apparent temperature). Across 15 European cities, an increase in apparent temperature of 1°C above a threshold temperature unique to each city was associated with a >3% increase in mortality in Mediterranean cities and a 2% increase in mortality in northern European cities (Baccini et al., 2008).

Humans are tropical animals, having evolved with a high eccrine sweat gland density and relatively small amount of hair compared with other primates, allowing the ability to dissipate large amounts of heat during bouts of intense physical activity in warm climates (Hanna & Brown, 1983). Although a comprehensive treatment of the physiology of heat stress is beyond the scope of this review, others have reviewed this topic in depth (Balmain, Sabapathy, Louis, & Morris, 2018; Cramer & Jay, 2016; Kenney & Munce, 2003; Périard, Eijsvogels, & Daanen, 2021). Briefly, in addition to eccrine sweating, thermoregulatory responses to heat stress involve increased skin blood flow, cardiac output, pulmonary ventilation, and reduced renal and splanchnic blood flow. A wealth of laboratory data confirms the mechanistic underpinnings of age-associated differences in thermoregulatory function, including decrements in sweating, attenuated skin blood flow, excess cardiac stress, etc. (Greaney, Kenney, & Alexander, 2016; Greaney, Stanhewicz, Proctor, Alexander, & Kenney, 2015; Holowatz, Thompson-Torgerson, & Kenney, 2007; Minson & Kenney, 1997; Stanhewicz, Alexander, & Kenney, 2013). Thus, in older adults for whom function of various organ systems (e.g., the heart, kidneys, lungs, etc.) are often compromised, extreme and/or prolonged heat stress may result in physiological demands that exceed the capabilities of the aging human body. Such increased physiological strain contributes to excess heat-related morbidity and mortality.

Although it is well-established that high ambient temperatures are associated with increases in heat-related morbidity, mortality, and hospital admissions, especially in those aged over 65 years (Conti et al., 2005; Kovats & Hajat, 2008), exploring the associations between heat waves or extreme heat events and human health can be complex and inconsistent given the lack of a globally promulgated set of characteristics to define a heat wave (Xu, FitzGerald, Guo, Jalaludin, & Tong, 2016). Multiple measurements of the ambient environment (e.g., dry-bulb temperature, Heat Index, Wet Bulb Globe Temperature, etc.), the statistical distribution of those measurements (e.g., mean, maximum, minimum, range), the extreme event threshold (absolute vs. relative; e.g., 35°C or 95^th percentile), and duration of an event (one vs. consecutive days) all lead to differing analyses. Relative risk (RR) at temperatures higher than a location’s minimum mortality temperature can represent the impact of heat on mortality, but minimum mortality temperatures are typically very mild and dependent on latitude and altitude (Q. Yin, Wang, Ren, Li, & Guo, 2019). Therefore, it is difficult to determine a critical limit above which there is an increase in morbidity and/or mortality due to extreme heat and humidity. As Sun et al. (2021) (and references therein) note in the largest study examining the impact of heat on emergency department visits in the United States, there is “no clear threshold” in determining or predicting excess morbidity or mortality due to extreme heat.

To establish environmental thresholds above which risk of heat-related morbidity and mortality start to increase, particularly in older adults and other vulnerable populations, ongoing experiments in our laboratory are aimed at examining the combinations of temperature and humidity (critical environmental limits) above which a stable body core temperature (T_c) cannot be maintained (i.e., heat stress becomes uncompensable). Thus, the purpose of the current review is to (1) examine the most recent (past three years) literature regarding heat-related morbidity and mortality and discuss projections of future heat-related morbidity and mortality based on climate model data, and (2) describe the ongoing Pennsylvania State University Human Environmental Age Thresholds (PSU HEAT) project that is designed to experimentally establish critical environmental limits for the maintenance of heat balance above which young and aged adults and other vulnerable populations may be at increased risk of heat-related morbidity and mortality.

REVIEW OF THE RECENT LITERATURE

Review Methodology

Here, we provide an update to the review from Åström, Bertil, and Joacim (2011) to further examine the impact of heat waves on morbidity and mortality on the elderly. We follow the same methodology with three mentionable caveats:

1. Our review of PubMed covers the same time span (three years) but identifies studies initially published between January 1, 2018 and December 31, 2020, a one decade shift forward.

2. We searched PubMed with the same key words as the previous work with one change. Mortality, morbidity, elderly, and temperature are all included in the search. We expand upon the “heat wave” search term in include “heat wave OR heatwave OR extreme temperature event OR extreme heat event” given the increasing usage in the literature of the latter two terms.

3. Projected future heat-related morbidity and mortality based on climate model data are included in addition to analyses of historical data.

The same publication inclusion criteria as those of Åström et al. (2011) were used; namely that studies were included only if they examined the associations between high temperature and morbidity/mortality, focused on the elderly either as the main population or as a sub-population, and presented results as RR, incidence risk ratio (IRR), odds ratio (OR) or similar measure. If raw values were not provided in the publication’s text or included tables, best estimates were made from the figures in which the data were depicted. 114 studies resulted from the search terms used in the PubMed database, with 35 of those studies included in this review. Of the 35 retained studies, 20 detailed past morbidity, 13 examined past mortality, and 2 focused on future mortality, a category which Åström et al. (2011) did not include.

Morbidity

Unsurprisingly, the literature review indicates that extreme heat continues to be detrimental to the health of the elderly and that this is apparent across the world (Table 1). In a study in Perth, Australia, general ambulance call out rates were 11 times higher during heatwave events for residents over the age of 60 years (Patel et al., 2019). Numerous heatwave definitions led to a 4–18% increase in heat-related hospitalization rates for those aged 80+ years in Brazil (Zhao et al., 2019). Similar results were reported in Ho Chi Minh, Vietnam (Dang et al., 2019) and the United States (Layton et al., 2020). Only one study, conducted in Tasmania, did not report an increased risk in heat-related hospitalization (Campbell, Remenyi, Williamson, White, & Johnston, 2019). The impacts of heat waves on morbidity appear to be influenced, to some degree, by temporality. A country-wide assessment in the United States (Liss & Naumova, 2019) found that the first and second heatwaves of a warm season were associated with higher RR of hospitalization than subsequent events, though regional differences were apparent.

Table 1.

Studies of the associations between heat waves/high temperature and morbidity published between 2018 and 2020.

Reference	Region	Exposure and threshold	Outcome variable	Change in outcome variable
Sherbakov et al. (2018)	California, USA, 1999 – 2009	Month-specific heatwave vs. non-heatwave day	RR of hospitalizations for various symptoms related to temperature	Respiratory disease, ~1.07; CVD with diabetes, ~1.05; Respiratory with diabetes, ~1.13
van Loenhout et al. (2018)	The Netherlands, 2002–2007	Daily max. temperature compared to average and MMT of 21°C	RR of ER admissions for heat-related, respiratory, and circulatory diseases as well as fractures of the femur (control group) after one day of 32°C heat	Heat-related: 65–84 y, 1.18; 85+ y, 1.19. Respiratory: 65–84 y, 1.08; 85+ y, 1.10. Circulatory 65–84 y, 1.00 (not sig.); 85+ y, 1.02
Song et al. (2018)	Beijing, China, 2009–2012	Heatwave vs. non-heatwave day	ER visits for respiratory disease during heatwaves in the 95–98th percentiles and the 99th percentile	95–98th %ile: ~1.2–1.3;. 99th %ile: 1.4–1.7 based on length of heatwave
Hopp et al. (2018)	1943 counties in the U.S., 1999–2010	Heatwave vs. non-heatwave day	RR of hospital admission for 50 different diagnoses	11 diagnoses with sig increases in RR. Highest (heat stroke), 22.55
Xiaole Liu et al. (2018)	Beijing, China, 2013–16	99th vs. 90th percentiles of maximum apparent temperature	RR of hospital admission for acute myocardial infarction over 21 day period	No sig increase in RR
Xuena Liu et al. (2019)	Jinan, China, Summer 2010	Heatwave vs. non-heatwave day	OR of daily hospitalization for mental illness	3.034 for old vs. young
Zhang et al. (2019)	Cangnan, China, July 2010 to October 2012	Heatwave vs. non-heatwave day	Daily hospitalization for respiratory ailments	Strongest: Lag 5, 1.412 for total respiratory disease; Lag 4, 1.794 for upper RT infection. Also sig: upper respiratory at Lag 0, 1.686
Dang et al. (2019)	Ho Chi Minh City, Vietnam, 2010–2013	Heatwave days vs. MMT	RR of daily hospitalization	1.28
Zhao et al. (2019)	1,814 cities in Brazil, 2000 – 2015	Heatwave vs. non-heatwave day	Daily all-cause hospitalization	80+ y was only group with increase in hospitalization regardless of heatwave definition. 4–18% increase
Borg et al. (2019)	Adelaide, Australia; July 2003 - March 2014	Heatwave vs. non-heatwave day	RR of daily emergency department and hospitalizations for urinary disease	ED (sig): Total urinary disease, 1.063; kidney failure, 1.227; acute kidney injury, 1.217 Inpatient (sig): Pyelonephritis, 1.3
Adeyeye et al. (2019)	New York State, USA, 2008–2012	1°C change in maximum temperature	RR of daily emergency department and hospitalizations for heat stress, dehydration, acute kidney failure (AKF), and cardiovascular diseases (CVD)	Heat stress: 65–84 y, 1.386; 85+ y, 1.4; Dehydration: 65–84 y, 1.021; 85+ y, 1.022; AKF: 65–84 y, 1.014; 85+ y, 1.016
Liss and Naumova (2019)	United States, 1991 – 2006	Heatwave vs. non-heatwave day	Hospitalization for heat stroke	First heatwave of year, 10.4; second, 11.4; fourth, 4.78
Xu et al. (2019)	Brisbane, Australia, 2005 – 2013	Heatwave vs. non-heatwave day	RR or hospitalization for diabetes	1.01 (not sig)
Patel et al. (2019)	Perth, Australia, 2006 – 2015	Heatwave vs. non-heatwave day	RR of ambulance calls	60+ y, 11.5
Parry et al. (2019)	Sydney, Australia 2001 – 2013	Heatwave vs. non heatwave with added effect of PM10	RR of daily emergency hospital admission for cardiovascular ailments	Cardiac arrest (max temp, lag 1, w/ PM10), 1.85; (mean temp, lag 1, w/ PM10), 1.65
Campbell et al. (2019)	Tasmania, Australia, 2008 – 2016	Severe and extreme heatwaves vs. non-heatwaves	OR of daily emergency department data	1.06 (non-sig)
Brennan et al. (2019)	Ireland, June 24 - July 3, 2017/18	Control event vs. heatwave event (>25C for 5 days)	Serum sodium samples	2.5 times higher likelihood of hypernatremia during heatwave vs. control
García-Lledó et al. (2020)	Mardid, Spain, June 2013 - June 2017	MMT of 18°C vs. 35°C	IRR of ST-segment elevation myocardial infarction cases	1.03 (non-sig)
Sohail et al. (2020)	Helsinki, Finland 2001 – 2017	Heatwave vs. non-heatwave day	RR of daily hospital admissions for all respiratory diseases	22.46 for all respiratory in 75+ y
Layton et al. (2020)	USA, 2007 – 2012	Heatwave vs. non-heatwave day	RR of heat-related hospitalizations	1.21 – 1.37 depending on medication type

MMT, minimum morbidity temperature; RR, relative risk; IRR, incidence risk ratio; OR, odds ratio; ER, emergency room; RT, respiratory tract; PM₁₀, particulate matter with diameter of less than 10 microns; CVD, cardiovascular disease.

Several diseases and associated symptoms were associated with increased risk of emergency medical service, emergency room visits, or hospitalization risk. A comprehensive study of heat-related illness in United States Medicare enrollees found 11 diagnoses associated with significantly increased RR for hospitalization over the period 1999 – 2010. Fluid loss and associated sequelae (septicemia, urinary tract infections, and kidney failure) made up a majority of the significant diagnoses during heatwaves (Hopp, Dominici, & Bobb, 2018). Similar results were found in an Irish heatwave event in 2018; older adults were 2.5 times more likely to experience hypernatremia due to water loss than a control group in thermoneutral meteorological conditions (Brennan, Murray, O’Shea, & Mulkerrin, 2019). These same associations with water loss were found in a study across New York State where increases in hospitalizations and emergency department visits for heat stress, dehydration, and acute kidney failure were significant and similar between age groups 65–84 years and 85+ years (Adeyeye et al., 2019). In Adelaide, Australia, when defining heatwaves using the Excess Heat Factor (see Appendix for definition) rather than absolute temperature, risk of emergency department visits increased for older adults with kidney failure, acute kidney injury, and chronic kidney disease (Borg et al., 2019).

Deleterious impacts of heatwaves on respiratory diseases were also apparent. In The Netherlands, heat and heatwave effects were most pronounced on the first day of the heat wave, leading to increased heat-related illness and respiratory disease-related emergency room admissions in those aged 65 years and older (van Loenhout et al., 2018). Similar results were observed in Finland (Sohail, Kollanus, Tiittanen, Schneider, & Lanki, 2020) and China (Song et al., 2018; Zhang et al., 2019). In California, respiratory hospitalizations were increased in residents over the age of 65 years during heatwave events even after taking seasonal acclimatization into account (Sherbakov, Malig, Guirguis, Gershunov, & Basu, 2018). Additionally, diabetes as a co-morbidity increased the RR of hospitalization for respiratory- and cardiovascular-related illness. However, an Australian study examining the impact of heat on diabetes hospitalizations alone found no significant increase in risk (Xu et al., 2019).

Lesser morbidity risk was associated with cardiovascular disease during heatwaves unless accompanied by a co-morbidity. Heatwaves in Sydney, Australia were associated with increases in hospitalization for cardiovascular conditions only when coupled with high PM₁₀ (i.e., particulate matter with a diameter less than 10 microns) episodes (Parry, Green, Zhang, & Hayen, 2019). A study in Beijing, China showed no significant increase in hospitalizations due to acute myocardial infarction between the 90^th and 99^th percentiles for daily mean temperature, daily minimum temperature, or daily minimum apparent temperature (Xiaole Liu et al., 2018). Similar findings were observed in Madrid, Spain (García-Lledó et al., 2020). Conversely, a recent systematic review and meta-analysis suggested that every 1°C rise in temperature above “reference temperatures” was associated with a 0.5% increased risk of morbidity due to arrhythmias, cardiac arrest, and coronary heart disease (J. Liu et al., 2022). Although the data are conflicting regarding the impact of extreme heat events on cardiovascular morbidity, there are clear increases in cardiovascular-related mortality during heatwaves (Khatana, Werner, & Groeneveld). It has been speculated that the lack of increases in morbidity, despite significant increases in mortality, may reflect a rapid progression from cardiovascular events to death before they can be admitted to a hospital (Åström et al., 2011; Linares & Díaz, 2007).

An important emerging view on detrimental heat-health effects has focused on the impact of extreme temperature on mental health. A one-summer, case-crossover study in Jinan, China noted that older adults were three times more likely than young adults to be admitted to the hospital for mental illness during extreme heat episodes (Xuena Liu, Liu, Fan, Liu, & Ding, 2019).

Mortality

Acclimatization.

Heat acclimatization elicits physiological adaptations that improve thermoregulatory function (e.g., reductions in resting T_c, lower heat rate, and increased sweating and cutaneous vasodilation) and may reduce risk of heat-related morbidity and mortality. Temporal changes in the association between extreme heat and mortality were apparent in studies on both seasonal and long-term time scales (Table 2). On the seasonal scale, a study of four heatwaves between 2013 and 2017 in Istanbul, Turkey showed effects of acclimatization and/or early-season harvesting on mortality (Can et al., 2019). Two of the investigated heatwaves occurred in August, well into the warm season, and were associated with non-significant risk for heat-related mortality. Conversely, two heatwaves occurring in June and July were associated with increased RR of death during the event. In Slovenia, longer-term effects were observed by comparing heatwaves that occurred in 2003 and 2015. Circulatory system-associated mortality was significantly higher in 2015 compared to 2003. While heat waves were longer in duration and occurred at higher temperatures, Slovenian life expectancy also increased in the latter period. These two factors likely both contributed to the increased circulatory system-associated mortality in 2015 compared to 2003. Additionally, all-cause mortality in those aged 75+ years during the 2015 event was significantly higher relative to non-heat wave control events, while the increase in all-cause mortality during the 2003 heatwave was not significant (Perčič, Kukec, Cegnar, & Hojs, 2018).

Table 2.

Studies of the associations between heat waves/high temperature and past mortality published between 2018 and 2020.

Reference	Region	Exposure and threshold	Outcome variable	Change in outcome variable
Perčič et al. (2018)	Slovenia; 2003 and 2015	Heatwaves vs. non-heatwaves in summers of 2015 compared to 2003	Mortality: all-cause, circulatory, respiratory in 75+	2015 sig: 75+ y all cause, 1.11; circulatory, 1.17; 2015 vs. 2003 sig: circulatory, 1.34
Cheng et al. (2018)	Australia, 1988 to 2011	Heatwave vs. non-heatwave day	All-cause mortality	28% increase in mortality on average across the country during heatwave events
Pyrgou and Santamouris (2018)	Cyprus, 2007 – 2014	Heatwave vs. non-heatwave day	Respiratory and circulatory mortality	Higher RR in urban areas: 65–69 y males, 2.38; 65–69 y females, 2.57; 70–74, 2.49
P. Yin et al. (2018)	272 Chinese cities, 2013–2015	Heatwave vs. non-heatwave day	RR of all-cause mortality	In 75+ y age group, all heatwave definitions sig. Highest: 97.5th %ile 2-day duration, ~1.10; 3-day, ~1.15 (estimated from figures)
Yang et al. (2019)	31 Chinese cities, 2007 – 2013	Heatwave vs. non-heatwave day	All-cause mortality, 65–74 and 75+	Lag 0: 65–74 y, 1.05; 75+ y, 1.08; Lag 0–2: 65–74 y, 1.08; 75+ y, 1.12; Lag 0–10: 65–74 y, 1.07; 75+ y, 1.17
Dang et al. (2019)	Ho Chi Minh City, Vietnam, 2010–2013	Heatwaves vs. MMT	RR of all-cause mortality	1.43
Gronlund et al. (2019)	8 Michigan cities, 1990–2007	Heatwave vs. non-heatwave day	RR of all-cause mortality	75+ y: 32.5°C, 1.01; 35°C, 0.98 (both non-sig)
Singh et al. (2019)	Varanasi, India, 2009 – 2016	Heatwave vs. non-heatwave day	RR of all-cause mortality	1.2 (6.83% increase in mortality risk for 1°C summer temperature increase)
Aboubakri et al. (2019)	Kerman Region, Iran, 2005 – 2017	Heatwave vs. non-heatwave day	RR of all-cause mortality	95th %ile: 4 days, 1.195; 98th %ile: 2 days, 1.43; 98th %ile: 3 days, 1.579
Can et al. (2019)	Istanbul, Turkey, 2013 – 2017	4 heatwave events vs. reference periods	All-cause mortality	In 75+ age group, sig RRs: HW1, 1.16; HW3, 1.38. Non-sig RRs: HW2, 1.03; HW4, 0.93
Kang et al. (2020)	South Korea, 2011 – 2017	Heatwave vs. non-heatwave (varying threshold points for urban vs. rural)	RR of all-cause mortality	Total max RR (98th %ile), 1.13; urban, (95th %ile), 1.13, rural (98th %ile), 1.20
Kim et al. (2020)	119 urban districts, Korea, 2008,2017	Heatwave vs. non-heatwave day	All-cause mortality	Lag-cumulative HW-mort risk: 11.6%, Lag-cumulative HW added effects: 7.4%
Sharafkhani et al. (2020)	Urmia, Iran, 2010 – 2016	H1: Heatwave vs. non-heatwave (95th of mean temp for 2 days), H2: 3 days of Tmax or Tmin at or above 90th percentile and max temp exceeding monthly mean by 5°C	All-cause mortality	For 75+ y at Lag 0, Main effect: H1, 1.64; H2, 1.53. Added effect: H1, 1.27; H2, 1.16. Non-sig for 65–74 y grouping

MMT, minimum mortality temperature; RR, relative risk.

Location.

Location is another determinant of heat-related mortality risk, likely due to socioeconomic factors such as access to and quality of healthcare, ability to mitigate extreme heat effects (i.e. access to air condition or electric fan use), and influence of the nearby built environment. A 1°C increase in mean summer temperatures and heatwave events were both associated with increased mortality in Varanasi, India (Singh, Mhawish, Ghosh, Banerjee, & Mall, 2019). Conversely, in Michigan, USA, no significant increases in all-cause mortality were found during heatwaves (Gronlund, Cameron, Shea, & O’Neill, 2019). Differential risk also occurs within countries; heat-related mortality is apparent in both urban and rural areas, although results are inconsistent as to which setting is associated with greater risk. In Cyprus, urban living was associated with higher RR in respiratory and circulatory mortality in comparison to rural settings, likely due to the urban heat island effect on temperatures and pollutants (Pyrgou & Santamouris, 2018). However, the maximum RR for all-cause heat-related mortality was higher in rural areas of Korea compared to urban areas, although mortality significantly increased during heat waves in both settings and maximum RR in rural areas occurred at a higher temperature threshold (Kang et al., 2020). Heterogeneity in mortality risk within urban areas also exists, as elderly who were more socially isolated in urban areas in Korea were more likely to die during heatwaves than those with more social connections (Kim, Lee, Kim, & Cho, 2020).

Age.

When age cohorts among the elderly were examined, most results suggest an increasing trend in mortality risk with increasing age. Heatwaves in Australia led to an average 28% increase in mortality in populations aged 75+ years, with significant risk across all regions and climate zones of the continent (Cheng et al., 2018). Across 272 cities in China, mortality risk during heatwaves defined by multiple exposure thresholds and durations was only significant in the population aged 75+ years with the largest risk occurring in multi-day heatwaves during which the daily mean temperature was above its 97.5^th percentile (P. Yin et al., 2018). Another study conducted in 31 different cities across China showed increased risk of mortality during heatwaves across several different threshold and duration definitions in all populations aged 65+. However, risk was exacerbated with increasing lag time (up to 10 days) post-heatwave event, particularly in those aged 75+ years (Yang et al., 2019). Studies in Iran similarly demonstrated increasing trends in RR with heatwave threshold and duration (Aboubakri, Khanjani, Jahani, & Bakhtiari, 2019) and the added heatwave risk to those aged 75+ years (Sharafkhani, Khanjani, Bakhtiari, Jahani, & Entezarmahdi, 2020).

Future Projections

Two studies used future temperature estimates from climate models, in combination with past population mortality RRs during heatwaves, to determine how a warming climate might affect mortality in the future (Table 3). A study projecting mortality during heatwave events in the city of Jakarta, Indonesia suggested that mortality will be increased by a factor of 1.44 in 2060 under the RCP8.5 “business as usual” future climate projection due to an increase in August maximum temperatures of over 1°C (see Appendix for definition of Representative Concentration Pathway (RCP)). In comparison, if the RCP2.6 scenario (warming of ~0.5°C) were followed and strong climate mitigation actions were taken, projections for increased mortality were lowered to 1.25. Importantly, neither model accounts for future population aging (Varquez, Darmanto, Honda, Ihara, & Kanda, 2020); thus, mortality factors in that study may be underestimated.

Table 3.

Studies of the relationship between heat waves/high temperature and future mortality published between 2018 and 2020.

Reference	Region	Exposure and threshold	Outcome variable	Change in outcome variable
Varquez et al. (2020)	Jakarta, Indonesia. 2050s vs. 2010s	Daily maximum temperature above optimum August temperature of 29.4°C	Future mortality increase factor	RCP2.6, 1.25; RCP8.5, 1.44
Chen et al. (2021)	Taiwan, 2060 vs. 2018	Daily mean temperature above 30°C (95th percentile)	Increases in all-cause mortality	1–4 per 100k in 2018 to 21–99 per 100k by 2060 across 6 regions

RCP, Representative Concentration Pathway.

Chen et al. (Chen et al., 2021) investigated changes to heat-attributable mortality across six different regions of Taiwan given the impact of rising temperature projections from 2018 to 2060. Present-day heat-attributable mortality in older adults associated with ambient temperatures at or above 30°C ranged from 1 – 4 deaths per 100,000. Under RCP8.5 conditions, projected mortality increased to 21 deaths per 100,000 in the northern region of New Taipei and 99 deaths per 100,000 in the southern region of Tainan, with population demographic changes controlled for in the analysis. The increases in heat-related mortality in these cities is likely due, at least in part, to an increase in the annual number of days at or above 30°C in the future. The projected increase in number of days at or above 30°C from 2018 to 2060 was 5.5 days to 12.9 days in New Taipei and 4.1 days to 22.2 days in Tainan.

Summary

In contrast to the review by Åström et al. (2011), there were more heat-related morbidity studies than mortality studies. This is likely due to increased prevalence and quality of ambulance call, emergency department (ED), and hospitalization data in recent years, allowing for more confidence in heat-related morbidity studies as data collection and categorization and methodological approaches have been improved upon. It is also well-established that heat-related mortality is likely severely undercounted due to co-factors in cause of death in addition to what can be noted on death certificates while there is much more leeway in morbidity codes (Josseran et al., 2009).

Two included studies assumed historical risk of mortality and applied the RR values to future climates and populations. Although this is likely the best current practice for predicting future risks to extreme heat, this approach does not take into account future adaptation to heat, either physiologically or behaviorally, or how humans will respond to heat that is not currently part of the normal distribution. Cultural human adaptation has been noted as the primary cause for the reduction in heat-related mortality across the United States (and likely other first-world countries) (Hondula, Balling, Vanos, & Georgescu, 2015; Sheridan & Dixon, 2017) due to the proliferation of air conditioning and better heat-health warning systems. However, adaptive capacity is highly variable across the world. Additionally, there are questions as to whether adaptive measures will continue to hold for future temperature distributions and unprecedented extremes (Petkova et al., 2017).

This review likely does not encompass all recently published works relating to the effects of extreme heat on morbidity and mortality. Indeed, even our review sample included studies that examined a range of temperatures (with no focus on extreme heat) or did not include significance testing but demonstrated higher rates of illness with higher temperatures (e.g., Davis et al. (2020)). Recent work focuses on indoor heat exposure, whether tied to extreme heat episodes or not, and its relation to morbidity and mortality is occurring (O’Lenick et al., 2020).

Although epidemiological and modeled data are consistent regarding the deleterious impacts of climate change on heat-related morbidity and mortality, particularly in older adults, the precise environments (i.e., combinations of temperature, humidity, etc.) above which risk begins to increase remain unclear. As such, empirical examinations of the environmental limits to human heat balance can (1) establish baseline critical combinations of ambient temperature and humidity for a heterogeneous cohort of healthy young men and women and (2) contrast those critical conditions with those for older adults and other vulnerable populations. That is the purpose of the PSU HEAT project, which uses a progressive heat stress protocol to investigate age-related changes in the upper limits of heat stress compensability (i.e., critical environmental limits) in humans.

CRITICAL ENVIRONMENTAL LIMITS

To establish critical environmental limits to human heat balance (i.e., the balance between heat gain and heat loss, allowing a relatively stable T_c), a progressive heat stress protocol has been used to determine those thermal environments above which heat balance cannot be maintained (i.e., uncompensable heat stress) for a given metabolic heat production. Heat balance is only possible when heat loss mechanisms equal or exceed total body heat gain. A seminal study on this topic was published by Lind (Lind, 1963a, 1963b, 1970), in which subjects completed a series of ~1 h exposures to various thermal environments during treadmill walking. Those experiments demonstrated that T_c equilibrates proportionally to metabolic heat production and is maintained at a relatively steady state across a wide range of environments (compensable heat stress). Beyond these environments, ambient conditions combine with metabolic heat production to raise T_c (uncompensable heat stress). The delineation between compensable and uncompensable thermal environments was termed the “upper limit of the prescriptive zone.”

Subsequently, Belding and Kamon developed a time-intensive protocol (Belding & Hatch, 1963) to determine critical ambient water vapor pressures (P_crit) at 36°C during exercise at various intensities and air movement velocities. A less time-intensive version of that protocol was developed to determine the P_crit at air temperatures ranging from 30°C to 52°C during treadmill walking at intensities ranging from 25% to 43% of maximal aerobic capacity (V̇O_2max) (E Kamon & Avellini, 1976; E. Kamon, Avellini, & Krajewski, 1978). Further refinements were made to the protocol by Kenney et al. (Kenney, Mikita, Havenith, Puhl, & Crosby, 1993; W. Larry Kenney & Michael J. Zeman, 2002) to minimize the required number and duration of tests and to better define environmental isotherms in hot, dry environments (W. Larry Kenney & Michael J. Zeman, 2002) using a controllable environmental chamber to systematically manipulate ambient dry-bulb temperature (T_db) and/or water vapor pressure (P_a). This final approach separates experiments in which heat balance is limited by sweat evaporation (i.e., warm-humid conditions; P_crit experiments) from those in which maximal sweating capacity and skin blood flow limitations are the primary limiting factors (i.e., hot-dry conditions; critical dry-bulb temperature, T_crit experiments) (W Larry Kenney, Dale E Hyde, & Thomas E Bernard, 1993; Kenney et al., 1988; W. L. Kenney & M. J. Zeman, 2002; Tasha J Kulka & W Larry Kenney, 2002). As humidity increases, heat dissipation becomes limited by a narrower water vapor pressure gradient that is unfavorable for the evaporation of sweat, resulting in increased “wasted” sweating (sweat that is produced but not evaporated). In hot-dry conditions, sweat is freely evaporated and therefore evaporative heat dissipation is limited by the ability to increase sweat production. Additionally, if ambient temperature exceeds skin temperature, dry heat gain further contributes to net heat gain vs. heat loss.

Figure 1 depicts a representative tracing of the protocol used to identify critical environmental limits to heat balance. T_c, dry-bulb temperature (T_db), and ambient water vapor pressure (P_a) are continuously measured throughout the experimental trial. Ambient T_db and P_a are held constant for the first 30 min as T_c equilibrates to an elevated, relative steady state proportional to metabolic heat production (a slow rise in T_c is observed in some unacclimated subjects, constituting positive heat storage (W. Larry Kenney & Michael J. Zeman, 2002). After 30 min, either T_db or P_a is systematically increased in a stepwise fashion. During T_crit trials, ambient P_a is held constant as T_db is increased by 1°C every 5 min. Conversely, during critical P_crit trials, ambient T_db is held constant as P_a is increased by 1 mmHg every 5 min. The critical T_db or P_a is characterized by an upward inflection of T_c from steady state, which is selected graphically from the raw data. Our laboratory has selected the T_c inflection point using visual inspection and segmental linear regression analysis, demonstrating excellent agreement (ICC = 0.99) between the two methods (Wolf, Bernard, & Kenney, 2022). The average T_db or P_a for the 2 min preceding the inflection point is defined as the critical T_db or P_a, respectively.

Figure 1.

Representative tracing of the time course of core temperature (T_c; open circles, top panel), ambient water vapor pressure (P_a; closed circles, bottom panel), and ambient dry-bulb temperature (T_db; gray circles, bottom panel) for a minimal activity trial with increasing P_a. The T_gi inflection point represents the combination of environmental conditions above which heat stress becomes uncompensable and a stable core temperature can no longer be maintained. In this case, the T_gi inflection point (i.e., critical P_a, P_crit) occurs at P_a = 27mmHg.

We have recently demonstrated excellent between-visit (ICC = 0.98) and intra-rater (ICC = 0.93) reliability of this experimental approach (Cottle, Wolf, Lichter, & Kenney, 2021). Similarly, we have shown that uncompensable heat stress occurs at the same combination of T_db and P_a regardless of which environmental variable is being manipulated (T_db, ICC = 0.95; P_a, ICC = 0.96) (Cottle et al., 2021). Thus, these sophisticated environmental chamber-based studies to determine environmental limits for heat balance are highly reliable and reproducible.

PSU HEAT (Penn State Human Environmental Age Thresholds) Project

Using the progressive heat stress protocol described herein, our laboratory has a long history of identifying critical environmental limits in a variety of populations, metabolic intensities, and clothing ensembles (Dougherty, Chow, & Larry Kenney, 2010; E Kamon & Avellini, 1976; Kenney, 2020; W. L. Kenney, D. E. Hyde, & T. E. Bernard, 1993; W. L. Kenney, D. J. Mikita, et al., 1993; W. Larry Kenney & Michael J. Zeman, 2002; T. J. Kulka & W. L. Kenney, 2002; Wolf, Folkerts, Cottle, Daanen, & Kenney, 2021). However, little work has been done to identify those limits in vulnerable populations such as the aged. Similarly, most previous studies using this experimental paradigm have been performed at an intensity of 30% V̇O_2max, a work rate that reflects the intensity associated with an 8-h work day in many industrial settings (Bonjer, 1962), as well as many self-paced activities in young, healthy adults. In this context, past data from our laboratory have established critical environmental limits for young men and women (W. Larry Kenney & Michael J. Zeman, 2002) and older unacclimated women (Kenney, 2020) during exercise at 30% V̇O_2max (Figure 2).

Figure 2.

A standard psychrometric chart showing empirically derived mean critical environmental limits (symbols and solid lines) for 6 young, heat acclimated women (blue circles), 10 young, unacclimated women (red circles), and 10 unacclimated older women (open circles) exercising at 30% V̇o_2max. The dashed line represents the lower bound of the 95% confidence interval associated with the older group (dashed line) [adapted from (Kenney, 2020)].

The overarching goal of the PSU HEAT project is to identify critical environmental limits to heat balance in aged adults and other vulnerable populations at metabolic rates that are relevant among free-living adults (i.e., during rest and minimal physical activity). To establish critical environmental limits and construct isothermal lines across a wide range of thermal environments, participants are exposed to progressive heat stress in six distinct environments; three P_crit trials during which T_db is held constant at either 36, 38, or 40 °C, and T_crit trials during which P_a is held constant at either 12, 16, or 20 mmHg.

As the initial phase of the project, recent data from our laboratory (Wolf, Cottle, Vecellio, & Kenney, 2021) demonstrated psychrometric (i.e., properties of atmospheric air that influence heat balance such as temperature and humidity) limits in these six environments (Figure 3) for healthy, young subjects (18 – 34 yrs) during light activity at two intensities reflecting the metabolic demand of minimal physical activity (cycling against zero resistance at a cadence of 40–50 rpm) or light ambulatory activity (walking on a motor-driven treadmill at a speed of 2.2 miles per hour and grade of 3%) (Ainsworth et al., 2000; Das Gupta, Bobbert, Faber, & Kistemaker, 2021). Combinations of environmental conditions that are below and to the left of the limit lines for each metabolic rate are compensable, and therefore relatively low risk for prolonged exposures. Combinations that are above and to the right of the limit lines are uncompensable, and therefore may increase risk of heat-related illness. Also included are the lower bounds of the 95% confidence intervals for each metabolic rate, which theoretically provide upper limits below which most young, healthy adults can maintain a steady state, albeit elevated, T_c at these low metabolic rates.

Figure 3.

A standard psychrometric chart showing empirically derived mean critical environmental limits (symbols and solid lines) for light ambulatory activity (red circles) and minimal activity (blue circles) trials. The larger dashed lines with arrows at the top left portion of the curves are isothermal lines from biophysical modeling of heat exchange. Smaller dashed lines denote the lower bounds of the 95% confidence interval for each condition. Critical environmental limits were significantly lower (i.e., shifted downward and to the left; P < 0.001) in light ambulatory compared to minimal activity across all environmental conditions tested [adapted from (Wolf, Cottle, et al., 2021)].

Using this method, we also recently compared empirically-derived thresholds (Vecellio, Wolf, Cottle, & Kenney, 2022) for the maintenance of heat balance to the theorized (and highly-publicized) 35°C wet-bulb temperature threshold for human adaptability (Sherwood & Huber, 2010). Our findings suggested that the upper limits for maintenance of heat balance were significantly lower than the theorized 35°C wet-bulb temperature threshold in all tested environments ranging from warm-humid to hot-dry (Figure 4). Furthermore, we demonstrated that the wet-bulb temperature threshold above which heat balance cannot be maintained varied across environments, particularly in hot-dry conditions, providing evidence against the notion of a single wet-bulb temperature threshold above which heat-related risk increases. Notably, these data were collected in young, healthy participants only, providing a “best case” scenario; wet-bulb temperature thresholds for the maintenance of heat balance are likely to be even lower for older adults and other vulnerable populations. That phase of data collection is ongoing at this time.

Figure 4.

Empirically-derived critical wet-bulb temperature limits for three critical water vapor pressure (P_crit; 36, 38, and 40°C) and three critical dry-bulb temperature (T_crit; 20, 16, and 12 mmHg) trials. The dashed line represents the limit for human adaptability as theorized by Sherwood and Huber (Sherwood & Huber, 2010). In all cases, the upper limit for maintenance of heat balance was lower than the theoretical adaptability limit [reprinted from (Vecellio et al., 2022)].

It is important to note that this approach is primarily reliant on the biophysics of heat exchange (i.e., heat gain from the environment and metabolic heat, and heat loss via sweating and cutaneous vasodilation), while the relatively small amount of variance in critical environmental limits can be explained by human biological variation. Heat acclimatization or acclimation is an important determinant of the ability to maintain heat balance by improving thermoregulatory function, as mentioned previously, and increases critical environmental limits (Kenney, 2020). As discussed previously, the definition of a heatwave varies and is typically reported as a percentile of the daily mean temperature. As such, variation across populations living in different climates may influence critical environmental limits to some degree, although this has not yet been specifically investigated.

CONCLUSIONS

Given the concomitant warming climate and aging global population, research examining the impact of extreme heat on older adults and other vulnerable populations is becoming increasingly critical. This review aimed to highlight the most recently (past three years) published data regarding heat-related morbidity and mortality. As expected, the recent literature demonstrates that extreme heat continues to be increasingly detrimental to the health of the elderly and that this is apparent across the world.

The currently available literature provides important insight into the impacts of extreme heat on morbidity and mortality across populations and regions, but is limited in providing information with regard to the specific environmental conditions above which risk begins to increase. Characterizing the combinations of environmental conditions above which increased risk of heat-related illness occurs is critical to protect vulnerable people by providing data for the development of evidence-based alert communication, policy decisions, triage for impending heat events, and implementation of other safety interventions. To this end, the purpose of the PSU HEAT project is to empirically derive critical environmental limits in older adults and those with various comorbidities.

Appendix A –. Weather and climate terms

Apparent temperature: a perceived temperature measure which includes not only air temperature and humidity, but also the effects of wind speed and radiation (Steadman, 1984). This measure has been modified over time but makes up the basis for the simplified equations we now use for the Heat Index and Wind Chill.

Dry-bulb temperature: air temperature

Excess Heat Factor (EHF): An index used to determine if heat is excessive or abnormal for a given time period. Daily mean temperature over a three-day period is compared to the 95^th percentile of daily mean temperature over that time in the historical record. Additionally, the heat event is compared to the previous 30 days to determine how much acclimatization may have taken place in the local population. These two measures are then combined to determine the EHF. Equations and more information: (Nairn & Fawcett, 2015)

Relative humidity: Percentage of the air that is saturated (moist) based on the ambient dry-bulb temperature.

Representative Concentration Pathway (RCP): These represent the increased radiative forcings that the climate system will endure in the year 2100 based on future projected scenarios of greenhouse gas emissions. The most used climate forcing scenarios are RCP2.6, RCP4.5, and RCP8.5 with the numbers after representing the amount of extra radiative forcing (i.e., RCP8.5 means there is an additional 8.5 Watts per square meter of radiative forcing in the climate system). Increased radiative forcing is strongly correlated with increased global average temperatures. For reference, an RCP8.5-induced climate change is projected to produce a temperature increase of 2.6 – 4.8°C by the year 2100. More information: (2014; van Vuuren et al., 2011)

Water vapor pressure: The partial pressure exerted by the mass of water vapor in an air mass

Wet-bulb temperature: The resultant air temperature that an air mass would have if all of the water it contained was evaporated. At 100% relative humidity, the wet-bulb temperature is equal to the dry-bulb temperature.