Evaluating the 35°C wet-bulb temperature adaptability threshold for young, healthy subjects (PSU HEAT Project)

Abstract

A wet-bulb temperature of 35°C has been theorized to be the limit to human adaptability to extreme heat, a growing concern in the face of continued and predicted accelerated climate change. Although this theorized threshold is based in physiological principles, it has not been tested using empirical data. This study examined the critical wet-bulb temperature (T_wb,crit) at which heat stress becomes uncompensable in young, healthy adults performing tasks at modest metabolic rates mimicking basic activities of daily life. Across six experimentally determined environmental limits, no subject’s T_wb,crit reached the 35°C limit and all means were significantly lower than the theoretical 35°C threshold. Mean T_wb,crit values were relatively constant across 36°C –40°C humid environments and averaged 30.55 ± 0.98°C but progressively decreased (higher deviation from 35°C) in hotter, dry ambient environments. T_wb,crit was significantly associated with mean skin temperature (and a faster warming rate of the skin) due to larger increases in dry heat gain in the hot-dry environments. As sweat rates did not significantly differ among experimental environments, evaporative cooling was outpaced by dry heat gain in hot-dry conditions, causing larger deviations from the theoretical 35°C adaptability threshold. In summary, a wet-bulb temperature threshold cannot be applied to human adaptability across all climatic conditions and where appropriate (high humidity), that threshold is well below 35°C.

NEW & NOTEWORTHY This study is the first to use empirical physiological observations to examine the well-publicized theoretical 35°C wet-bulb temperature limit for human to extreme environments. We find that uncompensable heat stress in humid environments occurs in young, healthy adults at wet-bulb temperatures significantly lower than 35°C. In addition, uncompensable heat stress occurs at widely different wet-bulb temperatures as a function of ambient vapor pressure.

INTRODUCTION

In their most recent report, the Intergovernmental Panel on Climate Change stated that global temperatures have increased by ∼1°C since the preindustrial era, primarily due to anthropogenic climate change (1). This increase in global mean temperature is accompanied by higher magnitude temperature increases on some regional scales (2), along with increased heatwave frequency, duration, and magnitude (3). Although the number of heatwaves is already on the rise, future generations will experience many more extreme temperature events than the present (4). As dry-bulb (air) temperatures (T_db) increase, there is a thermodynamic basis for concurrent humidity increases via the Clausius–Clayperon relation, as for every 1°C increase in temperature, a parcel of air can hold 7% more water vapor (5). Accordingly, the risk of humid heat stress becomes larger in the face of continued climate change (6). Humid heat stress reduces the body’s most efficient way to dissipate heat, i.e., the evaporation of sweat. Hence, the combination of extreme ambient heat and humidity, often quantified using the wet-bulb temperature (T_wb), prevents human heat loss to the environment and can lead to heat-related illness and even death, especially in vulnerable populations.

Sherwood and Huber (7) were the first climate scientists to propose a T_wb adaptability limit for humans to environmental heat stress. Following basic physiological principles, a threshold of T_wb = 35°C was established as the point where consistent exposure would negate the human body’s natural cooling processes via both convection and evaporation of sweat and induce hyperthermia. Although Raymond et al. (8) have reported a few instances of hourly T_wb values >35°C in recent observations, most maximal T_wb values on Earth have been in the 30°C –31°C range. However, climate models have predicted that regions such as the Middle East could experience T_wb values that regularly exceed 35°C by the end of the century (9, 10).

Therefore, the aim of this study was to evaluate the theoretical T_wb = 35°C survivability threshold with data collected as part of the PSU HEAT (Human Environmental Age Thresholds) project from young, healthy adults. Specifically, we determined critical environmental limits in terms of T_wb, above which steady-state core temperature (T_c) cannot be maintained within the confines of a controllable environmental chamber. This analysis involved subjects moving at low metabolic rates to replicate the baseline activities associated with everyday life. We hypothesized that the critical T_wb (T_wb,crit) would be lower than the theoretical limit of 35°C. Second, we hypothesized that T_wb,crit would be variable depending on combinations of temperature and humidity due to differences in sweat evaporation and heat gain (radiation and convection) in hot-dry versus warm-humid environments.

METHODS AND PROCEDURES

Data were collected at Pennsylvania State University with all procedures approved by the Institutional Review Board (IRB). All test subjects gave informed consent during an initial screening visit. Detailed information about testing procedures and measurements are written in detail in the companion paper by Wolf et al. (28). All subjects from Wolf et al. (TB500) were a part of the study detail here. However, individual trials from three of the subjects were removed for this analysis due to missing mean skin temperature (${\overline{\text{T}}}_{\text{sk}}$) data, one each from the 20 mmHg, 36°C, and 40°C experimental protocols. The exclusion of these participants did not affect the statistics of the subject sample or subsequent variable analyses. A brief summary of the testing procedures is provided here. Subject characteristics are presented in Table 1.

Table 1.

Experimental subject characteristics (24 subjects; 11 male/13 female)

Characteristic	Means ± SD	Range
Age, yr	24 ± 4	18–34
Height, m	1.73 ± 0.1	1.57–1.98
Weight, kg·m−2	71 ± 12	52–98
AD, m2	1.84 ± 0.20	1.50–2.31
AD/kg, m2·kg−1	0.026 ± 0.002	0.022–0.029
V̇o2max, mL·kg−1·min−1	49 ± 12	30–79

During experiments, subjects wore a standardized attire consisting of a t-shirt, shorts, socks, and sneakers. Female participants also wore sports bras. Subjects free-pedaled a cycle ergometer at a low intensity of ∼10 W designed to characterize activities of daily living (11). There were six experimental protocols included in this study: three critical water vapor pressure (P_a) experiments at 36°C, 38°C, and 40°C (P_crit) and three critical T_db experiments at 12, 16, and 20 mmHg (T_crit). After a 30-min acclimation period, P_a or T_db was increased by 1 mmHg or 1°C every 5 min until a clear inflection in T_c was observed, which determined the critical environmental loci of (T_db, P_a). Those loci were then translated to T_wb using a psychrometric chart and recorded as T_wb,crit. Core temperature was measured with gastrointestinal temperature telemetry capsules (VitalSense, Philips Respironics, Bend, OR) that were ingested by subjects 1–2 h before reporting to the laboratory. ${\overline{\text{T}}}_{\text{sk}}$ was measured continuously (iButton, Whitewater, WI) at the chest, upper arm, inner thigh, and calf. Whole body ${\overline{\text{T}}}_{\text{sk}}$ was calculated using a weighted-mean (${\overline{\text{T}}}_{\text{sk}}$) of the four measurement sites (12).

Calculated Variables

Dry heat gain was calculated at the T_c inflection point based on the clothing ensembles participants wore during the experimental protocols using ASHRAE (13) standards. The intrinsic clothing insulation ($R_{\text{cl}})$ was calculated as

$$R_{\text{cl}}=0.155{\text{ W/m}}^{\text{2}}\left(I_{\text{cl}}\right),$$

where $I_{\text{cl}}$ is the clothing insulation factor set to 0.27 clo based on the participants’ standard ensemble. The clothing thermal efficiency ($f_{\text{cl}}$) of the ensemble was calculated as

$$f_{\text{cl}}=1.0+0.3(I_{\text{cl}}).$$

Finally, dry heat gain through convection and radiation ($C+R)$ was calculated as a function of the air-skin temperature gradient and defined as

$$C+R=\frac{{\text{T}}_{\text{db}}-{ \overline{\text{T}}}_{\text{sk}}}{R_{\text{cl}}+1/(f_{\text{cl}}h)},$$

where ${\text{T}}_{\text{db}}$ and ${\overline{\text{T}}}_{\text{sk}}$ are the dry bulb and mean skin temperatures at the time of T_c inflection and $h$ is combined convective and radiative heat transfer coefficients of 4.7 and 3.4 W/(m²°C), respectively.

Statistical Analyses

Independent sample t tests were used to determine differences between mean values among experimental protocols due to their varying sample sizes. To account for multiple comparisons among relative humidity (RH), dry heat gain, and T_wb,crit in the 6 experimental protocols (a total of 15 interacting comparisons), significance was accepted at P = 0.003. The three T_crit and P_crit means were also tested against each other for significant differences with significance being accepted at 0.05/3 or P = 0.017. One sample t tests were performed to determine differences between each of the experimentally determined T_wb,crit means and the 35°C theoretical limit for human adaptability to extreme heat (α = 0.05). To examine relations among variables, linear least squares regression was performed and R² and P values (α = 0.05) were reported. All tests were performed using the Python Software Foundation (Python Language Reference, v. 3.6). Data are reported as means ± SD except in Fig. 1, which is presented as a box-and-whisker plot with individual data points.

Figure 1.

Critical wet-bulb temperature values for the study’s six experimental protocols.

RESULTS

The physiological characteristics of the study’s participants are presented in Table 1. Subjects were recruited to be representative of the population in this age group with respect to body size, adiposity, and aerobic fitness. There were no subject sample differences in age, height, weight, Dubois surface area (A_D), A_D/kg, or V̇o_2max among trial conditions (all P ≥ 0.05).

Mean T_crit and P_crit values for the protocols are presented in Table 2. During T_crit experiments, lower clamped P_a values were associated with higher critical T_db values and there were statistical differences among the three protocols. However, there was less variance in P_crit values among the three clamped T_db conditions and no statistical differences were present. All RH values for the six experimental protocols were statistically different from one another except for 36°C versus 38°C protocols (P = 0.08). Taken together, combinations of T_db, P_a, and RH indicate distinct thermal regimes for T_wb,crit categorization. Specifically, higher T_wb,crit values were associated with warm-humid environments whereas lower values of T_wb,crit were tied to hot-dry environments.

Table 2.

Critical environmental limits for the study’s six experimental protocols

Protocol	36°C	38°C	40°C	20 mmHg	16 mmHg	12 mmHg
No. of participants	8 (3 M/5F)	8 (5 M/3F)	8 (3 M/5F)	8 (6 M/2F)	9 (4 M/5F)	9 (4 M/5F)
Tcrit, °C				44.04 ± 0.23	47.48 ± 2.02	50.57 ± 1.65
Pcrit, mmHg	29.54 ± 2.37	30.03 ± 2.40	27.74 ± 2.52
RH, %	66.25 ± 5.72*	60.83 ± 5.40*	50.24 ± 4.58	28.81 ± 2.70	20.14 ± 1.56	12.70 ± 1.50

Values are presented as means ± SD. Mean T_crit values all are statistically different from one another whereas no statistical differences are present among the mean P_crit values. *Differences existed between all mean RH values except for between the 36°C and 38°C experimental protocols (P = 0.08). F, female; M, male; RH, relative humidity.

The T_wb,crit in each of the three T_crit experiments (12 mmHg: 25.75 ± 0.48°C; 16 mmHg: 27.12 ± 0.54°C; 20 mmHg: 27.82 ± 0.71°C) were lower than the T_wb,crit in any of the P_crit experiments (36°C: 30.34 ± 0.97°C; 38°C: 30.96 ± 0.97°C; 40°C: 30.45 ± 1.06°C; Fig. 1). Among T_crit experiments, T_wb,crit at 12 mmHg was lower than that at both 16 and 20 mmHg (both P < 0.001). There was no statistical difference between the T_wb,crit values for the 16 and 20 mmHg protocols (P = 0.046). There were no differences in T_wb,crit among the three P_crit experiments (36°C vs. 38°C: P = 0.24; 36°C vs. 40°C: P = 0.83; 38°C vs. 40°C: P = 0.36). Importantly, the T_wb,crit for all six experimental protocols were significantly different from the reported 35°C T_wb theoretical limit for human adaptability to extreme heat (Fig. 1).

Interactions between ${\overline{\text{T}}}_{\text{sk}}$ and T_wb,crit are presented in Fig. 2. Higher ${\overline{\text{T}}}_{\text{sk}}$ at the time of T_c inflection was associated with lower T_wb,crit values (R² = 0.54, P < 0.001; Fig. 2A). In all cases, ${\overline{\text{T}}}_{\text{sk}}$ at the time of T_c inflection was higher than 35°C. ${\overline{\text{T}}}_{\text{sk}}$ increased at a faster rate in the hot-dry protocols than in the warm-humid (R² = 0.37, P < 0.001; Fig. 2B).

Figure 2.

Relation between critical wet-bulb temperature and mean skin surface temperature (A) and rate of change in mean skin surface temperature (B) for the six experimental protocols.

Dry heat gain at the T_c inflection point was reflective of ambient environmental conditions, such that ${\overline{\text{T}}}_{\text{sk}}$ was higher in hot-dry protocols and lower (approaching zero) in warm-humid protocols (Table 3). Dry heat gain across critical environmental conditions were all significantly different from each other except for between the 12 and 16 mmHg protocols (P = 0.01). Conversely, there were no significant differences in whole body sweat rate among the six experimental protocols (Table 3).

Table 3.

Summary table of dry heat gain (via convection and radiation) and sweat rate for the study’s six experimental protocols

	36°C, 29.5 mmHg	38°C, 29.8 mmHg	40°C, 27.7 mmHg	44.0°C, 20 mmHg	47.3°C, 16 mmHg	50.6°C, 12 mmHg
Dry heat gain, W m−2	−1.51 ± 3.00	8.34 ± 1.72	18.60 ± 2.78	41.46 ± 9.67	61.38 ± 10.91*	76.95 ± 11.43*
Sweat rate, gm−2 h−1	97.61 ± 65.33	183.14 ± 113.42	159.87 ± 63.91	111.98 ± 35.59	142.98 ± 66.20	171.82 ± 98.25

Values are presented as means ± SD. *Differences existed between all mean dry heat gain values except for between the 12 mmHg and 16 mmHg experimental protocols (P = 0.01). There was no statistical difference in mean sweat rates across conditions.

DISCUSSION

Our results indicate that the theoretical T_wb = 35°C adaptability limit to climate change—introduced by Sherwood and Huber (7) and used in subsequent papers to determine future regions of livability (9)—overestimates real-world conditions that lead to uncompensable heat stress in young, healthy adults during minimal physical activity. In controlled experiments, critical wet-bulb temperatures ranged from 25°C to 28°C in hot-dry environments and from 30°C to 31°C in warm-humid environments. Sherwood and Huber (7) reasoning was contingent on the assumption of a maximum T_sk of 35°C to allow for heat to be moved away from the core of the body, which is typically within a half-degree of 37°C. However, our data suggest that ${\overline{\text{T}}}_{\text{sk}}$ typically exceeds 35°C after a short duration in ambient thermal environments above 36°C, even at very low metabolic rates, with the effect being more pronounced in hot-dry conditions.

In fact, ${\overline{\text{T}}}_{\text{sk}}$ often exceeded T_c by the time of T_c inflection during T_crit trials, which according to thermodynamic theory reverses the thermal gradient from the skin toward the core. The higher magnitude and faster rising ${\overline{\text{T}}}_{\text{sk}}$ are due to larger increases in dry heat gain in the hot-dry protocols compared with the warm-humid protocols, in conjunction with no difference in sweat rate across the six experimental protocols. With free evaporation occurring in the hot-dry protocols due to the large gradients in vapor pressure between the skin and environment, subject participants did not increase sweating (and thus evaporative) rate to compensate for the relatively higher dry heat gains.

As stated under results, distinct T_wb,crit thermal loci were present in the data set. Higher and more constant T_wb,crit values, closer to the 35°C theoretical limit yet still statistically different from it, were found in warm-humid environments whereas T_wb,crit values in hot-dry environments were nearly 10°C lower than the literature-proposed limit. These results indicate that not only is the 35°C theoretical threshold untenable under real-world testing, that ambient environmental control on T_wb,crit dictates that one universal wet-bulb temperature cannot be used to quantify human thermal tolerance across the world. Future adaptability and survivability work should incorporate the heterogeneous relations between climate and T_wb,crit via a geographic lens to provide a more realistic regional and global risk to continued extreme heat associated with climate change.

The critical environmental limits reported herein document that areas of the planet already experience wet-bulb temperatures associated with uncompensable heat stress on a more regular basis than previously theorized (7, 8). Intervention strategies such as electric fan use and air conditioning allow for survivability in these extreme environments, though they inhibit the ability to acclimatize and/or adapt (14). Still, some caveats apply for their use to combat extreme heat. The World Health Organization has advised against electric fan use at ambient T_db above 35°C, subsequently tied to T_wb values <35°C, due to increased dehydration and increased convective heat gain (15). However, biophysical modeling has shown that fans can effectively be used at much higher T_db values (though T_wb values were likely still less than 35°C) given that fans would augment evaporative cooling (16). Laboratory studies have shown the same, especially in young, healthy subjects (17, 18).

The T_wb,crit values in this study are applicable to young, healthy individuals meaning that the current risk to more vulnerable populations is even higher than previously thought. Notably, the elderly are at increased risk due to decreased thermoeffector responsiveness to heat stress (19, 20), medication-induced degradation of body cooling capacity (21), and biobehavioral alterations, which further inhibit heat tolerance (22). This has been realized in excess deaths among the elderly during the 1995 Chicago, USA (23) and 2003 European (24) heatwaves in addition to many others. The importance of continuing to study their interactions with the environment are noted in both clinical (25) and environmental literatures (26). T_wb,crit values for less heat tolerant populations will likely be lower than the values presented here and more commonly found in not only today’s climate, but in future climates as well, and form the scope of the ongoing PSU HEAT Study.

Limitations

Although data were collected over the calendar year to account for acclimatization effects, all experiments were done in State College, PA, which experiences a “warm summer-humid continental” (Dfb) climate according to the Koppen–Geiger climate classification system (27). Acclimatization and adaptation in warmer climates are important to improving the physiological response to extreme heat. Repeatability with subjects living in regions with tropical (class A) or dry (class B) climates, which typically experience higher warm-season extreme temperature and humidity values, would be useful to verify the critical values found in this study.

The environmental chamber used for this study did not include any considerable source of radiative heat input, neglecting an important source of heat gain for humans in outdoor conditions. Conversely, airflow was also limited in the chamber causing a lack of forced convection to aid in evaporation of sweat, which is the body’s main cooling mechanism in extreme heat. In outdoor environments with increased likelihood of forced convection, there is the chance that more sweat could be evaporated and delay the time to T_c inflection, likely allowing for subjects to inflect at higher critical wet-bulb temperatures. It is therefore unclear how additional radiative heat load and forced convection in combination would alter T_wb,crit.

Perspectives and Significance

In this paper, empirical physiological data were used to determine the validity of the theorized human adaptability limit to rising temperatures due to climate change. In all six of the experimental protocols, critical wet-bulb temperatures were significantly lower than the 35°C threshold proposed in the literature (7) and popularized in the lay press. Larger deviations from the 35°C threshold, some as high as 10°C, were found in hot-dry environmental conditions. Subjects in these protocols experienced higher ${\overline{\text{T}}}_{\text{sk}}$, increased dry heat gain, with no statistical difference in sweat rates compared with subjects in the more warm-humid environments, where critical wet-bulb temperatures were nearly constant between 30°C and 31°C. Two conclusions are therefore apparent: 1) The theoretical 35°C wet-bulb temperature threshold does not hold up under experimental testing and 2) there is likely not one critical threshold that can be set, especially so in lower-humidity environments. Future studies should examine the role of acclimatization on heat tolerance as well as how the impact of these conditions would affect critical wet-bulb temperatures in vulnerable populations such as the elderly.

GRANTS

This research was supported by the National Institute on Aging Grant T32 AG049676 to The Pennsylvania State University (to D.J.V.) and National Institutes of Health Grant R01 AG067471 (to W.L.K.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

D.J.V. and W.L.K. conceived and designed research; D.J.V., S.T.W., and R.M.C. performed experiments; D.J.V. analyzed data; D.J.V., S.T.W., R.M.C., and W.L.K. interpreted results of experiments; D.J.V. prepared figures; D.J.V. drafted manuscript; D.J.V., S.T.W., R.M.C., and W.L.K. edited and revised manuscript; D.J.V., S.T.W., R.M.C., and W.L.K. approved final version of manuscript.