Exertional heat illnesses represent the greatest threat to the health and performance of distance runners in environmental conditions of heat and humidity(1,5,12). Distance running elevates Tc in proportion to relative exercise intensity (7), heat and humidity restrict dry and evaporative heat loss (15,22), and the interaction of these and predisposing factors, such as illness and lack of heat acclimatization, increase the risk of heat illness (1,19). Single postrace rectal temperature measurements have traditionally quantified Tc responses to outdoor distance running (7,17,21,25-27). Immediate postrace values in the range of 38.0-41.1°C have frequently been reported (3). Attempts at continuous measurement of Tc during distance running have been restricted to serial (e.g., 9- and 10-min intervals) rectal temperature measurements (4,16,18). To our knowledge, only one study has successfully recorded rectal temperature continuously (i.e., 1-min intervals) in four subjects during a 5-km run (11). Serial measurements from highly trained distance runners have revealed fluctuating steady-state Tc (4,16) as high as 41.6-41.9°C(16) and linearly increasing Tc to approximately 40°C at race finish (18). Although the magnitude of Tc elevation following outdoor distance running has been well quantified (3), the continuous temporal nature of Tc in a moderately sized sample of runners undertaking a mass-participation race has not previously been investigated.
The invasive and obtrusive nature of rectal thermometry, which requires a wire connection from the sensor (typically inserted 10-15 cm beyond the anal sphincter) to the data recording device (24), represents a significant practical difficulty for the continuous measurement of Tc during outdoor running. An alternative method of thermometry suitable for ambulatory applications is the ingestible telemetric temperature sensor, which offers a wireless and valid measure of Tc during steady-state, increasing, and decreasing Tc conditions (24). The primary aim of this study was to continuously measure Tc using ingestible telemetric temperature sensor technology in a sample of runners competing in a 21-km mass-participation running event in environmental conditions of heat and humidity representing a high risk of heat illness (1). Furthermore, we aimed to quantify the physiological strain experienced by runners as evidenced by Tc and HR responses. In this regard, we applied the cumulative heat strain index (CHSI), which combines the thermoregulatory load, described by the hyperthermic area (i.e., Tc integral), and the circulatory load, described by heart beat accumulation (9). Finally, we investigated the relationship between thermoregulatory responses and fluid balance.
Twenty-three male volunteers provided written informed consent to participate in this institute ethics committee-approved study after reading a document describing the nature, benefits, and risks of the study. Core temperature data for 18 subjects are presented due to excretion of the temperature sensor before the race in one subject and incomplete recordings of Tc in four subjects. Corresponding HR data were gained for 15 of these 18 subjects due to incomplete HR recordings in three subjects. Table 1 provides the physical characteristics of the 18 subjects. Volunteers were soldiers in the Singapore Armed Forces participating in the 2003 Singapore Army Half-Marathon as part of their training. They were considered fully heat-acclimatized due to their upbringing, completion of basic military training, and continued active military training in Singapore's tropical environment. Although volunteers were trained for endurance during military activity, they were not distance running athletes, with specific training for distance running occurring on 2.4 ± 0.8 occasions per week, resulting in an accumulated weekly distance of 21 ± 16 km.
Ingestible telemetric temperature sensor system.
On the evening prior to the race (approximately 8-10 h before race start), volunteers, in the presence of a researcher, ingested a telemetric temperature sensor (CORTEMP™ COR-100 Wireless Ingestible Temperature Sensor, HQ Inc, Palmetto, FL). The sensors are 20 mm long, 10 mm in diameter, and contain a temperature-sensitive quartz crystal oscillator that vibrates at a frequency relative to its surrounding temperature. Temperature signalsare transmitted by radio waves to an external temperature-recording device(CORTEMP™ CT2000 Miniaturized Ambulatory Recorder, HQ Inc, Palmetto, FL). The sensors are energized by an internal silver-oxide battery, have aprotective silicone coating, and are passed through the gastrointestinal tractin approximately 8-92 h. Sensors were calibrated on the day of swallowing to ± 0.1°C using a heated water bath at known water temperatures of 35 and 43 (± 0.1) °C.
All measurements took place on the day of the 2003 Singapore Army Half-Marathon. Subjects reported at 0430 h. They were immediately asked to produce a urine sample for the analysis of urine specific gravity using a digital refractometer (UG-1, Atago, Tokyo, Japan). Nude body mass was measured to the nearest 0.1kg using a digital scale (Seca 881, Seca Vogel & Halke GmbH & Co, Hamburg, Germany). Subjects were fitted with a chest band and wrist-watch HR monitor (Polar Vantage, Polar Electro Oy, Kempele, Finland) and dressed in running singlet, shorts, socks, and shoes. A telemetric check was performed to ensure the temperature sensor was residing within the volunteer and transmitting a signal. Finally, the ambulatory Tc data-recording device was placed in a sealed waterproof bag, fitted into a padded pouch, and worn on a lightweight harness around the waist with the data recorder positioned in the lumbar region. The total weight of the data-recording device and harness was 301 g.
Following preparation, volunteers consumed water ad libitum in a prescribed 1.5-L water bottle until race start. All fluid intake and urine output was measured before the race. During the race, fluid stations positioned 1.5-2.0 km apart provided water and carbohydrate-electrolyte fluid (Gatorade) in standard volume cups. Fluid intake during the race was estimated by asking volunteers to completely finish any drinks taken and to note and recall the total number of drinks consumed. Immediately following the race, volunteers towelled dry, and nude body mass was recorded. The change in body mass, corrected for fluid intake and urine loss, but not accounting for metabolic fuel oxidation, metabolic water gain, or respiratory water losses,was used as an estimate of sweat loss. Dry bulb, wet bulb, and globe temperatures; and the wet bulb globe temperature index (WBGT = 0.1 (Tdry bulb)+ 0.7 (Twet bulb) + 0.2 (Tglobe)) were measured at minute intervals throughout the race using a portable climate monitoring device (Questemp°15 Area Heat Stress Monitor, Quest Technologies, WI) positioned at the start/finish area. Ambient water vapor pressure and relative humidity were calculated from the relationship between dry bulb and wet bulb temperature using a simplified psychrometric chart (15). Wind velocity was not measured, although conditions could be described as calm.
Core temperature and HR data were recorded at 15-s intervals. Data were downloaded and plotted at minute intervals against time allowing determination of peak, final (i.e., at race finish), and mean (HR only) values. Maximum Tc elevation (i.e., Δ peak Tc) was calculated as peak Tc - preexercise Tc. Core temperatureelevation at race finish (i.e., Δ final Tc) was calculated as final Tc - preexercise Tc. Core temperature rate of rise during the initial 30 min of running (i.e., rate30 in degrees Celsius per hour) was quantified as the slope of a linear regression line (i.e., regression coefficient as degrees Celsius per minute × 60 min) fitted to the Tc data from 0 to 30 min when Tc was increasing linearly. Average rate of metabolic heat production for the whole race was estimated based on the assumption that heat liberation during running approximates 4 kJ·kg−1 body mass·km−1 (8,22), and thus the rate of heat production is equal to the product of the runner's body mass (kg), the running speed (m·s−1), and the approximately 4 J produced per kilogram of body mass per meter run (8,22). Preexercise body mass and average speed for 21 km were used in the computation. Split times during the race were not recorded, and therefore our measure of average heat production is not sensitive to the changes in running speed/heat production that would have occurred during the race. The cumulative heat strain index, which is a combination of circulatory and thermoregulatory loads, and is a function of the excess heart beat count over the initial state and the integral of Tc, was calculated at minute intervals as follows (9):
where Σhb = accumulation of heart beats over the period of running (t, min); fc0 = initial heart rate (taken as 70 bpm); fc0·t = heart beat count over time at the level of the initial state; circulatory strain is gained by subtracting fc0·t from Σhb; and thermoregulatory strain is calculated as the integral of core temperature(∫Tc). Finally, the two components are complementary rather than additive.
The Singapore Army Half-Marathon is an annual mass-participation 21-km road race. "Fun runs" over 5- and 10-km distances also take place at the same event. All volunteers ran in the 21-km competitive event starting at 0600 h. The event is organized primarily for servicemen and reservists in the Singapore Armed Forces, but it is also open to runners from the general public. Approximately 20,000 servicemen and 40,000 reservists and civilian sparticipate in the various distance events.
Values are reported as means ± SD and range for the 18 runners. Statistical significance was accepted as P ≤ 0.05. The change in Tc, HR, and CHSI responses over time (i.e., minutes 0, 30, 60,90, and final) were analyzed with separate single-factor repeated-measures ANOVA. Significant main effects were analyzed with Tukey's post hoc test. Pearson's product-moment correlation coefficient (r) and the coefficient of determination (r2) were used to examine the strength of the relationships between thermoregulatory responses and running time/average metabolic heat production, and thermoregulatory responses and fluid balance variables.
Wet bulb globe temperature (WBGT) at race start (0600 h) was 26.0°C, representing a high-risk (WBGT 23-28°C) of heat illness (1). At the time of the last finishing volunteers (0826 h), WBGT had increased to 29.2°C, representing a very high risk (WBGT > 28°C) of heat illness (1). Environmental conditions were characterized by high wet bulb temperature (25.9 ± 0.3; 25.6-27.3°C) in relation to dry bulb temperature (27.2 ± 1.0; 26.3-30.6°C) with corresponding high humidity (87 ± 5; 90-75%) and ambient water vapor pressure (3.3 ± 0.1; 3.2-3.7 kPa).
Magnitude of Tc responses.
Mean Tc at 30 min (39.2 ± 0.3; 38.7-39.8°C),60 min (39.6 ± 0.6; 38.5-40.6°C), 90 min (39.7 ± 0.7; 38.3-41.3°C), and final (39.9 ± 0.8; 38.3-41.7°C)were significantly higher (P < 0.01) than the 0-min value(37.7 ± 0.3; 37.1-38.4°C). Mean Tc at 60 min (P < 0.05), 90 min (P < 0.05), and final (P < 0.01) were significantly higher than the 30-min value. The entire sample of runners recorded peak Tc > 39°C; 56% (N = 10) recorded Tc > 40°C; and 11% (N = 2) recorded Tc > 41°C. For final Tc, 89% (N = 16) recorded Tc > 39°C; 44% (N = 8) recorded Tc > 40°C; and 6% (N = 1) recorded Tc >41°C. Ten runners achieved their highest Tc during the race, whereas eight runners achieved their highest Tc at race finish. Figure 1 illustrates the individual Tc responses throughout the race, and Table 2 illustrates the magnitude, rate, and duration of the Tc responses. Core temperature rate of rise during the initial 30 min of running (rate30) was positively associated with Δ peak Tc (r2 = 0.29, P < 0.05), Δ final Tc (r2 = 0.38, P < 0.01), Tc integral (r2 = 0.46, P < 0.01), and CHSI at 60 min (r2 = 0.62, P < 0.01), 90 min (r2 = 0.62, P < 0.01), and final (r2 = 0.31, P < 0.05).
Running time, heat production, morphology, and Tc responses.
Table 1 illustrates individual and mean finishing time, average running speed for 21 km, average metabolic heat production for 21 km, and morphological characteristics. Neither finishing time nor heat production demonstrated a significant relationship (P > 0.05) with any Tc variable (e.g., peak Tc, final Tc, Δ peak Tc, Tc integral). Preexercise Tc demonstrated a slight positive relationship (approaching significance) with finish time (r2 = 0.22, P = 0.06). Body morphology variables (i.e., height, mass, BMI, and AD) failed to demonstrate any significant relationship (P > 0.05) with any Tc variable.
Heart rate responses.
Heart rate at 30 min (182 ± 7; 166-193 bpm), 60 min (179 ±9; 164-194 bpm), 90 min (181 ± 9; 164-200 bpm), and final (181 ± 13; 150-202 bpm) were significantly higher (P < 0.01) than the 0-min value (142 ± 20; 86-169 bpm) but did not differ (P > 0.05) at each time point during running. Peak HR was positively associated with peak Tc (r2 = 0.52, P < 0.01), final Tc (r2 = 0.42, P < 0.01), and Δ peak Tc (r2 = 0.26, P < 0.05); with mean HR positively associated with peak Tc (r2 = 0.32, P < 0.05) and final Tc (r2 = 0.27, P < 0.05).
Cumulative heat strain index.
Mean CHSI (N = 15) values increased significantly over time (P < 0.001) and were 80 ± 16 (58-103)units at 30 min, 507 ± 101 (373-691) units at 60 min, 1339 ± 332 (768-1889) units at 90 min, and 2541 ± 996 (1005-4557) units at race finish. Values were significantly different at each time point (P < 0.001). Figure 2 illustrates the individual CHSI responses throughout the race. Final CHSI was positively associated with CHSI at 60 min (r2 = 0.55, P < 0.01) and 90 min (r2 = 0.67, P < 0.01); finishing time (r2 = 0.58, P < 0.01), peak Tc (r2 = 0.77, P < 0.001), Δ peak Tc (r2 = 0.71, P < 0.001), time to peak Tc (r2 = 0.40, P < 0.01), post Tc (r2 = 0.50, P < 0.01), and Δ post Tc (r2 = 0.36, P < 0.05).
Fluid balance responses.
Table 3 illustrates the individual and mean fluid balance responses to the half-marathon. Sweat rate was positively associated with body mass (r2 = 0.61, P < 0.01), body surface area (r2 = 0.58, P < 0.01), and average metabolic heat production (r2 = 0.40, P < 0.01); and negatively associated with the surface area to body mass ratio (r2 = 0.46, P < 0.01). Core temperature responses demonstrated no significant relationship (P > 0.05) with absolute change (kg) in body mass (peak Tc r2 = 0.01; final Tc r2 = 0.02; Tc integral r2 = 0.03) and relative (% dehydration) change in body mass (peak Tc r2 = 0.00; final Tc r2 = 0.00; Tc integral r2 = 0.00). Furthermore, no significant relationship was observed between any fluid balance variable and any core temperature or CHSI response.
The major finding of the present study is establishing the magnitude of Tc elevation that runners will voluntarily achieve during mass-participation distance races in heat and high humidity without medical consequence. Over half of our sample achieved temperatures greater than 40°C; two runners achieved Tc greater than 41°C, with the highest individual value being 41.7°C. This magnitude is greater than previous field-based observations on asymptomatic distance runners (3,7,17,21,25-27). Mean final Tc was 39.9 ± 0.8°C versus mean postrace rectal temperatures in previous studies of 38.7°C (17), 38.9°C (21), and 39.0°C (25). The highest individual response observed in the present study (i.e., 41.7°C) was also greater than the highest individual response observed in most (e.g., 40.3°C) (17), 40.5°C (21), 40.8°C(27), and 41.1°C (26,27), but not all, previous studies (e.g., 41.9°C) (16).
Metabolic rate (V̇2) is an important determinant of Tc elevation during distance running (7,21). Noakes et al. (21) reported that absolute metabolic rate (L·min−1) in the last 6 km of a 42-km marathon was the most important predictor of postrace Tc, yet only accounted for 28% of the variation in postrace Tc. In laboratory and field measurements, Davies et al. (7)demonstrated that relative exercise intensity (%V̇2max) rather than absolute metabolic rate is more closely related, in a curvilinear manner, to steady-state and postrace Tc. Furthermore, Davies (6) illustrated the complimentary influence of ambient temperature on the exercise intensity-Tc relationship. Dry bulb temperatures over the range 5-21°C influenced Tc during treadmill running at 85% V̇2max but not at 65% V̇2max. At the higher intensity, Tc increased exponentially when ambient temperature exceeded 20°C (6). Due to the severity of the environmental conditions in the present study (i.e., 26.3-30.6°C dry bulb, 25.6-27.3°C wet bulb, and 90-75% relative humidity), which represent a greater restriction to dry and evaporative heat loss than many previous studies (3,17,21,25,27), we hypothesize that the exercise intensity-Tc relationship will have been exponential at intensities less than 85%V̇2max. Thus, we hypothesize that at moderate relative exercise intensities our runners will have achieved comparatively higher Tc responses than previous studies due to the severity of the environmental conditions encountered. This could explain the magnitude of Tc responses at modest average running speeds (i.e., 8.6-12.0 km·h−1) in the current study, whereas previous studies (25,27) have only observed comparable temperatures (e.g., 40.2-41.1°C)in the fastest runners producing average running speeds of 15.2-16.0 km·h−1 and presumably high relative exercise intensities.
We further hypothesize that interindividual differences in relative exercise intensity and intraindividual variations in running pace during the race determined the magnitude and temporal nature of Tc responses in the present study, respectively (see Fig. 1). Due to the absence of split time and metabolic data, we are unable to confirm whether %V̇2max determined the magnitude of Tc elevation and whether variations in running pace were associated with the temporal variations in Tc illustrated in Figure 1. Nevertheless, our heart rate data provide indirect support linking variations in running pace to the temporal variations in Tc responses. For example, concomitant reductions in heart rate and Tc, indicating decreasing running pace and metabolic rate, were observed in runners 2, 8, 17, and 18. Conversely, concomitant increases in heart rate and Tc, indicating increasing running pace and metabolic rate, were observed in runners 3, 6, 8, 12, 14, and 15.
A more accurate description of the severity of the hyperthermic exposure is provided by the total thermal area, calculated from Tc elevation and the duration of Tc elevation (10,13), rather than the magnitude of Tc elevation alone. In this regard, we determined the area under the hyperthermic curve (i.e., Tc integral), but unfortunately comparable data for exercising humans are lacking in the scientific literature. In the study of Maron et al. (16), one highly trained marathon runner maintained Tc between 41.6 and 41.9°C for the final 44 min of a 163-min, 42-km marathon race. To our knowledge this is the highest recorded Tc in an asymptomatic distance runner. The corresponding Tc integralabove 40°C was calculated as 102°C·min−1 (10), which is approximately twofoldhigher than values calculated for the two runners in our study achieving Tc > 41°C (i.e., runner 15 = 47°C·min−1, and runner 18 = 53°C·min−1). In the present study, Tc integral was combined with heart beat accumulation to compute the cumulative heat strain index (CHSI) (see Fig. 2 and Table 2). The CHSI represents a novel approach for assessing the total physiological strain of subjects exposed to exercise-heat stress (9). A lack of comparative CHSI data for distance running prevents any meaningful comparisons of our data to previous studies (e.g., (9)). Further studies utilizing this index are required to place the current data in context. Our data indicate that Tc rate of rise during the first 30 min of running could explain 46% of the variationin Tc integral, 62% of the variation in CHSI at 60 and 90 min, and 31% of the variation in final CHSI. A high rate of rise in Tc is therefore associated with a greater hyperthermic exposure, suggesting that pacing in the early part of the race is an important strategy in the avoidance of exertional heat illnesses.
Laboratory studies of humans exercising at a fixed work rate in heat suggest that fatigue occurs and exercise is terminated when a critically high Tc, reported as approximately 40°C, is attained (23). Over half of our sample (56% peak Tc > 40°C, 11% > 41°C)exceeded this value of Tc previously observed as limiting exercise performance in Scandinavian males (23).We do not believe the Singaporean sample/population is inherently different from other populations, whereby a tolerance to high Tc is afforded,since several other field-based studies have reported Tc > 40 and41°C in asymptomatic distance runners (16,25-27).One of the main methodological differences between field-based race situations and the laboratory studies referred to above is that the former are self-paced and exercising subjects are free to increase or decrease their running pace and therefore regulate their heat production. In the latter, an inability to maintain the required work rate is classified as exhaustion and exerciseis terminated (23). In the 10 runners achieving Tc > 40°C, our HR data do not support a reduction in running pace when a Tc of 40°C was achieved. Nevertheless, Table 2 indicates that 7 of the 10 runners displaying peak Tc > 40°C achievedtheir peak Tc at race finish, whereas only one of the eight runners displaying peak Tc < 40°C achieved their peak Tc atrace finish. This may indicate that approximately half of the sample adopted a strategy to increase pace, heat production, and heat storage towards the end of the race, resulting in their highest Tc being achieved atrace finish. Alternatively, it may suggest that time with Tc >40°C is finite and a pacing strategy that results in Tc >40°C is mainly adopted when race finish is anticipated. The latter explanation may suggest a role for Tc in regulating runners pacing strategy.
In agreement with previous studies (17,21), estimated fluid balance variables (e.g., fluid intake, % dehydration) demonstrated no significant relationship with the magnitude of Tc responses. Estimated sweat rates averaged 1.5 L·h−1, with combined prerace fluid intake and estimated race intake replacing less than half of total sweat losses (see Table 3). Of note, Table 3 illustrates that the two runners (i.e., 11 and 15) with the lowest estimated sweat rates (0.85 L·h−1) also had the highest preexercise urine specific gravity measurements (1.025 and 1.023, respectively). Table 2 illustrates that the runners peak Tc (i.e.,40.6 and 41.7°C, respectively) were among the highest of the sample. It is possible that the reduced sweat rate and elevated Tc in these runners was secondary to preexercise hypohydration and subsequent hyperosmolality, reduced blood volume, and reduced skin blood flow (2). The method of estimating fluid balance in the present study suffers from alack of control in precisely quantifying fluid intake during running, sincethe method relies on runners accurately recalling the frequency of drinks consumed. We assumed all weight losses postexercise were due to sweat losses, which also adds an unknown error into the estimation because an unknown proportion of weight loss is accounted for by metabolic fuel oxidation and respiratory water losses (20). Furthermore, the unaccounted gain of metabolic water during the half-marathon will have resulted in an overestimation of dehydration (20).
This study represents the first report of continuous measurement of thermoregulatory data in a sample of runners competing in a mass-participation race in environmental conditions representing a high risk of heat illness. Ingestible telemetric temperature sensors demonstrated utility for continuous measurement of Tc during mass-participation running. Successful application of this technology has highlighted the magnitude and duration of Tc elevation that runners will voluntarily achieve during mass-participation distance races in heat and high humidity without medical consequence.
The authors express their thanks to the volunteers and commanding officers from the following units of the Singapore Armed Forces for their participation and support of this research: Officer Cadet School, Basic Military Training Centre, Bedok Camp, Katib Camp, Maju Camp, and School of Physical Training. The authors thank Margaret Yap and Teh Chai Hong for their role in data collection and Mr. Andrew Dalton and Dr. Alasdair G Thin for their assistance in the analysis of cumulative heat strain index data. The authors have no conflictsof interest.
1. American College of Sports Medicine Position stand. Heat and cold illnesses during distance running. Med. Sci. Sports Exerc.
2. Armstrong, L.E., C. M. Maresh, C. V. Gabaree, et al. Thermal and circulatory responses during exercise: effects of hypohydration, dehydration, and water intake. J. Appl. Physiol.
3. Cheuvront, S.N., and E. M. Haymes. Thermoregulation and marathon running. Biological and environmental influences. Sports Med.
4. Christensen, C.L., and R. O. Ruhling. Thermoregulatory responses during a marathon. A case study of a woman runner. Br. J. SportsMed.
5. Costill, D.L. Physiology of marathon running. JAMA
6. Davies, C.T. M. Influence of skin temperature on sweating and aerobic performance during severe work. J. Appl. Physiol.
7. Davies, C.T. M., J. R. Brotherhood,and E. Zeidifard. Temperature regulation during severe exercise with some observations on effects of skin wetting. J. Appl. Physiol.
8. Dennis, S.C., and T. D. Noakes. Advantages of a smaller body mass in humans when distance running in warm, humid conditions. Eur. J. Appl. Physiol.
9. Frank, A., M. Belokopytov, Y. Shapiro, and Y. Epstein. The cumulative heat strain index
- a novel approach to assess the physiological strain induced by exercise-heat stress. Eur. J. Appl. Physiol.
10. Fruth, J.M., and C. V. Gisolfi. Work-heat tolerance in endurance-trained rats. J. Appl. Physiol.
11. Fuller, A., T. Oosthuyse, S. K. Maloney, and D. Mitchell. Evaluation of miniature data loggers for body temperature measurement during sporting activities. Eur. J. Appl. Phys.
12. Gisolfi, C.V., and J. R. Copping. Thermal effects of prolonged treadmill exercise in the heat. Med. Sci. Sports Exerc.
13. Hales, J. R. S., R. W. Hubbard, and S. L. Gaffin. Limitation of heat tolerance. In: Handbook of Physiology Section 4: Environmental Physiology
, M. J. Fregly and C. M. Blatteis. (Eds.) Oxford: Oxford University Press, 1996, Vol. 1, pp. 285-355.
14. International Union of Physiological Sciences Thermal Commission Glossary of terms for thermal physiology. J. Therm. Biol.
15. Kenney, W.L. Heat flux and storage in hot environments. Int. J. Sports Med.
16. Maron, M.B., J. A. Wagner,and S. M. Horvath. Thermoregulatory responses during competitive marathon running. J. Appl. Physiol.
17. Maughan, R.J., J. B. Leiper,and J. Thompson. Rectal temperature after marathon running. Br. J. Sports Med.
18. Millard-Stafford, M.L., P. B. Sparling, L. B. Rosskopf, and L. J. DiCarlo. Carbohydrate-electrolyte replacement improves distance running performance in the heat. Med. Sci. Sports. Exerc.
19. Nadel, E.R., C. B. Wenger, M. F. Roberts, J. A. J. Stolwijk, and E. Cafarelli. Physiological defenses against hyperthermia of exercise. Ann. N. Y. Acad. Sci.
20. Noakes, T.D. Lore of Running
, Champaign,IL: Human Kinetics, 2001,pp. 204-207.
21. Noakes, T.D., K. H. Myburgh,and J. Du Plessis. Metabolic rate, not percent dehydration predicts rectal temperature in marathon runners. Med. Sci. Sports Exerc.
22. Nielsen, B. Olympics in Atlanta: a fight against physics. Med. Sci. Sports Exerc.
23. Nielsen, B.,and L. Nybo. Cerebral changes during exercise in the heat. Sports Med.
24. O'Brien, C., R. W. Hoyt, M. J. Buller, J. W. Castellani, and A. J. Young. Telemetry pill measurement of core temperature in humans during active heating and cooling. Med. Sci. Sports Exerc.
25. Pugh, L. G. C. E., J. L. Corbett, and R. H. Johnson. Rectal temperatures, weight losses, and sweatrates in marathon running. J. Appl. Physiol.
26. Robinson, S. Temperature regulation in exercise. Pediatrics
27. Wyndham, C.H., and N. B. Strydom. The danger of an inadequate water intake duringmarathon running. S. Afr. Med. J.