Thermal Response to Running Across the Sahara Desert: Data for Three Men

SHORT COMMUNICATION Thermal Response to Running Across the Sahara Desert: Data for Three Men David W. DeGroot and W. Larry Kenney D EGROOT DW, K ENNEY WL. Thermal response to running across the Sahara Desert: data for three men. Aviat Space Environ Med 2008; 79:909 13. Background: There are limited data concerning the physiological responses to long-duration exercise collected under field conditions, and less data under harsh environmental conditions. This paper describes select environmental conditions and physiological responses of three runners attempting to run across the entire Sahara desert over a 111-d period. Methods: The runners started in Saint-Louis, Senegal, and we collected data on 2 d at the start of the expedition. Core temperature was measured via telemetry pill (T pill ), heart rate via Polar monitor, and metabolic rate (M) was estimated from two equations. The Pandolf equation uses movement speed and grade while Berglund s equation predicts M from heart rate and dry-bulb temperature. Data are presented as mean 6 SD (range). Results: The runners intermittently ran 8.0 km z h 2 1 over 6 h during Night (warm-humid) conditions and 6.9 km z h 2 1 over 7 h 40 min during Day (hot-dry) desert conditions. Mean T pill was similar for both days (37.8 6 10.34 vs. 37.82 6 0.50 C) while range was greater during the day (Day: 36.69 38.91 vs. Night: 37.11 38.48 C). Heart rate was 128 6 16 (72 156) and 119 6 17 (75 147) bpm for Night and Day, respectively. M mean was 299 6 66 (65 418) W z m 2 2 and 364 6 117 (58 542) W z m 2 2 during Night and 239 6 60 (67 356) and 244 6 139 (54 464) W z m 2 2 during Day, estimated by Berglund and Pandolf, respectively. Conclusions: During Day, the athletes ran slower than during Night, though T pill was similar, indicative of the greater environmental strain. Mean predicted M was similar between equations, though maximum and minimal values were more extreme and rate-of-change dynamics faster according to Pandolf s equation. Keywords: telemetry pill, core temperature, thermoregulation, heat, metabolic rate. T HERE ARE LIMITED data concerning the physiological responses to long-duration aerobic exercise collected under field conditions, and even less longduration data under harsh environmental conditions, such as a desert environment. With the advent of temperature telemetry pills (T pill ) for the wireless recording of core temperature (T c ) along with heart rate and the availability of global positioning satellite (GPS) receivers for the accurate calculation of speed, grade, and altitude, collection of such data has become possible. Data collected under field conditions will likely differ from laboratory data in several respects: 1) due to time of day effects, environmental conditions [dry-bulb (T db ), wet-bulb (T wb ), and mean radiant temperatures (MRT) and relative humidity] will fluctuate; 2) unless the field study is conducted under strictly controlled conditions, terrain conditions, grade, and movement speed may vary as well; and 3) when the direct measurement of oxygen consumption is not possible due to logistical or practical difficulties, metabolic rate (M) may need to be estimated; predicted M may differ between equations ( 3,9,13 ). A common equation used to estimate M is Pandolf s equation, which requires knowledge of an individual s movement speed, the surface grade, and weight of any load carried ( 13 ). This equation has been validated by other laboratories ( 9 ) and is used in several models of thermoregulation ( 10,14 ). Another equation to predict M was developed by Berglund, which only requires knowledge of heart rate (HR) and T db (3 ). To our knowledge there are no data comparing M predicted by these two equations, which differ greatly in required input variables. Considering that movement speed may change rapidly, but HR would be slower to respond, the dynamics of metabolic rate changes may differ between equations. Recently, three accomplished ultra-endurance athletes embarked on a trek in which they ran across the entire Sahara Desert together, beginning in Saint-Louis, Senegal, and finishing in Cairo, Egypt, 111 d later, traveling ~80 km per day. These men have frequently either won or finished among the leaders in numerous desert races, including the Marathon des Sables and the Badwater Ultra-marathon. The present paper represents part of an advisory effort that was initiated to conduct physiological testing on, and provide advice to, the runners. Consequently, we were invited to accompany the runners at the start of their expedition, providing us with the opportunity to describe their physiological responses. Therefore, the purpose of this study was to collect the environmental and physiological data necessary to describe the responses of three elite endurance runners during the first several days of their attempt to run across the entire Sahara desert. Additionally, data required for the Pandolf and Berglund M prediction From The Pennsylvania State University, University Park, PA. This manuscript was received for review in July 2007. It was accepted for publication in June 2008. Address reprint requests to: David W. DeGroot, Ph.D., Research Physiologist, Thermal and Mountain Medicine Division, U.S. Army Research Institute of Environmental Medicine, Natick, MA 01760; david.degroot@amedd.army.mil. Reprint & Copyright by the Aerospace Medical Association, Alexandria, VA. DOI: 10.3357/ASEM.2178.2008 Aviation, Space, and Environmental Medicine x Vol. 79, No. 9 x September 2008 909

equations were collected and the prediction equation results were compared. METHODS Subjects The three runners had the following individual characteristics: runner 1, age 43 yr, height 1.81 m, weight 79.4 kg, Dubois body surface area (BSA) 1.99 m 2 (5), maximal oxygen consumption ( V o 2max ) 4.33 L z min 21 and 54.5 ml z kg 21 z min 21, maximal heart rate (HR max ) 187 bpm and body fat 11.3%; runner 2, age 37 yr, height 1.72 m, weight 69.0 kg, BSA 1.81 m 2, V o 2max 4.01 L z min 21 and 58.1 ml z kg 21 z min2 1, HR max 177 bpm, and body fat 15.8%; and runner 3, age 30 yr, height 1.64 m, weight 57.8 kg, BSA 1.62 m 2, V o 2max 3.54 L z min 21 and 61.2 ml z kg 21 z min2 1, HRmax 185 bpm, and body fat 10.6%. Baseline Testing Approximately 3 mo prior to the expedition the runners visited the Gatorade Sports Science Institute in Barrington, IL, for 2 d of consultation meetings and baseline testing. The experimental procedures were explained to the subjects and verbal and written informed consent were obtained. Height (m) was measured to the nearest 0.5 cm and body mass (kg) was measured to the nearest 0.1 kg. Body composition was estimated using the BodPod air displacement plethysmography system (Life Measurement, Inc, Concord, CA). V o 2max (MOXUS system, AEI Technologies, Pittsburgh, PA) was measured using a modified Balke protocol, during which the subject self-selected speed and grade was increased 2% every 2 min. The test was terminated upon volitional exhaustion, and a valid V o 2max was considered when a subject achieved commonly accepted criteria ( 1 ). Heart rate was measured via Polar monitor (Polar Electro Inc., Lake Success, NY) throughout the test. Field Observations This was an observational study only and no attempt was made to influence start or stop times of any run, stop and restart times within a given run, or movement speed or distance. The only exception was for occasional stops requested by other researchers accompanying the expedition. Complete data sets for all three athletes were obtained on two runs, one taking place after sunset (Night) and the other the following day (Day). During Night, the runners traveled 44.1 km in 6 h 10 min, finishing at 0220 local time, which after accounting for total stop time of 39 min averaged 8.0 km z h 21. Approximately 9 h after the conclusion of Night, the runners started the Day run at 1120 local time. Total run time was 7 h 50 min, ending at 1910, covering 38.3 km, and after accounting for 136 min of stop time, average run speed was 6.9 km z h 21. Stop time was greater for Day due to stops requested by other researchers, to replenish fluid supplies and to eat, and an extended stop due to quadriceps muscle cramps in one runner. On each day the running surface was either asphalt or hard-packed dirt. Environmental Monitoring T db, T wb, and globe (T g ) temperatures were measured and %RH was calculated with a portable data-logging heat stress monitor (QuesTemp 34, Quest Technologies, Oconomowoc, WI) and air velocity (V air ), along with duplicate measurement of T db and %RH were recorded with a handheld personal weather monitor (Kestral 4000 pocket weather tracker, Nielsen-Kellerman, Boothwyn, PA). Data were automatically recorded once per minute for both instruments. MRT was estimated using the following equation ( 8 ): MRT4(1 ` 0.222 * V 0.5 air ) * (T g 1T db ) ` T db. Physiological Measurements Approximate 7 h prior to the first run, each subject ingested a temperature telemetry capsule (CorTemp, HQ Inc, Palmetto, FL) for measurement of core temperature ( 4,12 ). The presence of the telemetry capsule was verified as needed; two of the three subjects ingested new telemetry capsules ~1.5 h prior to the start of the Day run. During the run, subjects wore a Polar heart rate sensor and T pill and HR data were recorded by the CorTemp datalogger once per minute. Prior to further data reduction, 5-min averages were calculated. Average, minimum, and maximum HR and T pill were calculated. Prior to each run, hydration status was estimated using morning bodyweight compared to baseline bodyweight and urine specific gravity, where urine specific gravity. 1.029 was considered hypohydrated ( 2,11 ). Movement Speed One of the subjects wore a GPS receiver (Magellan explorist 600, Thales Navigation, San Dimas, CA) continually and time and location data were converted to movement speed (MapSend Worldwide V1.30, Magellan, Santa Clara, CA). As this was not a competitive run, the runners remained together throughout each run and the movement speed for one was assumed for all three. Metabolic Rate The Pandolf equation predicts M from movement speed, grade, and load carriage: M (Watts z m 22 ) 4 [1.5 * W ` 2.0 * (W ` L) * (L/W) 2 ` h * (W ` L) * (1.5V 2 ` 0.35V * G)]/BSA Where W is bodyweight in kg, L is load carried, h is the terrain coefficient (1.2 for asphalt or hard-packed surfaces, as in the present study), V is movement speed, and G is the fractional grade. This equation was originally validated for slow movement speeds (, 2.2 m z s 21 or 7.9 km z h 21 ) and a recent independent validation suggests that the equation is suitable for running speeds up to 10.2 km z h 21 ( 9 ). Unlike Pandolf s equation, the equation developed by Berglund ( 3 ) predicts metabolic rate in MET units from heart rate ratio (HRR; activity HR:resting HR) and T db : MET40.68 ` 4.69 * (HRR11)10.052 * (HRR11) * (T db 120). 910 Aviation, Space, and Environmental Medicine x Vol. 79, No. 9 x September 2008

The limits of the equation are 20 C T db 40 C and 1.2 HRR 2.1. In order to directly compare M prediction results, the results of Berglund s equation in METS were converted to W z m 2 2 by the equation W z m 2 2 5 METS * 58.2, where 58.2 W z m 2 2 5 1 MET ( 6 ). Based on each subject s V o 2max and the equation M (W z m 22 )4(0.23 * RER`0.77) * (5.873 * V o 2 ) * (60/A d ) (7), maximal metabolic rate (M max ) was calculated. The predicted peak (highest value attained during any one 5-min period; denoted as M peak ) and mean (M mean ) metabolic rate during the runs were expressed as a percentage of M max as estimated from the V o 2max data. Statistics Descriptive data are presented as mean 6 SD (range). Due to the small sample size, further statistical analysis is not appropriate and only limited qualitative conclusions are presented. RESULTS Complete Tpill, HR, and environmental data were collected on two consecutive days. T db and MRT were considerably higher and RH lower during Day due to time of day effects. Sunset was at 1832 on Day, approximately 40 min before the conclusion of the run. Urine specific gravity data indicated that the runners were euhydrated during both runs. Fig. 1 (Night) and Fig. 2 (Day) present the A) environmental conditions, B) individual and mean absolute T pill, C) individual and mean DT pill, D) mean HR, and E) estimated M for each equation. Baseline T pill was 37.48 6 0.36 and 37.01 6 0.60 C, during Night and Day, respectively. Mean T pill was very similar (Night 37.8 6 10.34 and Day 37.82 6 0.51 C), though the range was greater during Day (36.69 38.91 vs. 37.11 38.48 C). At Night, the mean HR was 128 6 16 (72 156) bpm, corresponding to 70 6 9% HR max. The mean HRR was 2.02 6 0.25. During Day, mean HR was 119 6 17 (75 147) bpm or 64 6 10% HR max, and the mean HRR was 1.91 6 0.30. M mean was 299 6 66 (range 65 418) W z m 22 calculated by Berglund s equation and 364 6 117 (58 542) W z m 22 by Pandolf s equation during Night and 239 6 60 (67 356) and 244 6 139 (54 464) W z m 22, respectively, during Day. According to Berglund s equation, the runners were exercising at 40 6 9 and 32 6 8% of M max during Night and Day, respectively. M peak was 60 6 5 and 50 6 3%, respectively. Pandolf s equation yielded higher values, such that M mean was 49 6 16 and 33 6 18% and M peak was 77 6 3 and 67 6 3, for Night and Day, respectively. DISCUSSION The present study presents select physiological responses of three elite ultra-endurance runners over 2 d at the start of a 111-d run across the entire Sahara Desert. The Night run was characterized by warm, humid conditions in the absence of solar radiation, while the Day Fig. 1. Night run. A) Environmental conditions expressed as mean 6 SD in 2-h intervals. B) Group mean and individual T pill data expressed as absolute values. C) Group mean and individual T pill data expressed as change from baseline. D) Mean heart rate. E) Metabolic rate predicted by Pandolf s equation (closed circles) and by Berglund s equation (open circles). The runners stopped at 2031 (2-min stop), 2107 (10 min), 2228 (1 min), 2247 (8 min), 2348 (10 min), and at 0117 (8 min). run, begun ~9 h later, was conducted under hot, dry conditions and full solar radiation. Dry-bulb temperature differences were mostly due to time of day effects, while lower humidity during the day could be attributed to both time of day effects and the runners progressing inland away from the Atlantic Ocean and the Senegal River. In order to limit the rise in T pill and avoid exertional heat injury, the runners self-selected a slower pace during the Day run. Fig. 1 B and C indicate that the runners Tpill increased during the initial ~2 h of the Night run and then dropped to ; baseline levels thereafter, even though exercise (and therefore metabolic heat production) continued. However, baseline T pill was elevated in all three runners by 0.47 C during Night compared to Day. This may be due Aviation, Space, and Environmental Medicine x Vol. 79, No. 9 x September 2008 911

Fig. 2. Day run. A) Environmental conditions expressed as mean 6 SD in 2-h intervals. B) Group mean and individual T pill data expressed as absolute values. C) Group mean and individual T pill data expressed as change from baseline. D) Mean heart rate. E) Metabolic rate predicted by Pandolf s equation (closed circles) and by Berglund s equation (open circles). The runners stopped at 1200 (14-min stop), 1255 (13 min), 1343 (12 min), 1425 (10 min), 1512 (30 min), 1609 (7 min), 1642 (35 min), 1825 (12 min), and at 1854 (3 min). to time of day effects, but a more likely explanation concerns the runner s activities prior to the start of the Night run, which consisted of sitting outdoors or in warm vehicles for several hours while waiting to cross the border from Senegal into Mauritania. This passive heating increased their baseline T pill and may partially explain the similar T pill at baseline and at the end of the run. At the same time, T db was dropping and %RH increasing; the net effect of these changes on heat loss capacity is unknown and difficult to estimate via partitional calorimetry without knowledge of skin temperature. A typical T pill response was seen during Day, in which T pill increased at the onset of exercise and any fluctuations appear attributable to varying metabolic rate due to rest periods. For example, the runners stopped for 30 min at 1512 and during this time HR and T pill dropped; when running resumed, HR and T pill rose to levels similar to before the rest break. In an attempt to quantify the exercise intensity, we selected two metabolic rate prediction equations, the Pandolf equation ( 13 ) and an equation developed by Berglund ( 3 ) which has not been compared to other equations or to measured V o 2. Unfortunately, due to logistical limitations it was not possible to directly measure V o 2 during the runs and, therefore, we were unable to compare predicted vs. measured values. However, based on the maximal metabolic rate determined during the V o 2max test, we calculated the mean and peak metabolic rate at which the athletes were running. These data indicate that the runners were exercising at an average of 32 49% of peak M, which is reasonable given the fitness level of the runners, the duration of the runs, and the environmental conditions. Based on the calculated mean M, the two equations appear to agree during Day, but not during Night. However, even a casual glance at Fig. 1E and 2E indicates substantial differences between the results which are likely due to the different required inputs and the dynamics of the temporal responses of those inputs. Pandolf s equation uses movement speed, along with load carriage and terrain grade, to estimate M. Load carriage was relatively constant, varying only due to fluid volume changes in the hydration system worn by each runner. Grade was assumed to equal zero, as the net elevation change for each run was negligible. Movement speed, on the other hand, varied from 0.0 km z h 2 1 to a peak of 11.0 km z h 2 1 for a short period during the Night run, and can change very rapidly. For example, at 1642 during Day, the runners stopped and M according to Pandolf dropped from 427 to 81 W z m 2 2 within 5 min. However, oxygen consumption and hence metabolic rate do not immediately return to basal levels due to elevated body temperature, respiratory rate and heart rate, replenishment of oxygen stores, removal of lactic acid, and other possible factors; therefore, the rapid decline in metabolic rate predicted by Pandolf s equation is not likely. Berglund s equation, on the other hand, uses HRR to predict M, along with T db to adjust for the increased M in a hot environment ( 3,15 ). Heart rate remained elevated during rest breaks likely due to continued activation of heat loss mechanisms and drive for increased skin blood flow; activity below the detectable limits of the GPS device may also have contributed. During sustained steady-state exercise periods, it appears that Berglund s equation under-predicts M compared to Pandolf s equation. However, Berglund s equation was developed using data from cycle ergometer exercise performed by subjects of average fitness level. Due to the increased HR: V o 2 relation during cycle ergometry compared to treadmill exercise, a lower M would be expected when this equation is applied to treadmill exercise, which is in agreement with our data. The possible confounding effects of the subject s high fitness level and the duration of the exercise bouts on the accuracy of 912 Aviation, Space, and Environmental Medicine x Vol. 79, No. 9 x September 2008

Berglund s equation are unknown. Our data suggest that a validation study of Berglund s equation is warranted prior to application to walking or running data or to subjects of above-average fitness level. In summary, this paper presents 2 d of data from three elite ultra-endurance athletes during a trek across the Sahara Desert. The runners completed two long-distance runs separated by 9 h of recovery. T pill, HR, and M were within normal physiological limits. The differing predicted metabolic rate dynamics between Pandolf s and Berglund s equations suggest a need for validation of these equations during intermittent exercise. ACKNOWLEDGMENTS The authors thank the runners for their exceptional efforts under harsh conditions and for providing us with the opportunity to study them during their expedition. Beth Stover Mooradian, M.S., and John Eric Smith, Ph.D., of the Gatorade Sports Science Institute assisted with data collection in Senegal and Mauritania and their contributions are appreciated. At the U.S. Army Research Institute of Environmental Medicine, William Santee, Ph.D., loaned us the weather monitoring equipment and Larry Berglund, Ph.D., assisted with the metabolic rate prediction equations. This project was funded by the Gatorade Sports Science Institute. Authors and affiliations: CPT David W. DeGroot, Ph.D., Research Physiologist, Thermal and Mountain Medicine Division, U.S. Army Research Institute of Environmental Medicine, Natick, MA, and W. Larry Kenney, Ph.D., Department of Kinesiology and Intercollege Graduate Degree Program in Physiology, The Pennsylvania State University, University Park, PA. REFERENCES 1. ACSM s guidelines for exercising testing and prescription, 7th ed. Baltimore: Lippincott Williams and Wilkins; 2006. 2. Armstrong LE, Maresh CM, Castellani JW, Bergeron MF, Kenefick RW, LaGasse KE, Riebe D. Urinary indices of hydration status. Int J Sport Nutr 1994; 4:265 79. 3. Berglund LG. Heart rate as an indicator of metabolic rate in hot environments. Proceedings of the 30th annual conference on engineering in medicine and biology; 1977 Nov 5-9; Los Angeles, CA. Piscataway, NJ: Engineering in Medicine and Biology Society; 1977. 4. Byrne C, Lim CL. The ingestible telemetric body core temperature sensor: a review of validity and exercise applications. Br J Sports Med 2007 ; 41 :126 33. 5. Dubois D, Dubois E. A formula to estimate the approximate surface area if height and weight be known. Arch Intern Med 1916 ; 17 :863 71. 6. Gagge AP, Burton AC, Bazett HC. A practical system of units for the description of the heat exchange of man with his environment. Science 1941 ; 94 :428 30. 7. Gagge AP, Gonzalez RR. Mechanisms of heat exchange : biophysics and physiology. In: Fregly M, Blatteis C, eds. Handbook of physiology, section 4: environmental physiology, volume 1. Oxford, UK : Oxford University Press ; 1996 :45 84. 8. Goldman RF. Prediction of human heat tolerance. In: Folinsbee L, Wagner JA, Borgia J, Drinkwater BL, Gliner J, Bedi J, eds. Environmental stress: individual human adaptations. New York : Academic Press ; 1978 :53 69. 9. Hall C, Figueroa A, Fernhall B, Kanaley JA. Energy expenditure of walking and running: comparison with prediction equations. Med Sci Sports Exerc 2004 ; 36 :2128 34. 10. Kraning K, Gonzalez RR. A mechanistic computer simulation of human work in heat that accounts for physical and physiological effects of clothing, aerobic fitness, and progressive dehydration. J Therm Biol 1997 ; 22 :331 42. 11. Leiper JB, Pitsiladis Y, Maughan RJ. Comparison of water turnover rates in men undertaking prolonged cycling exercise and sedentary men. Int J Sports Med 2001 ; 22 :181 5. 12. O Brien C, Hoyt R, Buller M, Castellani J, Young A. Telemetry pill measurement of core temperature in humans during active heating and cooling. Med Sci Sports Exerc 1998 ; 30 :468 72. 13. Pandolf KB, Givoni B, Goldman RF. Predicting energy expenditure with loads while standing or walking very slowly. J Appl Physiol 1977 ; 43 :577 81. 14. Pandolf KB, Stroschein LA, Drolet LL, Gonzalez RR, Sawka MN. Prediction modeling of physiological responses and human performance in the heat. Comput Biol Med 1986 ; 16 : 319 29. 15. Sawka M, Wenger C, Pandolf K. Thermoregulatory responses to acute exercise-heat stress and heat acclimation. In: Fregly M, Blatteis C, eds. Handbook of physiology: environmental physiology. New York : Oxford University Press ; 1996 : 157 185. Aviation, Space, and Environmental Medicine x Vol. 79, No. 9 x September 2008 913