Change in body mass accurately and reliably predicts change in body water after endurance exercise
- First Online:
- Cite this article as:
- Baker, L.B., Lang, J.A. & Larry Kenney, W. Eur J Appl Physiol (2009) 105: 959. doi:10.1007/s00421-009-0982-0
- 557 Views
This study tested the hypothesis that the change in body mass (ΔBM) accurately reflects the change in total body water (ΔTBW) after prolonged exercise. Subjects (4 men, 4 women; 22–36 year; 66 ± 10 kg) completed 2 h of interval running (70% VO2max) in the heat (30°C), followed by a run to exhaustion (85% VO2max), and then sat for a 1 h recovery period. During exercise and recovery, subjects drank fluid or no fluid to maintain their BM, increase BM by 2%, or decrease BM by 2 or 4% in separate trials. Pre- and post-experiment TBW were determined using the deuterium oxide (D2O) dilution technique and corrected for D2O lost in urine, sweat, breath vapor, and nonaqueous hydrogen exchange. The average difference between ΔBM and ΔTBW was 0.07 ± 1.07 kg (paired t test, P = 0.29). The slope and intercept of the relation between ΔBM and ΔTBW were not significantly different from 1 and 0, respectively. The intraclass correlation coefficient between ΔBM and ΔTBW was 0.76, which is indicative of excellent reliability between methods. Measuring pre- to post-exercise ΔBM is an accurate and reliable method to assess the ΔTBW.
KeywordsHydration statusProlonged exerciseFluid replacementDeuterium oxideFourier transform infrared spectroscopy
Sweating rates can vary from less than 0.5 to over 2.5 L h−1 depending upon exercise intensity, fitness level, and heat acclimation state (Sawka 1992; Sawka et al. 2007). Because of this considerable variation, it is important that each athlete know their individual sweating rate to avoid the consequences of drinking too little or too much of fluid. Measuring pre- to post-exercise body mass change (ΔBM) is often used as a simple method to assess sweat loss and presumably hydration status. The ACSM (Sawka et al. 2007) recommends this practice to determine athletes’ individual fluid replacement needs.
However, debate has arisen regarding the relation between BM loss and actual fluid deficit, i.e., dehydration level. According to Noakes et al. (2005), athletes can lose ≥3% of their BM during endurance events without experiencing a net loss in total body water (TBW). That is, not all of the weight lost during exercise is due to fluid loss that needs to be replaced to maintain pre-exercise TBW. According to Noakes et al. (2005), a 70-kg athlete gains ~1.9 kg via endogenous water production (via metabolism and release from glycogen) and loses ~0.6–0.8 kg via non-sweat sources (oxidation of glycogen and fatty acids) during an endurance event. Thus, Noakes et al. (2005) proposed that a total BM loss of ≥2.2 kg (3% BM) could occur in a 70-kg athlete without a change in TBW, i.e., in a state of euhydration.
Conversely, Sawka et al. (2007) and Cheuvront et al. (2007) suggest that the actual difference between ΔBM and ΔTBW is minimal and that ΔBM provides a reasonable estimate of sweat losses and hydration status. These authors indicate that while it is likely that ΔBM overestimates sweat losses to some extent, the overall impact of these factors (endogenous water production and fuel combustion) on fluid balance and fluid replacement recommendations during exercise are minimal.
The mass of endogenous water gain and non-sweat sources of BM loss (oxidation of glycogen and fatty acids) and their effect on overall fluid balance during endurance exercise has been estimated previously (Maughan et al. 2007; Rogers et al. 1997). However, no study has directly compared the ΔBM with the ΔTBW in male and female endurance athletes after prolonged running in a hot environment. Considering the impact of fluid balance on health and performance, the accuracy and reliability of using pre-to post-exercise ΔBM measurements to predict hydration status merits further investigation. Therefore, the purpose of the present study was to determine the relation between the ΔBM and the ΔTBW using the deuterium oxide (D2O) dilution technique in male and female endurance athletes after prolonged running. We hypothesized that the ΔBM would accurately reflect the ΔTBW. Additionally, the validity of other commonly used markers of hydration status, including serum osmolality (Sosm), urine osmolality (Uosm), and urine specific gravity (Usg) were assessed by comparing them with ΔTBW.
Materials and methods
Men (N = 4)
Women (N = 4)
All subjects (N = 8)
28 ± 4
28 ± 6
28 ± 5
74 ± 10
58 ± 6
66 ± 11
180 ± 9
169 ± 8
175 ± 10
48 ± 2
36 ± 2
42 ± 2
Body fat (%)
11 ± 3
16 ± 3
13 ± 5
VO2max (ml kg−1 min−1)
59 ± 3
49 ± 4
54 ± 6
Training volume (miles week−1)
36 ± 12
34 ± 10
35 ± 11
All subjects had been engaged in ≥12 weeks of regular running before participation in the study and maintained a consistent training schedule until completion of all experimental trials. Subjects were instructed to eat their typical pre-race diet the evening before each trial and to abstain from heavy exercise, alcohol, and caffeine the 24 h before each trial. Diet logs were kept by the subjects to facilitate consistent food and fluid consumption for 24 h before each trial. Subjects reported to the laboratory at 0700 hours on the morning of test days after an overnight fast. Immediately upon arrival, a blood sample was obtained to confirm normal baseline hydration status. Subjects were considered euhydrated when serum osmolality was <290 mOsmol (Sawka et al. 2007).
Subjects drank fluid or no fluid to (1) maintain BM (0%), (2) increase BM by 2%, (3) decrease BM by 2%, or (4) decrease BM by 4% in separate trials. Because this is a companion paper to a study conducted to determine the effect of Na+ intake on serum Na+ concentration (Baker et al. 2008), subjects drank a 6% carbohydrate solution with 0 (Na+0), 18 (Na+18), or 30 (Na+30) mmol/L of Na+ for each level of ΔBM (the osmolality of the beverages were 267, 287, and 279 mOsmol/kg, respectively). Thus, there were a total of 12 possible separate trials. Experimental trials were assigned in random order. To allow sufficient time for clearance of D2O dose between trials, data from trials that were scheduled at least 3 weeks apart were used in this study. Data from a total of 62 trials were analyzed for this paper (4–10 trials per subject).
Subjects had an 18-gauge Teflon catheter placed in an antecubital vein, voided their bladder, and then entered an environmental chamber set at 30ºC and 40% rh. Next, the subject sat quietly for 30 min before the baseline blood sample was obtained. Next, the subject’s pre-experiment BM was measured to the nearest 0.05 kg using a Seca 770 scale. All BM measurements during the experiment were taken with the subject wearing lightweight running shorts, sport bra (women), thin socks, and running shoes. Next, the subject ran for seven 15-min bouts (70% VO2max) each separated by 2 min of rest (2 h of interval running total). Twelve minutes into each running bout a venous blood sample was obtained. At the end of the 2-h interval running protocol, subjects voided their bladder, and then had their BM measured. Next (15 min after the end of the final interval running bout), subjects were asked to run at a speed corresponding to 85% VO2max until exhaustion.
During each rest period, the subject was toweled off and then had their BM measured. Urine samples were collected during rest periods as needed so that BM measurements did not include bladder volume. During the 0% ΔBM trials, subjects drank fluid volumes during the rest periods to maintain their initial BM. During the −2 and −4% ΔBM trials, fluid was restricted until the subjects reached their target BM. If the subjects’ BM fell below their target BM, they ingested enough fluid to maintain the desired %∆BM. During the +2% ΔBM trials, subjects drank the necessary fluid volumes to gain 2% of their BM by the end of the 2 h of interval running. The 2% gain in BM was titrated over the 2-h interval running protocol so that BM gain was achieved gradually (to maximize retention and minimize gastrointestinal discomfort).
During the first 10 min of recovery, subjects walked at 2.5 mph for a gradual cool down. Next, subjects sat quietly in the environmental chamber (30°C and 40% rh) for a 50-min recovery period to allow fluid compartments to stabilize. The subjects’ BM was measured at 10, 30 min, and end of the recovery period (post-experiment BM). Subjects drank fluid or no fluid during recovery to maintain the desired %ΔBM. Urine samples were collected at the beginning (as needed) and at the end of recovery. During exercise and recovery, fans were placed around the subject to promote evaporation of sweat and minimize the amount of sweat trapped in their clothing and shoes.
Sweat patches (PharmChem, Inc.) were placed on the forearm, chest, back, forehead, and thigh of subjects during the second rest period. The patches were removed when an adequate sample was obtained. The patches were then placed in an air-tight plastic tube (Sarstedt Salivette) and then centrifuged at 4°C for 15 min. The sweat was pooled from all five sites on each subject, aliquoted into a cryovial, and refrigerated until analysis.
TBW was quantified using the D2O dilution technique described by Schoeller (1996). The subjects’ natural background D2O concentration was determined from the blood sample drawn at 0700 hours (i.e., pre-dose serum D2O concentration). After voiding their bladder and having their BM measured, the subjects consumed a 30 g dose (measured to the nearest 0.01 g) of 99.9% D2O (Cambridge Isotope Laboratories, Inc.). Next, 100 mL of distilled water was then poured into the same cup and given to the subject to ensure that all D2O was ingested. After D2O dosing, subjects rested for 3 h to allow for equilibration of D2O with body fluids (i.e., according to the equilibration procedure validated by Schoeller 1996). The subjects were not allowed to consume any food or fluids and were instructed to collect all urine voided during this equilibration period.
At the end of the 3-h equilibration period, subjects were allowed to eat a dry, low-sodium snack of their choice (content and calories were consistent among trials within subjects). Then, a post-equilibration blood sample was collected to determine pre-experiment TBW. Post-experiment TBW was determined from the blood sample drawn after recovery. When calculating pre- and post-experiment TBW, corrections were made for lost dosage of D2O in urine, blood, sweat (each fluid collected and measured directly), and breath vapor (estimated according to Mitchell et al. 1972 and Wong et al. 1988), as well as nonaqueous hydrogen exchange (Schoeller et al. 1985).
When serum [D2O] measurements were not available (e.g., when there was not enough sample volume or the sample was contaminated during the D2O extraction procedure (see below), the pre-dose urine [D2O] or post-equilibration urine [D2O] was used in place of the pre-dose serum [D2O] or post-equilibration serum [D2O], respectively, to complete the calculation of Npre.
When serum [D2O] measurements were not available (e.g., when there was not enough sample volume or the sample was contaminated during the D2O extraction procedure (see below), the pre-dose urine [D2O] or post-experiment urine [D2O] was used in place of the pre-dose serum [D2O] or post-experiment serum [D2O], respectively, to complete the calculation of Npost.
D2O was extracted from serum, urine, and sweat samples according to an equilibration procedure developed and validated by Davis et al. (1987). First, 1.5 mL of sample was placed into the center well and 1.5 mL of distilled water into the surrounding moat of a covered Conway diffusion dish (Bel–Art Products). Then, an airtight top was placed on the diffusion dish and secured by parafilm and tape. Diffusion of H2O and D2O occurred through the vapor phase during incubation of the samples for 48 h at 37°C. The dishes were then cooled to room temperature and the fluid in the outer moat (1.5 mL of the original water phase) was aliquoted into an airtight vial with minimal dead space for subsequent analysis. The D2O concentration in the extracted samples was doubled to calculate the amount contained in the original sample before the equilibration procedure. Periodically, standards with a known D2O concentration were equilibrated and analyzed to determine the D2O recovery rate. The D2O concentration of the extracted sample was 48–52% of the original sample, confirming that the equilibration process was complete.
The D2O concentration in serum, urine, and sweat was measured in duplicate using a PC-controlled Fourier Transform Infrared Spectrometer (Bruker IFS 66/s, OPUS 6.0) equipped with an MCT-A detector and a sealed liquid cell (100 μm path length, CaF2 windows, Specac Model #20502,). Temperature regulation of the liquid cell was accomplished using a water heating jacket (Specac Model #20710) coupled to a water bath (Neslab RTE-111 with Digital Plus controller). Cell temperature was monitored using a T-type thermocouple placed in contact with the cell window. Sample introduction and evacuation was accomplished via custom fitted Luer lock terminated Teflon tubing (Hamilton #90619). This technique permitted rapid sample changes without removal of the cell and as such eliminated errors associated with cell position. A 4 mm internal aperture was used to minimize MCT nonlinearities associated with a high photon flux.
The following acquisition and processing parameters were used: 1,000 scans, 32 cm−1 resolution, Norton–Beer medium apodization function, and a zerofilling factor of 16. A cell temperature of 25.0 ± 0.10°C was selected to be near the laboratory ambient temperature for optimal control and to facilitate analysis of room temperature samples. Temperature equilibration of the cell started 20 min before measurements. The cell was rinsed (3 mL distilled water), purged, and allowed time for temperature equilibration before injection of the next sample. A temperature-controlled cell was used because minor variations in cell temperature produce variations in absorbance (Byers 1979; Lukaski and Johnson 1985).
Quantification of the D–O stretching region was done by applying a linear baseline function from 2,715 to 2,400 cm−1 and then integrating from the peak maximum to 2,715 cm−1 (OPUS 6.0, H-Type Method). This method was used to avoid integration errors associated with a variation in ambient CO2 levels (peak at 2,350 cm−1).
D2O concentrations were calculated from a linear standard curve. Standards were prepared on the morning of each day of analysis by gravimetric dilution of known quantities of D2O (99.9%, Cambridge Isotope Laboratories, Inc.) in distilled water and ranged from 50 to 700 ppm. During each day of analysis the D2O standard solutions were analyzed twice at random intervals to monitor and confirm the integrity of the system. Standards were periodically sent out for isotope ratio mass spectrometry analysis (Isotech Laboratories, Inc.) to confirm D2O concentration. Distilled water was also used as the spectral reference for all measurements. The coefficient of variation (CV) for duplicate measurements in the same assay of sample aliquots was 0.5% (the CV was the same when a subset of samples were run in triplicate).
Blood and urine analysis
Venous blood samples (9 mL each) were drawn without stasis. A 2-mL aliquot was transferred into an EDTA-treated test tube and immediately analyzed for hematocrit (microhematocrit centrifugation) and hemoglobin (Hemacue Hb 201+) in triplicate. The remaining 7-mL aliquot was transferred into a serum separator tube, allowed 30–60 min to clot, and then centrifuged at 4°C for 15 min. The percent change in plasma volume from baseline (ΔPV) was calculated from hematocrit and hemoglobin (Dill and Costill 1974). Serum and urine were analyzed for osmolality (freezing point depression, Advanced DigiMatic Osmometer Model 3D2) in triplicate. It should be noted that D2O depresses the freezing point of aqueous solutions; however, given the small amount of D2O (30 g) distributed in 42 kg of TBW, the effect of D2O on measured serum and urine osmolality should be minimal. Urine samples were also analyzed for specific gravity (Refractometer, Atago A300CL).
An intraclass correlation coefficient was used to determine the reliability of ΔBM as a predictor of ΔTBW (as described in Shoukri and Pause 1999). The slope and intercept of the regression line and line of identity of the scatter plot for ΔBM versus ΔTBW were compared to determine the accuracy of ΔBM as a predictor of ΔTBW. The standard error of the estimate (SEE) was calculated for the regression of ΔBM on ΔTBW to determine the accuracy with which ΔBM can predict ΔTBW during prolonged exercise. A paired t test was used to determine whether there was a significant difference between ΔBM and ΔTBW. The effect size for the BM and TBW methods were calculated for each individual subject. Effect size was calculated using the following equations for each subject: BM effect size = (mean pre-experiment BM − mean post-experiment BM)/(mean SD for pre- and post-experiment BM); TBW effect size = (mean pre-experiment TBW − mean post-experiment TBW)/(mean SD for pre- and post-experiment TBW), where mean pre- and post-experiment BM and TBW represent the average among trials (−4, −2, 0, +2% ∆BM).
Regression analyses and Pearson correlations were used to describe the relations between ΔTBW and each of the predictors of ΔTBW (i.e., ΔBM, Sosm, Uosm, and Usg). A two-way (beverage vs. %∆BM) analysis of variance with repeated measures was used to determine significant differences in run time to exhaustion and in the pre- to post-experiment change in Sosm, Uosm, Usg, and %ΔPV among trials. The significance level for all statistical tests was set to alpha = 0.05. All data are presented as mean ± SD unless otherwise indicated.
ΔBM versus ΔTBW
In the present study, the subjects’ actual %ΔBM was +1.8 ± 0.4, −0.2 ± 0.2, −2.1 ± 0.2, and −3.3 ± 0.6% during the target +2, 0, −2, and −4% ΔBM trials, respectively. The subjects’ absolute ΔBM ranged from −3.10 to +1.50 kg over all trials.
Effect sizes for BM and TBW for each subject
Biological markers of ΔTBW
Relations between ΔTBW and predictors of hydration status
Decrease TBW by 3%
Decrease TBW by 6%
Increase TBW by 3%
ΔTBW = 0.9377 BM + 0.1071
ΔTBW = −0.1717 Sosm + 48.7
285 ± 3
ΔTBW = −0.0044 Uosm + 1.66
462 ± 208
ΔTBW = −150 Usg + 151
1.012 ± 0.006
There were no differences in baseline Sosm, Uosm, or Usg among trials. The mean Sosm, Uosm, and Usg at baseline were 285 ± 3 mOsmol/kg, 462 ± 208 mOsmol/kg, and 1.012 ± 0.006, respectively. During the +2% ΔBM trials, the decrease in Sosm from baseline was significantly smaller with Na+18 (−4 ± 3 mOsmol/kg) and Na+30 (−5 ± 4 mOsmol/kg) versus Na+0 (−9 ± 3 mOsmol/kg). There were no other statistical differences in Sosm, Uosm, or Usg among beverages within ΔBM levels.
The ∆PV from pre-to post-experiment was +4.7 ± 3.6, +2.8 ± 3.3, −3.3 ± 3.0, and −7.5 ± 3.1% in the +2, 0, −2, and −4% ΔBM trials, respectively. There was a statistically significant difference in ΔPV among all levels of ΔBM except 0 versus +2% ΔBM. There were no statistically significant effects of beverage Na+ concentration on ΔPV.
The run times to volitional fatigue were 26 ± 17, 22 ± 13, 14 ± 8, and 8 ± 7 min during the +2, 0, −2, −4% ΔBM trials, respectively. Because performance was not the research question of interest here, these results will not be mentioned any further in this paper; however, they are discussed in more detail elsewhere (Baker et al. 2008).
The major findings from this study were: (1) pre- to post-exercise ΔBM is an accurate and reliable method to estimate the ΔTBW in men and women after prolonged running in the heat and (2) ΔBM is a better method to predict ΔTBW than Sosm, Uosm, and Usg.
ΔBM versus ΔTBW
The ACSM recommends that athletes estimate their sweating rates by measuring their BM before and after exercise (Sawka et al. 2007) because ΔBM provides a reasonably accurate estimate of the acute ΔTBW during exercise (Cheuvront et al. 2007; Sawka et al. 2007). The present study shows no statistical differences between methods, suggesting that the pre- to post-exercise ΔBM is an accurate and reliable predictor of an acute ΔTBW during prolonged exercise in the heat. The slope and intercept of the relation between the ΔBM and the ΔTBW (Fig. 1) were not significantly different from 1 and 0, respectively, indicating no statistical difference between methods. The SEE for the regression of ΔBM on ΔTBW indicates that ΔBM accurately predicts ΔTBW within 1.06 kg. Further, the intraclass correlation coefficient (0.76) indicates that there is excellent reliability (Shoukri and Pause 1999) between the ΔBM and the ΔTBW. These results are consistent with Bartok et al. (2004) who found that the average ΔBM and the average ΔTBW (measured by the D2O dilution technique) were comparable (−2.4 and −2.2 kg, respectively) in wrestlers after exercising in a hot environment without fluid replacement.
There has been some debate regarding the use of ΔBM as an index of ΔTBW and hydration status. Maughan et al. (2007) have written a theoretical analysis on the errors in the estimation of hydration status from ΔBM. The potential sources of error include (1) loss of BM due to substrate oxidation, (2) water of oxidation, and (3) release of water bound to muscle glycogen. Substrate oxidation results in the formation of CO2, which is lost in expired air, and water, which is added to the TBW pool. The rate of CO2 and metabolic water production depend upon exercise intensity and the relative contribution of carbohydrate and fat, while the rate of water release from glycogen depends on the amount of muscle glycogen utilized and the amount of water released from each gram of glycogen (Maughan et al. 2007).
Noakes et al. (2005) suggest that a 70-kg athlete can lose ≥2.2 kg (≥3% BM) during endurance events without experiencing a net loss in body water. According to Noakes et al. (2005), during an endurance event, a 70-kg athlete will lose ~0.6–0.8 kg from fuel oxidation. In addition, the athlete will gain ~0.4 kg via metabolic water production and ~1.5 kg via release of water as glycogen is utilized for fuel. Further, Noakes et al. (2005) assume that this water bound to glycogen represents a store of water that can be lost as sweat but which does not contribute to a reduction in the exchangeable TBW pool. Thus, according to Noakes et al. (2005), a total BM loss of ≥2.2 kg (≥3% BM) could occur in a 70-kg athlete without a ΔTBW.
This debate centers on whether endogenous water production plays a significant role in the overall ΔTBW during prolonged exercise. Studies attempting to determine the amount of water stored with glycogen have been inconclusive, so it is unknown exactly how much water is released as a result of glycogen utilization during prolonged exercise (Sherman et al. 1982). Further, Cheuvront et al. (2007) and King et al. (2008) suggest that water stored with glycogen is already part of the TBW pool and therefore release of water from glycogen does not increase TBW. Additionally, metabolic water production is partially offset by respiratory water loss during exercise (Cheuvront et al. 2007; Consolazio et al. 1963; Mitchell et al. 1972), resulting in water turnover with minimal net ΔTBW (via these factors). In the present study, the total calculated metabolic water gain was 195 g (Brooks and Mercier 1994; Consolazio et al. 1963) and the respiratory water loss was 187 g (Mitchell et al. 1972).
Loss of BM due to oxidation of muscle and liver glycogen and fatty acids stored in adipocytes does represent non-sweat sources of BM loss during exercise, i.e., would cause a decrease in BM without effecting TBW. Rogers et al. (1997) calculated the weight loss resulting from fuel oxidation in male athletes over the course of a 10 h ultra-endurance race. The weight change due to CO2 loss was estimated to be −68 g h−1. This rate of CO2 loss would amount to ~0.2–0.4 kg (≤0.6% of BM in a 70-kg athlete) over the course of a 3–5 h race (e.g., 42-km marathon). Additionally, Cheuvront et al. (2007) suggest that CO2–O2 exchange contributes ≤1% of body mass loss from pre-to post-exercise. Thus, a pre- to post-exercise BM deficit would be expected to slightly overestimate the actual decrease in TBW (by ≤1%). However, if BM measurements are taken in the field (precluding the measurement of nude BM) the loss of BM due to CO2–O2 exchange is likely countered by the mass of sweat trapped in clothing (Cheuvront and Haymes 2001). Further, athletes are likely to ingest carbohydrate during an endurance event, which will partially offset the decrease in BM due to fuel oxidation and CO2 loss.
Although there were no statistical differences between ΔBM and ΔTBW, perhaps one caveat of the present study results, in practical terms, is the relatively wide range in the differences between methods (−2.58 to +2.54 kg). Thus, it is important to consider the potential sources of variability that may account for this range, as well as, the ±1.06 kg SEE between ΔBM and ΔTBW. It is possible that some of the difference between these methods is caused by endogenous water production and/or non-sweat sources of weight loss (from the oxidation of carbohydrate and fatty acids). However, these factors would cause ΔBM to overestimate sweat losses, i.e., the decrease in BM would be consistently and significantly greater than the decrease in TBW. This systematic overestimation was not observed in the present study. The differences between methods were distributed normally around the mean and the average difference between ΔBM and ΔTBW was 0.07 kg, suggesting only a slight tendency for ΔBM to overestimate ΔTBW. Thus, because there was no indication that the decrease in BM was consistently or significantly greater than the decrease in TBW, we conclude that the data in the present study do not support the notion that 3% BM loss is equivalent to no net ΔTBW (Noakes et al. 2005). Moreover, it is probable that most of the variability in the difference between methods can be simply accounted for by the random measurement error associated with the D2O measurement. While the CV for each individual D2O measurement was low (0.5%), the CV for the overall calculation of ΔTBW would be higher (i.e., there would be an additive effect). Considering all of the steps involved in the TBW calculations, the overall CV is approximately 2.5% which would be equivalent to approximately ±1 kg of body water (in an individual with 42 kg of TBW). Similarly, in a study by Gudivaka et al. (1999), the mean, standard deviation, and range for pre-to post-experiment changes in TBW (induced by intravenous infusion of lactated Ringer solution or oral administration of a diuretic agent) measured by the dilution technique was 1.7, 0.8, and 0.1–3.1 kg, respectively.
Biological markers of ΔTBW
The isotope dilution technique is considered to be the most accurate method to measure ΔTBW (Armstrong 2007; Schoeller 1996). However, the expensive and time-consuming nature of this technique limits isotope dilution as a practical means to assess ΔTBW in most field and laboratory situations. Given the importance of fluid balance in exercise performance and health, athletes need a practical, yet accurate and reliable method to assess their hydration status in the field. In addition to ΔBM, other techniques often used to assess hydration status include Sosm, Uosm, and Usg (Sawka et al. 2007). Table 3 presents the regression and correlation analyses of each predictor versus ΔTBW. ∆BM is most highly correlated with ΔTBW (0.77, P < 0.05), and thus, is the best predictor of the change in hydration status during exercise. Although the correlations are not as high as with ΔBM, post-experiment Sosm (0.68), Uosm (0.61), and Usg (0.63) are also significantly correlated with ΔTBW.
Table 3 describes the predicted changes in BM, Sosm, Uosm, and Usg at various levels of %∆TBW. The regression analyses predict that for a 66 kg athlete with 42 kg of TBW (mean values for subjects in the present study), a 2% decrease in BM (or 3% decrease in TBW) from pre- to post-exercise would have a Sosm of 291 mOsmol/kg, a Uosm of 664 mOsmol/kg, and a Usg of 1.020. These values are relatively consistent with the indices of hydration status provided in the ACSM position stand on exercise and fluid replacement (Sawka et al. 2007); i.e., Sosm ≥ 290 mOsmol/kg, Uosm ≥ 700 mOsmol/kg, and Usg ≥ 1.020 are indicative of dehydration. These biomarkers can also be useful for detecting when an athlete is overdrinking relative to their sweat losses. According to Table 3, the indicators of a 2% gain in BM due to overdrinking include a Sosm of 276 mOsmol/kg, a Uosm of 91 mOsmol/kg, and a Usg of 1.003.
Summary and practical recommendations
In summary, measuring the pre- to post-exercise ΔBM in male and female endurance athletes is an accurate and reliable method to assess the ΔTBW after prolonged running in the heat. Athletes can use pre- to post-exercise ΔBM to obtain a reasonable estimate of their sweat loss and their hydration status. Therefore, this practice can confidently be used to estimate athletes’ fluid replacement needs during and after exercise. While measures of serum and urine concentration (e.g., Sosm, Usg, and Uosm) are not as highly correlated with ΔTBW as ΔBM, they can still be useful in predicting hydration status when pre- and post-BM measurements are not available.
The authors are grateful to the subjects for their participation in this study. Additionally, we thank Cynthia Bartok for her advice on TBW procedures and calculations, Josh Stapleton for his assistance with FTIR procedures, Mosuk Chow and Dennis Passe for their statistical consultation, Jane Pierzga, John Jennings, Matt Kenney, Jose Flores, Ben Miller, Doug Johnson, and Randy McCullough for their technical assistance, and the General Clinical Research Center nursing staff for their medical support. Support for this study was provided by the Gatorade Sports Science Institute and the General Clinical Research Center Grant MO1 RR010732.