Fluid Balance During and After an Ironman Triathlon

Ironman ultradistance triathlons involving a 3.8 km swim, a 180 km cycle, and a 42.2 km marathon run have increased greatly in popularity in the last 15 years. Despite the popularity of these events, there has been little field research to measure the fluid intake of triathletes during these events. Speedy et al. 1 reported the estimated fluid intakes of four athletes who developed hyponatremia during an Ironman triathlon. These intakes ranged between 6.2-16 L. These authors have also reported a mean weight loss, considered a measure of hydration status, of 2.1 kg or a 2.9% loss of body weight among healthy finishers in the 1996 New Zealand Ironman triathlon.

Although there are well-established guidelines for appropriate fluid intakes for shorter athletic events, 2 to the best of our knowledge there has been only one report on the actual fluid intakes of athletes in an ultradistance triathlon: Applegate et al. 3 reported a high rate of fluid ingestion (1.5 L/h) at the Hawaiian Ironman. There has only been one study investigating how weight changes relate to actual hydration status in these ultradistance events. 4 The paucity of such data make it difficult to provide athletes with firm recommendations on what is an appropriate fluid intake for these ultradistance triathlons, in which athletes are often exercising at a low intensity but for 12 or more hours. Indeed, ultradistance athletes need to be cognizant not only of the risks of dehydration if they drink too little, but must also avoid overdrinking as fluid overload is the likely etiology of symptomatic exercise-induced hyponatremia, a common complication of ultradistance triathlons.

Accordingly, the aim of this study was to measure weight changes, fluid intake, and changes in serum sodium concentration in 19 triathletes competing in the 1997 New Zealand Ironman triathlon, a large international ultradistance event.

METHODS

Twelve male and 7 female athletes were studied prospectively to determine their fluid intake during the 1997 New Zealand Ironman triathlon. This triathlon involves a swim of 3.8 km, a cycle of 180 km, and a run of 42.2 km. Ambient air temperature at 1200 h was 21°C, with a relative humidity of 91%. Water temperature was 20.7°C.

Subjects were recruited by examining all entry forms for the race. Athletes who lived in Auckland City, where the race was held, and who estimated their race time to be between 10.5 and 14 hours were invited to participate in this study. Subjects gave their informed consent, and ethical approval for the study was obtained from the North Health Ethics Committee. All subjects were healthy, with no significant medical illnesses, as expected in competitors in this arduous event.

One to 1.5 hours prior to the start of the race, subjects had 20 ml of blood drawn by venipuncture and assayed for serum sodium concentration, hemoglobin (Hb), and hematocrit (Hct). This procedure was repeated upon completion of the race, and at 0800 h on the morning after the race (approximately 12 hours after completing the race). All venipunctures were performed in the sitting position. Assay for serum sodium concentration was carried out on the day after the race with Hitachi 747 or 737 analyzers (Boehringer Mannheim, Mannheim, Germany), using standard methods and the manufacturer's reagents, on serum that had been collected into silicone-gel separator tubes and stored at 4°C after centrifugation within 1 hour of collection. Plasma sodium assays on any subjects needing medical care post-race were performed on-site using a Nova Ion Selective Electrode analyzer (Waltham, MA, U.S.A.) on lithium-heparin anticoagulated samples.

A calibration difference of 2 mmol/L was detected between the on-site Nova Ion Selective Electrode and the Hitachi analyzers in both this study and in a much larger study (n = 373) 1 performed using the same analyzers in the same race. The results from the Hitachi analyzers were therefore scaled down by 2 mmol/L to ensure comparability with the specimens analyzed on-site. Routine hematological assays for Hb and Hct were carried out on the day after the race with Technicon H1, H2, or H3 analyzers (Technicon Corporation, Tarrytown, NY, U.S.A.) on EDTA-anticoagulated samples that had been stored at 4°C. The laboratory performing the analyses was accredited for medical testing to ISO9001 by International Accreditation New Zealand. Plasma volume changes were calculated using the formula derived from the Dill and Costill's equation.

All subjects were weighed approximately 1 hour prior to the race, at the swim-bike transition, at the bike-run transition, immediately after finishing the race, and at 0800 h on the morning after the race. Subjects were weighed in minimal clothing and without shoes or bicycle helmets. Weights were measured on calibrated Seca scales (Seca, Hamburg, Germany) placed on a hard level surface.
Food and drink were freely available at support stations every 12 km on the cycle course and every 1.8 km on the run. Fluid choices included water, Coca-Cola, and a sports drink (Powerade) containing 8% carbohydrate and 10 mmol/L of sodium. Athletes were instructed prior to the race to keep a careful mental note of how much fluid and which fluid they were consuming during the race, expressed in terms of number of drink bottles on the cycle and cups of fluid on the run. Athletes were questioned while they were racing to determine their fluid intakes. Interviewers ran or cycled with the athletes for a short distance to obtain this information, and also questioned the athletes at the transitions. Interviews took place at the swim-cycle transition, at 100 km on the cycle course, the cycle-run transition, at three points on the run course (6.2 km, 19.9 km, and 34.7 km), and immediately after finishing the race.

Drink cups at the run support stations were of two sizes (predominantly 355 ml, with a few 250 mL cups also used). Also used were 750 mL cycle bottles. The drink cups and cycle bottles were filled by race support station staff, and the volumes with which they were filled were not measured. It was retrospectively estimated by the athletes (by interview within 3 days of the race) that the drink cups on the run were half filled (to contain approximately 175 mL of fluid) and that the cycle bottles were filled to contain approximately 700 mL of fluid.

Fluid losses (total of sweat, urine, and respiratory) during the cycle and run splits were calculated by subtracting weight change from fluid intake during these sections of the race. Rates of fluid loss were calculated by dividing total fluid loss by the time taken to complete each leg of the race. Weight of intake of solid nutrients and weight changes from any fecal loss were not included in these calculations. Weight loss from use of metabolic fuel was likewise disregarded for these calculations.

Statistical Analysis

Differences were tested by means of nonparametric tests, using the sign test for paired comparisons. Nonparametric analysis was used because of the small sample size and the nonnormal distribution of the outcome. Correlations between weight changes and serum sodium concentrations and change in serum sodium concentrations were calculated using Pearson's correlation coefficient.

RESULTS

One subject (male) failed to finish the race, withdrawing after finishing the cycle, with dizziness, vomiting, and mild hyponatremia ([Na] = 134 mmol/L), and was not included in the analysis, thus leaving 18 athletes available for analysis. Their median age was 35 years (range 23-49), and weight was 67 kg (range 57.5-93.5).
Subjects completed the race with a median race finishing time of 12.3 hours (range 10.9-15.3). The median time from race finish to post-race venipuncture was 13.5 minutes (range 7-32). Sixteen subjects lost weight during the race, the median weight change over the entire race was -2.5 kg (-4 to +1.5), which equates to a relative loss body weight of -3.5% (-6.1 to +2.5), this change being statistically significant (p < 0.0006).
Weight changes were measured after the swim, bike, and run sections, and at 0800 h the following morning (recovery). Full data on weight were not available on two subjects.

Subjects sustained median weight losses of -1.0 kg (range -2.0 to +0.5) during the swim (p = 0.3) and -2.0 kg (-3.5 to +1.5) during the run (p < 0.0002). However there was a significant median weight gain of +0.5 kg (-1.0 to +3.0) during the cycle (p = 0.03). Subjects also lost weight during recovery, with a significant median weight loss from race finish to the following morning of -1.0 kg (-4.5 to +1.0) (p < 0.03). As a result, all athletes lost weight over the period from before the race start to recovery the following morning (median = -2.8 kg, range -1.5 to -5).
Fluid intakes of subjects over the race are shown in Table 1. Median hourly fluid intake over the whole race was 716 ml/h (range 421-970). The median intakes of water and Powerade were 275 ml/h (0-567) and 271 ml/h (0-618), respectively. Median hourly fluid intakes were significantly higher on the bike 889 ml/h (601-1,310) than on the run 632 ml/h (238-1,129) (p = 0.0327). Median calculated fluid losses during cycling were 808 ml/h (range 469-1,083) and during running were median = 1,021 ml/h (range 404-1,801).

There were no significant differences between pre-race and post-race serum sodium concentrations (median =140 mmol/L versus 138 mmol/L) (p = 0.12), or between post-race and recovery serum sodium concentrations (median = 138 mmol/L versus 137 mmol/L) (p = 0.17). However there was a significant difference between pre-race and recovery serum sodium concentrations (median = 140 mmol/L versus 137 mmol/L) (p = 0.03).
Full data for hematocrits and hemoglobins were not available on three subjects. There was a significant increase in plasma volume over the race (median = +10.8%, p = 0.0005). As a result there was a significant fall in hematocrit during the race (median = -2.5%, range -7 to 1, p = 0.009), and a significant fall in hemoglobin during the race (median = -7 g/L, range -18 to 1, p = 0.014). However there was only a minimal change in plasma volume during the recovery period (median = -1.1%, range -12.4 to 14.2, p = NS).

Five subjects (one male and four females) developed hyponatremia with post-race serum sodium concentrations between 128-133 mmol/L. Two of these subjects (one male and one female) sought medical attention. The median pre-race weight of these five athletes was 67 kg and they had a median weight loss during the race of 0.5 kg, or 0.7%. Median fluid intake for these athletes was 714 ml/h (range 421-766). However they had a high fluid intake on the cycle (median = 913 ml/h; range 631-1,207). Median fluid intake over the race was water = 4,630 ml, Powerade = 3,325ml, Coke = 50 ml.

There was a significant inverse correlation between post-race serum sodium concentration and relative body weight change (r = -0.76, p = 0.0003), with subjects with the higher serum sodium concentrations losing the most weight, and those with hyponatremia losing only minimal weight or gaining weight (Figure 1). There was a similar inverse relationship between the change in serum sodium concentration pre-race to post-race and the relative weight change (r = -0.68, p = 0.0029): a lowering of serum sodium concentration was associated with the least weight loss or weight gain