Controversy exists regarding optimal fluid guidelines for athletes engaging in different sports. Most published recommendations emphasize the detrimental consequences of dehydration (Convertino 1996, Coris 2004, task force heat statement, Cheuvront 2003, Coyle 2004) while more recent reports warn of the morbid consequences of hyperhydration (Consensus Statement, Noakes BMJ, Noakes CJSM). Accordingly, individualized recommendations emphasizing a balance between the two extremes have evolved (Casa USATF, Gatorade and Powerade calculators). These revised guidelines, however, continue to promote static recommendations for dynamic athletic situations.Marathon running is a dynamic situation requiring a constant adjustment to a changing homeostatic imbalance. Real time assessments of fluid and sodium homeostasis are physiologically represented by changes in plasma osmolality. Thirst is stimulated when plasma osmolality increases 1-2%; after maximal antidiuresis can no longer prevent further elevations in plasma osmolality (Robertson 1984, McKinley 2004 New physiol sci, Verbalis 2003). Interestingly enough, concomitant performance decrements and cardiovascular strain are also documented when baseline body fluid losses exceed 2% (Coyle 2004).
The failure of athletes to replace 100% of bodyweight losses from ad libitum fluid intake has been well-described as “involuntary” (Nose 1988, Greenleaf 1992) or “voluntary” dehydration (Greenleaf 1965, Cheuvront 2001, Hubbard 1984); devaluating thirst as a “poor” indicator of body fluid needs. However, the combination of laboratory research with recent field data suggests that the body defends plasma osmolality and NOT bodyweight during prolonged endurance exercise. Since it is well established that in most land mammals thirst is stimulated in response to changes in tonicity at rest (Robinson 1984, Verbalis 2003, McKinley 2004, Stricker 1988, Rolls 1980, Phillips 1985), it seems reasonable to conclude that thirst would also be the premier physiological and dynamic motivator governing fluid and sodium balance during exercise as well.
This article will focus on the physiology of normal fluid balance as the ultimate guide towards the evolution of optimal fluid recommendations. Four physiological considerations of fluid and sodium balance will be detailed followed by five practical recommendations. The neuroendocrine regulation of homeostasis will be emphasized in defense of the behavioral drives for thirst and sodium palatability during dynamic activities such as prolonged endurance exercise. The behavioral drives to maintain fluid and sodium balance are evolutionary stable, essential for the safety and survival of the species and deeply rooted within our genotype (Fitzsimmons 1998).
The American College of Sports Medicine’s (ACSM) current guidelines promote a fluid intake of 600 – 1,200 ml/h; a range largely based on laboratory studies performed on elite male athletes (Convertino 1996). The popularity of marathon running increased markedly at the time these guidelines were released, with charity running groups enticing more recreational runners into the sport. Slower, more inexperienced, female athletes heeded the upper limit of the ACSM guidelines and developed exercise-associated hyponatremia (EAH) by ingesting more fluid than they could excrete over the course of a marathon run (Hew 2003, Almond 2005). In 2001, the International Marathon Medical Directors Association (IMMDA) approved guidelines lowering the “acceptable” range to 400 – 800 ml/h to adjust the upper limit and protect smaller athletes (mainly female) from overdrinking (Noakes CJSM).
Recently, mathematical models have proposed that these ranges may not be ideal to cover the current population of runners participating in marathon events (Montain 2006). Close inspection of a marathon field today personifies the variety of shapes, sizes and speeds that need to be taken into consideration when formulating fluid intake guidelines. For example, the average weight of 7,299 runners registering for the 2005 Comrades Marathon was 73 kg. The lightest competitor weighed 43 kg while the heaviest runner weighed 119 kg. The fastest runner averaged 16.4 km/h (winning time 5:27) while the slowest runners ran at a 7.4 km/h pace (the official 12 hour cut-off pace). The 11,728 runners who completed the 89.2 km Comrades Marathon ran, on average, 49 minutes slower during the second half of the race than they ran the first half of the race with a minimum and maximum range of -11 (negative split) and 180 minutes (positive split) respectively. The wide variation between split times exemplifies the late-race functional deterioration that has been described elsewhere (Cade 1992) and highlights the fact that marathon running is largely a dynamic situation, encompassing participants with widely diverse physical attributes and fitness levels. A much wider spread of running speeds and bodyweights would be expected in a standard 42.2 km marathon, as the Comrades Marathon is more than double the distance, requires a qualifying time for entry, has a defined cut-off and held over a grueling terrain; all of which precludes the participation of unfit and novice runners. Conversely, relatively unfit and untrained participants can often complete a standard marathon, amplifying the participant extremes.
The three main factors governing fluid loss are mass (bodyweight), running speed (metabolic rate) and ambient temperature (Costill, Cheuvront 2002). Race day temperatures in the New York City Marathon have ranged as much as 1ºC to 29ºC and have fluctuated as much as 17ºC from start to finish (ref 62 in Cheuvront 2001 review article). Therefore, it would be foolhardy to try and formulate ANY blanket “range” that would encompass this wide spectrum of weather conditions, bodyweights and running speeds that characterize modern day marathon participation.
Thirst is a subjective sensation characterized by a deep-seated desire for water (Robertson 1984; Fitszimmons 1998, Stricker 1988) and is generally associated with oral sensations such as dryness, irritation and an “unpleasant taste” in the mouth (Phillips 1985, Rolls 1980, Engell 1985). Thirst has been quantified using geometric and visual analogue scales (Rolls 1980, Engell 1985, Robertson 1984, Wright 2004) and documented to increase when plasma osmolality increases 1-2% above baseline levels; well within the normal physiological range of ~275-295 mOsm/kg (Valtin 2002, Robertson 1984, Phillips 1985).
Plasma osmolality (POsm) is maintained within this narrow range to protect intracellular volume. Serum sodium concentrations [Na+] mirror plasma osmolality (Kratz 2005) because sodium is the principle solute of extracellular fluid; thus [Na+] and plasma osmolality are viewed as interchangeable in this discussion, although the appropriate calculation is: POsm = 2 x [Na+] + [BUN] + [glucose] (all in mmol/L) (Verbalis 2003).
The maintenance of intracellular volume is of vital importance for cell function and survival. Hypertonicity causes intracellular dehydration; with a bodyweight loss of 5.5% during prolonged severe exercise eliciting a reduction in red cell volume by 3.2% (Astrand 1964). A reduced intracellular volume can reduce the rates of glycogen and protein synthesis (Shirreffs 2000) and hypertonicity beyond serum sodium levels of 160 mmol/L can cause encephalopathy and death (Riggs 2002). Conversely, hypotonicity causes intracellular expansion. Although high cell volume can stimulate glycogen and protein synthesis, excessive and acute cell swelling from serum sodium levels below 125 mmol/L can lead to noncardiogenic pulmonary edema, hyponatremic encephalopathy and death (Ayus 2000).
Thirst and arginine vasopressin (AVP) secretion are intimately intertwined: synergistically lowering elevations in plasma osmolality by stimulating fluid intake and promoting antidiuresis. The threshold for AVP release (~280 mOsm/L) is normally below that for thirst (~290 mOsm/L), despite significant variation between individual set points (Robertson 1984). This evolutionary design liberates individuals from constantly seeking water, as thirst is only stimulated when antidiuresis is maximal and hypertonicity exceeds the capacity of the kidneys to cope with rising tonicity. The strength of an intact thirst mechanism is exemplified in patients with nephrogenic diabetes insipidus, where plasma osmolality is maintained despite the intake and excretion of up to 20 liters of fluid per day (Robertson 1984). Oropharyngeal metering (Figaro 1997, Salata 1987, Seckl 1986) and stomach fullness (Rolls 1980, Phillips 1985, Costill 1970) provide inhibitory feedback to terminate drinking quicker than the return of POsm to normal levels perhaps as a safety measure to prevent overdrinking.
The strength and precision of the thirst mechanism is demonstrated both at rest and during exercise. An increase in POsm, in response to hypertonic saline infusion, promoted a dipsogenesis proportional to the increase in POsm; with 82% of the total volume of water consumed within the first 5 minutes of rehydration with the immediate cessation of intake due to “stomach fullness” (Phillips 1985). In a separate fluid deprivation study, 65% of total water intake was consumed during the first 2.5 minutes of rehydration (Rolls 1980). Subjective ratings of dry mouth and plasma protein concentrations returned to baseline levels in 2.5 minutes, plasma volume returned to baseline levels within 5 minutes and serum sodium levels started decreasing in 5 minutes; returning to baseline values within 12.5 minutes. This initial drinking bout was also interrupted by sensations of “stomach fullness”. During prolonged endurance exercise, athletes respond to graded levels of hypohydration with ad libitum fluid intakes sufficient enough to restore plasma osmolality, replace sweat losses equally and attenuate thermal and circulatory strain (Cheuvront 2001, Armstrong 1997).
Thus, humans respond to increases in [Na+] with AVP secretion, thirst and water intake that occur when POsm is well within the normal range. The precision and magnitude of the thirst drive is related to ancestral areas of the brain that are associated with vegetative systems such as the hunger for air, food, micturition and pain (Egan 2003). To assume that the thirst drive would be an “inaccurate index” of fluid balance during exercise (the dynamic homeostatic imbalance alternatively described as the “flight or fight” response) would seem contradictory to the evolutionary design and survival of our species.
Although there is concern that thirst due to intracellular dehydration or extracellular volume depletion may be confused with the sensations of a dry mouth during exercise, thirst due to xerostomia has not been shown to lead to polydipsic hyponatremia (Wright 2004). The sensation of a dry mouth can promote more frequent drinking episodes, but total fluid intake is similar, or counter regulated by increased urinary output, in tested clinical and experimental situations (Steggerda 1941). Recent studies on xerostomia, thirst, and interdialytic weight gain have not factored in osmolality and therefore do not provide convincing evidence that a dry mouth without changes in POsm can contribute to excessive fluid intakes and fluid overload hyponatremia (Brunstrom 2000, Brunstrom 2002, Bots 2004, Bots 2005)
The reluctance of man to voluntarily replace fluids at amounts equivalent to sweat losses was detailed in 1933, when Dill hypothesized that humans only drunk as much fluid to maintain constant osmolarity of the extracellular fluid (Dill 1933). This “failure” of man to replace 100% of bodyweight losses has been subsequently recognized as “voluntary” or “involuntary” dehydration; minimizing the importance of thirst as an adequate indicator of body fluid needs (Greenleaf 1965, Greenleaf 1992, Nose 1988).
Field study data indicates that the actual maintenance of bodyweight will facilitate reductions in serum sodium concentrations from pre to post race (Table 1). Athletes who lose less than 2% of bodyweight experience a decrease in [Na+] (Glace 2002, Gerth 2002). Athletes who gain weight progress from normonatremia to hyponatremia (Twerenbold 2003). Plasma osmolality is maintained within the normal range (± 3 mmol/L) when percent bodyweight losses are between 2-4% (Hew Butler, Stuempfle 2003, Nelson 1989, Refsum 1973, Cohen 1978, Noakes 1976, Kavanagh 1977, Maron 1975). Athletes who lose more than 4% of bodyweight demonstrate derangements in [Na+] beyond the normal range (Gastmann 1998, Rocker 1989, Beckner 1954, Riley 1975, Astrand 1964). Thus, the cumulative data in Table 1 suggests that the body defends plasma osmolality over bodyweight during prolonged endurance activity lasting between 2.7 to 38.2 hours, and can maintain intracellular volume best between 2-4% bodyweight losses. Acute bodyweight changes below 2% or above 4% cannot be compensated for internally and may lead to dysregulation and clinical symptomatology.
Concordant with these field data, three laboratory studies replacing fluids at 100% of bodyweight losses confirm a reduction in serum sodium concentrations during prolonged endurance exercise lasting between 2 and 6 hours (Barr 1991, McConnell 1997, Vrijens 1999). Although these decreases in serum sodium concentrations were within the normal physiological range, fluid replacement at or above 100% does not appear to offer any performance benefits (McConnell 1997, Maresh 2001). In fact, six cyclists recorded their fastest average time (124 ± 6 minutes) during an 80 km time trial with ad libitum fluid replacement (Dugas, in preparation). These cyclists performed significantly slower when fluid replacement was below ad libitum replacement (129 ± 7) and similarly when fluid replacement was at or above ad libitum (125 ± 5).
Advocates of replacing 100% of bodyweight losses argue that cardiovascular drift – the downward drift in central venous pressure and stroke volume with concomitant rise in heart rate – commences with a 1% decrease in fluid losses (Montain and Coyle 1992). Subsequent studies normalizing blood volume while inducing hypohydration have confirmed, however, that cardiac drift during exercise can and often does result from factors other than hypovolemia (Heaps 1994, Montain and Coyle 1992). The corresponding increase in heart rate from progressive dehydration is accompanied by a proportional decrease in stroke volume which serves to maintain cardiac output (Heaps 1994). These reductions in stroke volume are more related to elevations in core temperature, circulating catecholamines and heart rate than due to the redistribution of blood flow or circulating blood volume (Gonzalez-Alonso 2000). Only beyond ~ 3% dehydration is cardiac output significantly diminished from the reduction in circulating blood volume (Gonzalez-Alonso 1998, Gonzalo-Alonso 2000) which is beyond the threshold for which AVP and thirst are stimulated.
Equations predicting ad libitum fluid intake from multiple regression analyses identify plasma osmolality as the primary factor driving voluntary fluid intake during exercise or under stressful environmental conditions; with plasma volume and subjective symptomatology making secondary contributions (Greenleaf 1992, Engell 1985). Thus, under acute situations involving sweat sodium losses, the body will defend osmolality and intracellular volume through fluid shifts from the extracellular (primarily interstitial) fluid compartments (Maw 1998, Costill 1976, Sanders 2001) until ~4% bodyweight losses through sweating accrue; after which intracellular fluid volume is compromised and an external fluid supply is necessary.
Athletes rehydrating with either water or a sodium containing beverage restore 68% and 82% of fluid losses, respectively, after an exercise-induced dehydration of 2.3% (Nose 1988). Changes in free water clearance followed changes in plasma osmolality and only rehydration with water restored plasma osmolality to control (isotonic) levels. Conversely, extracellular fluid space was only fully restored through consumption of the sodium containing beverage. Ad libitum fluid intake did not match 100% bodyweight losses during 3 hours of rehydration with either beverage (hence the “involuntary” dehydration), exemplifying the immediate defense of the body to preserve plasma osmolality and then intravascular volume over bodyweight.
Metabolic water formation, from the combustion of fat and carbohydrate combined with the liberation of glycogen bound water, may contribute an internal water source to shifting fluid compartments during prolonged or intense physical activity (Kozlowski 1964, Pastene 1996, Rogers 1997). The activation or inactivation of sodium stores has also been hypothesized as a mechanism aiding in the defense of POsm during prolonged endurance exercise (Noakes 2005, Milledge 1982). Data from 2,135 weighed competitive athletic performances demonstrate that athletes who overhydrate, underhydrate or euhydrate during prolonged endurance races generally maintain serum sodium concentrations within the normal range (Figure 1 reprinted with permission; Noakes 2005). Despite this wide variation in weight changes, 80% of the athletes in this large and diverse pool maintained POsm and intracellular volume during prolonged continuous athletic activity.
Thus, the combination of laboratory work with field data begs a reinterpretation of the phrase “voluntary dehydration” to more accurately reflect the preservation of plasma osmolality (and intracellular volume) over bodyweight, secondary to sweat NaCl losses during endurance exercise. The body does not defend bodyweight over plasma osmolality during an acute exercise bout and the immediate replacement body fluids at 100% of bodyweight losses is not beneficial to performance. Thirst and AVP are stimulated when POsm increases 1-2%; well within the physiological range of homeostatic compensation. Performance and cardiovascular decrements are clearly documented when dehydration exceeds 2-4%, however this range is well beyond the protective capacity of the synergistic AVP and thirst mechanism whose primary physiological interaction is to preserve osmolality within a narrow physiological range. Only when plasma osmolality is stabilized will the body gradually restore plasma volume by increasing sodium consumption and fluid intake over the next 24 hours (Takamata 1994, Stricker 1988).
The primary rationale for the inclusion of sodium into rehydration beverages is to aid in the restoration of plasma volume (Shirreffs 2000) rather than prevent the development of exercise-associated hyponatremia (Consensus Statement). The presence of a sodium appetite is equivocal in humans (Verbalis 1990, Beauchamp 1990, McCance 1936), but studies on sodium palatability and preference have revealed significant associations with sodium loss and deficiency (Kochli 2005, Yeomans 2004, Wald 2003, Leshem 1997, Leshem 1998, Leshem 1999, Takamata 1994, Beauchamp 1990). A peculiar theme echoing throughout these studies is the delayed expression of the attractiveness of sodium containing items after sodium loss, particularly in studies involving sweat sodium losses after exercise (Leshem 1999, Takamata 1994).
The study by Takamata et al confirms that the body protects osmolality over extracellular volume following seven hours of intermittent exercise at 35ºC. Elevated serum sodium concentrations following exercise increased the palatability for water and decreased the pleasantness of concentrated sodium beverages during the first hour of rehydration. Plasma osmolality returned to baseline levels 3 hours after rehydration had begun, at which time the palatability rating for sodium containing beverages significantly increased over baseline levels. The palatability of highly concentrated sodium beverages continued to increase at 6, 17 and 23 hours following the cessation of exercise which correlated with significant increases in aldosterone secretion, but not AVP or POsm. The authors concluded that “increased H2O palatability is only associated with osmotically induced thirst and thus contributes to body fluid osmoregulation. In contrast, extracellular (ECF) thirst accompanied by an increased Na+ preference appears after a delay of many hours after osmoregulatory responses occur and may contribute to ECF volume regulation”. Thus, the palatability for sodium is delayed following exercise until plasma osmolality is restored to normal levels from the ingestion of and preference for plain water. Only after plasma osmolality returns to baseline levels - through the ingestion of plain water to dilute elevated serum sodium levels - will the body seek sodium (by a palatability increase) to return plasma volume to baseline levels in the succeeding hours following the exercise bout. This is presumably facilitated by the natriorexogenic and dipsogenic actions of aldosterone and angiotensin II, which are known to stimulate these actions (Fitzsimmons 1998). This phenomenon has also been indirectly supported in exercising students, with low sweat rates, whom expressed a decreased avidity for sodium when given high doses of salt before exercise (Wald 2003) and has also been demonstrated in rats made hypovolemic by polyethylene glycol injection (Stricker 1988).
Since sodium is clearly necessary to restore and maintain plasma volume after sweat sodium losses from exercise, the beneficial effect of sodium ingestion during exercise is a topic of heavy debate. Studies have shown that consumption of hypotonic sodium containing beverages do not prevent the development of hyponatremia in athletes replacing 100% fluid loses or less (Barr 1991, Vrijens 1999, Meyers 1995) or in polydipsic psychiatric patients (Goldman 1994, Reeves 2004) because most of the sodium ingested is rapidly lost through the urine (Vrijens 1999, Vieweg 1985). In conditions of fluid overload, however, a mild blunting of serum sodium decline occurs with consumption of sodium containing beverages. This blunting does not eliminate the development of hyponatremia in athletes who continue to overdrink, however (Twerenbold 2003, Baker 2005).
There are no studies documenting a performance, physiological or clinical benefit of sodium supplementation during exercise (Speedy 2002, Hew-Butler 2006). On the contrary, there are several studies linking sodium ingestion with negative physiological and performance effects (Pitts 1944, Ladell 1955, Konikoff 1986, Hargreaves 1989, Robertson 2004). Konikoff et al supplemented normal dietary intakes with 10.2 g (173 mmol) of sodium three days prior to a 2 hour exercise bout (60% VO2Max on a cycle ergometer) with fluid intake matching sweat rate. These authors concluded that the “salt loading” did not have a beneficial effect on temperature and fluid balance and in fact, elicited undesirable effects including significant increases in bodyweight, heart rate and rectal temperature (Konikoff 1986). The extra sodium was rapidly excreted in the urine, suggesting that the body had no need for the extra salt. Similarly, Pitts and Consolazio administered 9 g of NaCl to 11 subjects just prior to a ten mile march. Water was replaced at equal volumes to sweat loss every 2 miles. The supplemental sodium ingestion also leads to elevated heart rates and rectal temperatures along with complaints of “gastrointestinal uneasiness” (Pitts 1944). Performance in a VO2Max test was impaired after rapid infusion of 30/kg of normal saline 30 minutes prior to the exercise bout (Robertson 2004). There was a consistent increase in exercise ventilation after every exercise level following rapid saline infusion as well as significant reductions in forced vital capacity and forced expiratory volume. All subjects felt more fatigued at maximal effort and reported sensations of “increased leg tightness” which the authors’ hypothesize were due to increased intramuscular tissue pressure and resultant edema from the rapid saline infusion.
Thus, sodium supplementation has no documented advantages when consumed during exercise, and in fact may have detrimental cardiorespiratory and thermoregulatory effects during exercise and on performance. Since sodium is necessary for plasma volume restoration in the 24 hours following exercise - after plasma osmolality has acutely normalized – sodium containing food and beverages should be freely available and ingested according to palatability and tolerability post exercise.
Marathon running is a serious endeavor that requires a personal commitment to sustained physical training. The hazards of marathon running have been highlighted by the recent deaths of four novice female runners from exercise-associated hyponatremia. Understanding individual fluid requirements before tackling a 42.2 km race is pivotal towards ensuring the successful completion of such an event.
Although our society revolves around rules and algorithms to guide us through different situations, individuals should not be confined to these rules in a dynamic setting. There are no shortcuts toward great achievement, and marathon running is no exception. Clinicians and scientists must resist handing out “blanket advice” to individuals seeking unrealistic answers but rather encourage athletes to explore, understand and be flexible toward their own needs. By providing guidelines and advice on how to appropriately understand individual fluid replacement needs, we can eliminate future fluid balance problems by avoiding the temptation to globalize one rule for every situation.
Table 1. Field Study Data: Serum Sodium and Bodyweight Change
| Study | Race distance (km) |
[Na+] pre-race (mmol/L) |
[Na+] post-race (mmol/L) |
[Na+] change (pre-post race) |
Bodyweight loss(%) |
Finish time (hours) |
| Twerenbold2003(N=13) | ~41 run | 137 ± 1 | 133 ± 2 | -4 | 2 ± 1weight gain | 4 |
| Glace 2002(N=13) | 160 run | 144 | 140 | -4 | 1 | 26 |
| Gerth 2002(N=51) | 100 run | 137 ± 5 | 131 ± 2 | -6 | 1 | 14 |
| Hew-Butler (unpublished) (N=33) |
109 cycle | 139 ± 3 | 138 ± 3 | -1 | 2 ± 1 | 5 ± 1 |
| Stuempfle 2003(N=20) | 161 snow race | 141 ± 1 | 138 ± 2 | -3 | 2 | 38 ± 7 |
| Nelson 1989(N=45) | 42 run | 139 ± 0 | 142 ± 0 | 3 | 3 | 4 |
| Refsum 1973(N=41) | 90 ski | 142 | 141 | -1 | 3 | 8 |
| Cohen 1978 | 42 run | 139 ± 2 | 142 ± 2 | 3 | 3 | 3 |
| Noakes 1976(N=13) | 160 run | 144 ± 1 | 140 ± 3 | -4 | 3 | 16 ± 6 |
| Kavanagh 1977(N=9) | 42 run | 146 ± 2 | 148 ± 2 | 2 | 3 | 4 ± 1 |
| Hew-Butler (unpublished)(N=82) |
56 run | 139 ± 3 | 138 ± 3 | -1 | 4 ± 1 | 6 ± 1 |
| Hew-Butler 2006(N=413) | 226 triathlon | 141 ± 2 | 141 ± 3 | 0 | 4 ±2 | 12± 2 |
| Maron1 975(N=6) | 42 run | 141 ± 1 | 141 ± 1 | 0 | 4 ±1 | 3 ±0 |
| Gastmann 1998(N=9) | 453 triathlon | 138 ± 4 | 133 ± 4 | -5 | 5 | 26 |
| Rocker 1989 | 42 run | 144 | 149 | 5 | 5 | 3 |
| Beckner 1954 | 42 run | 141 | 156 | 7 | 5 | No time |
| Riley 1975 | 32 - 42run | 143 ± 1 | 148 ± 1 | 5 | 5 | 4 |
| Astrand 1964(N=6) | 85 ski | 144 | 152 | 8 | 6 | 9 |
*UPDATE: The 1996 Exercise and Fluid Replacement position statement of the American College of Sports Medicine (ACSM) is cited herein as the "current guidelines" of the ACSM. Note that in 2007, after the release of this IMMDA statement, the ACSM issued a substantially revised Exercise and Fluid Replacement position that moved away from blanket fluid intake guidelines to a stance suggesting a more individual, customized approach.