Plasma aldosterone concentration and cardiovascular response to low sodium intake in horses in training



    Corresponding author
    1. Department of Animal Nutrition and Management, Department of Anatomy, Physiology and Biochemistry, Swedish University of Agricultural Sciences, Uppsala, Sweden.
    Search for more papers by this author

    1. Department of Animal Nutrition and Management, Department of Anatomy, Physiology and Biochemistry, Swedish University of Agricultural Sciences, Uppsala, Sweden.
    Search for more papers by this author
  • C. KVART

    1. Department of Animal Nutrition and Management, Department of Anatomy, Physiology and Biochemistry, Swedish University of Agricultural Sciences, Uppsala, Sweden.
    Search for more papers by this author



Reason for performing the study: Horses in training lose large amounts of sodium but little is known about the cardiovascular response to low sodium intake.

Objectives: To investigate the effect of low sodium intake on plasma aldosterone (pAldo) concentrations and the cardiovascular system of athletic horses, and to identify markers of low sodium intake.

Methods: Seven Standardbred geldings in training (trained twice a week) were randomly offered a standardised diet supplemented (NaS, 58 mg Na/kg bwt) and not supplemented (NaN, 3 mg Na/kg bwt) with NaCl for 5 weeks in a changeover design. Blood samples were taken once a week and in Week 5, before and following an exercise test until 22.30 h and analysed for blood sodium (bNa), total plasma protein (TPP), pAldo, troponin I and packed cell volume (PCV). Blood pressure (BP) was measured and pulse wave recorded at rest with high definition oscillometric-technique (HDO). ECG and echocardiography were recorded. Water intake was measured before and on the day of exercise and voluntary saline intake was measured for 2 days after each period. Faecal samples were taken weekly and analysed for sodium and potassium content.

Results: The pAldo and the PCV was higher in NaN compared to NaS. There were no differences between diets in BP, ECG, plasma troponin I and echocardiogram but HDO pulse amplitude tended to be smaller on diet NaN. Water intake was lower on diet NaN and saline intake higher. The response to exercise in bNa, pAldo, PCV and TPP was different on the 2 diets. Faecal potassium/sodium ratio was higher on NaN than on NaS.

Conclusion: This study shows that 5 weeks of low sodium intake increased plasma aldosterone concentration and PCV but no alterations in heart function was observed. Faecal potassium/sodium ratio could be used to assess sodium status in horses.


Nonsupplemented diets to horses are marginal in sodium content (Anon 2007) and, since equine sweat is rich in sodium (McCutcheon et al. 1995), depletion might occur in unsupplemented horses in training. A marginal sodium diet will reduce the extracellular volume (Lindner et al. 1983) and possibly also increase the packed cell volume (PCV) and the plasma aldosterone concentration in horses in training (Jansson and Dahlborn 1999). The scope of the present study was to investigate potential effects of increased aldosterone concentrations due to marginal sodium intake in horses in training.

Aldosterone acts on the equine kidneys as well as on the gut (Clarke et al. 1992; Jansson et al. 2002) by increasing sodium reabsorption at the expense of potassium excretion. It has earlier been suggested that urine analysis of the creatinine/sodium ratio can be used to assess low sodium intakes in horses (Meyer and Stadermann 1990). Based on observations of changes in the sodium and potassium ratios of faecal samples collected during 3–6 h periods following exercise and aldosterone treatment it has also been suggested that a faecal sample could be used to assess sodium status of horses (Jansson 1999), but this has not yet been confirmed. Faecal samples are easier to collect than urine samples and would therefore be a simple, quick collection method suitable for trainers and veterinarians.

It has also been shown that aldosterone causes enhanced myocyte apoptosis in laboratory animals (Mano et al. 2004) and that an aldosterone antagonist as spironolactone has a protective effect (Burniston et al. 2005). Studies in rats have also shown that increased plasma aldosterone levels induced by salt restriction in volume-overloaded condition, resulted in increased myocardial fibrosis (Mori et al. 2009). In addition, it is known that the survival of human heart failure patients is increased if treated with aldosterone-blockade (Pitt et al. 1999). The effect of long-term exposure to high aldosterone concentrations caused by low salt intake on the cardiovascular function of horses is not known. However, it has recently been shown that 15–20% of a group of Standardbred and Thoroughbred horses had mildly increased post race cardiac troponin I levels, a biomarker regarded to be specific to detect cardiac injury (Nostell and Häggström 2008). Inversion of the T wave in lead V1, V2 and V3 has also been observed in Standardbred horses with impaired exercise tolerance (Persson and Forsberg 1986; Kvart 1989). The reason for this troponin I elevation and T wave inversion has never been clarified and investigation of a possible association with salt deficiency seemed justified. The aim of the present study was to investigate effects of5 weeks of low sodium intake on the plasma aldosterone concentration and cardiovascular system of horses in training and to evaluate the use of a faecal sample as a marker of low sodium intake. In addition, voluntary hydration with water and saline was investigated. The hypothesis was that a low sodium intake will result in increased aldosterone levels and an altered cardiovascular response, and that a faecal sample could be used to assess the sodium status.

Materials and methods

Animals, experimental design and diet

Seven Standardbred geldings (470–615 kg, 6–9 years) in training were used. All horses had been kept under the same conditions and same diet (supplemented daily with 0.07 g of NaCl/kg bwt) for at least 7 months prior to the study. They were housed in outdoor boxes on wooden litter and were kept in a sand/clay (0.3 g sodium/kg dry matter) paddock for 9 h/day, except for days with exercise. The horses were randomly offered a standardised diet supplemented (NaS) and not supplemented (NaN) with NaCl for 36 days in a changeover design (4 horses offered NaS in Period 1 and the other 3 in Period 2). With the exception of the sodium content in diet NaN, the diet met the nutrient requirements according to NRC (Anon 2007) and consisted of 1.48 kg grass haylage (dry matter 72%, 12.3% crude protein), 0.65 kg oats, 0.24 kg lucerne pellets, 0.11 kg wheat bran, 18.5 g sugar, 12 g chalk and vitamin E and selenium supplementation per 100 kg horse and day. Diet NaS was also supplemented with 13.9 g NaCl/100 kg horse. In total, diet NaS provided 58 mg Na/kg bwt and diet NaN 3 mg Na/kg bwt. The potassium content of the diet was 17.9 g/kg. The horses were fed at 07.00 (oats and lucerne), 08.00 (25% of the forage allowance), 17.00 (75% of the forage allowance, oats, wheat bran, salt, sugar, selenium and vitamin E) and 22.00 h (oats and lucerne). On the day of the exercise test in Week 5, the horses were fed haylage at 16.30 and concentrate at 19.15 h. Water was offered from buckets and an automatic waterer in the box and from a trough in the paddock. The day before and at the day of the exercise test water was offered solely from graded buckets in each box.

After each experimental period a 7 day recovery period was performed. On the first 2 days of this period NaN horses continued on this diet and NaS horses were still offered NaCl supplementation with the feed (80 g/day corresponding to a similar Na intake (64 mg/kg bwt) as on NaS) and all horses were offered saline (9 g NaCl/l). The saline was offered from graded buckets, 18 l/24 h to horses on NaN and 9 l/24 h to horses previously on diet NaS, and voluntary intake was measured. On the remaining 5 days all horses were offered 215 g NaCl/day with the feed which, together with the saline intake, was an amount estimated to cover for the deficit in horses previously subjected to diet NaN (maintenance plus observed sweat losses in Period 1). Water was offered ad libitum from graded buckets.

The horses were trained twice a week, once on a terrain track (8 km trot at heart rate <200 beats/min) and once on an uphill oval field track, including exercise at heart rates >200 beats/min. The latter training session was used as a standardised exercise test and measurements and samples were collected in connection to this. The exercise consisted of 4 km warm-up in trot (7 m/s), 1100 m and 700 m in faster trot (10 and 11 m/s, respectively) and then 500 m slow trot (6 m/s) back to the stable. All horses were exercised individually at the same individual time of the day (09.00–15.00 h) throughout the experiment. At 30 min post exercise the horses were returned to their boxes where water was available.

The study was performed in September to November in 2008 outside Uppsala, Sweden and the ambient temperature at exercise days varied from 0–18°C. The study was approved by the Uppsala Ethical Committee.

Sampling and analyses

Daily water intake was measured the day before and at the day of the exercise test and on days when saline was offered. Skin turgor was assessed once every week (before exercise) by pinching at the thoracic inlet for approximately 1 s and counting the time required for normal position of the skin to be achieved. A value <1 s was set to 0.5. Bodyweight was measured before and after each training session. It was noted whether the horse had defaecated or not during exercise, in order to estimate the fluid loss during exercise. The sodium loss was then calculated by using a rough estimation that 80% of the fluid loss during exercise was due to sweat losses (Hodgson et al. 1993) and a sweat sodium content of 3.2 g/l (McCutcheon et al. 1995).

In Week 5, ECG was recorded 3 h after exercise to measure heart rate, negative amplitude of the T-wave and Q-T interval from onset of the Q-wave to end of the T-wave. Cardiac rhythm was evaluated from ECG recorded by lead V1, V2 and V3 with a telemetric system1 also capable of recording on memory card for later analysis on a computer with Televet analysis software. Recordings were performed before, during and until 30 min after end of exercise once weekly. Blood pressure was measured at rest with high definition oscillometric (HDO) technique2 with a cuff placed at the root of the tail. The horses had been accustomed to this procedure weekly for 6 weeks before the study began. With this equipment a difference in pulse amplitude between occlusion and maximal level can be measure and reflects the arterial wall movement during the measurement. Heart rate and breathing frequency was also measured manually (by phonendoscope and counting breaths) 15 min post exercise in Week 5.

Echocardiogram was recorded the day after the exercise test in Week 5 on both diets with a ultrasonic machine (Vivid 3)3 with a 1.7–3 mHz probe with the following parameters measured as recommended by Patteson (1996): RVWd, RVDd, IVSd, LVIDd, LVPWd, RVWs, RVDs, IVSs, LVIDs, LVPWs, ES, FS%, Ao, LA, PA.

A blood sample was taken once every week 3 h after exercise when increases in troponin I are likely to be detected (Nostell and Häggström 2008). In Week 5, blood samples were taken before, at the end of exercise (intensive part on the track), 15 min and 3 h after exercise. Additional blood samples were taken at 19.30 and 22.30 h. Blood samples were drawn by Vacutainer technique in Weeks 1–4 and in Week 5 by a catheter inserted to a jugular vein during local anaesthesia (Carbocain 20 mg/ml)4 in the morning. Blood samples analysed for sodium, potassium, chloride, PCV, plasma troponin, total plasma protein (TPP) and plasma aldosterone concentration were collected in 10 ml Li-heparainised tubes. Analyses of blood sodium, potassium, chloride and PCV was performed within 10 min using an i-STAT1 analyser5 and EC8+ cartridges. The blood samples were then centrifuged and the plasma frozen (-20°C) until later analyses. Plasma samples intended for troponin I analysis were moved to −80°C within 48 h. Plasma concentration of troponin I were measured by enzyme immunoassay6 with a analytic sensitivity of 0.022 µg/l. The method has been evaluated for equine plasma in Standardbred horses with satisfying result (Nostell and Häggström 2008.) TPP was measured with a refractometer (Architect)7. Plasma aldosterone concentration was determined after extraction of fat and proteins (acetone and petroleum ether extraction) by use of a RIA kit (Coat-a-Count)8. The samples for the standard curve were extracted in the same way. The quality control was run using MultiCalc software version 2.09.

Feed samples were collected every week and a fresh faecal sample (approximately 200–300 g) was taken from the box once a week (at 08.00–10.00 h before exercise). Feed and faecal samples were analysed for sodium and potassium content by inductively coupled plasma optical emission spectrometry (SS-EN 14538:2006)10 after boiling samples in nitric acid (7 mol/l).

Statistical analysis

All data were subjected to analysis of variance (GLM procedure in SAS 9.1)11 using the following model; Yi j k=μ+αijkl+ (βγ)j k+ ei j k l where Yi j k is the observation, μ the mean value, αi the effect of animal, βj the effect of diet, γk the effect of sample/week, εl the effect of period (βγ)j k the effect of interaction between diet and sample/week and ei j k l the residuals; ei j k l∼IND (0, δ2). The P value for significance within and between treatments was <0.05. Post hoc analysis was made by a Tukey test (significance P<0.05). Values are presented as least square mean ± s.e.


General and weekly observations

The exercise induced weight loss was 5.4 ± 0.2 and 5.5 ± 0.2 kg per training session on NaN and NaS, respectively (not significant). The weight loss in training sessions without defaecation (33% of the observations) was 4.4 ± 0.3 kg. This fluid loss corresponded to a total sodium loss of approximately 113 g (280 g of NaCl) during each of the experimental periods.

Water intake was lower on diet NaN than on diet NaS (19.6 ± 0.6 vs. 22.5 ± 0.6 l/day). PCV was higher in NaN compared to NaS (37 ± 0.5 vs. 35 ± 0.5%) but there were no difference in TPP (66 ± 0.5 vs. 66 ± 0.5 g/l) and skin turgor (1.2 ± 0.1 vs. 1.0 ± 0.1 s). There were no differences in bodyweight between diets (550 vs. 549 kg on diets NaN and NaS, respectively, s.e. 0.7). No health problems were observed during the experiment except for on diet NaN (Day 25) when one horse had a mild post exercise rhabdomyolysis. The clinical signs (unwilling to move, stiff movement) diminished after 4–5 h and no treatment was given. The horse was moving and behaving normally the following morning.

In the recovery period, the intake of saline was higher (8.7 ± 1.1 vs. 4.0 ± 1.1 l/day) and the intake of water was lower (14.4 ± 1.1 vs. 19.3 ± 1.1 l/day) in horses previous offered diet NaN.

Effects on weekly aldosterone levels and faecal Na and K content

The 3 h post exercise plasma aldosterone concentration was elevated on NaN in Weeks 1–4 and there was a tendency (P = 0.07) in Week 5 (Fig 1). The faecal sodium content and sodium/potassium ratio was lower on diet NaN compared to diet NaS and the potassium content and the potassium/sodium ratio was higher (Fig 2). There was an effect of individual in all variables but the potassium/sodium ratio. Individual potassium/sodium ratios ranged from 0.9–5.4 in diet NaS and 4–155 in diet NaN. However, a value of 4 was only observed once in one individual on this diet and the rest of the observations were >10.

Figure 1.

The plasma aldosterone concentration (LS mean±s.e.) 3 h post exercise in 7 Standardbred horses in training on a diet not supplemented (triangles) and supplemented (diamonds) with NaCl for 5 weeks. * denote significant difference between diets.

Figure 2.

Faecal sodium (Na) and potassium (K) concentrations and their ratios (LS means±s.e.) in 7 Standardbred horses in training on a diet not supplemented (squares) and supplemented (diamonds) with NaCl for 5 weeks. * denote significant difference between diets.

Effects on ECG, blood pressure, HDO pulse amplitude, echocardiogram and troponin I

No arrhythmias were recorded at rest, during exercise and until 30 min after exercise or at 3 h post exercise. There were no differences in heart rate, Q-T interval and negative T amplitude on ECG recorded 3 h post exercise (Table 1). There was no difference in blood pressure but HDO pulse amplitude difference (arterial wall movement) tended to be larger in NaS than in NaN (Table 1). All horses were within normal limits on echocardiogram (Zucca et al. 2008) and no differences could be detected between diets. The troponin I concentration was low (<0.022 µg/l) on both diets. However, in the horse with a post exercise rhabdomyolysis on diet NaN the troponin I level was increased 3 and 10 days after the episode (0.026 and 0.16 µg/l, respectively) but it was not elevated (<0.022 µg/l) immediately after (sample collected 2–3 h after the onset of clinical signs).

Table 1. Cardiovascular findings (BP, blood pressure) in 7 Standardbred horses in training on a diet not supplemented (NaN) and supplemented (NaS) with NaCl for 5 weeks (least square means ± s.e.)
  • a 


  • b 

    mm (1 mV = 10 mm),

  • c 

    c sec,

  • d 

    d mmHg.

  • 1 

    Recorded 3 h post exercise in Weeks 1–5;

  • 2 

    2 Recorded with HDO-technique before exercise in Week 5.

  • #

    P = 0.053.

Heart rate on ECG1a41 ± 140 ± 1
Negative T wave amplitude1b1.8 ± 0.21.3 ± 0.2
Q-T interval1c0.50 ± 0.0030.50 ± 0.003
Heart rate during BP2a38 ± 139 ± 1
Systolic BP2d129 ± 4122 ± 4
Diastolic BP2d83 ± 474 ± 4
Mean arterial BP2d99 ± 492 ± 4
Difference in pulse amplitude238 ± 348 ± 3#

Effects of exercise in Week 5

On diet NaN the plasma aldosterone concentration increased and the blood sodium concentration decreased after exercise and significant differences between diets were observed at 22.30 h (Fig 3). On diet NaS, a significant increase in plasma aldosterone was only observed in the evening samples and there were no changes in the blood sodium concentration. TPP was elevated on diet NaN at 19.30 h compared to diet NaS, and on diet NaS TPP was back to pre-exercise levels within 3 h whereas the pre-exercise level was not regained before 22.30 h in NaN (Fig 3). The pre-exercise PCV was regained at 19.30 h on diet NaN but not before 22.30 h in diet NaS, when it was also higher than in diet NaN (Fig 3). There were no differences in blood potassium (data not shown) and chloride between diets except for at 3 h when the chloride concentration was higher in diet NaN (99.5 vs. 98.2 mmol/l, s.e. 0.3) and 19.30 h when it was higher in diet NaS (99.7 vs. 98.6 mmol/l, s.e. 0.3). There were no differences in heart rate and breathing frequency 15 min post exercise (heart rate: 58 vs. 56 beats/min (s.e. 1.5) on diet NaN and NaS respectively, and breathing frequency: 29 vs. 20 breaths/min (s.e. 6) on diet NaN and NaS, respectively.)

Figure 3.

The plasma aldosterone concentration, sodium (Na) concentration, packed cell volume and total plasma protein concentration (LS means ± s.e.) before and after exercise (end of exercise (EX), 15 min and 3 h post exercise) in 7 Standardbred horses in training on a diet not supplemented (triangles) and supplemented (diamonds) with NaCl for 5 weeks. All horses were fed haylage at 16.30 and concentrate at 19.15 h. Additional blood samples were taken at 19.30 and 22.30 h. Filled markers indicates significant difference from before exercise and * denote significant difference between diets.


Based on the estimated sweat losses the horses lost 23 g of sodium per week in addition to their daily maintenance requirements (Anon 2007) of 20 mg/kg bwt in the present study. To maintain sodium balance a 550 kg horse would therefore need 100 g of sodium per week. However, this might be an over-estimation since sodium losses have been estimated to be 20% lower in sodium depleted ponies due to a reduced sweat rate and slightly reduced sweat sodium concentration (Lindner et al. 1983). Nevertheless, in this study a 550 kg horse was offered 220 g of sodium per week on diet NaS (in the concentrate, with no leftovers) but only 11.5 g on diet NaN and sodium balance was therefore probably maintained on diet NaS, whereas a negative balance was the case on diet NaN. To protect the volume of the extracellular fluid, the negative balance was probably counteracted by a depletion of the sodium content of the digesta and body tissues. Thirty-five days of sodium deficiency in ponies has earlier been shown to reduce total body sodium content with 22% and the gastrointestinal sodium content by >75% (Lindner et al. 1983).

The low sodium intake on diet NaN resulted in an increase in the plasma aldosterone concentration that could be observed post exercise as early as Week 1. The elevation in Week 1 on diet NaN was marked even in horses previous subjected to diet NaS and the supplementation in the recovery week (140–354% higher than on diet NaS). However, in Week 5 the elevation was not significant until 22.30 h. Interestingly, the effect on aldosterone coincided with a numerical increase in the faecal sodium content compared to the weeks before, and may reflect an increased release and availability of sodium from body tissues at this time. However, a lack of difference in plasma aldosterone concentration also confirms the observation made by Jansson and Dahlborn (1999) that horses with varying sodium intake may not display a large difference in the plasma aldosterone concentration at daytime but horses with a low sodium intake have marked increased levels between 21.30 and 04.30 h. It is widely accepted that there is a circadian rhythm in plasma aldosterone concentration in humans and that it can be amplified by sodium restriction (Katz et al. 1975). The reason for this variation in the release of aldosterone is unclear but is interesting in the perspective that aldosterone might cause myocyte apoptosis (Mano et al. 2004). It could be speculated that a temporary high elevation is less toxic than a constant elevation. Our hypothesis, that increased aldosterone levels caused by low sodium intake could cause cardiac damage, was not supported by any general increase in the plasma troponin I concentration or changes in the echocardiogram. One horse showed however, a gradual increase in troponin I after Day 25 on diet NaN (after a short episode with clinical signs of rhabdomyolysis). Therefore the question still remains, can low salt intake trigger damage to cardiac muscle?

Although sodium and potassium plays a major role in the electrophysiology of the heart the hypothesis that low sodium intake might increase the risk for arrhythmias before, during or after exercise or cause T wave inversion in resting ECG could not be accepted by this study. A more severe sodium depletion, possibly in combination with other stress factors i.e. endotoxaemia, might be necessary to induce arrhythmias or other ECG changes.

Five weeks of low sodium intake affected fluid homeostasis and circulation in the present horses. It was detected by a decreased water intake, an increased PCV and a tendency (P = 0.053) to a reduced difference in pulse amplitude during HDO measurements. The lowered water intake probably reflects the reduced capacity to maintain body fluid volume due to a loss of the major extracellular cation. The increased PCV was highly significant (P = 0.008) and is in accordance with reports of cellular hypervolaemia during sodium deprivation (Andersson 1971; Salzman et al. 1990) and observations of other horses with low sodium intake (Jansson and Dahlborn 1999). The tendency to reduced difference in HDO pulse amplitude reflects a reduction in the arterial wall movement and possible explanations for this finding are a reduced peripheral blood flow or an altered elasticity of the arteries. However, during these circumstances a reduced peripheral blood flow seems to be the most plausible explanation. Five weeks of low sodium intake, however, caused no significant differences in blood pressure, TPP, skin turgor, and post exercise heart and breathing rate. A lack of difference in heart rate is, however, also in accordance with previous observations on severely sodium depleted ponies (Lindner et al. 1983). Skin turgor has recently also been shown to be an unreliable method to assess hydration status (Pritchard et al. 2007). Unfortunately, body condition seemed to increase in all horses during the experiment, which may have ruled out the possibilities of observing any changes in bodyweight.

In contrast to diet NaS, where no or limited changes in blood sodium and plasma aldosterone were observed following exercise, plasma aldosterone increased immediately in connection with exercise on diet NaN and the blood sodium concentration was lowered 15 min post exercise and in the evening samples. This shows that the response to exercise was different on diet NaN compared to diet NaS and indicates altered fluid and sodium shifts. In addition, on diet NaS the exercise induced increase in TPP was recovered within 3 h whereas on diet NaN it was not recovered, not even in the 22.30 h sample. This might reflect a limited/retarded capacity to restore fluid lost from the circulation. However, since there were no general difference in TPP between diets, it is likely that a fluid shift occurred later. The increase in TPP at 19.30 h on diet NaS, indicating that a temporary reduction in the plasma volume, could be due to a high rate of feed intake causing a more rapid secretion of gastrointestinal fluids, which temporarily reduces the plasma volume (Clarke et al. 1990; Jansson and Dahlborn 1999). It could also be due to a fluid shift into the gastrointestinal tract due to a high osmolality. The reason for the difference in PCV observed in the evening samples is not clear.

As expected, the voluntary intake of saline was higher following diet NaN than following diet NaS. However, there was a large individual variation in the voluntary sodium intake as has been observed earlier in horses in training (Jansson and Dahlborn 1999). Two individuals showed a comparatively strong appetite and consumed all of the saline offered, irrespective of the previous diet. One of these individuals consumed 18 l rapidly. The rest of the horses showed less strong appetite and this emphasises the need for dietary supplementation of sodium.

It has been suggested that urine analysis of the creatinine/sodium ratio can be used to assess low sodium intakes in horses (Meyer and Stadermann 1990). Based on this study and the observations made earlier by Jansson (1999), we also suggest that the faecal potassium/sodium ratio could be used to identify individuals not supplemented with adequate amounts of sodium. Individuals with a ratio >10 can be regarded as having a sodium deficient diet. Measurement of plasma aldosterone concentration could probably also be used to assess sodium status in individual horses but this needs further investigations.

In conclusion, this study shows that 5 weeks of low sodium intake increases plasma aldosterone concentration and PCV, and delays the recovery of total plasma protein concentration post exercise. No alterations in heart function could be detected except in one horse where increased troponin I levels were found, indicating that further investigations of the relationship between low sodium intake and cardiac damage would be of interest. The study also shows that the faecal potassium/sodium ratio could be used to assess sodium status in horses.


Thanks to trainer Anna Svensson and Sophie Maurer for assistance during the study and Gunilla Drugge (Department. of Anatomy, Physiology and Biochemistry, SLU) and Eva Heldesjö Blom (Akademiska Hospital, Uppsala, Sweden). Thanks also to Åke Rosengren and trainer Jim Frick. The project was supported by Stiftelsen Svensk Hästforskning.

Conflicts of interest

The authors have declared no conflicts of interest.

Manufacturers' addresses

1 Televet ECG,, Germany.

2 S+B medVET GmbH, Germany.

3 GE Healthcare, Little Chalfont, Buckinghamshire, UK.

4 Astra AB, Södertälje, Sweden.

5 Abbott Laboratories, Abbott Park, Illinois, USA.

6 Abbot Laboratories, Diagnostic Division, Abbott Park, Illinois, USA.

7 Atago, Sur-Ne, Tokyo, Japan.

8 DPC, Los Angeles, California, USA.

9 Wallac, Turku, Finland.

10 Ametec Spectro Analytical Instruments, Kleve, Germany.

11 SAS Institute Inc. Cary, North Carolina, USA.