Archived Comments for:
The prevalence of exercise-associated hyponatremia in 24-hour ultra-mountain bikers, 24-hour ultra-runners and multi-stage ultra-mountain bikers in the Czech Republic
Re: The prevalence of exercise-associated hyponatremia in 24-hour ultra-mountain bikers, 24-hour ultra-runners and multi-stage ultra-mountain bikers in the Czech Republic.
Ivan Chernev, Appalachian Regional Healthcare
26 November 2014
Dear Editor,
We read with great interest an article by Chlíbková et al named “The prevalence of exercise-associated hyponatremia in 24-hour ultra-mountain bikers, 24-hour ultra-runners and multi-stage ultra-mountain bikers in the Czech Republic” recently published in the Journal of the International Society of Sports Nutrition [1]. This article investigates a series of ultra-endurance races held in the Czech Republic. Although the authors collected substantial amount of valuable data from these races, we have significant concerns regarding the statistical analysis and conclusions of that study.
First, we believe that these studies were not cross-sectional as stated by the authors but rather longitudinal. Measurements were taken at the beginning and the end of the race with the purpose of identifying new cases of EAH (incidence of EAH), which technically makes these studies longitudinal. In contrast, prevalence includes the new and existing cases combined for a period of time or a particular section of time. Having said that, we will use the term incidence instead of prevalence when referring to the new cases of EAH in this study.
Second, the authors did not provide information regarding the appropriateness of the sample sizes. Statistical power analysis (not reported by the authors) is important to determine if the sample sizes are appropriate and sufficient for inferential statistics. While most literature on determining sample size and power indicates that low power is associated with higher chances of Type II error (failure to find significance when one exists), insufficient statistical power may be also associated with higher chances of Type I error (finding statistically significant effects when such do not exist) [2]. The sample sizes reported in the article appear relatively small and without the relevant statistical power information, those may be too small and inadequate for any inferential testing. For the calculation of the overall incidence of EAH, the authors sampled 58 out of 461 athletes. One formula for calculating the appropriate sample size for an accurate proportion or percentage estimation, N=P(100%-P)/(SE)2, approximates that for a proportion of the size reported in this study (5%), with a confidence interval of ±5%, the study requires a sample size of 73 participants [4]. Furthermore, breaking down the sample into smaller samples (R1, R2, R3, and R4) in order to calculate incidence of each separate group is also inadequate and statistically not appropriate. Additionally, the authors hypothesized that “the prevalence of EAH would be higher in 24-hour races compared to a MTB stage race”. Thirty-nine 24-hour race participants were compared to 14 athletes in the multistage race. Comparison of two largely different sample sizes is statistically inappropriate and should be a subject to an appropriate adjustment [3]. Similarly, they hypothesized that “the incidence of EAH would be higher in runners compared to cyclists” and compared 27 cyclists vs. 12 runners. Again, two different groups with significantly different sizes which may also be too small for an appropriate estimation.
Third, the authors hypothesized that “body mass loss in all races would have no influence on race performance”. The authors concluded that change in body mass was negatively related to the race performance in R2. However, no performance data in R3 and R4 was presented at all. No correlation analysis of body mass and race performance was presented for any of the races. In addition, the correlational analysis presented in the article for fluid intake and race performance for R1 and R2 utilized samples sizes, which are too small and statistically inappropriate. For correlation studies with a statistically significant correlation of approximately 0.5 as reported in the current study, a minimum of 29 participants for achieving statistical power of 0.8 and a sample size of 37 or more participants to achieve statistical power of 0.9 are recommended [5]. The reported results could be linked to low statistical power and higher chances of Type I error (finding statistically significant effects when such do not exist) [2].
Fourth, the conclusions in the abstract are not supported by the evidence in that study. The authors concluded that “lower plasma [Na+] and development of EAH may be attributed to overdrinking, a pituitary secretion of vasopressin, an impaired mobilization of osmotically inactive sodium stores, and/or an inappropriate inactivity of osmotically active sodium”. However, two out of the three athletes with EAH reported fluid intake in the recommended by IMMDA range of 0.4-0.8 l/h and in all athletes the post-race body mass was lower than the pre-race body mass. In addition, the blood and urine parameters were inconclusive. Similarly, they attributed the development of the EAH to the impaired mobilization of osmotically inactive sodium stores and/or an inappropriate inactivation of osmotically active sodium. Subsequently, in the discussion part they stated that this cannot be determined from the present study.
Fifth, it is not clear from the article if any of the EAH athletes had altered mental status. In the results section the authors reported that “the ultra-runner (EAH-A-R2) reported in a 24-hour running race R3 reported symptoms like antidiuresis, headache, flushing, irritability, dizziness, myalgia, disorientation, lethargy, swelling and mental status change”. Subsequently, in the discussion section they stated that “post-race, all finishers, both hyponatremic and normonatremic, presented without symptoms of altered mental status”. Based on serum sodium classification by Noakes et al, all athletes in this study had mild (biochemical) hyponatremia. It is not a surprise that none of the patients required any medical treatment. Usually, these cases are treated with observation and fluid restriction.
Finally, inappropriate use of reference was detected. In the background section the authors stated that 50% of the participants in an Alaskan cold weather race presented symptoms of EAH upon finishing the race stating an article by Stuempfle at al [6]. This study did not investigate for symptoms of EAH. Furthermore, they did not detect any cases of EAH despite significant decrease of serum sodium concentration.
In conclusion, we believe that the authors should revise their data and conclusions and present it in a more accurate manner. Due to the relatively small overall sample size and discrepancies in group sizes, a descriptive analysis of the data may be more appropriate. Until then, the interpretation and conclusions of this study should be approached very cautiously due to the above mentioned reasons.
References
[1] Chlíbková D, Knechtle B, Rosemann T, Zákovská A, Tomášková I. The prevalence of exercise-associated hyponatremia in 24-hour ultra-mountain bikers, 24-hour ultra-runners and multi-stage ultra-mountain bikers in the Czech Republic. J Int Soc Sports Nutr 2014; 11(1):3
[2] Christley RM: Power and error: increased risk of false positive results in underpowered studies. Open Epidemiol J 2010; 3:16-19.
[3] Campbell MJ, Julious SA, Altman DG: Estimating sample sizes for binary, ordered categorical, and continuous outcomes in two group comparisons. BMJ 1995; 311(7013):1145-1148.
[4] Fox N, Hunn A, Mathers N. Sampling and sample size calculation. The NIHR RDS for the East Midlands/Yorkshire & the Humber 2007 Available at: http://www.webpages.uidaho.edu/ed571/571-Modules/M3/NIHS-Sampling_Sample_Size_calculation.pdf
[5] Machin D, Campbell MJ, Tan SB, Tan SH: The Correlation Coefficient. In Sample size tables for clinical studies. 3rd edition. Edited by Machin D, Campbell MJ, Tan SB, Tan SH. New Jersey: John Wiley & Sons; 2009:153-157.
[6] Stuempfle KJ, Lehmann DR, Case HS, Hughes SL, Evans D. Change in serum sodium concentration during a cold weather ultradistance race. Clin J Sport Med 2003; 13(3):171-175.
Ivan Chernev, M.D.
Clinical Assistant Professor of Physical Medicine and Rehabilitation
West Virginia School of Osteopathic Medicine, Lewisburg, West Virginia, USA
Attending Physician in Physical Medicine and Rehabilitation
Beckley Appalachian Regional Healthcare, Beckley, West Virginia, USA
Zornitsa Georgieva, MS
Graduate Student Researcher
College of Education and Human Services
West Virginia University, Morgantown, West Virginia, USA
Re: The prevalence of exercise-associated hyponatremia in 24-hour ultra-mountain bikers, 24-hour ultra-runners and multi-stage ultra-mountain bikers in the Czech Republic.
26 November 2014
Dear Editor,
We read with great interest an article by Chlíbková et al named “The prevalence of exercise-associated hyponatremia in 24-hour ultra-mountain bikers, 24-hour ultra-runners and multi-stage ultra-mountain bikers in the Czech Republic” recently published in the Journal of the International Society of Sports Nutrition [1]. This article investigates a series of ultra-endurance races held in the Czech Republic. Although the authors collected substantial amount of valuable data from these races, we have significant concerns regarding the statistical analysis and conclusions of that study.
First, we believe that these studies were not cross-sectional as stated by the authors but rather longitudinal. Measurements were taken at the beginning and the end of the race with the purpose of identifying new cases of EAH (incidence of EAH), which technically makes these studies longitudinal. In contrast, prevalence includes the new and existing cases combined for a period of time or a particular section of time. Having said that, we will use the term incidence instead of prevalence when referring to the new cases of EAH in this study.
Second, the authors did not provide information regarding the appropriateness of the sample sizes. Statistical power analysis (not reported by the authors) is important to determine if the sample sizes are appropriate and sufficient for inferential statistics. While most literature on determining sample size and power indicates that low power is associated with higher chances of Type II error (failure to find significance when one exists), insufficient statistical power may be also associated with higher chances of Type I error (finding statistically significant effects when such do not exist) [2]. The sample sizes reported in the article appear relatively small and without the relevant statistical power information, those may be too small and inadequate for any inferential testing. For the calculation of the overall incidence of EAH, the authors sampled 58 out of 461 athletes. One formula for calculating the appropriate sample size for an accurate proportion or percentage estimation, N=P(100%-P)/(SE)2, approximates that for a proportion of the size reported in this study (5%), with a confidence interval of ±5%, the study requires a sample size of 73 participants [4]. Furthermore, breaking down the sample into smaller samples (R1, R2, R3, and R4) in order to calculate incidence of each separate group is also inadequate and statistically not appropriate. Additionally, the authors hypothesized that “the prevalence of EAH would be higher in 24-hour races compared to a MTB stage race”. Thirty-nine 24-hour race participants were compared to 14 athletes in the multistage race. Comparison of two largely different sample sizes is statistically inappropriate and should be a subject to an appropriate adjustment [3]. Similarly, they hypothesized that “the incidence of EAH would be higher in runners compared to cyclists” and compared 27 cyclists vs. 12 runners. Again, two different groups with significantly different sizes which may also be too small for an appropriate estimation.
Third, the authors hypothesized that “body mass loss in all races would have no influence on race performance”. The authors concluded that change in body mass was negatively related to the race performance in R2. However, no performance data in R3 and R4 was presented at all. No correlation analysis of body mass and race performance was presented for any of the races. In addition, the correlational analysis presented in the article for fluid intake and race performance for R1 and R2 utilized samples sizes, which are too small and statistically inappropriate. For correlation studies with a statistically significant correlation of approximately 0.5 as reported in the current study, a minimum of 29 participants for achieving statistical power of 0.8 and a sample size of 37 or more participants to achieve statistical power of 0.9 are recommended [5]. The reported results could be linked to low statistical power and higher chances of Type I error (finding statistically significant effects when such do not exist) [2].
Fourth, the conclusions in the abstract are not supported by the evidence in that study. The authors concluded that “lower plasma [Na+] and development of EAH may be attributed to overdrinking, a pituitary secretion of vasopressin, an impaired mobilization of osmotically inactive sodium stores, and/or an inappropriate inactivity of osmotically active sodium”. However, two out of the three athletes with EAH reported fluid intake in the recommended by IMMDA range of 0.4-0.8 l/h and in all athletes the post-race body mass was lower than the pre-race body mass. In addition, the blood and urine parameters were inconclusive. Similarly, they attributed the development of the EAH to the impaired mobilization of osmotically inactive sodium stores and/or an inappropriate inactivation of osmotically active sodium. Subsequently, in the discussion part they stated that this cannot be determined from the present study.
Fifth, it is not clear from the article if any of the EAH athletes had altered mental status. In the results section the authors reported that “the ultra-runner (EAH-A-R2) reported in a 24-hour running race R3 reported symptoms like antidiuresis, headache, flushing, irritability, dizziness, myalgia, disorientation, lethargy, swelling and mental status change”. Subsequently, in the discussion section they stated that “post-race, all finishers, both hyponatremic and normonatremic, presented without symptoms of altered mental status”. Based on serum sodium classification by Noakes et al, all athletes in this study had mild (biochemical) hyponatremia. It is not a surprise that none of the patients required any medical treatment. Usually, these cases are treated with observation and fluid restriction.
Finally, inappropriate use of reference was detected. In the background section the authors stated that 50% of the participants in an Alaskan cold weather race presented symptoms of EAH upon finishing the race stating an article by Stuempfle at al [6]. This study did not investigate for symptoms of EAH. Furthermore, they did not detect any cases of EAH despite significant decrease of serum sodium concentration.
In conclusion, we believe that the authors should revise their data and conclusions and present it in a more accurate manner. Due to the relatively small overall sample size and discrepancies in group sizes, a descriptive analysis of the data may be more appropriate. Until then, the interpretation and conclusions of this study should be approached very cautiously due to the above mentioned reasons.
References
[1] Chlíbková D, Knechtle B, Rosemann T, Zákovská A, Tomášková I. The prevalence of exercise-associated hyponatremia in 24-hour ultra-mountain bikers, 24-hour ultra-runners and multi-stage ultra-mountain bikers in the Czech Republic. J Int Soc Sports Nutr 2014; 11(1):3
[2] Christley RM: Power and error: increased risk of false positive results in underpowered studies. Open Epidemiol J 2010; 3:16-19.
[3] Campbell MJ, Julious SA, Altman DG: Estimating sample sizes for binary, ordered categorical, and continuous outcomes in two group comparisons. BMJ 1995; 311(7013):1145-1148.
[4] Fox N, Hunn A, Mathers N. Sampling and sample size calculation. The NIHR RDS for the East Midlands/Yorkshire & the Humber 2007 Available at: http://www.webpages.uidaho.edu/ed571/571-Modules/M3/NIHS-Sampling_Sample_Size_calculation.pdf
[5] Machin D, Campbell MJ, Tan SB, Tan SH: The Correlation Coefficient. In Sample size tables for clinical studies. 3rd edition. Edited by Machin D, Campbell MJ, Tan SB, Tan SH. New Jersey: John Wiley & Sons; 2009:153-157.
[6] Stuempfle KJ, Lehmann DR, Case HS, Hughes SL, Evans D. Change in serum sodium concentration during a cold weather ultradistance race. Clin J Sport Med 2003; 13(3):171-175.
Ivan Chernev, M.D.
Clinical Assistant Professor of Physical Medicine and Rehabilitation
West Virginia School of Osteopathic Medicine, Lewisburg, West Virginia, USA
Attending Physician in Physical Medicine and Rehabilitation
Beckley Appalachian Regional Healthcare, Beckley, West Virginia, USA
Zornitsa Georgieva, MS
Graduate Student Researcher
College of Education and Human Services
West Virginia University, Morgantown, West Virginia, USA
Competing interests
none