Constant performance in balance and proprioception tests across the menstrual cycle – a pilot study in well trained female ice hockey players on hormonal contraception

Abstract Introduction It has yet to be explained why female athletes appear to suffer more often from non‐contact ligament injuries during the first half of the menstrual cycle. Fluctuations in balance, caused by impaired proprioception due to increased fluid retention, may be relevant factors contributing to this anomaly in distribution. The aim of this study was therefore to uncover relationships between dynamic stability, proprioception and fluid retention in association with the menstrual cycle as a possible explanation for the observed changes in injury rates. Methods Nine healthy, female, well trained ice hockey players on hormonal contraception with regular cycles were tested at least twelve times during one menstrual cycle. Bioimpedance analysis was applied to investigate body composition and fluid distribution. A joint position sense test was performed to characterize changes in proprioception, while unexpected perturbations on a balance board were used to obtain measures for dynamic postural control. Results No distinct changes in proprioception, dynamic stability and fluid retention were identified across the menstrual cycle in this population. Weak correlations were found between parameters for dynamic stability and proprioception. Conclusion Variations in proprioception, dynamic stability and fluid retention seem unlikely to be major contributors to injury risk peaks, at least in this population of healthy trained athletes using hormonal contraception.

association between estradiol or progesterone levels, self-reported fluid retention scores also peaked at the onset of menses. 35 The aim of this study was therefore to identify possible relationships between postural control, proprioception and fluid retention and to observe variations along the menstrual cycle in female athletes on hormonal contraception. If these relationships can be identified, they might provide a better explanation for the observed changes in injury rates along the menstrual cycle. In turn, this might help female athletes and their coaches to address these issues by adjusting their training accordingly.
As subtle changes in performance are both more relevant to as well as easier to identify in elite athletes than in the normal population, the study was conducted with a homogenous group of well-trained female ice hockey players on hormonal contraception. We hypothesized, that total body water would increase around menses, being accompanied by a reduction in joint position sense and measures of dynamic postural control.

| METHODS
The study was performed with nine healthy female well-trained (national level) ice hockey players (age: 22 ± 7 years; height: 168 ± 5 cm; weight: 61 ± 6 kg). They were tested during the general preparation period between May and July, while training 10 ± 2 hours per week. The athletes continued with their normal training regime during the study. All players were using hormonal contraception containing a combination of two active substances, estrogen and progesterone. All contraceptives contained Ethinylestradiol as first active substance. When orally taken the concentration of Ethinylestradiol was 0.02-0.03 mg per pill (eight players) and 2.7 mg per vaginal ring (one player). The second active substance was Levonorgestrel (0.1-0.15 mg per pill; 5 players), Chlormadinone (2 mg per pill, two players), Dienogest (2 mg per pill, one player) or Etonogestrel (11.7 mg per vaginal ring, one player).
Local institutional review board (Humboldt University Berlin, Faculty of Humanities and Social Sciences) approval was obtained.
Written, informed consent was given, and the study was performed in compliance with the Declaration of Helsinki.
After two pre sessions, which were used to familiarize the subjects with the dynamic stability test to minimize training effects, the subjects were tested twelve to fourteen times during one full menstrual cycle for bioimpedance measures, dynamic stability and sense of joint position. The participants entered the study at a random point in their menstrual cycle, to exclude sequence effects. The first day of the last menses, the next menses and if necessary the second next menses were recorded. The test days were retrospectively allocated to a day in the menstrual cycle, ranging from day 1 as the onset of menses to day 28 as the day before onset of menses.

| Bioimpedance measurement
For bioimpedance measurements the subjects stood bare feet and lightly dressed on the scale (InBody 720, Biospace Co., Korea). The subjects adopted an upright posture, with each foot being placed on two foot electrodes (heel and forefoot), each hand being in contact with two hand electrodes (fingers and thumb) and the arms held in an~15°angle to the upper body.

| Proprioception
In order to obtain a quantifiable measure for proprioception, a joint position sense (JPS) test was performed, which records the awareness of the joints static location in space. 37 A JPS test for the upper limb was developed, allowing marker-less assessment of three angles by image recorded angulation. The subjects were asked to adopt the fol-

| Dynamic postural control
In order to quantify dynamic postural control, the balance platform Posturomed (Haider Bioswing GmBH, Germany) was used. The Posturomed is widely applied as both training device 38 and diagnostic tool. 39 A good inter-and intra-day reliability has been reported for the dynamic balance test on the Posturomed with intra-class correlation coefficients ranging between 0.713 and 0.970. 40 This vertically suspended, unstable balance platform, which is mounted to eight springs and has been described in detail elsewhere, 41

| Statistical analysis
A linear mixed-effects model (LMM) was applied to detect statistically significant changes in the investigated parameters over the course of the menstrual cycle using Matlab (R2016a, 64 Bit, The Mathworks, Natick, USA). The LMM considers fixed effects (β k ) describing population parameters, and random effects (b ik ) which are associated with intrasubject differences over time. 42 A biquadratic LMM which also includes a constant, linear, squared and cubic term was created using the function fitlmematrix. Taking into account the small sample size the function REML (restricted maximum likelihood method) was used to estimate the variance components. The model coefficients were tested combined using the hypothesis test coefTest with H 0 :

| Body composition
The

| Proprioception
The deviation from the anticipated target joint position was largest for the angle between upper arm and longitudinal body axis (β) with a mean deviation of 13.0 ± 6.1°, followed by the angle between forearm and upper arm (α) with a mean deviation of 4.9 ± 1.9°, and the angle between longitudinal body axis and gravitational axis (γ) with a mean deviation of 1.8 ± 0.3°. None of the angles showed any significant variation during the menstrual cycle (α: P=0.81; β: P=0.52; γ: P=0.88; Figure 3).

| Dynamic postural control
Since there were no significant differences between the dominant and the non-dominant leg for all measured parameters, the mean of both  Figure 4).

| Relationships between postural control, proprioception and body water
No correlations were detected between body water and the parame-

| DISCUSSION
In contrast to our hypothesis, no significant variations in body water, proprioception or dynamic stability were detected along the   [47][48][49] Interestingly, this appears to change when women suffer from premenstrual symptoms. Postural sway is greater and kinaesthesia reduced in women with premenstrual symptoms compared to women without premenstrual symptoms. 49 Postural sway is also greater in the luteal compared to the follicular and ovulatory phase in women suffering from premenstrual symptoms. 50 Since aerobic exercise has been reported to reduce premenstrual symptoms, including fluid retention, 51   One limitation of the study is that it was not possible to differentiate between oral contraceptive users and non-users, as we were not able to recruit enough non-contraceptive users. This also means that the results of this study are predominantly applicable to athletes using a hormonal method of contraception. Also, hormone levels were not determined, and thus no correlations between hormone levels and the investigated parameters were analyzed. However, previous studies already failed to show these correlations as described above, so we refrained from investigating them. Furthermore, changes in the investigated variables still might have been occurred, being too small to be detected with the small sample size and by measuring only one menstrual cycle.

| PERSPECTIVE
Our ice hockey players constant performances in the executed tests suggests that most healthy young women on hormonal contraception are unlikely to be affected by major fluctuations in proprioception, dynamic stability and fluid retention along the menstrual cycle. It appears that the effect of hormonal changes along the menstrual cycle, often given as an explanation why a study population is restricted to men, might have been overestimated in general.
In conclusion, regarding the parameters investigated in our study, one should be able to include healthy young women on hormonal contraception in a study population, without having to fear hormone associated fluctuations. We furthermore conclude that fluctuations in balance or proprioception within the menstrual cycle are either non-existent or so small, that they should not have to be taken into account by athletes and their coaches when planning training or competition.
However, fluctuations might still occur in individual cases or in association with other factors, such as e.g. the premenstrual symptom.
The challenge remains to identify the specific ones at risk to suffer from menstrual cycle associated performance impairment or injury.
Simple cause and effect relations failed so far to provide a sufficient explanation for menstrual cycle phase associated injuries. Thus, future studies might have to apply complex analytic methods including multiple independent variables to uncover the factors responsible for fluctuations in injury rate and performance.

ACKNOWLEDGEMENTS
We would like to thank Lisa Ebach and Arno Schroll for their help with the data analysis and the Physiotherapy Department of the Olympiastützpunkt Berlin for their support.

DISCLOSURE STATEMENT
In accordance with ethical obligations as researcher, the authors report no conflicts of interest that may affect the research reported in the enclosed paper. We declare that the results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation.