1 Background
Over the last three decades, there has been a rise in the number of women participating in exercise, from physical activity to elite sport, attributable to the increasing development of, and investment in, women’s professional sport [
1‐
4]. Specifically, the percentage of women competing at the Olympic Games has increased from 26% in Seoul in 1988 to 45% in Rio de Janeiro in 2016 [
5]. Furthermore, Tokyo 2021 is set to be the most sex-balanced Games in history, with the same number of medals available for men and women, which is projected to see women participation in the Games rise to 49% [
5]. Performance-based research in women has not kept pace with the exponential rise in participation [
6,
7]. Indeed, it would be naive to assume that all research in men can be directly applied to women, given the anatomical, physiological and endocrinological differences between the sexes [
4,
8‐
10]. As such, sportswomen will benefit from sex-specific research and guidelines, which consider the effects of women’s physiology, such as the menstrual cycle (MC), on performance [
8,
11].
The MC is an important biological rhythm, whereby large cyclic fluctuations in endogenous sex hormones, such as oestrogen and progesterone, are observed [
12‐
14]. The fairly predictable (and measurable) fluctuations in oestrogen and progesterone across the MC create significantly different transient hormonal profiles, which are used to differentiate between MC phases [
15,
16]. As such, the MC is commonly divided into three phases, (1) the early follicular phase, characterised by low oestrogen and progesterone, (2) the ovulatory phase, characterised by high oestrogen and low progesterone, and (3) the mid-luteal phase, characterised by high oestrogen and progesterone [
17]. Although the primary function of these hormones is to support reproduction, research has highlighted that the changing concentrations of oestrogen and progesterone across the MC also exert a myriad of diverse and complex effects on multiple physiological systems, including cardiovascular, respiratory, metabolic and neuromuscular parameters [
12,
18,
19], which could have subsequent implications for exercise performance [
15,
20‐
23].
There are a range of suggested mechanisms by which the cyclical fluctuations in oestrogen and progesterone across the MC might affect performance. Specifically, oestrogen is thought to have an anabolic effect on skeletal muscle [
24,
25] and has been shown to play a role in substrate metabolism changes through increased muscle glycogen storage and increased fat utilisation [
26]. Additionally, progesterone is thought to have anti-oestrogenic effects [
21]. As such, it is plausible that changes in exercise performance might be observed due to the different hormonal profiles across the MC [
15,
20‐
23]. To date, the effects of fluctuations in oestrogen and progesterone across the MC on exercise performance are conflicting, with studies reporting improved performance outcomes during the early follicular [
27‐
29], ovulatory [
30] and mid-luteal [
31,
32] phases; whereas, others have shown no changes in exercise performance between MC phases [
33‐
39]. Therefore, it is evident that a consensus is yet to be reached regarding the effects of the MC on exercise performance. Subsequently, no evidence-based guidelines for managing exercise performance across the MC currently exist for either exercising women, nor for practitioners working with elite sportswomen.
Given the recent increase in the number of women participating in exercise and the lack of consensus regarding the effects of the MC on exercise performance, there is a growing need to determine the effects of the fluctuations in oestrogen and progesterone across the MC on exercise performance. To our knowledge, this is the first meta-analysis to critically examine existing studies investigating changes in exercise performance across the MC, in eumenorrheic women. Additionally, this review is the first of its kind to appraise the quality of previous studies using robust assurance tools. The information provided by this meta-analysis can be used to inform practical recommendations for athletes, practitioners and researchers interested in managing exercise performance across the MC.
4 Discussion
The aim of this review was to examine if MC phase affects exercise performance in eumenorrheic women. The results indicate that on average, exercise performance might be trivially reduced during the early follicular phase of the MC when compared with all other MC phases. Performance was consistent between all other MC phases. In addition to the estimated trivial average effect, results from the meta-analysis models showed relatively large between-study variance indicating that research design, participant characteristics and type of performance measured might influence any effect. Furthermore, most studies that were included in this meta-analysis were classified as “low” in quality, and as such, the confidence in the evidence reported in this meta-analysis is also low, and should be interpreted with caution. Due to the trivial effect size, the large between-study variation and the number of poor-quality studies included in this review, general guidelines on exercise performance across the MC cannot be formed; rather, it is recommended that a personalised approach should be taken based on each individual's response to exercise performance across the MC.
There are a range of suggested mechanisms by which the lower levels of oestrogen and progesterone seen in the early follicular phase of the MC might negatively affect the exercise performance. Although a detailed mechanistic review is beyond the scope of this review, the following points can be noted. First, oestrogen is known for its anabolic effects [
24,
25], as well as its role in regulating substrate metabolism through increasing glycogen uptake and sparing glycogen stores. Additionally, it has been shown to have antioxidant and membrane stabiliser properties, which might offer protection against exercise-induced muscle damage and reduce inflammatory responses [
26]. Further, oestrogen is thought to have neuroexcitatory effects, whereby it reduces inhibition and increases voluntary activation [
19]. Therefore, when oestrogen rises during the late follicular and ovulatory phases and remains elevated in the mid-luteal phase, it is plausible that this might affect muscular performance [
24,
25] or maximal and submaximal intensity exercise performance [
26]. Moreover, progesterone is thought to have anti-oestrogenic effects [
21]; therefore, it could be speculated that the beneficial performance effects of oestrogen are likely to be greater in the late follicular and ovulatory phases when oestrogen is high without the interference of progesterone, compared to the mid-luteal phase when
both oestrogen and progesterone are high. This speculation is supported by the finding presented here that the biggest difference in performance was between the early follicular and late follicular phases of the MC. However, the average effect calculated was trivial and there was considerable overlap between each of the pairwise comparisons with the early follicular phase. Whilst the current meta-analysis cannot identify the mechanisms responsible, it does indicate that, on average, exercise performance might be reduced by a trivial amount in the early follicular phase of the MC compared with all other phases. Interestingly, our sister meta-analysis, on the effects of oral contraceptives (OCs) on exercise performance, showed that, compared with eumenorrheic women, OC users have on average slightly inferior exercise performance [
119]. Oral contraceptive use results in significantly downregulated concentrations of endogenous oestrogen and progesterone when compared with the ovulatory and mid-luteal phases of the MC [
71]. Indeed, the endogenous hormonal profile of OC users is comparable to the profile seen during the early follicular phase of the MC [
71]. Both meta-analyses show slightly impaired, group-level, exercise performance when
both oestrogen and progesterone are at their lowest, therefore collectively suggesting that exercise performance might be mediated by the concentration of endogenous ovarian hormones in some exercising women.
Within the literature to date, the most common comparison used when investigating the effects of the MC on performance was between the early follicular and mid-luteal phase. This is not surprising, as the difference in the hormonal milieu is typically at its greatest between these phases (early follicular when
both oestrogen and progesterone are low, and mid-luteal when
both oestrogen and progesterone are high) [
17]. As such, if performance was altered by synergistic fluctuations in oestrogen and progesterone levels, the comparison between these two phases would maximise the chance of observing an effect. This bi-phasic comparison, however, ignores the late follicular and ovulatory phases of the MC, when oestrogen is high, and progesterone is low. The network analysis indicated that the largest difference in performance might be expected between the early follicular and the late follicular phases of the MC, when
both oestrogen and progesterone are low and when oestrogen rises without a concurrent increase in progesterone. Therefore, the effects of oestrogen, without the interference of progesterone, might be overlooked if the late follicular or ovulatory phases are not included within the phase comparisons. Future studies should, therefore, consider multiple phase comparisons so that the effects of different ratios of oestrogen and progesterone can be explored. It should be noted, however, that the inclusion of multiple phase comparisons will result in more variability, and as such, more participants will be needed to conclude any potential effects.
Although this systematic review included 78 studies and 1193 women (range
n = 5–100), there were very few studies classified as “moderate” or “high” in quality, which implies that the confidence in the evidence used in this meta-analysis should be low. Specifically, only 24% of studies were allocated a quality rating of “moderate”, and only 8% of studies were allocated a quality rating of “high”. Our quality assessment approach included consideration of the methods used to identify and verify the MC phase in the included studies, which is considered to be key to the trustworthiness of the results obtained (i.e. Q1. was the MC phase confirmed using blood samples; Q2. was the MC phase confirmed using urinary ovulation detection kits?). Across the included studies there was large variability in the methods used to identify and then verify MC phase, namely calendar-based counting, BBT, MC history questionnaires, urinary ovulation detection kits, and salivary, urinary and serum measurement of both oestrogen and progesterone. Calendar-based counting is an indirect method to identify MC phase, whereby the self-reported onset of menses is set as day one, and the phases are then established by counting days from this point [
17]. This method, however, assumes that all participants with regular menstruation experience ovulatory cycles with a mid-cycle peak in oestrogen, which is not always the case [
120,
121]. As such, the use of calendar-based counting methods in isolation is not recommended when accurate identification of MC phase is required [
122]. Similarly, BBT is a widely used method for identifying ovulation, and the length of the follicular and luteal phases [
17], but this method does not provide information regarding actual hormone concentrations [
123], and temperature readings might also be influenced by a range of factors such as illness, stress, sleep patterns and medication [
124]; hence BBT in isolation is not considered a reliable method for MC phase verification [
17]. Studies using these aforementioned methods were downgraded on this basis. Indeed, very few studies used a combination of the recommended methods by Cable and Elliott [
10] and Janse de Jonge et al. [
17], which include the use of the calendar-based counting method in conjunction with urinary ovulation detection kits to assist in setting the timing of testing throughout the MC and to confirm the presence of an ovulatory cycle, followed by serum measurement of both oestrogen and progesterone levels to subsequently verify the phases of the MC. Given that the rationale for exploring the effect of the MC on performance is underpinned by changes in oestrogen and progesterone, it is essential that studies should accurately verify the acute changes in endogenous hormones during each phase of the MC to ensure that the intended phase is being examined. Overall, without blood analysis, it is unclear which hormone milieu is being investigated, thus making it difficult to draw accurate conclusions regarding changes in performance across the MC and to make direct comparisons between studies. These recommendations echo recent publications in the area of women’s physiology [
10,
17], demonstrating an increasing awareness for the nuances of this type of research, and collectively provide researchers with ample tools to make methodological decisions for future investigations. To limit the influence of low quality papers on the analyses, a sensitivity analysis was conducted with data obtained from studies that were classified as either “moderate” or “high” in quality [
19,
27,
31,
35,
37,
54,
57,
58,
67,
71,
75,
85,
90,
99,
106,
115]. Due to the limited amount of data available, only the pairwise meta-analysis comparing exercise performance during the early follicular phase of the MC with all other MC phases was conducted. The sensitivity analysis provided no evidence of any effect, with a relatively symmetric credible interval centred close to zero. Whilst studies that were allocated a higher quality rating were better able to identify and verify the MC phase, there was no association between study quality and average sample size. Given the reduced amount of data included within the sensitivity analysis and the low sample sizes, the result is consistent with the primary analyses and conclusion that if an average effect exists, it is likely to be trivial in magnitude.
The results from the meta-analysis models consistently showed large between-study variance, which might be attributable to several factors: (a) inconsistent research design, as shown by the network analysis that highlights the discrepancy in the number of phase comparisons made between studies; (b) poor methodological practices, as emphasised by the quality assessment, whereby the majority of studies included in the meta-analysis were classified as “low” (42%) in quality primarily due to inadequate MC phase identification and verification in many studies; (c) non-homogenous participant groups, as shown in Electronic Supplementary Material Appendix S5 participants in this meta-analysis ranged from sedentary, to healthy, to physically active to elite athletes; and (d) large variation in the type of performance outcome measured, as detailed in Electronic Supplementary Material Appendix S2. As such, the breadth of this research area, without the corresponding depth, makes it difficult to apply a meaningful, yet generalisable, interpretation of the current data.