Methodological Considerations for Studies in Sport and Exercise Science with Women as Participants: A Working Guide for Standards of Practice for Research on Women

Across the lifespan, circulating concentrations of oestrogen and progesterone change from puberty, through adulthood, until menopause, when a dramatic decline in the secretion of ovarian steroids occurs and cyclic oestrogen production is replaced by a low constant production by the ovaries. Furthermore, between puberty and menopause, circulating concentrations of oestrogen fluctuate fivefold and progesterone greater than 50-fold over a 21- to 35-day cycle (i.e., menstrual cycle). Indeed, these cyclical acute fluctuations in ovarian hormones across the menstrual cycle may be altered by external factors (e.g., hormonal contraceptives [HC], circadian variation, diet, exercise), clinical conditions (e.g., functional hypothalamic amenorrhea, polycystic ovary syndrome) and pregnancy (Fig. 1; [1]).

In addition to reproductive function, ovarian steroids have been shown to target a number of tissues (e.g., epithelial, connective, muscle and nervous; [2‐4]) and influence an array of biological processes (e.g., metabolism, ventilation, immunity, cognition, gastrointestinal, cardiovascular, autonomic regulation, and genitourinary function; [5‐13]). As such, different concentrations of circulating ovarian steroids throughout the lifespan and across the menstrual cycle (with and without dysfunction and/or manipulation) might significantly influence the mechanisms that control and regulate cell function and integrated physiologic adaptation in women.

Given the potential influence of ovarian hormones on multiple biological systems, it is reasonable to assume that the heterogeneic findings observed among studies conducted with women as participants, particularly those examining physiological responses to exercise and adaptations to training, may be attributable to the numerous hormonal profiles a woman may experience during her lifetime. As such, it is important that scientists consider the potential effects of ovarian hormones throughout the lifespan on exercise responses and sports performance and apply suitable methodology to their studies in order to control these effects and reduce the heterogeneic results. Nonetheless, for decades scientists have avoided conducting sport and exercise research with women as participants due, in part, to the complex methodological considerations required and/or the difficulty in interpreting the heterogeneic results often observed within and among studies. Consequently, research findings from sport and exercise performance and training studies with men only as participants are applied to women, which can result in a missed opportunity for women to reach their athletic and performance potential. In order to guide scientists and practitioners through the complexities of: i. identifying the maturation status and menstrual-cycle status of women participants, and ii. designing studies that provide insight into the difference in the physiological responses to exercise among the various hormonal profiles of women, guidance on the definition(s) of sampled populations is required and a working guide for standards of practice for research in this field is warranted.

Research into the effects of the menstrual cycle on the exercise response has being ongoing since at least 1876, when Professor Mary Jacobi won the Harvard University Boylston Prize Essay for her observations on menstruation and physical work and rest [14]. Despite a recent growing interest in research in sport and exercise science with women as participants, the overall research quantity in this area is still very small compared to research with men as participants [15]. Therefore, the effects of oestrogen and progesterone on women’s responses to exercise and adaptations to training are still not fully understood. As a result, there are no sport and exercise-related guidelines for training and nutrition which are customised for sportswomen (i.e., athletes and exercisers) and underpinned by sufficient, high-quality scientific evidence. Numerous reviews [e.g., 16‐20], books [e.g., 21‐23] and meta-analyses [e.g., 24,25] have summarised and examined data from this field spanning almost 50 years, but all have concluded that more, better quality studies are needed before a consensus on the effects of ovarian steroids on training and performance can be reached.

It is clear that more research focusing on areas unique to female physiology is required to inform best practice guidelines for training and nutrition [24, 25]. There are some studies, however, that have demonstrated clear and recurring findings of ovarian-steroid interference with biological processes that may affect athletic performance (i.e., responses to exercise and adaptation to, and recovery from, training), which highlight the importance of considering ovarian steroid profiles in sport and exercise science research. In particular, the manipulation of endogenous ovarian steroid levels with synthetic hormones (i.e., HC) has been repeatedly associated with increased markers of muscle damage in women after unaccustomed eccentric exercise [8, 26‐30]. Therefore, knowledge of the use of HC by athletes or research participants is imperative in understanding the recovery process after an acute bout of exercise or adaptation to training. There is a strong body of evidence indicating altered thermoregulation in women taking HC during exercise including altered skin blood flow and cutaneous vasodilation, as well as a delayed onset of sweating and a higher internal temperature threshold [31‐35]. As such, ovarian hormone interference via the administration of HC may have implications for exercise in the heat. Oestrogen is often associated with substrate oxidation and utilisation during prolonged exercise. The notion that elevated plasma oestrogen concentrations increase the reliance on fat as an energy source during endurance exercise is a consistent observation [6, 36‐38]. Despite these few convincing findings, whether disturbances to these physiological mechanisms result in a decrement to performance in strength, speed, and/or endurance events remains to be elucidated. It should be also noted that there is always at least one study that opposes the findings of the majority and as newer formulations of synthetic hormones are developed, there will be further ambiguity in the findings. Therefore, the importance of considering a woman’s current ovarian hormonal profile in sport and exercise science research is paramount.

Rather than viewing the dynamic profiles of ovarian steroids in women as a reason to exclude them from sport and exercise science research, future studies need to explore the specific effects of oestrogen and progesterone and related hormonal ageing on physiological function and ergo athletic performance across the lifespan in women. Female-specific research is needed as the number of women participating in sport and exercise is increasing in many countries, and parity in the Olympics and Paralympics is approaching (i.e., equal number of male and female competitors and medals). For example, in 1900 only 2.2% of the competitors in the Olympics were female; in 1952 the proportion was 10.5%, in 2000 it was 38.2% and in 2021 it is predicted to be 48.8%. If male-based research continues to be applied to females, there is a risk that the true potential of women will not be achieved as the potential influence of ovarian hormones remains unclear; herein exist the gap in the field and the fundamental statement of the problem within contemporary sport and exercise science. At the elite level, understanding the effects of oestrogen and progesterone on body systems and performance may produce the marginal gain needed for success. In addition, there is a need to understand the effects of training on hypothalamic, pituitary and ovarian hormones and related outcomes, such as body composition and bone [39]. At the recreational level, understanding the impact of changes in ovarian hormone concentration, experienced during the menstrual cycle or as a result of menstrual irregularities, hormonal contraceptive use, pregnancy or the menopause, on the ability and willingness to undertake exercise and physical activity may help reduce the disproportionate decline in sports participation seen at puberty in girls [40, 41]. This decrease in physical activity often extends into adulthood, and is compounded during pregnancy and the postpartum period [22], and as such has implications for health and well-being [42] for women of all ages. Among middle-aged women, menopausal hormonal changes are associated with deterioration of the body systems [43], performance and functioning [44], but further understanding is needed of the magnitude and timing of these changes as well as the role of physical activity and exercise to overcome or attenuate these changes.

In order to facilitate a new generation of quality research in sport and exercise science with women as participants that addresses the current gap in the field, direction is needed on the best and most accurate types of research design to employ in this field. In addition, guidance on the definition(s) of sampled populations is required in order to improve the uniformity of studies involving women. As such, this article is intended to serve as a working guide for standards of practice for research on women.

Synchronous with changes in performance, ovarian steroids exert many clinical effects, which have also received less attention in sport and exercise medicine-based research. Examples include hyponatremia, which is more prevalent in females than males [45] and heat stroke, which is influenced by ovarian hormonal concentrations [46]. Although this paper is focused on studies related to sport and exercise science, some of the approaches described herein may be also relevant for female-based sport and exercise medicine-based research.

In the early 2000s, articles started to provide critical commentary on the methodological approach used in studies investigating the effects of oestrogen and progesterone on different aspects of exercise performance [18, 47]. This began to raise an awareness of some of the technical flaws and issues within this field. Cable and Elliott [47] focussed on the methodological approach used in studies investigating the effects of ovarian steroids on muscle strength in women. They identified inconsistencies in terminology and research design, namely differences in the definition of reproductive status (e.g., inconsistencies in menstrual cycle phase definition), often leading to the grouping of non-homogeneous participants; and quantification of reproductive status, by using inaccurate and subjective methods. In addition, they highlighted that consideration needed to be given to characteristics of the hormonal milieu; such as, if the changes in ovarian steroids were acute (e.g., menstrual cycle) or chronic (e.g., menopause); within normal (e.g., menstrual cycle) or supra-physiological (e.g., pregnancy) ranges; caused by endogenous fluctuations (e.g., menstrual cycle) or exogenous (e.g., HC or hormone replacement therapy [HRT]) supplementation. The review by Janse de Jonge [18] focussed on methodological considerations in menstrual cycle research, including the verification of menstrual cycle phase and the timing of testing throughout the cycle. More recently, Janse de Jonge et al. [48] consolidated the approaches and issues observed specifically in menstrual-cycle studies and proposed methodological guidelines for undertaking menstrual cycle-based research in sport and exercise science. Furthermore, broad methodological considerations for studies in exercise endocrinology have been provided lately [49], including the critical consideration of sex and the menstrual cycle. The current statement builds upon these reviews by providing guidance for studies including women from across the lifespan.

The considerations outlined herein are based on previous research and expert opinion and are delivered alongside the rationale and pros and cons for each method. Where appropriate the primary source has been provided; however, some items have been specifically developed for this statement. For ease of reading we are using oestrogen throughout the paper, but recognise this encompasses oestradiol, oestrone and oestriol. For more detailed information on female endocrinology see Hackney [21].

The aim of this statement was to consolidate the opinions, lessons and evidence from experienced researchers in the area of women’s physiology, with a view to providing counsel for those beginning to pursue research in this very important area. Moreover, the considerations described herein are intended to help dispel the myths and poor practices commonly associated with this subject area, thus improving the quality of future research and allowing more meaningful comparisons to be made between studies that have employed consistent terminology and research designs. Overall, this statement aims to contribute to advancing the scientific community’s knowledge and understanding of women’s physiology in the context of sport and exercise science.

Much of the heterogeneity in this area of research is derived from the lack of uniformity when describing the populations used in studies with women as participants. While chronological age is often controlled and reported, maturation status and menstrual-cycle status (i.e., length, irregularities/dysfunction, manipulation) are more often not. Moreover, if menstrual-cycle status is even considered by researchers, the hormonal profile of the participants is often implied rather than confirmed [25] which may exacerbate the variability between studies.

In order to overcome these issues, consistency is needed in the following: (i) the terminology used to describe participants; and (ii) the inclusion/exclusion criteria, specifically the hormonal parameters, used to define eligible participants. This uniformity will allow meaningful comparison within and among studies (i.e., comparing like with like). Table 1 proposes a framework of terms and conditions, based on previous research and our collective experience and expertise in the field that could help to ensure consistency in women’s sport and exercise-based studies. Across the lifespan, the ovarian steroids start changing at puberty. Therefore, the stage of puberty needs to be classified consistently in research with younger populations. During the reproductive years there are five main factors that will influence the fluctuations in oestrogen and progesterone: i) fluctuations across the eumenorrheic menstrual cycle; ii) disruptions to the eumenorrheic menstrual cycle caused by illness or disease (e.g., polycystic ovary syndrome, endometriosis or other hypothalamic pituitary axis disruptions), low energy availability, the first two years after menses, and the time nearing menopause; iii) manipulation of the eumenorrheic menstrual cycle with exogenous hormones, both synthetic and natural (e.g., HC and HRT); iv) pregnancy (increasing ovarian steroids profile); and v) menopause (decreasing ovarian steroids profile). Researchers should strive to adopt all recommendations that are appropriate, thus facilitating good practice in this area.

The experimental design employed in studies with women as participants is often poorly considered, described, controlled and/or executed, which is mainly due to a lack of understanding of the hormonal changes experienced across the lifespan in women. In addition, studies involving women as participants usually require repeated measures due to the acute (e.g., ultradian, circadian and infradian rhythms) and chronic fluctuations (e.g., pregnancy, menopause, HC and HRT usage) in reproductive hormones over time, unlike in studies with men as participants who have relatively stable (i.e., outside of circadian variation) sex hormone profiles following puberty. Table 2 outlines some methodological approaches that could be employed at various stages of the life cycle (i.e., puberty, menstrual cycle and irregularities, hormonal contraception, pregnancy and menopause) in women.

One particular aspect of research design that has continued to cause debate amongst authors and disparity between studies is the nomenclature associated with the phases of the menstrual cycle and menopausal status [57]. In the simplest terms, the menstrual cycle can be divided into two phases that are separated by ovulation; follicular (begins at the onset of menses) and luteal (post ovulation). These phases can be further divided into the early, mid and late phases for each, with or without the inclusion of ovulation as a discrete phase. As such, studies involving eumenorrheic women report phases on a continuum from two to seven. Furthermore, the basis for this taxonomy is often unclear; i.e., the phases are linked to poorly defined or undefined ovarian steroid profiles. Rather than continue to dispute the name of menstrual cycle phases, a consensus is needed on the hormonal profiles of interest, such that testing timepoints and the determination of eligible data can be based on an a priori decision. Table 3 and Fig. 4 show four distinct hormonal profiles, which represent significant changes in both oestrogen and progesterone concentrations. These four profiles could be used to investigate the effects of ovarian steroids on physiology, health and performance.

Some methodological issues can affect all sport and exercise science studies with women as participants; these points are described in Table 4. One consideration is the use of “unclassified” women, viz. where the reproductive profile of the participants is not stated. In some studies, it is relevant to describe the menstrual characteristics of the participants, but not necessary to measure ovarian steroid levels. It is important to balance the internal and external validity of a study, such that the translational reach and impact of the data are not limited. The research design should be suitable to answer the research question. For example, if a study is designed to examine the effects of menstrual cycle phase on an outcome, then biochemical confirmation of the phase is warranted alongside self-reported menstrual characteristics (e.g., last menses), whereas if a study is designed to examine the physical/performance characteristics of a female sports team, then a description of the population (i.e., solely self-reported menstrual cycle phase) may be sufficient. As such, the research question should clearly indicate if the intention is to attribute an outcome to a specific hormonal profile or to observe the characteristics/responses of a clearly defined, mixed group of sportswomen or exercisers.

Herein cisgender females are described and defined based on ovarian steroid concentrations. Transgender females and women with differences in sexual development have not been characterised but require and deserve further clear research parameters to better understand their physiology. The diversity in female reproductive endocrinology makes it difficult to title and define women based on reproductive hormonal profile alone. As such, it can be challenging for researchers to decide on the inclusion/exclusion criteria for participant recruitment; determine the methods by which this criteria can be established; and use the eligibility criteria to inform the research design.

Example 3; a researcher wishing to conduct menopause-related study needs to decide which menopausal phases to include and how to define those phases; whether or not to include HRT users or those with ovarian concerns that have resulted in hormonal conditions similar to menopause; if post-menopausal women are included, how to define the time that the participants have been post-menopausal [72].

As such, we propose that:

These steps should help alleviate the ambiguity surrounding the classification of women previously seen in sport and exercise studies that has led to considerable sampling heterogeneity and has hindered meaningful comparison between studies.

The validity of studies, or lack thereof, in sport and exercise science using women as participants is rarely discussed. The considerations described herein are designed to increase the internal validity through adequate quality control as a result of stringent recruitment strategies (e.g., using predetermined population definitions and strict inclusion/exclusion criteria) and appropriate experimental design (e.g., a priori classification of hormonal profile associated with each testing timepoint and post priori exclusion of data not meeting the stipulated terms). These measures should help to ensure that the results reflect the truth in the population studied rather than methodological errors [73]. At present, given the lack of clarity on either the direction or magnitude of potential effects of ovarian steroids on physical performance, maximising internal validity should be considered a research priority. The external validity of studies involving women as participants is often more difficult to increase due to the large inter and intra-individual variability in ovarian hormone concentrations; by reducing some of this variation through precise recruitment and testing, the external validity is lessened, and thus generalisability is reduced. These issues can be somewhat overcome through field-based testing and by testing entire teams, which could help maximise the translational potential of performance-based research in women.

Until recently, women were less represented in elite sport. Women accounted for only 2.2% of the athletes in the Paris Olympics in 1900, but they are predicted to account for an estimated high of 48.8% in the next Olympic Games [74]. Therefore, the time is right to shift more attention to women as participants in sport and exercise science research, especially given that often the research findings that are based on male participants are neither applicable nor generalizable to women. It is not appropriate to exclude women from sport and exercise science research for any reason, especially on the basis of convenience, i.e. excluding women as participants due to the potential confounding effect of female hormone fluctuations, or cost, i.e., excluding women as participants because they can be more expensive than men to study due to the extra time (e.g., repeated measures across a menstrual cycle) and resources (e.g., blood samples for the determination of ovarian steroid levels) needed to produce high-quality data. Unfortunately, it is evident that many researchers have avoided using female participants with data indicating just 4 to 13% of articles from leading journals in sports and exercise science included only women as participants [15]. So even though some research has focussed on the effects of ovarian steroids on performance, health and physiology, the small quantity of research in this area limits our understanding of the topic. Furthermore, the heterogeneity in study populations (e.g., eumenorrheic, OCP user, post-menopausal) and large variation in research designs (e.g., definition of menstrual cycle phase, oral contraceptive pill consumption or withdrawal) has resulted in many conflicting findings in sport and exercise science research with women as participants. To improve our understanding of many female specific topics, undoubtedly more high-quality research is warranted to overcome the heterogeneity and ambiguity seen in the limited number of studies in this area to date. Women account for half the world-wide population, therefore a willingness to study women as participants in sport and exercise science needs to be fostered and funding needs to be made available to achieve equivalence in knowledge about the female sex.

Although this paper has covered many aspects of the reproductive ageing female (i.e., from puberty to the menopause), it has not covered every perturbation of the menstrual cycle [75] or every type of hormonal contraceptive [76] or all aspects of (post) menopausal hormonal therapy. In addition, medications known to influence the hypothalamic-pituitary-ovarian axis are beyond the scope of this statement [77].