[Editorial] Sex and gender in physiotherapy research

Evidence from preclinical and clinical research demonstrates that sex and gender can impact disease presentation, prognosis and response to treatment.¹ Sex refers to the biological attributes that define humans as male, female, or intersex, including chromosomes, gene expression, hormone levels and reproductive anatomy.² Gender refers to the socially constructed norms, behaviours and roles of girls, women, boys, men and gender-diverse people.² Gender is a continuum and not a static concept; gender can change over time. The mechanisms by which sex and gender impact health are complex and not yet fully understood, but include an interaction between epigenetic, genetic, environmental, social and behavioural factors, among others.³

Disaggregation of data by sex and/or gender helps to unpack the complex biological and psychosocial interactions that might influence disease presentation and progression. In a clinical setting, unpacking this interaction helps us to screen for disease more effectively, and choose the best evidence-based treatment course for patients. Differences between females/women and males/men have already been identified across several disciplines of physiotherapy, including neurology,⁴ musculoskeletal⁵ and cardiopulmonary.⁶

In clinical research, recruiting sex-diverse and gender-diverse populations is essential to enable disaggregation of data. Historically, health data have been collected on males and generalised to females.⁷ Such patterns are still observed in certain health fields such as pain research,⁸ exercise science⁹ and pharmaceutical trials.¹⁰ To remedy this, the governmental funding bodies of many countries have mandated equality in the collection, analysis and reporting of sex-specific and gender-specific health. It is still debateable whether or not these policies translate into research, so that the complex biological and psychosocial interactions of sex and gender with health can be unpacked.⁹ However, the first step in this process for clinical research must be the recruitment of sex-diverse and gender-diverse populations.

It is currently unknown whether physiotherapy research trials recruit sex-diverse and gender-diverse populations, or whether data are reported as sex and/or gender disaggregated. To gauge the current but pre-pandemic status of sex and gender reporting in physiotherapy trials, the reports of 250 randomised trials published in 2019 were randomly sampled from among those on the Physiotherapy Evidence Database (PEDro). Among these, four trial reports provided no information or conflicting information on sex and gender; these were not analysed further. Fifty-five trials studied females only or males only, and the remaining 191 trials included some degree of sex or gender diversity among the participants.

In the 55 trials that studied females only or males only, the minority were related to conditions that are female-specific (breast cancer n = 6, menopause n = 5, pregnancy n = 2, puberty n = 1) or male-specific (prostatitis n = 1). The other 39 trials involved participants with female sex only (n = 27), male sex only (n = 11) or female gender only (n = 1), even though the condition being investigated affects both females and males. Of these 39 trials, only 17 studies provided a rationale as to why they were investigating a single sex or gender; the remaining 22 studies failed to provide a rationale. Interestingly, the high proportion of female-only trials is in contrast to what has been seen in other health disciplines, where trials are more likely to be male-only⁹ or have a higher proportion of male participant inclusion.^11,12

Sex and gender diversity in physiotherapy research

The 191 trials with sex/gender diversity among participants included data on 19,390 participants. Nineteen trials, involving 1,049 participants, failed to specify the male/female split and so could not be included in the analysis of sex or gender ratio. For the remaining trials, the percentage of female participants was calculated. Among trials reporting the sex of participants, the mean percentage of female participants was 50% (SD 9), indicating that these trials typically had roughly equal participation of males and females. Among trials reporting the gender of participants, the mean percentage of female participants was 55% (SD 22). While this result again indicates roughly equal participation of males and females overall, the larger SD indicates that more trials would have had substantial imbalance. Also, no trials reported including any gender-diverse participants beyond binary males or females. Population-based studies estimate the prevalence of gender nonconformity to occur in up to 4.6% of the general population.¹³ That no trials included gender-nonconforming participants highlights a gap in current physiotherapy research. Previous research has identified that LGBTIQ+ individuals often have uncomfortable experiences with physiotherapists and physiotherapists lack knowledge specific to the needs of these individuals.¹⁴ These findings may reflect: poor physiotherapist training on gender diversity and inclusion; poor participation or recruitment of gender-diverse participants; and/or poor data collection and reporting regarding gender in research. It also indicates some deficiency in the evidence that physiotherapists have with which to tailor their clinical decision making to individual patients.

When describing demographic data, sex was used as a descriptor in 132 trials and gender as a descriptor in 59 trials. Eleven trials conflated sex and gender, using the terms interchangeably throughout the manuscript. Twenty-seven trials did not use either term, but instead provided a number or percentage of males and/or females (n/% of males/females n = 14, n/% of males n = 7, n/% of females n = 6). No trial included a definition of either sex or gender, or stated why one variable was chosen over the other to be collected from participants.

Sex and gender disaggregation in physiotherapy research

None of the 191 trials disaggregated their data by sex or gender. Some trials used statistical methods to account for potential sex or gender differences. Twenty trials used statistical adjustments for sex differences (18% of trials reporting the sex of participants) and five trials used statistical adjustments for gender differences (9% of trials reporting the gender of participants). While statistical adjustment for demographic variables is helpful, adjusting for sex and gender is insufficient in health research. Statistical adjustments do not show where specific differences between sex and gender exist, and understanding these differences is vital for clinical decision making.

Recommendations for future research

This analysis of physiotherapy trials has highlighted strengths and weaknesses in that research. The equal recruitment of males and females by sex and gender is encouraging and is something our profession should continue to strive to do in research. However, there was a lack of inclusion of gender non-conforming participants. Future physiotherapy research should look to recruit even or representative numbers of women and men (by gender), and also include gender-nonconforming participants.

Sex and gender were often used interchangeably, and no studies provided definitions of sex or gender, or a description of why either sex or gender was chosen over the other as a demographic characteristic. Researchers are encouraged to consider sex and gender variables in the planning stage of trials, and determine which characteristic may be more important to the primary research aim. Recruitment of participants by either sex or gender, or even both, should relate to any hypotheses around whether these characteristics may influence the outcomes of the trial. Gendered Innovations in Science, Health & Medicine, Engineering, and Environment¹⁵ provides comprehensive resources for health researchers to plan for analysing sex and gender differences, and the Canadian Institutes of Health Research offer free online training courses to assist researchers in appreciating the differences between and across sexes and genders.¹⁶

While some of the included trials statistically adjusted for sex or gender variables, no trials disaggregated their data according to sex or gender. Disaggregating data in this way ensures that differences in outcomes can be observed between and across sexes and genders, which is vital information for clinicians to more effectively screen for disease and choose the best evidence-based treatment options for presenting patients. However, we recognise that analysis of subgroups by sex or gender results in a loss of statistical power, which is an important consideration in clinical trials. Therefore, we encourage researchers to run analyses to determine possible interaction effects due to sex and/or gender, preserving statistical power, and present disaggregated data where such an interaction effect exists.

If a sex or gender effect is identified, it is important to determine whether it is a consistent finding. This could be assessed in future replications of the study. It could also be assessed easily in existing similar trials if they publish the individual participant data with sex/gender data. See, for example, Table 4 in the eAddenda of the trial by Scheer et al in this issue.¹⁷

The Journal of Physiotherapy encourages researchers to consider whether sex and/or gender may moderate the effects of the intervention. Where it is plausible that sex and/or gender may be treatment effect moderators, researchers should plan, a priori, to investigate this using appropriate analyses (eg, test for an interaction effect based on sex and/or gender). If a significant interaction is found, disaggregated data should be reported. We also encourage authors to adopt this approach when submitting research to other journals.

References

1 A.J. McGregor, et al.Biol Sex Differ, 7 (2016), pp. 61-72

2 Canadian Institutes of Health ResearchWhat is gender? What is sex?https://cihr-irsc.gc.ca/e/48642.html, Accessed 28th Jul 2021Google Scholar

3 Z. Wainer, et al.Med J Aust, 212 (2020), pp. 57-62 Download PDFView Record in Scopus

4 R.A. Haast, et al.J Cereb Blood Flow Metab, 32 (2012), pp. 2100-2107 Download PDFCrossRefView Record in Scopus

5 S.Z. George, et al.J Orthop Sports Phys Ther, 36 (2006), pp. 354-363 Download PDFCrossRefView Record in Scopus

6 J.A. Krishnan, et al.Arch Intern Med, 161 (2001), pp. 1660-1668 Download PDFView Record in Scopus

7 C.C. PerezInvisible women: Exposing data bias in a world designed for menRandom House (2019)Google Scholar

8 J.S. MogilNat Rev Neurosci, 21 (2020), pp. 353-365CrossRefView Record in Scopus

9 J.T. Costello, et al.Eur J Sport Sci, 14 (2014), pp. 847-851CrossRefView Record in Scopus

10 S.S. Richardson, et al.Proc Natl Acad Sci, 112 (2015), pp. 13419-13420 Download PDFCrossRefView Record in Scopus

11 V.S. Prakash, et al.J Womens Health, 27 (2018), pp. 1342-1348CrossRefView Record in Scopus

12 S. Feldman, et al.JAMA Netw Open, 2 (2019), Article e196700 Download PDFCrossRefView Record in Scopus

13 L. Kuyper, et al.Arch Sex Behav, 43 (2014), pp. 377-385CrossRefView Record in Scopus

14 M.H. Ross, et al.J Physiother, 65 (2019), pp. 99-105ArticleDownload PDFView Record in Scopus

15 Gendered Innovations http://genderedinnovations.stanford.edu/index.html, Accessed 9th Aug 2021Google Scholar

16 Canadian Institutes of HealthSex and Gender Training Moduleshttps://www.cihr-irsc-igh-isfh.ca/, Accessed 9th Aug 2021Google Scholar17A. Scheer, et al.J Physiother, 67 (2021)XXX–XXX

Source