Australian Journal Of Physiotherapy

100 Ross and Setchell: LGBTIQ+ people’s experiences of physiotherapy Box 1. Survey questions. Participants were encouraged to provide as much detail as possible in response to the following questions: Overall, how did you feel during your [physiotherapy] appointment/s? Please describe any situations that made you think about your gender and/or sexual preference [during appointments]? Please describe any conversations about your gender and/or sexual preference and why you did or did not discuss these. Please describe any physiotherapy experiences where you felt comfortable or uncomfortable about your gender or sexual preferences. Were any of your negative or positive LGBTIQþ and physiotherapy experiences affected by other factors such as your race, ability, age, religion, etc? How do you think the physiotherapist themselves felt during your appointment in relation to your gender or sexual preference? Did they mention their own gender and/or sexuality? During your physiotherapy appointment/s, has anyone ever assumed anything about your gender and/or sexuality? If so, please describe the assumptions and any effects on you. Please indicate if you would have felt comfortable or uncomfortable discussing your gender/sexual preference with your physiotherapists. Please describe any experiences you had with ‘coming out’ about any aspect of your gender and/or sexuality during any of your appointments with a physiotherapist. Do you have any suggestions of what could help your/other LGBTIQþ people’s experience of physiotherapy? Thinking broadly about LGBTIQþ people attending physiotherapy, would it be helpful if (Likert scale):  physiotherapists use gender-neutral language  there is an option for ‘non-binary’ on forms  physiotherapy workplaces displayed a small rainbow flag in the reception area  physiotherapy clinics’ images show a range of people of different genders and sexualities  physiotherapists receive LGBTIQþ diversity training  physiotherapists are more aware of health issues specific to LGBTIQþ people LGBTIQþ = lesbian, gay, bisexual, transgender, intersex, queer or related identities. To address the dearth of literature on the topic, this study aimed to recruitment wording was ‘Do you identify as LGBTIQþ? Have you identify how people who identify as LGBTIQþ experience physio- seen a physiotherapist? We want to hear from you.’ Participant therapy. Understanding any key concerns will provide physiothera- numbers were determined by the principle of saturation. Sufficient pists with opportunities to improve (where required) provision of saturation was considered to be reached when iterative analyses healthcare to LGBTIQþ people. during data collection demonstrated sufficient repetition and depth of concepts in the data related to the study aims.25 Therefore, the research questions for this mixed-methods survey were: Theoretical underpinnings 1. What are the experiences of people who identify as LGBTIQþ and This descriptive study was underpinned by the theory of rela- attend physiotherapy? tivism. That is: people have different experiences and there is no one knowable version of reality.26 This theoretical perspective allows the 2. How could those experiences be improved? possibility for no singular experience of people who identify as LGBTIQþ and attend physiotherapy, but rather there may be multiple Method experiences. This reasoning is consistent with understandings of pluralism (ie, many factors influence people’s experiences) and Design intersectionality (ie, these many factors affect each other).27 For this study, this means that people who identify as one aspect of LGBTIQþ People who identified as LGBTIQþ and had attended physiotherapy might have different physiotherapy experiences (eg, someone who were invited to complete an online survey about their experiences of identifies as a lesbian might have different experiences to a trans- physiotherapy. Recruitment was via social media, word of mouth, and gender man). Further, sexuality and gender might not be the only LGBTIQþ organisations in Australia. Snowballing was encouraged. factors influencing people’s physiotherapy experiences: other factors such as race or social class might also have an effect. Consistent with Data were collected using the online survey.24 This method was this theoretical approach, this study was designed to invite and detect chosen because data needed to be collected from a large number of divergent experiences. participants to suit the study aim of seeking an overview of the ex- periences of people identifying as LGBTIQþ with physiotherapy. Data analysis Study information was detailed online before providing consent and entering the survey. Participants remained anonymous throughout Data analysis was principally qualitative, with a small quantitative the survey. The survey was purpose-built and comprised questions component. For qualitative data, a descriptive thematic analysis related to demographic data, and about participant’s experiences of outlined by Braun and Clarke was used.26 Analyses were conducted physiotherapy relating to their identity as LGBTIQþ (see Box 1 for an by two physiotherapists, both of whom identify as LGBTIQþ: JS is an overview of questions). The survey had one quantitative section experienced qualitative researcher and MHR is trained in qualitative where participants identified (using a Likert scale) levels of support analysis. for strategies to improve LGBTIQþ experiences of physiotherapy. The analysis involved five iterative stages. First, MHR and JS Participants independently read the entire dataset and made preliminary notes. They then re-read the dataset and refined analyses into provisional Inclusion criteria were: aged  18 years, self-identifying as themes. Next, the authors discussed their independent analyses to LGBTIQþ, had attended physiotherapy, and English speaking. Key

Research 101 Table 1 Participants (n = 108) Table 3 Characteristics of participants. 39 (9) 19 to 75 Glossary of gender and sexual orientation terminology. Adapted from the Anti- Characteristic Discrimination Commission Queensland65 and the Human Rights Campaign.66 83 (77) Age (yr), mean (SD) range 16 (15) Term Definition Place of residence, n (%) 8 (7) Agender A term used to describe people who identify as being metropolitan/urban 1 (1) Asexual without gender regional Binder Having a general lack of interest in sex and sexual desire rural/remote 41 (38) Bisexual Tight fabric worn around the chest to obscure the shape of not stated 27 (25) breasts Living situation, n (%)a 18 (17) Butch Having an attraction to both males and females as sexual partner/s 17 (16) partners though not necessarily simultaneously, in the same housemate/s Cis-gender way or to the same degree alone 5 (5) Gay A word used to describe people who identify as lesbian and partner/s þ child/ren 3 (3) Gender binary whose appearance and behaviour are seen as traditionally child/ren 1 (1) masculine parent/s 1 (1) Gender identity A person whose gender identity and biological sex (assigned co-housing 1 (1) at birth) align partner in a co-operative household 1 (1) Femme A term to describe people who are primarily attracted to ‘fly-in, fly-out’ worker 1 (1) people of the same sex as them foster carers Fluid/Gender fluid A concept or belief that there are only two genders and that siblings 6 (6) Heterocentric one’s biological or birth-assigned gender will align with Household income (AUD), n (%) 18 (17) Homosexual traditional social constructs of masculinity and femininity , 20 000 14 (13) Intersex An inner sense of one’s gender, which could be neither, 20 000 to 49 999 18 (17) either, both, or moving around freely between or outside of 50 000 to 79 999 18 (17) Lesbian the gender binary 80 000 to 109 999 13 (12) Misgendering A word to describe people who identify as lesbian and 110 000 to 139 999 15 (14) Non-binary whose appearance and behaviour are seen as traditionally 140 000 to 169 999 Panromantic feminine  170 000 6 (6) Pansexual A person who does not identify with a single fixed gender or not stated has a fluid or unfixed gender identity Level of education, n (%) 36 (33) Queer Having a heterosexual bias post-graduate 33 (31) Sexual feeling for a person (or persons) of the same sex university 20 (19) Transgender A term used to describe people born with physical, tertiary 18 (17) Transmasculine/ hormonal or genetic features that do not fit medical norms high school Transfeminine for female or male bodies , high school 1 (1) A term used to describe women who are attracted to other Number of visits to a physiotherapist, n (%) women . 10 61 (57) Referring to a person in a way that does not reflect their 5 to 10 26 (24) gender identity 2 to 4 18 (17) Not identifying as either male or female 1 Having a romantic attraction to people of all genders 3 (3) A term used to describe someone who has the potential for a Participants were able to select multiple options. attraction to people of any gender though not necessarily simultaneously, in the same way or to the same degree solidify themes and subthemes. MHR then coded all data into these An umbrella term used by the LGBTIQþ community to categories. The fifth step was for the authors to further refine this describe a range of sexualities and gender identities that are analysis to finalise coding. Any discrepancies were included in the outside of heterosexual and gender binary norms A term used to describe people who do not identify with the Table 2 biological sex they were assigned at birth Participant gender and sexual orientation. A term used to describe people who do not identify with the biological sex they were assigned at birth – they identify with masculinity more than femininity or the reverse Characteristic Participants (n = 108) Gender, n (%) 67 (62) findings. To enhance rigour, an experienced qualitative researcher female 18 (17) external to the study checked final coding to ensure that results were male grounded in the data. The Consolidated Criteria for Reporting Quali- transgender male, transmasculine 8 (7) tative Data (COREQ)28 was also used for rigour; all relevant criteria non-binary 4 (4) were satisfied. Quantitative data (Likert scale results) were analysed cis-female 3 (3) using descriptive statistics. gender queer 1 (1) agender 1 (1) Results fluid 1 (1) transmasculine non-binary guy 1 (1) Participants trans/non-binary 1 (1) gender queer male 1 (1) In total, 114 people responded to the online survey, with 108 of femme 1 (1) these meeting all of the eligibility criteria. The participants’ ages butch female 1 (1) ranged from 19 to 75 years. The sample was diverse in terms of living situation, household income and education level (Table 1). Sexual orientation, n (%) 41 (38) lesbian 22 (20) The participants identified with over 13 genders, including: 67 gay 16 (15) females (62%), eight transgender males (7%) and four non-binary queer 15 (14) people (ie, identifying as neither male nor female, 4%) (Table 2). bisexual There was also large variability in sexual orientation, with 41 (38%) pansexual 4 (4) identifying as lesbian, 22 (20%) as gay and 16 (15%) as queer, heterosexual 3 (3) amongst 12 other orientations (Table 2). For readers who are un- bi/pansexual 2 (2) familiar with the terminology used in Table 2, a glossary is pre- bisexual/queer 1 (1) sented in Table 3. asexual 1 (1) panromantic 1 (1) queer lesbian 1 (1) homosexual 1 (1) Participants chose more terms to explain their gender and sexual orientations than presented here (gender n = 18, sexuality n = 17). Some similar terms are grouped.

102 Ross and Setchell: LGBTIQ+ people’s experiences of physiotherapy Table 4 physiotherapist] assumed I was genetic female.as a transgender Themes and subthemes identified in the qualitative analysis. female I was pleased by that’ (P63, transgender female). Themes Description of theme A few participants mentioned assumptions implicit within the Subthemes broader physiotherapy environment. For example, P50 (trans/non- binary, bisexual) said ‘the environment in general wasn’t outwardly Assumptions This theme encompasses queer- or trans-friendly.there were assumptions that customers Assumptions of sexuality participant descriptions of were cis-gender and straight’ and P6 (female, queer) she ‘felt like [it Assumptions of gender the assumptions that were was] an uber heterocentric space’. made about their sexuality and/or gender identity. Theme 2: Proximity/exposure of bodies The second theme was proximity/exposure of bodies, which Proximity/exposure of bodies This theme relates to the Physical proximity and/or touch issues that participants included subthemes of physical proximity and/or touch, and undressing Undressing and/or observing the body expressed when physiotherapy and/or observing the body. When physical proximity and/or touch were included aspects of physical part of physiotherapy sessions, some participants expressed a fear of proximity, touch, undress making therapists of the same gender as them uncomfortable. For and/or observation. example, P80 (female, lesbian) said ‘massages made me feel hyper aware about my sexuality. They were strictly professional on her Discrimination This theme encompasses behalf but I was still anxious that my sexuality might make her un- Reports of discrimination participant reports of comfortable’. Male-identified participants highlighted similar Fear of discrimination discrimination, or fear of discomfort, for example, P71 (gender queer male, gay) said ‘It wasn’t discrimination, for being until there was body contact did I become aware/think about my LGBTIQþ. gender and sexuality.[W]ould this be more comfortable if I was a straight male?’. Some female-identified participants also mentioned Lack of knowledge of transgender health issues This theme encompasses that they feared sexualisation from male physiotherapists for being no subthemes a lack of knowledge, lesbians/women. For example, one participant said she didn’t want to understanding or an be ‘.seen as buying into any ‘male lesbian fantasies’’ (P17, female, over-interest in queer). transgender-specific health issues. When their body was exposed or observed, participants discussed perceptions or fears of judgement for not fitting into normative LGBTIQþ = lesbian, gay, bisexual, transgender, intersex, queer or related identities. gender conventions. For example, P30 (fluid) said they ‘presented female but with unshaven legs [and] when the physio was so distant Participants’ experiences of physiotherapy (qualitative analysis) and cold in her manner, [they] started to wonder whether she was homophobic or transphobic’. Some participants reported choosing to Four themes were identified during analysis of the open responses hide their gender identity, for example, ‘when my back was the issue I (see Table 4). Themes were: ‘assumptions’, ‘proximity/exposure of would often have to take my shirt off and I’d wear a bra instead of a bodies’, ‘discrimination’, and ‘lack of knowledge about transgender binder to avoid questions’ (P27, agender). Participants also expressed health issues’. Participants are distinguished by participant numbers concern that their biological sex may be exposed while undressed or (eg, ‘P34’). being observed. For example, P20 (transgender male) said he ‘was worried they would know/discover I wasn’t biologically male [and] Theme 1: Assumptions would realise that I have female hips, etc. I didn’t really want to tell Many participants mentioned that physiotherapists made as- them I was trans.’ sumptions about them. There were two subthemes – assumptions Theme 3: Discrimination about sexuality and assumptions about gender identity. The third identified theme was discrimination. There were two Assumptions about sexuality were that physiotherapists incor- subthemes: reports of discrimination by physiotherapists or other rectly assumed participants were heterosexual. For example, P32 staff, and the fear of discrimination. Reports of discrimination included (female, bisexual) said the physiotherapist had ‘assumed my partner both explicit and implicit discrimination. Explicit discrimination was was male’ and another said her physiotherapist ‘assumed I was reported infrequently, but included overt homophobic remarks, or straight as I was being treated for a pregnancy-related condition’ repeated misgendering. P39 (male transgender) said ‘despite (P102, female, lesbian). Reactions to these assumptions were mixed. repeatedly stating my gender and having my gender legally recog- Many participants expressed frustration or annoyance - often because nised as male/listed as male in their systems, I was referred to as they then felt that they had to either disclose their sexual preference ‘female’ and ‘she/her’ by the therapist’ and another transgender (and risk discrimination) or lie (and hide aspects of their life). For participant stated that they told the physiotherapist ‘my pronouns are example, P81 (male, gay) said: ‘I hate the feeling of having to come not optional. She kept ‘forgetting’ despite my beard’ (P40). Partici- out all the time’ and another said she felt awkward having to ‘make a pants also experienced implicit discrimination, that is, the frequently decision (whether) to come out or not, or if it’s safe to’ (P107, female, assumed heterosexuality and/or gender discussed in Theme 1 above. lesbian). Other people (fewer) said they were not concerned by these assumptions: ‘It didn’t bother me, although it is incorrect, it is the Fear of discrimination was also commonly reported, and related to societal norm and people cannot help assuming these things some- both gender and sexuality. For example: ‘I was unsure of whether times’ (P33, female, bisexual). In contrast, some participants reported they would have an awkward or uncomfortable reaction and didn’t positive experiences when physiotherapists interacted with them in a want to create an uncomfortable situation while I was in a vulnerable way that counteracted assumptions about their sexuality, for or physically compromised position: flat on my face with my clothes example, P17 (female, queer) said: there was ‘general conversation off, or with the therapist manipulating a painful area’ (P50, about my partner.this helps to normalise things and put me at ease’. transgender/non-binary, bisexual). The second subtheme was assumptions about people’s gender Importantly this fear of discrimination resulted in some partici- identity. Generally, it was assumed that people fitted into gender pants saying they did not disclose their gender or sexuality, even binary categories of male and female. That is, they were assumed to when relevant. For example: ‘My gender affirmation surgery causes be ‘cis-gendered’. For example, P38 who identified as ‘non-binary’ many problems in movement and posture, I wanted to tell the and ‘panromantic’ experienced assumptions across both subthemes practitioner why, but wasn’t brave enough.[disclosure] would have saying ‘[physiotherapists] assumed I was a cis-gendered heterosexual explained a great deal about my condition and very likely have girl. Every. Single. Time. That’s the default, until I explain’. In contrast, improved the quality of my care’ (P32, female). assumptions about gender were received positively by transgender participants when they were assumed to be the gender with which they identified. For example, one participant said ‘[the

Research 103 0% Percentage of participants Physiotherapists use gender-neutral language 20% 40% 60% 80% 100% There is an option for 'non-binary' on forms Proposed recommendation Physiotherapy workplaces display a small rainbow flag in the reception area Physiotherapy clinics' images show a range of people of different genders/sexualities Physiotherapists receive LGBTIQ+ diversity training Physiotherapists are more aware of health issues specific to LGBTIQ+ people Really don't like this idea Don't like this idea Not sure about this idea Like this idea Really like this idea Figure 1. Participant responses to proposed recommendations to improve LGBTIQþ people’s experiences with physiotherapy. LGBTIQþ = lesbian, gay, bisexual, transgender, intersex, queer or related identities. Other participants similarly suggested they would have liked to pronoun that they identify with. This can boost their confidence and disclose, but felt it would risk the therapeutic relationship ‘if [the make them feel really good about themselves since it’s something physiotherapist] wasn’t supportive I would never be able to go back they’ve been wanting and fighting for, for so long.’ (P25, transgender to him and because his skills are so specialised I have limited options’ male). (P44, female, bisexual/pansexual). Discussion In contrast, a few participants said they had no fear of discrimi- nation, for example, ‘I have always felt completely comfortable with The key finding of this study is that physiotherapy interactions my sexual preferences and gender when seeing my physiotherapist. I and environments may lack inclusivity of LGBTIQþ people. This lack have never been given any reason not to feel this way’ (P33, female, is inconsistent with national and global policies that advocate for bisexual). One participant clearly articulated the benefits of feeling inclusive and respectful physiotherapy treatment of all people.29 The safe to discuss their LGBTIQþ status with their physiotherapist, for findings highlight that almost all participants reported experiences example: ‘We talk about our lives together.physios can be incredibly relating to at least one of the following themes: ‘assumptions’ about healing and inspiring guides to reconnecting with those essential sexuality and gender identity, ‘proximity/exposure of bodies’, parts of life’ (P16, cis-gender woman, bisexual). ‘discrimination’, and a ‘lack of knowledge of trans health issues’. Positive experiences were also reported. However, as with any survey Theme 4: Lack of knowledge about transgender health issues of healthcare experiences, it is possible that people experiencing The final theme was a lack of knowledge or misunderstanding of challenges were more likely to respond. The findings also suggest a number of potential ways to improve experiences of physiotherapy transgender-specific health concerns (eg, about hormone therapy, for people identifying as LGBTIQþ. surgeries, etc). For example, P40 (transmasculine, non-binary guy) said he believed that he ‘had to educate them and that they were This study found physiotherapists often made erroneous as- incorrectly blaming health issues on being trans’. Some participants sumptions about LGBTIQþ people’s gender and sexuality. These as- expressed significant distress about physiotherapists’ over-curiosity, sumptions were aligned with the underlying heteronormativity of for example, ‘[it] felt like it had nothing to do with anything and [I] physiotherapy that some participants mentioned, including as- felt they were being invasive about my personal life’ (P34, trans- sumptions that people are heterosexual and conform to a binary masculine, asexual). This was not always the case. One participant gender. This study shows that, like elsewhere in society,30,31 hetero- reported a positive experience: ‘my physio took my trans status in his sexuality and gender normativity are woven into physiotherapy as an stride, and otherwise continued to work with me exactly as he had expected and general norm. Assumptions of gender and/or sexuality previously’ (P21, female). are also pervasive elsewhere in healthcare,32,33 manifesting in a va- riety of ways as ingrained assumptions within healthcare systems, Participant’s level of support for strategies to improve care (Likert including presumptuous use of pronouns, patient forms with male/ scale results) female checkboxes only, and assumptions about the need for contraception.34–36 Unconscious heteronormative assumptions like Participants supported a number of proposed ways to improve those seen in this study unintentionally result in feelings of invisi- LGBTIQþ experiences with physiotherapy. The degree of support for six bility for LGBTIQþ people,37,38 and incorrect use of gender pronouns proposed recommendations is presented in Figure 1. (For the numerical can be distressing for transgender people and other people with non- data used to generate this figure, see Appendix 1 on the eAddenda). The binary gender. Similar to other research findings, participants in this proposed improvements that the participants particularly supported study supported mechanisms to address these issues, including the were: physiotherapists become more aware of health issues specific to use of inclusive intake forms16,38–40 (eg, ‘non-binary’ being an option) LGBTIQþ people (94% ‘like’ or ‘really like’ idea), physiotherapists un- and a willingness to use gender-neutral pronouns (eg, ‘they’ instead dertake LGBTIQþ diversity training (93% ‘like’ or ‘really like’ idea), and of ‘he’ or ‘she’) when appropriate. clinics display images with a range of people of different genders and sexualities (93% ‘like’ or ‘really like’ idea). The idea of displaying a Another key finding is perhaps more specific to the physical ther- rainbow flag (75% ‘like’ or ‘really like’ idea) and use of gender-neutral apies. The proximity and exposure of bodies during consultation language (72% ‘like’ or ‘really like’ idea) were comparatively less well brought up discomfort for some participants. As transgender and non- supported. Some participants expressed ambivalence about displaying binary people often experience judgement about their bodies,41 rainbow flags, stating the importance of physiotherapists having ‘a physiotherapy’s intimate nature (touching, undressing, close observa- good understanding of LGBTIQþ people. I would hate for it to be a token tion of the body) may at times render it uncomfortable. Similar gesture in place of actual support’ (P44, female, bisexual/pansexual). discomfort has been noted in physiotherapy settings with other They also highlighted the complexity of using gender-neutral language, ‘vulnerable’ populations.23 In order to avoid judgement, participants suggesting that some prefer correct use of their gendered pronouns: sometimes concealed their gender or withheld personal/health ‘some transgender people really like it when they are addressed as the

104 Ross and Setchell: LGBTIQ+ people’s experiences of physiotherapy information that may have been important for their well-being. identify as LGBTIQþ can experience challenges when attending Non-disclosure may impact psychological health42 and, when rele- physiotherapy. Some people fear receiving, or have experienced, vant to physiotherapy management, could be associated with poorer discriminatory care, which may have negative consequences, physical health outcomes. Self-care improves when people of diverse including patients withholding information important to their care, gender and/or sexuality feel comfortable to disclose LGBTIQþ or avoiding attending physiotherapy. Like any health professional, health issues to practitioners.43 Consistent with other areas of health- physiotherapists have a responsibility to care for the well-being of all care,19,35,39,44 the current findings recommend diversity training for patients, which includes providing a safe environment to discuss all physiotherapists in LGBTIQþ-specific health issues (particularly trans- relevant aspects of their bodies and lives. Improved education of gender health) to move towards greater trust for physiotherapists and physiotherapists and implementation of participant-suggested their clinics to be considered safe spaces. Although there is a paucity of changes offer ways forward for improving physiotherapy for physiotherapy-specific resources in this area, there have been recent LGBTIQþ people. developments, including an Australian educational video45 and non- academic articles in national physiotherapy publications.46,47 What is already known on this topic: Some people who identify as LGBTIQþ receive suboptimal care in a variety of This study found that many LGBTIQþ people experienced, or health professions, including medicine, nursing, mental health, feared, discrimination related to their sexuality and/or gender iden- and peri-natal care. However, there has been no research to tity while attending physiotherapy. For example, participants re- investigate how people who identify as LGBTIQþ experience ported resistance to use of preferred pronouns. Such discriminatory physiotherapy. behaviour is unacceptable, as both enacted and expected discrimi- What this study adds: People who identify as LGBTIQþ can nation have significant negative effects.48 Fearing discrimination experience challenges when attending physiotherapy, such as: suggests that physiotherapy may feel like a non-accepting environ- physiotherapists making incorrect assumptions about LGBTIQþ ment, but may also be related to previous negative experiences or patients; patients having concerns about revealing some infor- participants’ own anxieties. Although this study is the first to explore mation and receiving discriminatory care; and physiotherapists this in physiotherapy, LGBTIQþ people experiencing discrimination lacking knowledge about transgender health issues. Physiother- has been widely reported elsewhere in healthcare.38,49–53 Despite apists could become more aware of health issues specific to legislative advances in many countries,54 significant gaps remain in LGBTIQþ people and adopt other strategies to make their prac- delivering equitable healthcare for LGBTIQþ people.55 Access to and tice more inclusive of this population. utilisation of healthcare is adversely affected by experiences of discrimination like those seen in the current study.44,56,57 eAddenda: Appendix 1 can be found online at DOI: https://doi. org/10.1016/j.jphys.2019.02.002. The current findings highlight physiotherapists’ lack of under- standing of specific health issues for the LGBTIQþ community, Ethics approval: This study received ethics approval from the The particularly transgender health. A similar lack has frequently been University of Queensland Human Research Ethics Committee reported across healthcare professions.34,58,59 While the complexities (2018000797). of transgender healthcare and requirement for specialist training are acknowledged, there is a pivotal role physiotherapists could play, Competing interests: Nil. including in pelvic health.60 Consistent with other areas of health- Source of support: Nil. care,61 participants in the current study strongly supported training Acknowledgements: The researchers would like to thank the for physiotherapists in issues specific to LGBTIQþ health. A recent participants and the numerous people and organisations who helped systematic review of LGBTIQþ health inclusion in undergraduate and with study recruitment. Thank you to Laetitia Coles for providing professional healthcare training55 highlighted the need for develop- expert review of the analysis. ment of evidence-based curricula covering terminology, stigma and Provenance: Not invited. Peer reviewed. health issues specific to LGBTIQþ people.62–64 Inclusion of these Correspondence: Megan H Ross, School of Health and recommended areas in undergraduate and professional development Rehabilitation Sciences, University of Queensland, Australia; programs offers a way forward to improving physiotherapy experi- Email: [email protected] ences of LGBTIQþ people. References There are a number of methodological considerations when applying these findings beyond the context of this study. The study 1. Australian Bureau of Statistics. General Social Survey; 2014. cat. no. 4159.0. http:// was conducted in Australia, which has relatively progressive laws (eg, www.abs.gov.au/AUSSTATS/[email protected]/Lookup/4159.0MainþFeatures12014?OpenDo same-sex marriage was legalised in 2017) and attitudes towards cument. Accessed 27 July, 2018. people who identify as LGBTIQþ. The findings are likely to be most relevant to similar contexts. Further, the recruitment approach 2. Lea T, de Wit J, Reynolds R. Minority stress in lesbian, gay, and bisexual young (convenience sample) and uncertainty about uptake rate make it adults in Australia: Associations with psychological distress, suicidality, and sub- unclear how common the experiences of physiotherapy reported in stance use. Arch Sex Behav. 2014;43:1571–1578. this study are. For example, the results may disproportionately represent people who are active in online LGBTIQþ communities or 3. Bonvicini KA. LGBT healthcare disparities: What progress have we made? Patient who access LGBTIQþ services, as these were some of the foci of Educ Couns. 2017;100:2357–2361. recruitment strategies. Nevertheless, with the large sample and di- versity of participant characteristics, findings are likely to be fairly 4. Cochran SD, Mays VM, Alegria M, Ortega AN, Takeuchi D. Mental health and representative. It is also possible people who had experienced substance use disorders among Latino and Asian American lesbian, gay, and problems were more likely to respond to the survey. The study tried bisexual adults. J Consult Clin Psychol. 2007;75:785–794. to minimise this effect by using neutral wording in study recruitment material. Finally, as this study focused on physiotherapy patients and 5. Gilman SE, Cochran SD, Mays VM, Hughes M, Ostrow D, Kessler RC. Risk of psy- did not explore the LGBTIQþ identities of physiotherapists, it would chiatric disorders among individuals reporting same-sex sexual partners in the be beneficial to further research how this influences these National Comorbidity Survey. Am J Public Health. 2001;91:933–939. encounters. 6. Koh AS, Ross LK. Mental health issues: a comparison of lesbian, bisexual and Conclusion heterosexual women. J Homosex. 2006;51:33–57. Physiotherapists provide care for diverse people with unique 7. McNeil J, Ellis SJ, Eccles FJR. Suicide in trans populations: A systematic review healthcare needs. This study’s findings highlight that people who of prevalence and correlates. Psychol Sex Orientat Gend Divers. 2017;4:341– 353. 8. Haas AP, Eliason M, Mays VM, Mathy RM, Cochran SD, D’Augelli AR, et al. Suicide and suicide risk in lesbian, gay, bisexual, and transgender populations: review and recommendations. J Homosex. 2011;58:10–51. 9. Blosnich J, Foynes MM, Shipherd JC. Health disparities among sexual minority women veterans. J Womens Health. 2013;22:631–636. 10. Farmer GW, Jabson JM, Bucholz KK, Bowen DJ. A population-based study of car- diovascular disease risk in sexual-minority women. Am J Public Health. 2013;103:1845–1850. 11. Fredriksen-Goldsen KI, Kim HJ, Barkan SE, Muraco A, Hoy-Ellis CP. Health dispar- ities among lesbian, gay, and bisexual older adults: results from a population- based study. Am J Public Health. 2013;103:1802–1809.

Research 105 12. Case P, Austin SB, Hunter DJ, Manson JE, Malspeis S, Willett WC, et al. Sexual 42. Killian ML. The political is personal: relationship recognition policies in the United orientation, health risk factors, and physical functioning in the Nurses’ Health States and their impact on services for LGBT people. J Gay Lesbian Soc Serv. Study II. J Womens Health. 2004;13:1033–1047. 2010;22:9–21. 13. Bolderston A, Ralph S. Improving the health care experiences of lesbian, gay, 43. Greene R, Cox B, Gursky J, Rosendale N, Solomon B, Fox J, et al. Patients who feel bisexual and transgender patients. Radiography. 2016;22:e207–e211. safe to disclose lgbt related issues to their healthcare providers are more motivated to take care of their own health. J Gen Intern Med. 2013;28. S141–S141. 14. Conron KJ, Mimiaga MJ, Landers SJ. A population-based study of sexual orientation identity and gender differences in adult health. Am J Public Health. 2010;100:1953– 44. McCann E, Sharek D. Survey of lesbian, gay, bisexual, and transgender people’s 1960. experiences of mental health services in Ireland. Int J Ment Health Nurs. 2014;23:118–127. 15. Leonard W, Pitts M, Mitchell A, Lyons A, Smith A, Patel S, et al. Private Lives 2: The second national survey of the health and wellbeing of gay, lesbian, bisexual and 45. Dunbabin A, Eibl A, Keng A, Twyerould R. LBGTi know. Paper presented at: transgender (GLBT) Australians. Melbourne: The Australian Research Centre in Sex, Australian Physiotherapy Association Conference 2017. Sydney, Australia. https:// Health & Society, La Trobe University; 2012. www.youtube.com/watch?v=HNaANAAs_ns. Accessed 27 July, 2018. 16. Saulnier CF. Deciding who to see: lesbians discuss their preferences in health and 46. Harman K, Jones G. Creating a safe environment for trans populations in physio- mental health care providers. Soc Work. 2002;47:355–365. therapy. Physiotherapy Practice (Canadian Physiotherapy Association), Diversity in Practice Issue, 7, 2017. https://criticalphysio.net/2017/05/12/creating-a-safe- 17. Irwin L. Homophobia and heterosexism: implications for nursing and nursing environment-for-trans-populations-in-physio. Accessed 27 July, 2018. practice. Aust J Adv Nurs. 2007;25:70. 47. Mitchell M. Breaking the silence on LGBTIQ issues. InMotion (Australian Physio- 18. Shelton K, Delgado-Romero EA. Sexual orientation microaggressions: The experi- therapy Association), February, 40-45, 2018. http://www.printgraphics.net.au/ ence of lesbian, gay, bisexual, and queer clients in psychotherapy. J Couns Psychol. myfiles/InMotion_February_2018/index.html. Accessed 27 July, 2018. 2011;58:210–221. 48. Bockting WO, Miner MH, Swinburne Romine RE, Hamilton A, Coleman E. Stigma, 19. Rondahl G, Bruhner E, Lindhe J. Heteronormative communication with lesbian mental health, and resilience in an online sample of the US transgender popula- families in antenatal care, childbirth and postnatal care. J Adv Nurs. 2009;65:2337– tion. Am J Public Health. 2013;103:943–951. 2344. 49. Sharek DB, McCann E, Sheerin F, Glacken M, Higgins A. Older LGBT people’s ex- 20. Setchell J. What has stigma got to do with physiotherapy? Physiother Can. periences and concerns with healthcare professionals and services in Ireland. Int J 2017;69:1–5. Older People Nurs. 2015;10:230–240. 21. Nicholls DA, Holmes D. Discipline, desire, and transgression in physiotherapy 50. Platt LF, Lenzen AL. Sexual orientation microaggressions and the experience of practice. Physiother Theory Pract. 2012;28:454–465. sexual minorities. J Homosex. 2013;60:1011–1034. 22. Nicholls DA, Cheek J. Physiotherapy and the shadow of prostitution: the Society of 51. Adam J, Elizabeth P. Chronic illness in non-heterosexual contexts: an online survey Trained Masseuses and the massage scandals of 1894. Soc Sci Med. 2006;62:2336– of experiences. Fem Psychol. 2009;19:454–474. 2348. 52. Durso LE, Meyer IH. Patterns and predictors of disclosure of sexual orientation to 23. Setchell J, Watson B, Jones L, Gard M. Weight stigma in physiotherapy practice: healthcare providers among lesbians, gay men, and bisexuals. Sex Res Social Policy. Patient perceptions of interactions with physiotherapists. Man Ther. 2015;20:835– 2013;10:35–42. 841. 53. Chapman R, Wardrop J, Freeman P, Zappia T, Watkins R, Shields L. A descriptive 24. Punch K. Survey research: The basics. London, UK: Sage; 2003. study of the experiences of lesbian, gay and transgender parents accessing health 25. Caelli K, Ray L, Mill J. ‘Clear as mud’: Toward greater clarity in generic qualitative services for their children. J Clin Nurs. 2012;21:1128–1135. research. Int J Qual Methods. 2003;2:1–13. 54. World Health Organization. Addressing the causes of disparities in health service access 26. Braun V, Clarke V. Successful Qualitative Research. London, UK: Sage; 2013. and utilization for lesbian, gay, bisexual and transgender (LGBTQ) persons. Geneva: WHO; 27. Corus C, Saatcioglu B. An intersectionality framework for transformative services 2013. research. Serv Ind J. 2015;35:415–429. 55. McCann E, Brown M. The inclusion of LGBTþ health issues within undergraduate 28. Tong A, Sainsbury P, Craig J. Consolidated criteria for reporting qualitative research healthcare education and professional training programmes: A systematic review. Nurse Educ Today. 2018;64:204–214. (COREQ): a 32-item checklist for interviews and focus groups. Int J Qual Health Care. 2007;19:349–357. 56. Meyer IH. Prejudice, social stress, and mental health in lesbian, gay, and bisexual pop- 29. World Confederation for Physical Therapy. Policy statement: Ethical responsibilities ulations: conceptual issues and research evidence. Psychol Bull. 2003;129:674–697. of physical therapists and WCPT members. London, UK; 2017. http://www.wcpt.org/ ethical-principles. Accessed 5 September, 2018. 57. Bauer GR, Hammond R, Travers R, Kaay M, Hohenadel KM, Boyce M. “I don’t think 30. Röndahl G. Heteronormativity in health care education programs. Nurse Educ this is theoretical; this is our lives”: How erasure impacts health care for trans- Today. 2011;31:345–349. gender people. J Assoc Nurses AIDS Care. 2009;20:348–361. 31. Hird MJ. Sex, gender, and science. New York: Palgrave Macmillan; 2004. 32. Mule NJ, Ross LE, Deeprose B, Jackson BE, Daley A, Travers A, et al. Promoting LGBT 58. Johnson MJ, Nemeth LS. Addressing health disparities of lesbian and bisexual health and wellbeing through inclusive policy development. Int J Equity Health. women: a grounded theory study. Womens Health Issues. 2014;24:635–640. 2009;8:18. 33. Williams SL, Mann AK. Sexual and gender minority health disparities as a social 59. Rounds K, Burns Mcgrath B, Walsh E. Perspectives on provider behaviors: A issue: how stigma and intergroup relations can explain and reduce health dis- qualitative study of sexual and gender minorities regarding quality of care. Con- parities. J Soc Issues. 2017;73:450–461. temp Nurse. 2013;44:99–110. 34. Morrison S, Dinkel S. Heterosexism and health care: a concept analysis. Nurs Forum. 2012;47:123–130. 60. Miciak M, Setchell J, Ossai U, Kadelin Mogard D, Brown D. LGBTIQ issues: Increasing 35. Neville S, Henrickson M. Perceptions of lesbian, gay and bisexual people of primary awareness and action in physical therapy practice. Paper presented at: World healthcare services. J Adv Nurs. 2006;55:407–415. Congress in Physical Therapy 2017; Cape Town, South Africa. https://www.wcpt. 36. Sabin JA, Riskind RG, Nosek BA. Health care providers’ implicit and explicit attitudes org/wcpt2017/DS-08. Accessed 27 July, 2018. toward lesbian women and gay men. Am J Public Health. 2015;105:1831–1841. 37. McGeorge C, Stone Carlson T. Deconstructing heterosexism: becoming an LGB 61. Smith SK, Turell SC. Perceptions of healthcare experiences: relational and affirmative heterosexual couple and family therapist. J Marital Fam Ther. communicative competencies to improve care for LGBT people. J Soc Issues. 2011;37:14–26. 2017;73:637–657. 38. Fish J, Bewley S. Using human rights-based approaches to conceptualise lesbian and bisexual women’s health inequalities. Health Soc Care Community. 62. Cheng L-F, Yang H-C. Learning about gender on campus: an analysis of the hidden 2010;18:355–362. curriculum for medical students. Med Educ. 2015;49:321–331. 39. Rondahl G, Innala S, Carlsson M. Heterosexual assumptions in verbal and non- verbal communication in nursing. J Adv Nurs. 2006;56:373–381. 63. Echezona-Johnson C. Evaluation of lesbian, gay, bisexual, and transgender 40. Sinding C, Barnoff L, Grassau P. Homophobia and heterosexism in cancer care: the knowledge in basic obstetrical nursing education. Nurs Educ Perspect. experiences of lesbians. Can J Nurs Res. 2004;36:170–188. 2017;38:138–142. 41. Weinhardt LS, Stevens P, Xie H, Wesp LM, John SA, Apchemengich I, et al. Trans- gender and gender nonconforming youths’ public facilities use and psychological 64. Sekoni AO, Gale NK, Manga-Atangana B, Bhadhuri A, Jolly K. The effects of well-being: a mixed-method study. Transgend Health. 2017;2:140–150. educational curricula and training on LGBT-specific health issues for healthcare students and professionals: a mixed-method systematic review. J Int AIDS Soc. 2017;20:21624. 65. Anti-Discrimination Commission Queensland. LGBTI terminology. 25 September 2018. https://www.adcq.qld.gov.au/resources/lgbti-people/lgbti-terminology. Accessed 27 July, 2018. 66. Human Rights Campaign. Glossary of Terms; 2018. https://www.hrc.org/resources/ glossary-of-terms. Accessed 25 September, 2018.

Journal of Physiotherapy 65 (2019) 63–64 j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j p hy s Editorial Program evaluation within the research translation framework Natasha K Brusco a,b,c, Helena C Frawley b,c a Alpha Crucis Group; b Centre for Allied Health Research and Education, Cabrini Health; c School of Primary and Allied Health Care, Monash University, Melbourne, Australia A program evaluation is the systematic collection and analysis of unintended. The summative evaluation can occur at completion of data designed to examine the value of a program in terms of the project or substantially after its implementation.3 To capture both implementation, efficiency and effectiveness.1–3 In healthcare, a the process and summative evaluations, an established tool may be program evaluation can demonstrate improved patient and economic used, such as the Standards for Reporting Implementation Studies: outcomes or, if an improvement is not shown, the evaluation can StaRI checklist,13 to evaluate both the implementation strategy and potentially minimise avoidable waste for the health service by program effectiveness. reducing the provision of low-value care practices.4 The evaluation may also reduce avoidable waste in research by improving the rate of Core to the methodology that underpins the formative, process successful replication of basic research and translation to healthcare.5 and summative evaluation are three corresponding elements. First, Program evaluation provides a bridge between research and clinical there is the development of a theory of change to conceptualise how practice. The purpose of this editorial is to highlight that program change will occur. Also, a logic model captures the inputs (resources), evaluation lies within the research translation framework, as well as activities, outputs and outcomes.9 In addition, there is the to discuss the key elements of program evaluation methodology and development of primary and secondary evaluation questions, as they give examples for appropriate application of a program evaluation. relate to recommended implementation outcomes.12 Program evaluation is a part of the research translation framework Having reviewed what program evaluation is, the next step is to in healthcare, similar to many quality improvement activities that understand when to undertake a program evaluation. The impetus to aim to test implementation of research in a local setting before complete a program evaluation may include: the development of a widespread uptake. Program evaluation can be considered as a new program (especially when funding is contingent upon the knowledge transfer step within the phases of research translation; it completion of a program evaluation); awareness of new evidence; is reflected in the third stage of research translation, known as T3 concerns about the effect or cost of an established program; plans to (from guidelines to health practice), and in the fourth stage of scale-up an established program; and as a reaction to an external research translation, known as T4 (from health practice to population threat to a service or program. The latter impetus may include the health outcomes).6 Program evaluation includes both knowledge threat of funding cuts or amalgamation of services, and in this case a transfer (closing the ‘know-do’ gap between what is known and what program evaluation may be an attempt to prove or shore up is done)7 and implementation science (‘how to’ implement the perceptions of value. A program evaluation is particularly relevant knowledge).8 when the more common pre-intervention and post-intervention cohort or observational study designs are not appropriate. Program evaluation has three distinct components: formative, process and summative evaluations.3,9,10 The formative phase is In order to maximise the value of the program evaluation findings designed to help shape the evaluation by assessing the program design and reduce research waste via successful dissemination strategies, and piloting it prior to implementation across the health service. For the clinical team may benefit from having a researcher as a member example, the formative evaluation may include: a gap analysis to of the team. The researcher can provide valuable input from inception report on current state compared to evidence-based clinical guide- of the evaluation through to dissemination of findings, and provide lines; a needs assessment to report the perceived want or need from the necessary link with the human research ethics regulatory au- those influencing or impacted by the program; and an analysis of thority/institutional review board. Another important consideration barriers and facilitators to inform the implementation strategy. for program evaluation is the identification and involvement of stakeholders.14 The inclusion of stakeholders can serve to enhance an Process evaluation is designed to determine the extent to which a evaluation and its findings; however, challenges with obtaining and program was implemented according to plan. Should a program fail, it sustaining stakeholder engagement as well as the risk of equity and is imperative to distinguish between a defective program and a bias from stakeholder involvement have been identified and require defective implementation strategy. For example, the process evaluation careful consideration when undertaking a program evaluation.15 outcome may simply include measured adherence to a published implementation framework using a single metric such as that for In Victoria, Australia, through the state government’s Department stroke guidelines11 or more comprehensively consider success/lack of of Health and Human Services, there are opportunities for clinicians success to achieve the eight recommended implementation outcomes, to seek funding to undertake healthcare improvement initiatives including: acceptability, adoption, appropriateness, feasibility, fidelity, evaluated via a program evaluation. For example the ‘Better Care implementation cost, penetration, and sustainability.12 Victoria Innovation Fund’ targets the T3 phase of research translation (from guidelines to health practice) and the ‘Safer Care Victoria The summative evaluation may be more familiar to traditional Scaling Collaborations’ targets the T4 phase of translation (from researchers and clinicians, as this is an assessment of the overall health practice to population health outcomes).16 Both of these impact of the program based on the program objectives and includes funding schemes require a program evaluation. It is likely that similar analysis of the health and economic outcomes, both intended and schemes and requirements exist in other jurisdictions. https://doi.org/10.1016/j.jphys.2019.02.010 1836-9553/© 2019 Australian Physiotherapy Association. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).

64 Editorial There is a plethora of literature that describes the program Acknowledgements: Nil. Group, evaluation framework1–3,9,17,18 as well as published studies relating to Provenance: Invited. Peer reviewed. physiotherapy that report using a program evaluation.19–22 However, Correspondence: Natasha K Brusco, Alpha Crucis few of the published studies actually used a program evaluation Melbourne, Australia. Email: [email protected] definition and framework,19,22 and instead reported the findings using other study designs such as an observational study.20,21 This References indicates that while the term program evaluation is commonly used, many studies are not reported according to the framework. It is 1. Owen J. Program evaluation: forms and approaches. third edition. New York: unknown how often a program evaluation framework is being Guilford Press; 2007. utilised in the healthcare setting to evaluate a new or established program, implement new evidence, or scale-up an established 2. Chen HT. Practical program evaluation. London: Sage; 2014. program due to a likely gap in dissemination of the program 3. New South Wales Agency for Clinical Innovation. Understanding Program evaluation findings beyond the health service. This has the potential to impact widespread implementation and uptake. Evaluation: An ACI Framework; 2013. www.aci.health.nsw.gov.au. 4. Levinson W, et al. BMJ Qual Saf. 2015;24:167–174. Understanding how to bring a more theoretical analytic 5. Chalmers I, et al. Lancet. 2014;383:156–165. perspective to program evaluation would improve its acceptance 6. Khoury MJ, et al. Am J Epidemiol. 2010;172:517–524. within the research translation framework in healthcare. Program 7. World Health Organisation. Knowledge translation; 2018. http://www.who.int/ evaluation methodology is at risk of bias from poor content-related validity, low inter-rater reliability, and self-reported bias.23,24 ageing/projects/knowledge_translation/en/. Accessed August, 2018. However, the last decade has seen significant development in the 8. Khalil H. Int J Evid-Based Healthc. 2016;14:39–40. field of program evaluation, including the development of program 9. Department of Health and Human Services. Evaluation Guide: Centre for evaluation reporting guidelines10,25 to strengthen program evaluation design and ensure systematic reporting of results.26 Evaluation and Research. Victorian Government, 1 Treasury Place, Melbourne; 2017. https://intranet.dhhs.vic.gov.au/evaluation-and-research-support. From our experience in public health, private health and 10. Chacón Moscoso S, et al. Int J Clin Health Psychol. 2013;13:58–66. government sectors involved in research and clinician-led program 11. National Stroke Foundation. Implementing the Clinical Guidelines for Stroke evaluation, our perspective is that there has been a recent and welcome Management; 2011. www.strokefoundation.com.au. shift to integrate program evaluation into research design. With this has 12. Proctor E, et al. Adm Policy Ment Health. 2011;38:65–76. come a collaboration between researchers and clinicians to work 13. Pinnock H, et al. BMJ. 2017;356:i6795. together on the design, evaluation and dissemination phases of 14. Johnson K, et al. Am J Eval. 2009;30:377–410. research activity. We applaud this move towards recognising the value 15. Brandon PR, Fukunaga LL. Am J Eval. 2014;35:26–44. of program evaluation and the bridging of research and clinical practice. 16. Department of Health and Human Services. Better Care Victoria funded innovation projects; 2018. https://www.bettercare.vic.gov.au/innovation-fund. Accessed August, Ethics approval: N/A. 2018. Competing interests: The authors declare that they have no 17. Newcomer KE, et al. Handbook of practical program evaluation. Hoboken, USA: John competing interests. Wiley & Sons; 2015. Source(s) of support: Nil. 18. Fitzpatrick JL, et al. Program evaluation: Alternative approaches and practical guidelines. Pearson; 2004. 19. McAuley C, et al. Physiother Can. 2014;66:274–285. 20. Solomon P, Salfi J. Educ Health. 2011;24:616. 21. Dufour SP, et al. Physiother Theory Pract. 2015;31:29–38. 22. Bajnok I, et al. Contemp Nurse. 2012;42:76–89. 23. Brandon PR, Singh JM. Am J Eval. 2009;30:123–157. 24. Peck LR, et al. Am J Eval. 2012;33:350–365. 25. Portell M, et al. Psicothema. 2015;27:283–289. 26. Labin SN, et al. Am J Eval. 2012;33:307–338.

Journal of Physiotherapy 65 (2019) 113–116 Appraisal j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j p hy s Research Note: Adaptive trials The need for innovative clinical trials size or the ratio of allocation to treatment arms, although more complex adaptations are possible (Box 1). The essential feature is that The number of clinical trials published each year shows no sign of the rules for all adaptations must be specified in advance in the reaching a plateau,1 despite the cost of conducting a clinical trial protocol, and adaptations only occur if a threshold for change has rising five-fold over the past decade.2 Randomised clinical trials been met due to accruing trial data. The design of these studies (RCTs) can have a profound and immediate impact on health policy typically requires extensive simulations of potential trial outcomes to and clinical practice, but while activity and quality have increased, demonstrate acceptable trial operating characteristics such as the only around half of all trials are published.3 Even when RCT evidence type I (false-positive) error rate. exists, there may be a failure to adopt this as best practice if the results are considered non-applicable to the specific patient at hand, There are various types of adaptations that can be implemented in a or where there is mismanagement of commercial and academic randomised trial. These include: flexible sample size and trial duration, interests,4 or other biases in the design, management, or reporting.3 response adaptive randomisation with multiple treatment options Clinicians are often faced with the need to make treatment and/or doses, adding or dropping of treatments, subgroup evaluations decisions across a range of comorbidities, whereas evidence from and enrichment, seamless phase 2/3 designs, and adaptive endpoint RCTs is predominantly presented for a single medical indication selection.9,10 These adaptations can improve trial efficiency and reduce and frequently assumes homogeneity in participant responses. A the risk of failed or inconclusive results.11 Despite the increase in the one-size-fits-all approach to clinical trials is not in harmony with the number of reports of adaptive trials, adopting these innovative designs heterogeneity and complexity of modern diseases. While the pipeline in publicly-funded clinical trials has been slow. More investment is of new treatments is ever-increasing, the capacity to formally needed to further develop the statistical methods, software tools, and evaluate new treatments in RCTs is diminishing, due to the large guidelines for successful implementation. inherent costs, overburdened healthcare system and low participation rates.5 Almost half of publicly funded trials do not meet One reason for the slow adoption of innovative trial designs is that recruitment targets,6 increasing the costs and the risk of inconclusive existing competitive mechanisms for publicly funded trials often do results. not have the flexibility or the timeframe needed to support the necessary planning and simulations required at the design stage, or Although innovative solutions, such as adaptive designs, exist to the finances for ongoing statistical support for repeat interim improve trial efficiency, trialists rarely invest in the time and resources analyses and trial modifications. Guidelines are needed to aid to simulate a broad range of enrolment and outcome scenarios; this is trialists through the logistics of an adaptive design, and the structure, predominantly due to a lack of statistical expertise in trial simulation role, and conduct of Data and Safety Monitoring Boards.10 and the dearth of simple statistical tools with which to perform this Vandemeulebroecke established a framework for the discussion of function. Thus, most publicly funded clinical trials have simple adaptive designs based on five main points: feasibility, validity, parallel group designs (around 90% in the UK) that do not have inbuilt integrity, efficiency, and flexibility.12 flexibility to adapt to accruing trial evidence.7 A single conservative stopping boundary applied after 75% of target recruitment could The relative disadvantages of adaptive trials have been well correctly allow early trial termination for futility in up to 30% of trials,7 documented.10,13,14 Modifications in an ongoing trial, particularly saving participants from exposure to futile treatments and decreasing those related to the ratio of allocation to treatment arms, have raised trial costs. Sub-optimal and deleterious treatments continue to be some concerns about the potential to partially unblind or change the used, and RCTs are increasingly the preserve of large well-resourced type of participants recruited to the study over time. Therefore, pharmaceutical companies. Healthcare-embedded approaches that additional safeguards are necessary to prevent leakage of information simultaneously evaluate and implement the best treatment and maintain the integrity of the trial outside the sphere of option(s) might provide timely outcomes for participants that confidentiality of the Data and Safety Monitoring Board. are resource-efficient, cost-effective, minimally biased, and accessible to all. Trends in patient populations and in the underlying response to treatment are major issues for all trials. Adaptive designs can be Adaptive trials robust to moderate population trends;13 however, less is known about their sensitivity to time-dependent treatment responses and The United States of America’s Food and Drug Administration how accruing information, internal and external to the trial, may be (FDA) defines an adaptive trial as ‘a clinical trial design that allows for integrated to optimise trial operating characteristics. Many journal prospectively planned modifications to one or more aspects of the articles have provided overviews of adaptive trials, predominantly for design based on accumulating data from subjects in the trial.’8 The drug development and most recently by Bhatt et al,15 Pallmann et al,16 planned modifications are often defined as thresholds for early and Thorland et al.17 However, an important and, until now, largely stopping due to anticipated trial success or failure. Adaptive methods neglected potential for these innovative designs lies within generally result in changes to the trial parameters, such as the sample comparative effectiveness trials. Examples of adaptive comparative effective trials include ESETT for status epilepticus,18 PREPARE ALICE for influenza,19 and REMAP-CAP for community-acquired pneumonia in patients admitted to intensive care units.20 https://doi.org/10.1016/j.jphys.2019.02.004 1836-9553/© 2019 Published by Elsevier B.V. on behalf of Australian Physiotherapy Association. This is an open access article under the CC BY-NC-ND license (http:// creativecommons.org/licenses/by-nc-nd/4.0/).

114 Appraisal Research Note Box 1. Common adaptive trial designs (may be used in combination). 1. Sample size re-estimation Adjustments to sample size based on interim parameter estimates such as treatment effect and variance, number lost to follow-up, etc. 2. Continual assessment (dose-finding) Balances the need to collect sufficient information on each dose, with early stopping rules for treatment arms with poor performance or toxicity. 3. Response-adaptive randomisation Adjustments to treatment allocation ratios dependent on optimising participant outcome; often proportional to the interim performance of each treatment arm. 4. Population enrichment Adjustments to trial eligibility criteria to increase recruitment into subgroups that respond well to treatment at interim assessment. All adaptive designs need to minimise the potential risks associated with: (i) treatment effect estimates being less unreliable in early interim analyses, (ii) leakage of ongoing trial information from multiple interim analyses, (iii) investigator behaviour, participant response or underlying disease changing over time. Response adaptive randomisation drift’) of an important magnitude (ie, a change . 25% in the proba- bility of treatment outcome) appear to seriously inflate the risk of a Adaptive trials can gain substantial efficiency from response false-positive trial outcome.23 Response adaptive randomisation adaptive randomisation (RAR). This is the process where participants designs have been shown to be more efficient than multi-arm are allocated to treatment arms dependent on preceding participant multi-stage (MAMS) designs when there is a superior treatment, outcomes (Box 2). It has the ethical advantage of assigning fewer whereas MAMS designs are slightly more efficient than RAR designs participants to poorly performing treatments; however, it can only be when none of the treatments are effective.24 implemented when the treatment outcome (or a surrogate) is available within a timely fashion relative to the accrual time required Multi-arm-multi-stage and Bayesian adaptive platform trials for the entire trial. Concerns have been raised over reductions in statistical power in RAR designs to detect a difference between Both MAMS and Bayesian platform trials are concerned with any two treatment arms (when computed using traditional optimising outcomes for a condition rather than focused on any hypothesis testing procedures) and the deterministic nature of particular therapy.25,26 The key features of a MAMS trial are the response-adaptive algorithms.21,22 Solutions have been proposed for simultaneous comparison of several different treatment options multi-arm designs that address these issues by: (i) fixing the against a single control arm, a weighted randomisation ratio to ensure allocation to the control group; (ii) using equal allocation to allocation is discontinued to poorly performing treatment arms, and a treatment arms until a predefined trial or subgroup size is reached; frequentist statistical inference and flexible sample size.25 In 2012, the (iii) defining a minimum probability of allocation to each treatment results for the first MAMS trial – STAMPEDE – an open-label, five-stage, arm; (iv) applying probabilistic algorithms that apply to blocks of six-arm trial in prostate cancer,27 were published in Lancet Oncology, participants; and (v) using post-trial simulations to quantify the bias and identified celecoxib as an inferior option for men with and provide corrections.22 Only time trends in the data (‘patient hormone-sensitive and advanced prostate cancer. Box 2. Representation of response adaptive randomisation in an over-simplified adaptive design. A total of 30 individuals are recruited over time. After the first 15 individuals have outcome data, the first interim analysis is performed. Treatment A has predominantly poor outcomes (blue dots) and meets the threshold for futility after the first interim analysis (n = 15); therefore, allocation to this arm is stopped. Treatment B has an equal number of good and poor outcomes and it meets the threshold for futility at the second interim analysis (n = 30). Therefore, allocation to Treatment B is also stopped. This triggers the final analysis.

Appraisal Research Note 115 The key features of a Bayesian adaptive platform trial are the use represented as a probability, whereas in frequentist analyses, the of Bayesian statistical inference and comparison of several different level of certainty in hypothetical frequencies of data patterns is treatment options with innovative adaptive features. Typical represented as a probability.33 Bayesian inference does this by adaptations include a flexible sample size, different domains of combining the likelihood of the observation for the range of possible treatment simultaneously evaluated and in combination (each treatment differences with the prior probability (ie, prior to data domain with multiple treatment options), response adaptive collection) of those possible treatment differences.34 Bayesian randomisation, and evaluation of treatment responses in different inference provides a straightforward mechanism for updating the subgroups of participants.26 In 2016, the results for the signature estimates of the most probable range of treatment differences as the Bayesian platform trial I-SPY 2, a phase 2 trial of neoadjuvant therapy data accrue.18 Trial designs that are adaptive can unfold in many in breast cancer for subgroups of participants defined by genetic different permutations, depending on the accruing data, so it is hard signatures, were published and identified better options for women to estimate how ‘unlikely’ a particular set of results are, making with triple negative and HER21 breast cancer. 28,29 frequentist strategies particularly challenging (though not impossible). Therefore, the ability to update the probabilities for a Platform trials require extensive consultation between clinicians, range of possible treatment differences as new data accrue makes consumers, trialists and statisticians in the planning stage, to set the Bayesian inference very useful for adaptive studies. research priorities, and extensive, statistically complex, computer simulations in the design stage. The workload for the trial statistician is The ability to generate trial data using simulation is an essential considerable during the execution of a trial, which includes multiple skill when designing adaptive trials. Simulations may be used to interim analyses and potential additional trial simulations to assess design the trial, such as determining the timing and thresholds for planned adaptations. Globally, demand for these statistical skills are a repeat analyses, or estimating trial parameters, such as the sample major area of unmet need. These trials also require standardised size, or to explore the potential extent of bias in the estimate of eligibility criteria, trial endpoints and subgroup definitions, which are treatment effect. For frequentist inference they may also be used to documented in a master protocol that is designed to answer multiple explore whether the confidence intervals have the correct research questions.30 This ensures that treatment responses can be coverage. Some key parameters to consider when performing trial meaningfully aggregated across domains, across sites and over time. simulations are accrual rate, potential distribution of response in The master protocol sets out exactly what data are to be collected, the various treatment arms, number and timing of sequential including the primary and secondary endpoints, and the procedures (interim) analyses, and thresholds for treatment and trial success around how data are captured and managed, including the trial and futility. In addition, the statistician needs to program an governance and safety monitoring arrangements. The trial procedures extensive range (often between 50 and 100) of plausible and less and analyses for each therapeutic domain are documented in separate plausible trial outcome scenarios, identified by the trial team, to appendices. This modular structure allows for the domain-specific evaluate the trial operating characteristics. Depending on the appendices to be modified over time without changing the master complexity of the proposed adaptive trial and computer hardware, protocol; treatment options within a domain can be added or removed running a simulation for a single scenario may take anywhere according to pre-specified adaptation rules, and entire domains can be between a few minutes and several days. It is recommended that added or removed. Investment in the development of a master protocol anyone embarking on a program of adaptive trials starts with across multiple sites can provide a platform for the ready identification simple adaptive designs and builds up experience in the area over of potentially eligible study participants. time, or engages an experienced statistical team to support the program. Digital support for adaptive trials This Research Note has explored the strengths, risks and potential Data management is more challenging for adaptive trials complexity of adaptive trials. The logistical components surrounding compared to conventional trials, due to the need for timely capture of the implementation of adaptive trials require a highly integrated, accurate data for the frequently planned analyses. Participant multidisciplinary team encompassing clinicians, statisticians, trialists, samples may need to be laboratory analysed and the primary consumers, data managers and computer scientists. The challenges endpoint data made electronically available on an ongoing basis are considerable but adaptive trials have been demonstrated to deliver rather than stored and analysed at the end of the trial. Ideally, logic on the promise of simultaneously evaluating and implementing the checks need to be automated at the point of data entry and collection best treatment option(s) to address the complexity and science of tools need to be designed to minimise data entry errors. In common modern diseases. with conventional trials, high standards of data security and privacy must be ensured. The major components of a digital solution are data Competing interests: Nil. capture tools, a central or federated database, a treatment allocation Sources of support: For the manuscript production: JM is sup- tool (that can be updated if RAR is required), an analysis engine (to ported by a Telethon-Perth Children’s Hospital Research Fund streamline frequent planned analyses), and a decision engine. Digital research capacity building grant (CRIPTIC); AS holds an NHMRC TRIP solutions that manage the workflow may also be customised to Fellowship; BS and SB are supported by a Channel 7 Telethon Trust facilitate secure data sharing and provide interfaces to engage with Major Beneficiary grant; TS holds an NHMRC Career Development participants, clinicians and study co-ordinators. Lessons learnt from Fellowship (no 1111657) and a Raine Clinical Research Fellowship. the successful deployment of clinical registries offer insights into the Acknowledgements: Nil. types of digital infrastructure required to implement platform trials, Provenance: Invited. Not peer reviewed. such as online participant registration and consent, and automated Correspondence: Dr Julie A Marsh, Telethon Kids Institute, Perth, notifications and data entry for self-assessments.31,32 Finally, patient Australia. Email: [email protected] registry platforms can have multi-lingual support. Julie A Marsha,b, André Schultza,c,d, Benjamin R Savillee,f Bayesian inference and trial simulation Scott M Berrye and Thomas L Snellinga,g,h,i At the heart of an adaptive trial is a decision-making process that aWesfarmers Centre of Vaccines & Infectious Diseases, Telethon Kids occurs in light of the accruing data. Whilst this Research Note is not the place for a full discussion on the use of frequentist or Bayesian Institute, Perth, Australia inference in adaptive trials, it is worth noting the difference in bSchool of Population & Global Health, University of Western Australia, interpretation of the results between these two approaches. In Bayesian analyses, the level of certainty in the hypothesis is Perth, Australia cFaculty of Health and Medical Sciences, University of Western Australia Medical School, Crawley, Perth, Australia dDepartment of Respiratory Medicine, Perth Children’s Hospital, Perth, Australia

116 Appraisal Research Note eBerry Consultants, Austin, TX, USA 10. Kairalla JA, et al. Trials. 2012;13:145. fVanderbilt University Department of Biostatistics, Nashville, TN, USA 11. Saville BR, Berry SM. Clin Trials. 2016;13:358–366. 12. Vandemeulebroeke M. Biometrical J. 2008;50:541–557. gSchool of Public Health, Curtin University, Bentley, Perth, Australia 13. Coffey CS, Kairalla JA. Drugs R D. 2008;9:229–242. hDepartment of Infectious Diseases, Perth Children’s Hospital, Nedlands, 14. Berry D. J Clin Onc. 2010;29:606–609. 15. Bhatt DL, Mehta C. N Engl J Med. 2016;375:65–74. Perth, Australia 16. Pallmann P, et al. BMC Med. 2018;16:29. iMenzies School of Health Research and Charles Darwin University, 17. Thorlund K, et al. BMJ. 2018;360:k698. 18. Connor T, et al. J Clin Epidemiol. 2013;66:S130–S137. Darwin, Australia 19. https://www.phc.ox.ac.uk/phctrials/trial-portfolio/alic4e. (accessed 4/2/2019). 20. https://prepare.ersnet.org/lrmedia/2017/pdf/425.pdf. (accessed 4/2/2019). References 21. Hu F, Rosenberger W. J Am Statist Assoc. 2003;98:671–678. 22. Villars SS, Rosenberger WF. Biometrics. 2018;74:49–57. 1. Bastian H, et al. PLoS Med. 2010;7:e1000326. 23. Villars SS, et al. Pharm Stat. 2017;17:1–28. 2. Collier R. CMAJ. 2009;180:277–278. 24. Wason JM, Trippa L. Stat Med. 2014;33:2206–2221. 3. Heneghan C, et al. BMJ Evid Based Med. 2017;22:120–122. 25. Sydes MR, et al. Trials. 2009;10:39. 4. Bekelman JE, et al. JAMA. 2003;289:454–465. 26. Berry SM, et al. JAMA. 2015;313:1619–1620. 5. McDonald AM, et al. Trials. 2006;7:9. 27. James ND, et al. Lancet Oncol. 2012;13:549–558. 6. Sully BGO, et al. Trials. 2013;14:166. 28. Rugo HS, et al. N Engl J Med. 2016;375:23–34. 7. Sully BGO, et al. Trials. 2014;15:61. 29. Park JW, et al. N Engl J Med. 2016;375:11–22. 8. Adaptive Designs for Clinical Trials of Drugs and Biologics Guidance for Industry DRAFT 30. Woodcock J, LaVange LM. N Engl J Med. 2017;377:62–70. 31. Bellgard M, et al. Source Code Biol Med. 2013;8:21. GUIDANCE; September 2018. https://www.fda.gov/downloads/drugs/guidances/ucm 32. Lacaze P, et al. Intern Med J. 2017;47:1075–1079. 201790.pdf. (accessed 4/2/2019). 33. Greenland S, et al. Eur J Epidemiol. 2016;31:337. 9. Berry SM, et al. Bayesian Adaptive Methods for Clinical Trials. Florida, USA: CRC 34. Dmitrienko A, Wang MD. Stat Med. 2006;25:2178–2195. Press; 2010.