The “Trans Health Research Project” is a hospital and web-based cross-sectional survey that was built with input from the medical and TGD communities to assess the healthcare needs of and access barriers for transgender and gender diverse people (TGD) residents in two Brazilian states. It is an evidence-informed, policy-making initiative of the Universidade Federal do Rio Grande do Sul (UFRGS) in tandem with the Universidade de São Paulo (USP).

The project was submitted and approved by the Ethical Committee and Research Commission of Hospital de Clínicas de Porto Alegre. In addition, in the Ethical Committee of Hospital de Clínicas of Medicine School of Universidade de São Paulo and in the Research Commission of Ethical Committee of Universidade Federal do Rio Grande do Sul Psychology Institute. Participants were informed about the objectives and functioning of the research and a free and informed consent term was provided. Confidentiality and anonymity were assured; moreover, the participations were voluntary with the possibility to withdraw at any time. It was also informed the inexistence of direct benefits for participating in the survey, clarifying that it would not influence healthcare provision, in case of the participants that completed the questionnaire in hospitals.

Data were collected in two Brazilian states that pioneered the provision of specialized services for TGD since the Brazilian health policies for TGD were first implemented in Rio Grande do Sul, Brazil’s southernmost state, and São Paulo, a state in the southeast region. Both states have gender identity programs that provide GCS at university hospitals. Since the Brazilian Unified Health System (SUS) provides georeferenced care, patients seeking gender affirmation must access those procedures in the states in which they live. Patients in the two programs were voluntary invited to complete an electronic version of the survey. The questionnaire was also available on the internet through an online Facebook announcement during two time periods: July–October 2014 and January–March 2015. The announcement was displayed for users of Facebook who indicated the following characteristics in their profiles: lived in the states of São Paulo and Rio Grande do Sul; were 18 years or older; and “liked” pages on Facebook and joined groups or events related to transsexuality, TGD individuals and the LGBT movement. Facebook statistics indicated that the ad had 521,601 impressions (number of times the ad was shown on the site), with 7,226 ad clicks. The click rate indicated that 0.72% of the impressions drove individuals to the main page of the study. IP number of computers was registered in order to withdraw participants that completed the survey more than once.

The survey was modeled after the TransPULSE project. TransPULSE was one of the first large-scale studies that addressed the health needs and vulnerabilities of TGD and their barriers to healthcare access. The project aimed to improve the quality of life of TGD in Ontario, Canada by measuring levels of social exclusion and their impact on physical and mental health. The procedure for cross-cultural adaptation of the instrument for Brazilian populations was based on Borsa et al. (2012), according to the following steps: (1) contextual equivalence and review by expert committee; (2) translation; (3) evaluation by the target audience; and (4) evaluation by the original authors of the instrument. For this study, the survey was adapted for use in the Brazilian population by a group of health practitioners who worked in gender and sexuality diversity and who were assessed by Brazilian TGD community members.

The Brazilian survey was composed of 122 items grouped into 11 categories: 1- demographic characteristics; 2- motherhood/fatherhood; 3- physical health; 4- use of sexual hormones; 5- gender reaffirming procedures; 6- sexually transmitted infections and HIV; 7- experiences of discrimination in healthcare (violence experience, suicidal attempts), work and educational contexts; 8- dwelling; 9- income; 10- abusive use of alcohol and other drugs; and 11- sexual and mental health.

Gender identity was assessed using the two questions method (one related to the gender designated at birth and another asking how the participants describe their gender identity), and people were considered eligible for participation if they reported a gender different from that assigned to them at birth (Bauer et al., 2017). The organized Brazilian social movement prefers the use of the terms travesti, transsexual and trans person (man or woman) to the anglophone umbrella term transgender. Travesti is a culturally specific gender identity term for Brazilians who were designated as male at birth but affirm a female gender performance and bodily form, although they typically do not undergo neovaginoplasty. Their gender identity varies: most identify as male, some identify as women, and others simply identify as travesti. Based on their self-reported gender identity, participants were re-categorized as transgender women, transgender men or gender non-conforming people. Transgender women were those who were designated as male at birth but identified as women, transgender women or travestis. Transgender men were those who were designated as female at birth but identified as men or transgender men. Finally, gender non-conforming persons were those who identified with a gender identity outside the binaries (male–female), such as queer, non-binary, a-gender, etc.

The scale of support for transsexual identity was developed for the TransPULSE project (Bauer, 2012). The scale assesses the extent to which friends, family, and colleagues support the transsexual identity of the respondent. The answers were evaluated on a 4-point scale with the following response options: “they do not support at all,” “they do not support much,”, “they support a little,” “they strongly support,” and “not applicable.” The final score was determined by the number of answers indicating support: lower scores or scores up to 1 are classified as having very little or no support, scores between 1 and 2 are classified as having little support, scores between 2 and 3 are classified as having some support, and scores with values greater than 3 are classified as having strong support. Cronbach’s alpha value was 0.83.

Family support was calculated on a 3-point scale with the following response options: 1 or “no support,” 2 or “support a little,” and 3 or “support or support a lot.” In addition, participants could mark that they had no mother or father. Cronbach’s alpha for this sample considering the overall scale was 0.90. For this study, only support related from father/mother was considered. Furthermore, participants were asked the following question: “how often do you need to move away from your family or friends for being TGD?”. The answer choices were rated on a 5-point Likert scale ranging from 1 (never) to 5 (always). Finally, they were asked the following question: “do you already live without fixed housing?” If the respondent answered yes, they were asked about the context of their home.

The Brazilian version (Avanci et al., 2007) of the Rosenberg (1965) Self-Esteem Scale was used to evaluate the respondent’s self-esteem. This scale contains ten affirmations concerning high or low self-esteem rated on a 4-point Likert scale with the following response options: “I totally agree,” “I agree,” “I disagree,” and “I strongly disagree.” Higher scores on this scale indicate higher levels of self-esteem. Cronbach’s alpha for this sample was 0.86.

Seven hundred and one TGD participants completed the survey, and 423 met the inclusion criteria for this study. The average age of the participants was 27.06 years (95% CI [26.22, 27.90 years]; SD = 8.74 years; median 24 years), and the age range of the participants was 18–61 years. Additional sociodemographic data are presented in Table 1.

Central tendency and frequency statistics were calculated for the survey variables analyzed in this article. Relationships between outcomes and contextual variables were calculated using χ2 tests with the phi (φ) coefficient to determine the magnitude of the effects. The differences were considered to be significant when the p-value was smaller than 0.05, the 95% confidence interval did not include 0 and the effect size was larger than 0.15, which denote medium to strong effects of exogenous latent variables according to the general (Cohen, 1988).

Structural equation modeling (SEM) was used to investigate the fit of the theoretical model considering the associations between the stages of the gender affirmation procedure, the level of family support and self-esteem, the need to move out for being TGD and the occurrence of homelessness. The tested model included the following variables and associations: (1) steps on whether the process of gender affirmation would predict family support and a TGD individuals’ self-esteem; (2) whether a TGD individuals’ low self-esteem would correlate with their willingness to leave home and whether lack of family support would predict their willingness to leave home; and (3) whether willingness to leave home would predict homelessness. The correlation matrix and the Variance Inflation Factor (VIF) values among all variables in the model were calculated in order to test the required assumption that there was no multicollinearity impairing the SEM analyses.

Of the total of 421 respondents in relation to a parents’ support for one’s TGD identity, 29.45% (n = 124) of the sample answered that their parents did not at all support their status, 20.43% (n = 86) answered that their parents supported their status only a little, 20.43% (n = 86) answered that their parents supported them, 24.23% (n = 102) answered that their parents supported them completely, and 5.46% (n = 23) answered that they did not have a father and/or mother.

Welch’s adjusted F ratio was used in the one-way ANOVA, which demonstrated statistically significant differences between the groups: Welch’s F(2,82.62) = 9.06, p < 0.001, ω² = 0.04. Additionally, the Games-Howell post hoc test indicated that gender-diverse persons had lower parental support than transgender women (Δ = -0.62, 95% CI [-0.99, -0.25], p = 0.001) and transgender men (Δ = -0.42, 95% CI [-0.81, -0.03], p < 0.05). Nevertheless, transgender men and transgender women did not differ from each other regarding parental support.

There was also a significant difference in parental support regarding the medical gender affirmation status F(3,394) = 4.83, p < 0.05, η² = 0.04. Participants who had already completed the gender affirmation procedures did not differ from those who were in the process of completing them. However, those who had already performed the procedures had greater parental support than those who planned to complete the process (Δ = 0.47, 95% CI [0.11, 0.83], p < 0.05) and those who were in doubt or would not complete the process (Δ = 0.49, 95% CI [0.04, 0.93], p < 0.05). Finally, those who were in doubt or would not complete the process did not differ in support relative to those who did plan to complete the process (Table 2).

Considering only those who did not receive any support (M = 14.54, SD = 6.05) versus those who received little support, some support and high levels of support (M = 18.75, SD = 6.07), there was a difference between the levels of self-esteem (t(354) = -6.11, p < 0.001 d = 0.69). Scores below 15 suggest low self-esteem. The mean score of non transgender participants in Brazil was 18.9 (SD = 6.78) (Hutz and Zanon, 2011).

Of the total 410 participants who were asked if they needed to move away from their family because they were TGD, 164 (40.00%) answered yes. Besides, foram a total of 388 participantes who responded if they received any parental support, the majority (64.17%, n = 77) of participants who did not receive support needed to leave home, compared to 29.85% (n = 80) of participants who received some support. Family support significantly affected the need to leave home: χ2 (1, N = 388) = 40.52, p < 0.001, and φ = 0.32. The lack of family support increased the need to leave home by 4.21 times (odds ratio of 95% [2.67, 6.64]). In relation to these data, it is important to point out that some parts of the survey were optional, so the frequencies for each option varied as the total number of respondents for each question.

In addition, 13.00% (n = 13) of the participants answered that they were already without fixed housing, of a total of 416 respondents. Of those who did not receive any support from their parents, 23.57% (n = 29) stated that they did not have fixed housing, compared to 7.38% (n = 20) of those who already received some type of support. Family support significantly affected the frequency of episodes of homelessness (χ2 (1, N = 394) = 20.38, p < 0.001, φ = 0.23). The lack of family support increased the chance of living without fixed housing by 3.87 times (odds ratio 95% CI [2.09, 7.18]). More information on these outcomes can be found in Table 3.

For the SEM analyses, Table 4 presents the correlation matrix of the various factors included in the model: the gender affirmation procedures, family support, self-esteem, the need to move because of TGD identity and that lack of a home. As can be seen, no variables presented correlation values higher than 0.40. Furthermore, VIF values varied from 1.10 to 1.24. Since there were no correlations equal to or higher than 0.80 and VIF values did not exceed 10.0, there was no indication of multicollinearity and thus SEM models were tested. The SEM results demonstrated adequate fit indices for the model tested (χ2(4) = 7.44, p = 0.115; χ2/df = 1.86; CFI = 0.982; TLI = 0.955; RMSEA [90% CI] = 0.047 [<0.001 – 0.099]).

Table 5 presents the unstandardized and standardized coefficients and significance levels of the associations in the model. The association between the status of the gender affirmation procedures, family support and self-esteem was significant and indicated that the further TGD individuals advanced in gender affirmation, the more self-esteem and family support they would have. The association between family support and self-esteem indicated that family support was associated with higher self-esteem. Low family support was associated with the willingness to move from home due to one’s TGD status and there was also a significant correlation between low self-esteem and the willingness to move from home due to one’s TGD status. Finally, homelessness was associated with the willingness to move with a large effect size. Figure 1 depicts the path model among the variables tested in the SEM. An alternative model was tested replacing the correlation between self-esteem and willingness to move from home due to one’s TGD status for a regression path from self-esteem to willingness to move from home. The fit of this alternative model, although acceptable (χ2(4) = 9.51, p = 0.049; χ2/df = 2.38; CFI = 0.971; TLI = 0.928; RMSEA [90% CI] = 0.059 [0.003 – 0.109]), was poorer (i.e., significant χ2 and lower CFI) and, most importantly, the theoretical association between these variables reflects a correlation, in which lower self-esteem may lead to willingness to move out, but also having to move from home impacts negatively on one’s self-esteem. Therefore the model with the correlation seems better suited.

In summary, the results suggest the indirect effect of the impact of gender affirmation procedures on homelessness through the effects of family support and the need to move. The indirect effects on homelessness was –0.074 (p = 0.024) for gender affirmation procedures and –0.318 (p < 0.001). In the model, the combined independent variables accounted for 0.323 of the variance of the homelessness dependent variable.