Gender dysphoria and psychiatric comorbidities in childhood: a systematic review

ABSTRACT

Objective: To appraise the methodological quality of studies on the prevalence of psychiatric comorbidities for children presenting with gender dysphoria, including diagnosis and management.

Study design: A systematic review of 15 articles on psychiatric comorbidities for children diagnosed with gender dysphoria between the ages of two – 12 years.

Data sources: A systematic literature search of Medline, PsychINFO, CINAHL, Scopus and Web of Science for English-only studies published from 1980 to 2019, supplemented by other sources. Of 736 studies, 721 were removed following title, abstract or full-text review.

Results: Ten studies were retrospectively-oriented clinical case series or observational studies. There were few randomised, controlled trials. Over 80% of the data came from gender clinics in the United States and the Netherlands. Funding or conflicts of interest were often not declared. Mood and anxiety disorders were the most common psychiatric conditions studied. There was little research on complex comorbidities. One quarter of studies made a diagnosis by a comprehensive psychological assessment. A wide range of psychological tests was used for screening or diagnostic purposes. Over half of the studies diagnosed gender dysphoria using evidence-based criteria. A quarter of the studies mentioned treating serious psychopathology prior to addressing gender dysphoria.

KEY POINTS

What is already known about this topic:

1. Children with gender dysphoria are likely to experience profound psychological and physical difficulties.

2. Gender clinics around the world have different ways of assessing and treating children with gender dysphoria.

3. Children often rely on caregivers and health professionals to make treatment decisions on their behalf.

What this topic adds:

1. Children with gender dysphoria often experience a range of psychiatric comorbidities, with a high prevalence of mood and anxiety disorders, trauma, eating disorders and autism spectrum conditions, suicidality and self-harm.

2. It is vitally important to consider psychiatric comorbidities when prioritising and sequencing treatments for children with gender dysphoria.

3. The development of international treatment guidelines would provide greater consistency across diagnosis, treatment and ongoing management.

KEYWORDS:

• Children
• gender dysphoria
• diagnosis
• co-morbidity
• treatment

Introduction

Gender dysphoria is a psychological condition in which a person’s subjectively felt identity and gender are not congruent with their biological sex, often causing clinically significant impairment in social and other areas of functioning (American Psychiatric Association, 2013; Smith & Matthews, 2015; De Vries et al., 2011). People experiencing gender dysphoria often describe themselves as being transgender, or gender non-binary, which is the sense that their personal identity and gender do not correspond with their birth sex. Several changes in how gender dysphoria is defined, previously referred to as gender identity disorder, have been made in the current version of the Diagnostic and Statistical Manual of Mental Disorders (5^th ed.; DSM-5; American Psychiatric Association, 2013).

For children diagnosed with gender dysphoria, the DSM-5 diagnostic criteria outline that the child’s subjective experience of their own gender is different to their birth sex. The child may insist they are the other gender or have a strong dislike for their sexual anatomy. In some cases, children have been known to mutilate their genitals to relieve their inner distress (Smith & Matthews, 2015). The World Health Organisation had also recently updated the diagnosis from gender identity disorder to “gender incongruence in childhood” in the International Statistical Classification of Diseases and Related Health Problems (11th ed,; World Health Organization, 2020) using the same criterion for assessment and diagnosis as the DSM-5. Both updates suggest a global consensus that children with gender dysphoria are likely to experience profound psychological and physical difficulties, impacting negatively on their quality of life, mental wellbeing and capacity to integrate socially.

Children as young as two years old have been reported to present with symptoms of gender dysphoria (Cohen-Kettenis et al., 2002; Drummond et al., 2008). Some studies estimate 10% of children persist with the condition into adolescence and adulthood (Zucker, 2008), while later studies estimate between 12% and 27% of cases persist with the condition (Cohen-Kettenis, 2001; Drummond et al., 2008). The prevalence rates for gender dysphoria are generally collated across all age ranges; DSM-5 reports a conservative estimation of 0.01% of the natal male population and 0.003% of the natal female population experience gender incongruence (American Psychiatric Association, 2013).

A range of studies finds gender dysphoria is becoming increasingly prevalent and this trend is particularly highlighted in child populations. Included in these increased referrals are children presenting at younger ages, with the median age now seeking assessment and treatment estimated to be 8 years old (M = 11, SD = 2.16) (De Vries & Cohen-Kettenis, 2012). The reason for the increase in referrals is largely unknown. Some studies report factors such as an increase in role models like gender non-conforming celebrities, as well as a greater visibility of the gender-diverse community in the media (Chen et al., 2016). With the increase in referrals to gender clinics, a growing interest in how children with gender dysphoria are being clinically treated and managed is also occurring across the world (Kruekels & Cohen-Kettenis, 2011).

Children have no legal ability to provide informed consent for medical treatment or to engage in decision-making about their condition. As such, children often rely on caregivers and health professionals to make treatment decisions on their behalf (Drescher et al., 2016). Parents may not always make the right decisions for their child and a bioethical argument exists as to whether parental authority should not encompass denying children with gender dysphoria access to medical treatments (Priest, 2019). Gender clinics around the world have different ways of assessing and treating children with gender dysphoria. Some clinics strongly disagree with any medical intervention prior to adulthood and instead focus on psychological treatment for both the family and the child (De Vries & Cohen-Kettenis, 2012). Other treatment models include a social role transition to the affirmed gender, which is seen as a reversible means of managing the child’s distress until it is known whether the gender dysphoria is persisting into adolescence (Steensma et al., 2013). Other clinics work exclusively in a medical model and facilitate the administration of puberty suppressing hormones in children as young as 8 years old, relying on medical intervention to treat the symptoms of gender dysphoria (Spack et al., 2012). An Australian standards of care and treatment guideline was released to maximise the quality of care provision for children with gender dysphoria, recommending a move away from treatment based on chronological age, with the timing of medical transition based on the patient’s capacity to make informed decisions, including consideration of the severity of psychiatric comorbidities when making clinical decisions about treatment (Telfer et al., 2018).

The core symptoms of gender dysphoria in childhood rarely exist in isolation; more commonly the symptoms are exacerbated by psychosocial stressors and psychiatric disorders (Vrouenraets et al., 2015). There are several psychiatric comorbidities in populations of children with gender dysphoria, such as autism spectrum conditions (VanderLaan et al., 2015), anxiety and depressive disorders (Holt et al., 2014), eating disorders (Russell & Keel, 2002), self-harm and suicidality (Reisner et al., 2015), psychosis and posttraumatic stress disorder (Coleman et al., 2012). Most of the studies on comorbidities are conducted in gender clinics, where there may be a stronger clinical emphasis on gender dysphoria as the core source of the child’s suffering (De Vries et al., 2011). While treating gender dysphoria as the primary condition is important, psychiatric comorbidities often have a greater likelihood of persisting into the child’s future and complicating mental health outcomes throughout the medical transition (Steensma et al., 2011). This suggests it is vitally important to consider psychiatric comorbidities when prioritising and sequencing treatments for children with gender dysphoria (De Vries et al., 2011).

The absence of a global and coordinated approach to treatment for children with gender dysphoria triggered the formation of an American Psychiatric Association Taskforce in 2012, to determine whether sufficient credible evidence existed to create broad, global treatment guidelines (Byne et al., 2012). A recommendation of the Taskforce was to prioritise the psychological wellbeing of the child when assessing for medical treatments, by diagnosing and providing treatment for any psychiatric comorbidities (Byne et al., 2012). Despite the Taskforce’s recommendations, there appear to be no randomised controlled treatment outcome studies on psychiatric comorbidities for children with gender dysphoria. This means the highest level of evidence available for treatment recommendations can best be characterised as expert opinion. These opinions vary widely about what appropriate treatment is, and these decisions are often influenced by a theoretical orientation, as well as assumptions and beliefs regarding the origins, meanings and perceived rigidity or malleability of gender identity (Drescher & Byne, 2012). The aim of this review was to identify the method utilized and appraise the methodological quality of studies of the prevalence of psychiatric comorbidity among children presenting with gender dysphoria, including (1) was gender dysphoria diagnosed using DSM-5 criteria, (2) were psychiatric comorbidities diagnosed by a comprehensive psychological assessment, including the use of valid and reliable psychological tests, and 3) how psychiatric comorbidities are managed regarding treatment recommendations.

Method

Search strategy

This systematic review adhered to the guidelines detailed in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement (Moher et al., 2009). A systematic literature search was conducted using the following search terms (“gender dysphoria” OR “gender identity”) AND (comorbid* OR psychopathology) AND psych* AND child*. The search was undertaken in Medline, PsychINFO, CINAHL, Scopus and Web of Science. The reference lists from the selected studies were also screened for additional relevant studies. A grey literature search was conducted using the University of Canberra database, a multidisciplinary database that contains research reports, conference papers and dissertations covering the social and biomedical sciences. Clinical trial registries were also searched to identify trials at various stages, from registration of the protocol to trial completion. Additional websites known to contain research were also searched, including the Australian Psychological Society website, Google Scholar and Research Gate.

Studies published between 1980 and 2019 were selected for inclusion in the review. The year of 1980 was chosen because gender dysphoria was formally recognised as a diagnosis in the DSM-III at this time. Non-clinical case studies and studies which were not published in English were excluded from the review. Studies with children and adolescents were included in the initial searching, to identify studies that included populations of children aged two to 12 years old. The lower bound of 2 years old was selected as this was found to be when children begin to develop social awareness of gender roles (Martinot et al., 2012). The upper bound of 12 years old was decided as a cut-off age by using the Tanner Scale (Hembree, 2012), a measure of physical development commonly used in studies of children with gender dysphoria to contribute to age-appropriate medical decision-making.

Some studies had participants older than 12 years and these studies were still included in the review, if they also had participants 12 years and under in the study. The studies had to include participants who met the DSM-5 diagnostic threshold for gender dysphoria. Studies which included participants with sub-threshold symptoms were excluded. It was anticipated that many studies were likely to be non-randomised (observational) and uncontrolled. It was decided to include these studies as they were relevant to the research question. Review articles, commentaries and letters were excluded Please refer to Appendix A for additional details on the Prisma flow diagram.

Procedure

Studies meeting the inclusion criteria were examined. In the first instance, 726 studies were identified using a database search and an additional 10 studies were found searching the internet. Duplicates (n = 134) were removed using the Endnote program. A review of the titles led to 513 records being excluded. The main reason for the exclusion of these records was the study titles did not include any reference to psychiatric comorbidities and gender dysphoria. The remaining 89 articles were screened by abstract. There were 74 articles excluded based on: (1) the population was not relevant (adults) or (2) the study did not have any participants (a literature review) or (3) the study used a projective test such as the Rorschach as a measure of distress. A spreadsheet was created to record the reasons for exclusion for each of the full-text articles reviewed. Figure A1 presents a Prisma flow diagram and Table A1 summarises the systematic review of the 15 articles.

Assessment of quality and risk of bias

The study used a checklist based on the National Institute for Health and Care Excellence (NICE) quality appraisal checklist for quantitative studies (National Institute for Health and Care Excellence, 2017). This checklist had been created and adapted from the Graphical Appraisal Tool for Epidemiological studies (GATE) developed by Jackson et al. (2006) and revised for public health studies. The criteria used to assess the methodological quality of each study were: (1) how the population was selected, (2) whether a comprehensive psychological assessment was undertaken, (3) the duration and adequacy of study follow-up, (4) the proportion of participants lost to follow up, (5) whether there was blinding of key study personnel, (6) conflict of interest/funding disclosures, (7) whether gender dysphoria was diagnosed using DSM-5 criteria and (8) the age of the study.

While the risk of bias and the methodological quality of studies can be assessed using checklists or scales, these instruments often contain items or criteria that are not directly related to internal validity (Jüni et al., 2001; Moher et al., 2009). Therefore, this systematic review used a domain-based evaluation as recommended by the Cochrane Collaboration, in which critical assessments were made separately for different domains, such as a selection or attrition bias (see Appendix B). The domain-based method was developed between 2005 and 2007 by a working group of methodologists and reviewers, who found this was an optimal approach as it was not based on a “generally accepted” criterion found in a checklist or scale (Lundh & Gøtzsche, 2008). The articles in this systematic review were coded using symbols to represent either a low risk of bias (++), medium risk of bias (+) or a high risk of bias (-) (see Table B1). The process used to assess the methodological quality of the selected studies was also reviewed by two independent assessors at the University of Canberra as an additional reliability check.

Results

Participant characteristics

Five of the studies (33.3%) focussed on participants in the population of interest, children aged between two – 12 years old. A further five studies (33.3%) had participants with wider age ranges (between two – 20 years old). These studies contained a large number of participants in the 5 years and below age range, which is an important age range for children in terms of the onset and awareness of gender-linked behaviours (Martin et al., 2002). The remaining five studies (33.3%) had participants in the eight – 20 age range, meaning a third of the studies in the systematic review had a limited focus on the population of interest. Six of the studies were from the Netherlands (40.0%), six of the studies were from the United States (40.0%), one study (6.6%) was from Australia and one study (6.6%) was from the United Kingdom. The final study had a mix of participants from both Canada and the Netherlands (6.6%). Research from the Netherlands and the United States dominated the sample, with 12 of the 15 studies including data from either country. In terms of sample size, the range over the 15 studies varied between 39 and 2778 participants.

Study design and data collection method

Three of the studies (20.0%) used an experimental design (matched pairs and mixed factorial design, respectively) and one of these studies also included a quasi-experimental comparison between-groups, as well as a clinical case series. Ten studies (66.6%) were observational design, including cross-sectional (three studies), six clinical case series (one twin study) and one study (6.6%) was a longitudinal design. Two studies were quasi-experimental (13.3%), both using a between pre-existing groups design. All 15 of the studies included in the systematic review used convenience sampling as the method for obtaining participants. In terms of recruitment, 12 studies (80.0%) included children referred to a gender clinic. One of the studies (6.6%) recruited children through newspaper advertising, another study (6.6%) recruited from a National Longitudinal Study and the final study (6.6%) recruited participants from local schools.

The data on psychiatric comorbidity was collected using self-report measures in six studies (40.0%). Four studies (26.6%) used a combination of self-report measures and a clinical interview (with one of these studies including the collection of biodata, such as heart rate), three studies (20.0%) used a clinical interview only. Two studies (13.3%) reviewed clinical data only, such as patient files, including referral letters, clinical notes and reports. There was poor consistency across studies regarding declarations of conflicts of interest and funding disclosures. Seven studies (46.6%) did not state whether there were any conflicts of interest or whether funding was received for the study. Six of the studies (40.0%) did declare conflicts of interest and the receipt of funding. One study (6.6%) disclosed funding only, and one study (6.6%) disclosed a conflict of interest but did not mention funding.

Gender dysphoria and psychiatric comorbidity

In terms of the diagnosis of gender dysphoria, nine studies (60.0%) explicitly reported whether their participants were diagnosed with gender dysphoria using DSM-5 criterion. The remaining six studies (40.0%) mentioned that a clinical interview was conducted, without mentioning whether the participants met the DSM-5 criterion for gender dysphoria. In terms of psychaitric comorbidities, five studies (33.3%) measured internalising and externalising behaviours (mood, anxiety and aggression difficulties), five studies (33.3%) measured internalising and externalising behaviours as well as a range of other DSM-5 conditions, such as specific phobia, attention deficit hyperactivity disorder, oppositional defiant disorder, tic disorder, psychosis, eating disorders, autism spectrum disorder (ASD) and complex posttraumatic stress disorder. Two of the studies (13.3%) focussed solely on ASD, one study (6.6%) measured separation anxiety and depression, and the final study (6.6%) assessed against all DSM-5 clinical disorders (not including personality disorders).

To measure psychiatric comorbidities, five studies (33.3%) used the Child Behaviour Check List (CBCL). A further three studies (20.0%) used the CBCL and additional tools such as the Diagnostic Interview Schedule for Children (DISC-IV), Beck Depression Inventory (BDI), Children’s Global Assessment Scale (CGAS), Spielberger’s Trait Anger (STAI) and Spielberger’s Trait Anxiety (STAI). Two studies (13.3%) used no assessment tool, and the remaining four studies (26.6%) used only one assessment tool. The tool selected was not consistent across studies, ranges included the Coolidge Personality and Neuropsychiatric Inventory (CPNI), Diagnostic Interview for Social and Communication Disorders (DISCO-10), the DISC-IV or the Asperger Syndrome Diagnostic Scale (ASDS). The final study (6.6%) used two assessment tools, the National Institute of Health Patient Reported Outcomes Measurement Information System (PROMIS Scale) and the Harter Self Perception Scale. Most studies (80.0%) did not report the reliability coefficients of the selected assessment tools. However, the remaining studies did include these figures as an indicator of the robustness of the study. While the majority (60.0%) of studies used valid and reliable psychological tests, only four (26.6%) used a combination of clinical interview and psychological testing for clinical case formulation and diagnostic purposes.

Mental health outcomes were inconsistently operationalised across studies. For example, depression was assessed using a variety of tools, such as the CBCL which had scales orientated towards DSM-5 criterion, but it was not diagnostic for major depressive episode (Cohen-Kettenis et al., 2002). Other studies used the DISC-IV which is explicitly based on diagnostic level DSM-5 criteria for major depressive disorder (De Vries et al., 2011). Another study used the PROMIS Scale, which did not include the physiological aspects of depressive disorders, such as weight changes and sleep disturbance (Durwood et al., 2017). Many studies of mood disorders also used diverse clinical screening cut-off scores, such as “depressive distress in the last seven days” (de Vries et al., 2011; Durwood et al., 2017), “scores within a 90^th percentile” (Cohen-Kettenis et al., 2002), a “clinical diagnosis of major depressive episode” (Holt et al., 2014; Spack et al., 2012) or “symptoms within the last 12 months” (De Vries et al., 2011). There were also differing timeframes for assessment, such as “lifetime prevalence”, “distress in the last seven days” and “within the last 12 months”. Some studies used heterogeneous sub-populations of people with gender dysphoria, such as “predominantly homosexual” (Zucker, 2002), “Caucasian races only” (Aitken et al., 2016), “social ranking” of the parents (for example, either “professionals” or “unskilled labourers”) (Cohen-Kettenis et al., 2002) or participants considered “eligible for puberty supressing hormones” (De Vries et al., 2011).

Despite these inconsistencies, data consistently showed that children with gender dysphoria were burdened by mental health concerns. For example, 21.0% of a sample of 105 children had an anxiety disorder (De Vries et al., 2011), 7.8% of a sample of 204 children were found to have co-occurring ASD and 44.3% of a sample of 97 children were found to have significant psychiatric history, including 9.3% of the sample had attempted suicide. It was commonly understood that children with gender dysphoria often suffered from comorbidities that worsened during puberty, at which time they were generally at high risk of suicide (Lopez et al., 2016).

In terms of how psychiatric comorbidities were managed, six studies (40.0%) discussed the importance of referrals for support. One of these studies recommended that patients with comorbid ASD should be referred to a specialist clinic for assessment and treatment. Four of the studies (26.6%) recommended that referrals for the treatment of serious psychiatric comorbidities (for example, major depressive episode) should be actioned prior to treating gender dysphoria. One study (6.6%) discussed the importance of children having access to a mental health professional, but it was unclear whether this was for treatment of psychiatric comorbidities or for support through the process of medical transition of gender. Only one study (6.6%) explicitly stated referrals for treatment of psychiatric comorbidities were actioned.

Methodological quality and risk of bias

The risk of bias was assessed using guidance contained in the Cochrane Collaboration’s tool for assessing the risk of bias in studies (“Cochrane Handbook,” 2017). Seven studies (46.6%) were classified as having a high risk of bias (-), a further seven studies (46.6%) were classified as having a medium risk of bias (+) and one study (6.6%) was classified as having a low risk of bias (++). This final study was an experimental design which used the participants as their own controls. Fourteen of the 15 studies were assessed as having a medium to high risk of bias. All studies used convenience sampling. The data collected in five studies were up to 44 years old (33.3%), four studies (26.6%) had high rates of participant exclusion and no studies used blinding of key study personnel. Overall, the methodological quality of the studies was poor.

Discussion

This was the first systematic review of studies on the prevalence of psychiatric comorbidity among children presenting with gender dysphoria. The aim of the study was to assess the methodological quality of the studies (including risk of bias) and to investigate whether (1) gender dysphoria was diagnosed using DSM-5 criteria, (2) ascertain how psychiatric comorbidities were diagnosed (3) how psychiatric comorbidities were managed in terms of actioning referrals for treatment.

Methodological quality

Overall, the methodological quality of the existing data was poor which indicated a potential systematic bias in the literature on this topic area. Of the 15 studies reviewed, 14 were classified as either a medium or high risk of bias. Over half of the studies used a clinical case series design, which provided a retrospective look at a series of cases with features in common. This design can be suited to the study of rare or new psychiatric phenomenon, but it may not be of value for studies on psychiatric comorbidities in children with symptoms of gender dypshoria (Vandenbroucke, 2001) as many children around the world are already receiving medical treatment. It was a suitable research design for older studies undertaken when gender dysphoria was first included in DSM-III in the 1980s. However, it appeared the study designs in this topic area had not yet progressed to testing hypotheses, as no study used a fully blinded, randomised controlled trial design. In the hierarchy of research designs, the results of randomised, controlled trials were generally considered to be evidence of the highest grade, whereas observational studies were viewed as having less validity because they often overestimated treatment effects (Concato et al., 2000). This suggests the results of global studies may have very different explanations and recommendations regarding psychiatric comorbidities among children with gender dysphoria simply because of the limitations of the study designs (Costa & Colizzi, 2016).

All studies gathered their data using convenience, or non-probability sampling. Most of the studies used data from participants referred to a gender clinic. The assumption with this type of sampling was that the members of the target group were homogeneous (Etikan et al., 2016), suggesting there would be no difference in the research results obtained with a random sampling method. While this means of sampling is often affordable and convenient for the researcher (Heckathorn, 2011), none of the studies described how their samples differed from a sample that would be randomly selected. This represented a critical limitation in terms of not being able to generalise the study results to the broader population of children with gender dysphoria.

Another factor which limited the generalisability was the data being dominated by studies from the United States and the Netherlands, with over half of the studies not declaring who funded it or whether there was a conflict of interest. This raises concerns about whether the participants’ welfare was unduly influenced by a secondary interest, such as financial gains for the clinics or the desire for professional recognition by researchers (Bekelman et al., 2003). It was likely that Western countries conducted most of the studies in this topic area as their attitudes were progressive, in terms of concern for the welfare of people with gender dysphoria and a desire to reduce stigma and discrimination (Vance Jr et al., 2010). However, it was difficult to make this inference as funding and conflicts of interest were often not declared.

Clinical implications

Children with gender dysphoria often report a reduction in symptoms of depression and anxiety after receiving gender-affirming medical treatments, but they do not always show improvements in symptoms of gender dysphoria (Chew et al., 2018). The lack of follow-up and referrals for treatment of psychiatric comorbidities in the reviewed studies was concerning, given there may be psychological symptoms related to having gender dypshoria which are assumed to be treated once medical interventions are pursued.

Furthermore, the studies examined in the review did not make mention that previous twin studies had found a biological basis to gender dysphoria involving endocrine, neurobiological and genetic factors. For instance, an increased prevalence of gender dysphoria was observed among people who experienced atypical prenatal androgen exposure in utero, such as females with congenital adrenal hyperplasia (Fisher et al., 2016). Heritability studies also suggested a genetic component: 23% to 33% of monozygotic twin pairs were concordant for gender dysphoria (Diamond, 2013). While genetics may play a role in the development of gender dysphoria in children, it was not found to be the sole determinant which means that psychiatric comorbidities may also impact in some way on the development and course of gender dysphoria.

Limitations and strengths

The main strength of this paper was its focus as the first systematic review to summarise and critically assess the prevalence data on psychiatric comorbidities of children with gender dysphoria. The studies did not appear to have become methodologically stronger over the years, noting the paucity of randomised controlled trials available. The majority of studies were conducted in Western countries and the results may have been influced by the level of tolerance in society, the healthcare system, legislation regarding the rights of people with gender dysphoria and academic interest (Arcelus et al., 2015). The paucity of studies on children with gender dysphoria in developing countries may indicate an absence of clinical services, which would be likely to perpetuate psychiatric comorbidities (Claes et al., 2015). The study was also limited by the high heterogeneity of studies, noting the clear differences in the methodology of the studies included in the review.

Future research

Further research is needed in a few key areas. First, longitudinal data on the factors that may predict children who continue to present with symptoms of gender dysphoria into adolescence, versus those who do not (Steensma et al., 2011) may improve knowledge on whether psychiatric comorbidities mediate this relationship. Gender affirming hormone treatment often improves mental health outcomes but it does not always improve body image problems or symptoms of gender dysphoria (Chew et al., 2018), indicating further work may be important in terms of evaluating the efficacy of medical treatment for symptoms of gender dysphoria.

Qualitative studies on the experiences of children with gender dysphoria may also provide valuable information on the potential relationship between internalised transphobia and the development of psychiatric comorbidities. Furthermore, randomised controlled trials on the effectiveness of psychological intervention for complex psychiatric comorbidities, such as posttraumatic stress disorder, ASD and eating disorders are required, specifically how treatment of these conditions may relate to improved coping with gender dysphoria into adolescence. Studies into the relationship between gender dysphoria and mentalising ability for those with autism spectrum conditions may also be of value, to determine whether theory of mind deficits affect perceptions of gender identity.

Conclusion

The results of the study found that children with gender dysphoria often experience a range of psychiatric comorbidities, with a high prevalence of mood and anxiety disorders, trauma, eating disorders and autism spectrum conditions, suicidality and self-harm. Most importantly, this systematic review found that there was no standardised approach to the assessment, diagnosis, treatment and clinical management of children with gender dysphoria and psychiatric comorbidities, including a paucity of randomised control trials, which has implications clinical decision-making.

Noting children do not have capacity to consent to their own medical treatment, it was concerning to find that there were a range of different treatment models used around the world for children presenting with gender dypshoria, from early intervention medical treatments involving parental consent for treatment, to denial of medical treatment for children and adolescents. This may have implications for the level of psychological support that a child may receive depending on how the respective treatments for gender dysphoria are perceived. There was also an absence of studies from developing countries on gender dypshoria in childhood, suggesting implications may arise for the assessment and treatment of psychiatric comorbidities in non-gender clinics. Looking ahead, it will be essential for future researchers to expand on the findings of existing studies by using large, prospective longitudinal studies with sufficient follow-up, statistical power and the inclusion of well-matched controls.

Disclosure statement

No potential conflict of interest was reported by the authors.

Appendix A

Figure A1. Prisma 2009 flow diagram

Table A1. Summary of studies

Study	Aim of study	Population	Age (years)	Sample (N)	Country	Assessment type	Mental health parameters	Results
Aitken et al., 2016	Examine rates of self-harm and suicidality	Children with gender dysphoria and their siblings	3-12	2778 (572 gender-referred children, 425 siblings, 878 referred children and 903 non-referred children)	United States	Self-report data from parents (CBCL)	Self-harm and suicidality	Significantly higher rates of suicidality and self-harm among gender-referred and referred children than siblings and non-referred children
Cohen-Kettenis et al., 2002	Examine the similarities and differences of children with gender dysphoria from two different countries regarding demographic variables, social competence and behavioural problems	Children with gender dysphoria referred to a Canadian gender clinic between 1975-2000; children with gender dysphoria referred to a Dutch gender clinic between 1988-2000	3-12 (Canadian sample) 4-12 (Dutch sample)	358 children in Canadian sample, 130 children in Dutch sample	Canada and Netherlands	Self-report data from parents (CBCL)	Internalising and externalising psychopathology	No significant difference in behavioural problems between samples. All samples were in the clinical range for externalising and internalising behaviours. Significant difference between internalising and externalising behaviours for both groups; children are more likely to internalise their distress. Canadian sample: significantly greater amount of 6-11 had scores in the clinical range
Cohen-Kettenis et al., 2002	Investigate the quantity and type of emotional and psychological problems at the time of application for GRS	Children and adolescents with a diagnosis of gender dysphoria referred to a Dutch gender clinic	1-17	73 participants. CBCL data was used for 29 participants and DISC data for 44 participants	Netherlands	Clinical interview with child/adolescent and self-report data from both children and the parents (CBCL and DISC)	Internalising and externalising psychopathology, specific phobia, tic disorder, oppositional disorder, and anxiety disorder	Self-report scores for the child/adolescent group within the normal range of scores for internalising/externalising behaviours on the CBCL. Self-report scores for parent group indicate one third of sample met clinical range for internalising/externalising behaviours. One quarter of participants in the DISC group met DSM criteria for a psychological disorder
Coolidge et al., 2002	Examine the heritability of gender dysphoria and rates of comorbidity of depression and separation anxiety in a sample of children and adolescent twins with gender dysphoria	Children and adolescent twins recruited through newspaper advertisements	1-17	Parents of 314 twins; 96 monozygotic pairs and 61 dizygotic pairs	United States	Self-report data from CPNI	Separation anxiety and depression	Modest and significant correlation between gender dysphoria and depression Nonsignificant correlation between gender dysphoria and separation anxiety
De Vries et al., 2011	Examines psychiatric comorbidity in adolescents with gender dysphoria	Adolescents referred to a gender clinic and classified as either ‘delayed eligible’ for GRS or ‘immediately eligible’ for GRS	0-18	105 (53 natal males, 52 natal females) and their parents	Netherlands	Clinical interview and self-report: clinical interview with child and parent, and administration of CBCL and DISC-IV to the parents	Anxiety, mood and disruptive disorders	67.6% of the sample had no comorbid psychiatric disorder. 21% had an anxiety disorder, 12.4% had a mood disorder and 11.4% had a disruptive disorder. Compared to natal females, natal males (n = 53) suffered 2+ comorbid conditions
de Vries et al., 2011	Longitudinal examination of the impact of GRS on psychological outcomes	The first cohort of adolescents with gender dysphoria to receive puberty suppressing hormones at a gender clinic in 2004	1-17	55 adults who received puberty suppression during adolescents were assessed three times: before puberty, after administration of cross-sex hormones and one-year post GRS	Netherlands	Self-report: children and parents, using CGAS, BDI, TPI, STAI, CBCL and YSR	Internalising and externalising psychopathology	Significant improvement in global functioning, reduction in internalising and externalising behaviours Significant increase in depression scores Transmen had reduced anger, anxiety and externalising t scores Transwomen had more symptomatology
de Vries et al., 2010	Examine the incidence of ASD in gender dysphoric children and adolescents and describe their clinical characteristics	Children and adolescents with gender dysphoria referred to a gender clinic between 2004-2007	0-12	204 children with gender dysphoria referred to a gender clinic	Netherlands	Clinical interview using the DISCO-10, and collateral information gained from the parents and school	ASD only	7.8% (n = 16) of participants were found to have cooccurring ASD
Durwood et al., 2017	Longitudinal examination of the wellbeing of socially transitioned children with gender dysphoria	Children enrolled in a National longitudinal study (TransYouthProject) from 2015 to 2016	6-14	63 children and their parents, 63 controls and 38 siblings	United States	Self-report using child version of the PROMIS scale. Self-report using Harter Self-Perception Scale for Children	Internalising psychopathology (depression and anxiety) and self-worth	Socially transitioned children; depression and self-worth not significantly different from matched-control or sibling peers. Marginally higher non-significant rates of anxiety. Transitioned children showed higher rates of depression and non-significantly higher rates of anxiety. Parents rated transitioned children as greater anxiety (p = .002) and equivalent levels of depression (ns)
Hewitt et al., 2012	Describe the experience of hormone treatment in children diagnosed with gender dysphoria	Medical record review of children referred to a gender clinic between 2003 to 2011	0-17	39 children and adolescents	Australia	Clinical interview only	Anxiety, depression, suicidal ideation, obsessional disorder and ASD	21 participants reported symptoms of depression and anxiety. Three children were diagnosed with an obsessional disorder and two with ASD
Holt et al., 2014	Examines demographic variables and associated difficulties in children/adolescents with gender dysphoria	Children and adolescents referred to a gender service over a one-year period (2012)	5-17	218 children and adolescents	United Kingdom	Data extracted from patient files (referral letters, clinical notes and reports) and exported into a spreadsheet	Suicidal ideation, self-harm, ASD, ADHD, anxiety, psychosis, eating disorder, bullying, complex PTSD	Most common difficulties were bullying, low mood/depression and self-harm. Self-harm was significantly more common in natal females and ASD in natal males
Shumer et al., 2016	Examines clinical data on ASD in children diagnosed with gender dysphoria	Medical review of clinical data for children presenting to a gender clinic between 2007 to 2011	8-20	39 children and adolescents	United States	Screening test of clinical data using the ASDS	ASD only	23.1% participants presenting with gender dysphoria had possible, likely, or very likely Asperger syndrome
Spack et al., 2011	Describe clinical features of patients with gender dysphoria	Children and adolescents referred to a gender clinic over the period 1998 to 2010	4-20	97 children and adolescents	United States	Triage interview, clinical interview and self-report measures using a battery of psychological tests; CDI, RCMAS, Piers-Harris Children’s Self Concept Scale, Body Image Scale, Draw-A-Person test CDI (parent), CBCL, ASDS and Parenting Stress Index	Depression, GAD, bipolar disorder, pervasive developmental disorder and ASD, eating disorder, ADHD/ADD, panic disorder and PTSD	Forty-three patients (44.3%) presented with significant psychiatric history, including 20 reporting self-mutilation (20.6%) and suicide attempts (9.3%)
Wallien, Swaab, 2007	Examines whether children with gender dysphoria have increased physiological activity during stressful events	Children diagnosed with gender dysphoria and matched controls recruited from local elementary schools	7-12	50 children (25 with gender dysphoria and 25 matched controls)	Netherlands	Clinical interview, child self-report and parent self-report using CBCL Physiological measures; saliva cortisol collection, skin conductance and heart rate measurement	Internalising and externalising psychopathology	Children with gender dysphoria had more negative emotions and tonically elevated skin conductance. No differences between the groups in cortisol and heart rate
Wallien, Van Goozen, 2007	Investigates the prevalence and type of psychiatric comorbidity in children with gender dysphoria	Children referred to a gender clinic between 1998 to 2004. Children referred to an ADHD clinic between 1998 to 2004	4-11	120 children in the gender dysphoria group and 47 children in the (clinical comparison) ADHD group	Netherlands	Structured clinical interview of the parents using the DISC-IV	All DSM-IV disorders not including personality disorders (31 in total)	52% of children had one or more comorbid diagnosis Greater internalising (37%) than externalising (23%) psychopathology in boys and girls with gender dysphoria Odds ratios of having internalising or externalising comorbidity were 1.28 and 1.39 times higher, in the clinical comparison group than gender dysphoria group. 31% of children with gender dysphoria suffered from an anxiety disorder
Zucker, 2002	Examines demographic characteristics, behavioural problems and peer relations in children and adolescents diagnosed with gender dysphoria	Children and adolescents referred to a gender clinic for problems with gender identity development	3-16	358 children and 72 adolescents	Canada	Self-report of the mother using the CBCL	Internalising and externalising psychopathology	Adolescent sample showed significantly more general behavioural disturbance on the CBCL than the child sample although both age groups had, on average, mean scores that fell within the clinical range

N = entire sample involved in the study. Age = If the study included the population of interest (age range 0-12 years) and included additional participants >12 years, the study was still included as it provided information related to the research questions.

Abbreviations: ADD = attention deficit disorder. ADHD = attention deficit hyperactivity disorder. ASD = autism spectrum disorder. GAD = generalised anxiety disorder. GRS = gender reassignment surgery. PTSD = posttraumatic stress disorder

Appendix B

Table B1. Methodological quality of studies

Study	Conflict disclosure	Sampling method	Study type	Date of study	Gender dysphoria diagnosis	Reliability coefficients	Participant exclusion	Blinding	Risk of bias
Aitken et al., 2016	Funding disclosed Authors reported no conflicts	Convenience	Between pre-existing groups Gender-referred children formed a clinical case series Clinic-referred and non-referred children were matched pairs	1976 to 2015 (M = 1997)	Pre-existing DSM-5 diagnosis	Tool used was the CBCL: Correlation of items on Suicidality Scale, r = .53. Peer Relations Scale, α = .81	7.5% of gender-referred children did not complete CBCL 3.6% of the clinic-referred and 88% of non-referred children were excluded	None reported	- Convenience sampling: average date of assessment is > 20 years; no blinding of key study personnel
Cohen-Kettenis et al., 2002	Funding not disclosed No conflict of interest statement	Convenience	Cross-sectional	Dates not reported	DSM-5 diagnosis	Tool used was the CBCL None reported	60.3% were excluded from the assessment of behavioural and emotional problems as their parents had not completed to CBCL	None reported	- Convenience sampling; high rates of participant exclusion; no blinding of key study personnel
Cohen-Kettenis et al., 2002	Funding not disclosed No conflict of interest statement	Convenience	Between pre-existing groups comparison: both groups were clinical case series	Toronto sample: 1974 to 2000 (M = 1991) Dutch sample: 1988 to 2000 (M = 1996)	Pre-existing DSM-5 diagnosis	Tool used was the CBCL Peer Relations Scale α = .77	None reported	None reported	- Convenience sampling: average date of assessment is > 20 years; no blinding of key study personnel
Coolidge et al., 2002	Funding not disclosed No conflict of interest statement	Convenience	Twin study (case series)	Dates not reported	DSM-5 diagnosis	Tool used was the CPNI Gender Dysphoria Scale was .84 Anxiety Scale was .82 Depression Scale was .80 The statistic used to test reliability was not stated/reported	1.6% were excluded from the gender dysphoria analysis due to missing data	None reported	+ Convenience sampling; no blinding of key study personnel
de Vries et al., 2010	Funding disclosed Principle author reported having no conflicts	Convenience	Cross-sectional	2004 to 2007	No DSM-5 diagnosis	Tool used was DISCO-10 None reported	11.7% of the children excluded because of serious psycho-pathology	None reported	+ Convenience sampling; no blinding of key study personnel
de Vries et al., 2011	Funding disclosed Authors reported having no conflicts	Convenience	Cross-sectional	2002 to 2009	DSM-5 diagnosis	Tools used were CBCL and DISC-IV None reported	84% withdrew because gender dysphoria was not their primary clinical problem 9.6% excluded because parents had provided data in earlier studies 30.0% were excluded because of travel distance	None reported	- Convenience sampling; high rates of participant exclusion; no blinding of key study personnel
de Vries et al., 2011	Funding not disclosed No conflict of interest statement	Convenience	Longitudinal	2004 to 2011	No DSM-5 diagnosis	Tools used were CGAS, BDI, TPI, STAI, CBCL YSR: None reported	21.4% withdrew from the study or did not complete the assessment	None reported	- Convenience sampling; high rates of participant exclusion; no blinding of key study personnel
Durwood et al., 2017	Funding disclosed Principle author reported having no conflicts	Convenience	Between pre-existing groups: matched pairs comparison included	2015 to 2016	No DSM-5 diagnosis	Tools used were PROMIS Scale and Harter Self Perception Scale for Children Anxiety assessment for children, α = .86; parent 1, α = .94; parent 2, α = .91 Depression assessment for children, α = .86; parent 1, α = .88; parent 2, α = .89 Self-worth assessment for children, α = .67	None reported	None reported	+ Convenience sampling; no blinding of key study personnel
Hewitt et al., 2012	Funding not disclosed; authors reported no conflicts	Convenience	Clinical case study	2003 to 2011	No DSM-5 diagnosis	No tool used Clinical interview only None reported	None reported	None reported	+ Convenience sampling; no blinding of key study personnel
Holt et al., 2014	Funding disclosed; no statement on conflicts of interest	Convenience	Clinical case series	2012	No DSM-5 diagnosis	No tool used Data extracted from patient files None reported	28.1% of participants excluded because parents had gender dysphoria	None reported	+ Convenience sampling; no blinding of key study personnel
Shumer et al., 2016	Funding disclosed; authors reported having no conflicts of interest	Convenience	Clinical case series	2007 to 2011	No DSM-5 diagnosis	Tool used was ASDS None reported	None reported	None reported	+ Convenience sample; no blinding of key study personnel
Spack et al., 2011	Funding disclosed; authors reported they had no financial conflicts of interest	Convenience	Clinical case series	1998 to 2010	No DSM-5 diagnosis	Tools used were CDI, RCMAS, Piers-Harris Children’s Self Concept Scale, Body Image Scale, Draw-A-Person test CDI (parent), CBCL, ASDS and Parenting Stress Index None reported	None reported	None reported	- Convenience sampling; assessments 7 to 19 years old; no blinding of key study personnel
Wallien, Swaab, 2007	Funding not disclosed No conflict of interest statement	Convenience	Between pre-existing groups	1998 to 2004	DSM-5 diagnosis	Tool used was DISC-IV None reported	15.5% were excluded because they did not meet DSM criterion for gender dysphoria or had an IQ <75	Interviewers administered the DISC were blind to whether the participant had been referred for ADHD or gender dysphoria	+ Convenience sampling; assessments 13 to 17 years old
Wallien, Van Goozen, 2007	Funding not disclosed No conflict of interest statement	Convenience	Mixed factorial experiment: between groups were matched pairs	Dates not reported	DSM-5 diagnosis	Tool used was the CBCL None reported	None reported	None reported	++ Convenience sampling less of an issue as it is an experiment: no blinding of key study personnel. Used participants as their own controls (reduces random error).
Zucker 2002	Funding not disclosed No conflict of interest statement	Convenience	Clinical case series: sample split into child/adolescent based on age for between groups comparison	1975 to 2000	DSM-5 diagnosis	Tool used was the CBCL None reported	Unspecified number of participants were excluded for transvestic fetishism or autogynephilia 40.3% excluded due to not completing CBCL	None reported	- Convenience sampling: assessments are 17 to 42 years old; high rate of participant exclusion; no blinding of key study personnel

Coding. ++ = low risk of bias. + = medium risk of bias. – = high risk of bias.

Abbreviations. SRS = sex reassignment surgery. CBCL = Child Behaviour Check List. CPNI = Coolidge Personality and Neuropsychiatric Inventory. DISCO-10 = Diagnostic Interview for Social and Communication Disorders. DISC-IV = Diagnostic Interview Schedule for Children. UGDS = Utrecht Gender Dysphoria Scale. BIS = Body Image Scale. CGAS = Children’s Global Assessment Scale. BDI = Beck Depression Inventory. TPI = Spielberger’s Trait Anger. STAI = Spielberger’s Trait Anxiety. YSR = CBCL Youth Self Report. PROMIS Scale = National Institute of Health Patient Reported Outcomes Measurement Information System. ASDS = Asperger Syndrome Diagnostic Scale. CDI = Children’s Depression Inventory. RCMAS = Revised Children’s Manifest Anxiety Scale. ADHD = Attention Deficit Hyperactivity Disorder. DISC-IV = Diagnostic Interview Schedule for Children.