Introduction

Gender incongruence (GI) is the lack of accordance between experienced gender identity and birth-assigned gender [1]. This can lead to gender dysphoria (GD) and a possible wish for gender-affirming treatment [2]. Other terms used are opposite sex identification and transgender identity, consisting of a versatile definition of gender diversity including non-binary identification. The prevalence of GI/GD in the general population of children and adolescents is uncertain due to limited epidemiological data and different definitions of gender identity. An American study on students aged 11–13 years found that 1.3% of them identified as transgender [3], while 1.2% of high school students in New Zealand reported being transgender [4]. A large adolescent population survey from Finland reported 0.6% opposite sex identification and 3.2% non-binary/other gender identity [5].

Numerous studies have documented a sharp increase in young people with GD being referred to gender clinics worldwide over the last decade, with a higher frequency of AFAB seeking services compared to AMAB [6,7,8,9,10,11,12,13]. However, few studies on national samples of people seeking gender-affirming medical treatment have been published. A Dutch study on diagnostic- and treatment trajectories from the Center of Expertise on Gender Dysphoria included referrals up to 2018 [13]. Another study from the Gender Clinic in London, the UK, described the increase and change in sex ratio from 2000 to 2017 [7]. To our knowledge, no studies have published numbers of referrals to national gender clinics from the most recent years. It is an unanswered question whether there has been an increase in referrals for GI after the COVID-19 pandemic, as has been reported for psychiatric disorders [14].

Few studies have examined whether the characteristics of children and adolescents with GI have changed over time. An increase in psychiatric diagnoses over time could imply that GI more often is part of psychological difficulties. A decrease in the prevalence of psychiatric diagnoses among the referred could be due to a more open and accepting environment for gender exploration in parts of society, potentially reducing minority stress. Also, youth seen in recent years could differ from study-samples that constituted the Dutch Protocol [15]. The Dutch Protocol describe gender-affirming medical treatment for children and adolescents and is widely adopted among gender clinics worldwide [16, 17]. If there are differences in demographics and mental health among young people presenting to gender clinics now compared to earlier years, this could implicate the need for a different treatment approach. Children and adolescents included in the original Dutch Protocol had experienced distinct GI/GD since early childhood, were psychologically stable and had parental support [15]. If children and adolescents with GI differ from the original Dutch Protocol, there could be a need to focus on psychosocial support before considering gender-affirming medical treatment. Arnoldussen et al. assessed changes in young people referred to the Center of Expertise on Gender Dysphoria in Amsterdam, the Netherlands, during 2000–2016 [18]. They found no changes concerning demographic factors, but they found increased psychological functioning in referred patients over time. However, the study was based on parental and self-reported psychological symptoms and did not include referrals after 2016. Studies have increasingly revealed high levels of psychological difficulties among young people applying for gender-affirming treatment, however they are often limited by small case numbers and are mostly based on self-reported psychological symptoms [11, 18,19,20,21,22,23,24,25]. A Norwegian survey on living conditions found that transgender people report lower quality of life and more discrimination and psychological difficulties than other LHBTIQ-groups [26].

To better understand possible time changes, there is a need to describe demographic and mental health variables for earlier and more recently referred young people. The present study includes information on these variables for children and adolescents referred to the Norwegian National Center for Gender Incongruence (NCGI) over two decades. Oslo University Hospital (OUS) has operated, the only gender clinic in Norway since 1979 that offers both medical and surgical gender-affirming treatment including full gender transition.

The high prevalence of mental health problems among transgender people and the potentially irreversible treatment of young people, have generated discussions on the appropriateness of gender-affirming medical treatment and whether psychiatric complexity demands greater caution [27,28,29,30,31,32]. Studies from the Netherlands have been leading in research, but studies from other countries are called upon [33, 34]. Our Norwegian study is a useful complement to existing research owing to its large sample size, the reporting on psychiatric diagnoses and the population-adjusted numbers of referrals up to 2022. Increased knowledge on demographic background and mental health among children and adolescents with GI are important for more informed discussions on treatment approaches for this group of young people.

The aim of this study was threefold. First, we wanted to calculate the yearly population-adjusted numbers of children and adolescents referred to the NCGI in Norway from 2000 to 2022. Second, we wanted to describe the demographic characteristics and prevalence of psychiatric diagnoses, self-harm and suicide attempts among the referred. Third, we wanted to investigate possible time trends in assigned gender, age at referral, psychiatric diagnoses, self-harm and suicide attempts over two decades.

Methods

Design

Gender Incongruence Registry for Children and Adolescents

This study used data collected retrospectively for the Gender Incongruence Registry for Children and Adolescents (GIRCA) in Norway. In 2020, GIRCA was established by the NCGI for children and adolescents at OUS in Norway as part of a quality improvement project. OUS is a public non-cost hospital providing health services for all inhabitants. GIRCA includes all Norwegian applicants to the NCGI under the age of 18 years from 2000 onwards, with the first referred child registered in 2002. All referred persons were included regardless of whether they were given an appointment at the NCGI. The registry and study were approved by the OUS Data Protection Officer (DPO).

GIRCA includes information from a retrospective review of referral letters, supplemented by information that emerged during the assessments and treatment at the NCGI. Referrals to the NCGI were required from the patients’ local specialist healthcare services – the child and adolescent psychiatric outpatient clinics (CAPOCs) that are geographically spread throughout the country. In some cases (approximately 8%), referrals came from private or primary care psychologists. In cases where referrals came from general practitioners (approximately 8%), patients were not given an appointment, but were informed about the need for an assessment at the local CAPOC in case of a new referral.

Evaluations at the CAPOC involve exploring patients’ experience of GD and possible need for further assessment at the NCGI, along with psychological screening. The screening includes developmental history and description of the family situation, a semi-structured diagnostic interview covering most psychiatric disorders among children and adolescents (Kiddie-SADS-PL DSM-5) [35], the Achenbach system of empirically based assessment (ASEBA) [36] and screening for posttraumatic stress symptoms (CATS) [37]. Further assessments are carried out at the CAPOC when indicated, such as cognitive assessment [38] and assessment of autism spectrum disorders (ASD) [39].

GIRCA includes demographic variables, psychiatric diagnoses coded according to the International Classification of Diseases 10th Revision [1], information on self-harm, history of suicide attempts and information on treatment at the NCGI.

Sample and setting

The sample included all patients in GIRCA below 18 years of age at referral from the years 2000–2020, totalling 1258 patients. We added patients referred in 2021 (n = 237) and 2022 (n = 268) for population-adjusted numbers of referrals only.

Measures

Population-adjusted numbers of referrals, 2000–2022

Yearly population-adjusted numbers of referrals for AFAB and AMAB were estimated using population numbers for Norwegian inhabitants under 18 years of age, provided by Statistics Norway [40].

Demographics

We included the following variables from GIRCA: year and age at referral, birth-assigned gender (AFAB = 0, AMAB = 1), change of legal sex, current living situation, being adopted, having same-sex parents, proportion of immigrants, prior or ongoing self-harm and reported suicide attempts during lifetime (coded as yes = 1 and no = 0). Information on self-harm was registered regardless of severity. Only information on suicide attempts was included, excluding solely suicidal ideation.

Psychiatric diagnoses

GIRCA includes data on prior and current psychiatric diagnoses, retrieved from CAPOC referral letters according to ICD-10, assessed by a psychologist or psychiatrist. Diagnoses were supplemented at the NCGI after assessment with a structured diagnostic interview. All diagnoses used in this study were registered prior to any gender-affirming medical intervention at the OUS, to prevent bias of initiated treatment on their mental health. Prior and current psychiatric diagnoses were combined to describe the burden of psychiatric diagnoses among the referred.

Statistical analyses

Gender differences in demographics and psychiatric diagnoses were analysed using an independent t-test (mean differences) or a chi-square goodness-of-fit test/Pearson’s chi-square test, comparing the AFAB and AMAB groups. To examine the interplay between age at referral and gender, we applied Pearson’s chi-square test as presented in Fig. 1. These analyses were computed using SPSS version 28.

To calculate time trends in yearly population-adjusted numbers of referrals, we modelled changes in referrals over time using a binomial regression model in which we used splines to capture the effect of time in a flexible manner. Additionally, the model assumed that because of reduced working capacity at the CAPOCs and the health system in general, referrals during the main COVID-19 pandemic years of 2020 and 2021 were lower than they would have been without the pandemic. To forecast referrals, we assumed that the trend estimated by the regression model for the years 2000–2022 continues into the future.

Time trends in assigned gender, age at referral and psychiatric diagnoses from 2000 to 2020 were analysed using logistic regression analysis. Linear regressions were used for continuous outcomes, binomial regressions were used for counts with an upper limit, and logistic regressions were used for dichotomous outcomes. Time trends were assessed by simple regression with the calendar year as the predictor variable (2002 coded as 0, 2003 coded as 1 and so on) and demographics or diagnoses as the outcome variable. When exploratory data analyses suggested non-linear trends, those were modelled using splines. These analyses also adjusted for baseline differences in prevalence of diagnoses across gender assigned at birth.

Data on registered depressive disorders showed a non-monotonic trend. Therefore, we used a logistic regression model with thin plate splines to estimate the association between referral year and the prevalence of registered depression while also accounting for prevalence differences due to gender assigned at birth. These analyses were conducted in R 4.3 with the mgcv package. We did not find non-monotonic changes in the prevalence of other diagnoses.

Due to the difficulty of interpreting odds ratios in logistic regressions when the baseline risk is not close to zero, we also calculated average marginal effects. These quantify the average yearly change in diagnosis prevalence in percentage points. Additionally, we calculated statistics for interaction effects using average marginal effects. We calculated all marginal effects using the R package for marginal effects [41].

Bonferroni correction was used as a multiple comparison correction at p = 0.05, testing 17 different mental health variables, giving an adjusted alpha level of 0.003.

Changes over time in the number of diagnoses per referred person from 2000 to 2020 were estimated using a binomial regression model. This model also adjusted for gender assigned at birth. To estimate if the effect of time was different for AFAB and AMAB, we also modelled the interaction between these two variables in an additional analysis.

Results

Demographics

For all persons referred to the NCGI in 2000–2020, the proportion of AFAB was 68.4%. Mean age at referral was 14.4 years, ranging from 4 to 17 years (SD 2.76). AFAB were, on average, referred at a statistically significantly older age (14.6 years) than AMAB (13.8 years), t (1256) = 4.91, p ≤ 0.001. The chi-square test for the association between gender and age at referral revealed a shift in the gender difference in the proportion of referrals. As Fig. 1 reveals, there was predominantly AMAB individuals among the youngest referred, and a shift towards more AFAB individuals among the oldest, χ2 (13) = 74.66, p ≤ 0.001.

Fig. 1
figure 1

Age distribution. Proportion of persons assigned female gender at birth (AFAB) and persons assigned male gender at birth (AMAB) by age at referral to the NCGI

Full size image

Among the children and adolescents in the study, 5.2% reported other than a binary gender identity (male or female). More than half (53.3%) had changed their legal gender. One in 10 (10.6%) were immigrants, which is lower than the proportion in the general population (16.8%) [42]. The distribution of immigrants was 46.6% from Western countries, 36.6% from Asia and 17.3% from other parts of the world. Furthermore, 2.7% were adopted and 1.0% had parents of the same sex. Concerning living situation, 43.5% were living with both parents, compared to 63% of all 17 year olds in the general population [43]. For the rest of the sample, 22.7% had parents with shared custody; 16.2% were living with their mother only; 3.1% were living with their father only; 1.4% were living with a partner or other adults, and 5.6% had an unknown living situation. There were no significant gender differences in living conditions except for a higher proportion of AMAB living in either foster care, at an institution or by themselves (10.6% among AMAB compared to 6.0% among AFAB, χ2(1) = 8.11, p = 0.004).

Population-adjusted numbers of referrals

Table 1 presents referral numbers, population numbers and population-adjusted rates from 2000 to 2022 for AFAB and AMAB. As the table indicates, referral numbers were low during the first decade of the millennium. Figure 2 shows the observed number of referrals per 100,000 as solid lines and referrals estimated by the statistical model as dotted lines. The shaded regions for the years 2000–2022 show the 95% and 50% credible intervals for model-predicted numbers of referrals after removing the effect of the COVID-19 pandemic. For the years 2023–2025, the dashed line shows the expected number of referrals, obtained by projecting the 2023 trend into the future.

Table 1 Number of referrals and population adjusted rates per 100.000 of persons assigned female gender at birth (AFAB) and assigned male gender at birth (AMAB) < 18 years referred to the NCGI
Full size table
Fig. 2
figure 2

Population adjusted number of referrals to the NCGI 2000–2025. The shaded regions for the years 2000–2022 show the 95% and 50% credible intervals for model-predicted number of referrals after removing the effect of the COVID-19 pandemic. Model predictions are shown to illustrate that referrals will not necessarily continue to increase at the same rate as in the first post-pandemic years

Full size image

As Table 1; Fig. 2 indicate, there has been an increase in referrals over the last 20 years. The average rate of increase in referrals was higher for AFAB, with a sharp increase from the year 2014 onwards. As expected, there was a transient dip in referrals during the first COVID-19 pandemic year of 2020, which was more distinct among AFAB. For the coming years of referrals, the model predictions show a continued increase or stabilization among AFAB, rather than a decrease. Importantly, the very wide credible intervals indicate a high level of uncertainty for predictions.

Time trends in age and gender during 2000–2020

The increase in referrals has been more prominent among AFAB than AMAB (Fig. 2). We found no significant time trend in mean age at referral (β = 0.008 (0.022), p = 0.729), which indicates that the age at time of referral has remained relatively stable for both AFAB and AMAB.

Prevalence and gender differences in psychiatric diagnoses, self-harm and suicide attempts

Nearly two in three (64.5%) of the referred persons were registered as having one or more previous or ongoing psychiatric diagnoses, which was significantly more prevalent among AFAB (67.8%) than AMAB (57.4%), p ≤ 0.001. For a detailed presentation, see Table 2. The most prevalent diagnoses were depression (33.9%) and anxiety disorders (21.1%), which were significantly more prevalent among AFAB. Other frequently registered diagnoses included hyperkinetic disorders (14.0%), ASD (9.1%) and reaction to severe stress and adjustment disorder (8.3%). We found a significant gender difference in the prevalence of reaction to severe stress and adjustment disorder, where AFAB were more likely to be registered with this diagnosis.

Table 2 Psychiatric diagnoses, history of suicide attempt and prior or ongoing self-harm combined, for persons assigned female gender at birth (AFAB) and persons assigned male gender at birth (AMAB). Psychiatric diagnoses where n < 10, are not included in the table
Full size table

Current or prior self-harm was registered among 35.5% of the referred, and 12.7% had attempted suicide. There was a significant gender difference in the prevalence of self-harm (p ≤ 0.001), predominantly registered among AFAB individuals.

Time trends in psychiatric diagnoses, self-harm and suicide attempts during 2000–2020

Logistic regression was used to analyse the relationship between time trends and psychiatric diagnoses, suicide attempts and self-harm. There was a significant decrease in the probability of being registered with depression over time, OR = 0.91, 95% CI = 0.88, 0.94. For a detailed presentation, see Fig. 3. The average marginal effects show that the prevalence of registered depression diagnoses in the GIRCA sample decreased, on average, by 3.3% points (CI = − 4.2, − 2.4; p ≤ 0.001) per year. For AFAB, the reduction was 3.7% points (CI = − 4.9, − 2.6; p ≤ 0.001), and for AMAB, the reduction was 2.4% points (CI = − 3.7, − 1.2; p ≤ 0.001). The difference in time trends for AFAB and AMAB was not statistically significant (trend AFAB − trend AMAB = 1.3, CI = − 0.4, 3.0, p = 0.143).

Fig. 3
figure 3

Registered diagnosed depression among persons assigned female gender at birth (AFAB) and persons assigned male gender at birth (AMAB) from 2000–2020. Bands represents 95% confidence intervals

Full size image

Note that since most cases were referred during the last 10 years, the results of the time trend analysis are dominated by data from this period. For instance, the prevalence of depression has not decreased by around 3.3% points per year since 2000, but it has decreased by approximately this amount since 2010. No other registered psychiatric diagnoses or prevalence of suicide attempts or self-harm changed significantly during the two decades.

Differences over time in the number of psychiatric diagnoses

We found a significant negative time trend in the number of registered diagnoses per person, OR = 0.98 (CI = 0.97, 1.00, p = 0.02). This corresponds to an average decrease of 0.02 diagnoses per person per year. Adding an interaction term showed that the OR for AMAB was not significantly different from that for AFAB (1.01, CI = 0.98, 1.05, p = 0.4), even though it better revealed a small difference in the average marginal effects between AFAB (− 0.03, CI = − 0.05, -0.01) and AMAB (− 0.01, CI = − 0.03, 0.02). The difference for marginal effects was also not statistically significant (0.018, CI = − 0.01, 0.05, p = 0.26).

Discussion

This study use a sample of all children and adolescents applying for gender-affirming treatment in Norway at the NCGI. The study presents not only referral numbers, but population-adjusted rates over two decades and a prediction for the coming years. We found a sharp increase in referrals to the NCGI favouring AFAB over AMAB. Although the number of registered diagnoses per person decreased over time, psychiatric diagnoses were highly prevalent among referred children and adolescents with GI.

The first aim of this study was to calculate population-adjusted rates of referrals to the NCGI over two decades. By including all referrals and not solely patients under assessment, numbers were not biased by capacity limitations at the clinic. We found a considerable increase in referrals with a spike around 2014–2015, which is comparable to numbers presented in other countries [7, 8, 12, 13, 18]. Interestingly, we found no significant shift in the sex ratio in our sample, as the proportion of AFAB referrals dominated most of the years (Table 1). Other studies have reported a gender shift with proportional more referred AFAB over the last years [7, 8, 12, 13]. One possible explanation may be that Norway has a longer history of being more accepting of AFAB living as transmen, while AMAB could experience more scepticism when living as transwomen. This explanation was also hypothesized in a Dutch and Canadian study by Aitken et al. [6]. However, there were few referrals to the NCGI in Norway during the first 10 years of the millennium (108 referrals in total between 2000 and 2010), which could make the proportions more uncertain.

We found a dip in referrals during the first pandemic year of 2020, which is in line with a decline in mental health consultations in general among children and adolescents that year due to the decreased capacity in health facilities [14]. There was particularly a decrease in AFAB referrals. Referrals to the NCGI in Norway for 2021 and 2022 present higher numbers compared to pre-pandemic years, and our trend model suggests that the number of AFAB referrals is likely to increase or remain stable over the coming years, rather than to decrease. However, predicting future referral numbers implies high levels of uncertainty.

Mean age at referral did not change over time in our study. In “The Dutch Cohort”, there were fluctuations in mean age at assessment, but not a changing trend [18]. Our study found that AFAB were referred at a significantly older age than AMAB, which is consistent with findings from the Netherlands [13] and the UK [7].

Since Norway only has one public NCGI that receives referrals from the whole country, this study is one of few samples that includes all children and adolescents referred to a national gender clinic [7, 13]. Our study makes an important contribution to knowledge on the mental health burden by reporting validated psychiatric diagnoses given by psychiatrists or psychologists. Other clinical studies are mostly based on self- or parental reports through screening instruments [11, 18,19,20,21,22,23,24,25]. The reporting of psychiatric diagnoses, as seen in our study, is important as they may capture more severe mental health problems than symptom descriptions from questionnaires. The prevalence of registered psychiatric diagnoses was much higher in our study (64.5%) than in the general European population of children and adolescents (15.5%) [44]. Results from our study differ from the “Dutch Cohort” from 2002 to 2009, where most (67.6%) of the 105 adolescents with GD had no psychiatric disorders, although the prevalence of certain psychiatric diagnoses was high [45]. In the Dutch study, mental health outcomes were reported by parents of transgender youth and might not reflect the actual level of psychological distress. In addition, due to the small sample size, generalizability is uncertain.

A retrospective cohort study using American military healthcare data from 2010 to 2018 included more than 3,700 transgender youth and their siblings [46]. The study found that transgender youth had five times the odds of having a psychiatric diagnosis than their siblings, and seven times the odds of having suicidal ideation or self-harm. Norwegian children and adolescents with GI in our study also showed high levels of self-harm (35.5%) and suicide attempts (12.7%). Our study could not clarify if the high proportion of self-harm and suicide attempt is caused by prevalent numbers of psychiatric disorders, or if suicidal behaviour is caused by the experience of GD itself. However, the prevalence of self-harm in this sample (35.5%) is higher than in the general population (approximately 16%) [47]. Increasing prevalence of self-harm is reported in the general population [48]. Due to frequent suicidal behaviour, an assessment of suicidality among adolescents with GI is of high importance. Mortality among adolescents with GI should be further assessed using national health registers.

Our study found a reduction in the number of psychiatric diagnoses over time, with an average reduction of 0.02 diagnoses per year out of the 26 psychiatric diagnoses we considered. Arnoldussen et al. found an improvement in parental- and self-reported externalizing problems and peer relations when comparing recently referred youth with GD to earlier referrals in the Netherlands, although referrals were only included until 2016 [18]. The authors posited that the increase in referrals during the most recent years might have been caused by GD being more common than expected, rather than GD being part of psychological difficulties or a lower threshold of GD when seeking help. The decrease in diagnosed depression and number of psychiatric diagnoses over time found in our study, could imply that an increasing group of children and adolescents referred to the NCGI have less psychological difficulties. This could be due to more acceptance for gender exploration and gender diversity in parts of society. However, a contributing factor for the decline could be that persons referred the most recent years, have not yet been assessed with a structured psychiatric interview at the NCGI. However, all psychiatric diagnoses from CAPOC referral letters are included, also among persons referred the most recent years. Time trend analyses in psychiatric diagnoses should be repeated after sufficient time to verify or dismiss the time trends in psychiatric diagnoses. Frequencies of depressive disorders were still higher than in the general population [49], and the total mental health burden among children and adolescents with GI emphasizes the need for a biopsychosocial approach and local caretaking of psychological difficulties.

We found a co-occurrence of ASD among 9.1% of children and adolescents in this study, which is comparable to findings from the Netherlands (7.8%) [50] and a systematic literature review and meta-analysis (11%) from 2022 [51]. The prevalence of ASD is around 1–2% in the general population [52,53,54]. The co-occurrence of ASD could be a contributor to the dissatisfaction of pubertal changes and developing sexuality [55]. Thus, clinicians working at gender clinics should have the competence in child- and adolescent neurodevelopmental disorders to disentangle symptoms of GI and ASD and to better adjust their communication during assessments.

Notably, gender differences in psychiatric diagnoses in our study were consistent with population-based studies on psychiatric disorders among adolescents, showing that children and adolescents with GI follow the same pattern as their peers given their gender assigned at birth [56, 57]. The prevalence of mental health diagnoses from specialized healthcare services increased by 40% among all girls aged 15–17 years in Norway from 2008-2011 to 2016 [56]. We found significantly more registered psychiatric diagnoses among AFAB than AMAB in our study. Due to the heterogeneity of children and adolescents seeking gender-affirming treatment, there should be a continuous discussion among health politicians and clinicians, in collaboration with user organizations, to assure the best holistic approach to gender health care, as described in recently published statements and reports [58, 59].

Limitations

The study was limited by its retrospective design, as data were retrieved from a quality registry on GI based on a retrospective chart review and could therefore lack information that was not included in medical records. As intake at the NCGI requires an assessment at the local CAPOC, this routine could lead to more children and adolescents receiving psychiatric diagnoses compared to children and adolescents in the general population. The study included all children and adolescents consulting with the NCGI for GI, but could not determine the total population with GI.