The Demography of Sweden's Transgender Population: A Research Note on Patterns, Changes, and Sociodemographics

Abstract

We examine the prevalence of gender transitions in Sweden over time and document the sociodemographic characteristics of people transitioning in different periods. Using administrative data covering the transgender population from 1973 through 2020, we analyze two common events in a gender transition: the earliest diagnosis of gender incongruence and the change of legal gender. Our research note presents three main findings. First, the measured prevalence rates of diagnoses and legal gender changes are relatively low in all periods, although they have increased substantially since the early 2010s. Second, the recent increase in transition prevalence is most pronounced among people in early adulthood; in particular, young transgender men drive an increase in overall transition rates through 2018, followed by moderate declines in 2019 and 2020. Third, transgender men and women have substantially lower socioeconomic outcomes than cisgender men and women, regardless of the age at which they transition or the historical period. They are also considerably less likely to be in a legal union or reside with children. These findings highlight the continued economic and social vulnerability of the transgender population.

Introduction

Since the late twentieth century, transgender people have achieved greater societal recognition and improved access to gender-affirming health care (Schilt and Lagos 2017; Stryker 2017). Alongside these developments, the number of people who consider themselves transgender has increased in many countries (e.g., Lagos 2022). Yet, research on the demography of the transgender population remains scarce and fragmented. There is limited knowledge about the prevalence and timing of transition-related events, such as when people receive a diagnosis of gender incongruence or change their legal gender. Additionally, the changing sociodemographics of the transgender population are poorly documented and understood (Badgett et al. 2021; Lagos 2022; Witten 2014). Higher quality data are crucial to inform policies that address the needs of a vulnerable group at risk of discrimination, socioeconomic difficulties, and mental health struggles (Aldridge et al. 2022; Dhejne et al. 2016; McCann and Sharek 2016; Nolan et al. 2019; Schilt and Lagos 2017), as well as to contribute to discussions on how to measure transgender populations (Bates et al. 2022).

In this study, we analyze the prevalence and sociodemographic characteristics of transgender people in Sweden using administrative data from the period 1973–2020. We define the transgender population as all people observed with a diagnosis of gender incongruence in population-wide health records.¹ The data enable us to examine two common events in a gender transition: the year people receive their first diagnosis and, for a subset of those with a diagnosis, the year they change their legal gender. We also document such sociodemographic characteristics as civil status, education level, and labor earnings at the time of these two events.

Our work complements research that studies a country's transgender population using cross-sectional surveys in which people self-report their gender identity. Unlike those studies, we can provide a population-wide count of transgender people over time and examine age, cohort, and period trends in detail. One limitation, however, is that our data capture only the medical and legal aspects of gender transition, which implies that we study a segment of the broader gender-diverse community. Our results should be interpreted as representing people seeking medical care to affirm their identity and who receive a gender incongruence diagnosis.

Background and Previous Research

Prevalence estimates of the transgender population vary considerably by data source, sex assigned at birth, and age span (for an overview, see Collin et al. 2016; Goodman et al. 2019; Meier and Labuski 2013; Zhang et al. 2020). The lowest estimates of overall population prevalence come from administrative data on legal gender changes (e.g., 0.0138% in the Netherlands; Kuyper and Vanden Berghe 2017) and medical data related to gender-affirming care (e.g., 0.017–0.033% in an international review by Zhang et al. (2020)). The highest estimates come from census and survey data on self-reported gender identity: for example, 0.33% reported being transgender or nonbinary in Canadian census data (Statistics Canada 2022), 0.54% in census data from England and Wales (Office for National Statistics 2023), 0.44% in Scotland's census (National Records of Scotland 2024), and 0.4–2% across 19 studies in five high-income countries reviewed by Zhang et al. (2020). Transgender women (assigned male at birth) typically outnumber transgender men (assigned female at birth) in older data, but the numbers are now more similar (Collin et al. 2016; Goodman et al. 2019; Lagos 2022). The greater representation of transgender men among younger cohorts has also been found more broadly in qualitative interviews and mixed-methods studies (e.g., Hammack et al. 2022; Puckett et al. 2022). Recent data also reveal substantial age differences, with prevalence being about five times as high for transgender individuals among adolescents than among adults (e.g., Herman et al. 2017; Statistics Canada 2022).

Several important gaps exist in the literature. First, few researchers have taken a demographic approach to studying transgender populations, with a detailed focus on age, cohort, and period trends. An exception is Lagos (2022), who used U.S. survey data to examine changes across birth cohorts and found large increases in the share of transgender people born after 1984. Second, few articles include comprehensive sociodemographic information. Some evidence from U.S. surveys and Dutch and Danish administrative data indicates that transgender and gender-nonconforming people have worse economic outcomes than cisgender people, particularly cisgender men (Badgett et al. 2021; Carpenter et al. 2020, 2022; Geijtenbeek and Plug 2018; Lagos 2022). These differences are more pronounced for transgender people with a medical diagnosis and legal gender change than in self-reported data (Thomsen et al. 2024). However, the patterns for education, ethnicity, civil status, and household structure are less consistent across contexts, data sources, and comparison groups. Additionally, we know relatively little about changes in sociodemographic patterns over time.

Our study helps fill these gaps using Swedish administrative data that span several decades. Since 1972, Sweden has allowed transgender people to change their legal gender and receive gender-affirming care through the publicly funded health care system (Statens Offentliga Utredningar 2017). Before receiving care or changing legal gender, one must undergo extensive psychological evaluation at a specialized clinic and obtain a gender incongruence diagnosis (for additional context, see sections B and C in the online appendix). Thereafter, doctors also record a gender incongruence diagnosis when providing gender-affirming care, such as hormone therapy or surgery. In Sweden's administrative data, researchers can match diagnoses made during these visits to records of legal gender changes and sociodemographic characteristics. The data are population-wide and thus less likely to suffer from selective disclosure and low response rates. Yet, such data have scarcely been used in social science research, and most sociodemographic knowledge about the transgender population still originates from government reports (National Board of Health and Welfare 2020), cross-sectional surveys (Åhs et al. 2018), and limited descriptive statistics in medical studies (Bränström and Pachankis 2020).

Data and Methods

Sources and Measures

We use longitudinal, deidentified administrative data covering all residents of Sweden from 1973 to 2020. We observe two common events in a gender transition: the earliest diagnosis of gender incongruence and the change of legal gender.² To determine if and when people received a gender incongruence diagnosis, we use diagnoses recorded in the National Patient Register from the National Board of Health and Welfare. From 1973 to 2000, our data cover hospital stays, and from 2001 to 2020, they also cover visits to specialized care (see section A in the online appendix for details about diagnostic codes). We match all people with a gender incongruence diagnosis to records of legal gender changes from the Swedish Tax Agency. Finally, we obtain sociodemographic information for the full adult population using registers from Statistics Sweden. The study was done following approval from the Swedish Ethical Review Authority.

Our population is restricted to people aged 15–65 when studying transitions based on first diagnoses.³ When analyzing legal gender changes, we restrict the calculations to people aged 18–65 because minors cannot change their legal gender, and we use the same age bracket for sociodemographic characteristics because we have sociodemographic data only for those 18 or older. Our main results exclude people older than 65 because transitions are rare at older ages and including the elderly population would thus attenuate our findings.⁴

We use three types of measures to describe the transgender population in Sweden: cumulative measures, transition rates, and cohort-based prevalence measures. Our approach is event-based, defining a transition as either the first diagnosis of gender incongruence or the change of legal gender. In a strict statistical sense, people are considered cisgender before the observed transition event and transgender after.⁵ We refer to people assigned male at birth transitioning to female as “transgender women,” and people assigned female at birth transitioning to male as “transgender men.” Notably, these categorizations of people as cisgender/transgender and men/women may not align with how people self-identify or present their gender. For example, these categories exclude nonbinary gender identities and count people who transition without medical steps as cisgender.⁶Section A in the online appendix further explains our measurements and definitions.

We also document sociodemographic patterns in the transgender population compared with the cisgender population. We study people's birthplace, municipality type, civil status, coresidence with young children, years of education, and annual labor earnings (see Table S1 for exact definitions; all tables and figures designated with an “S” are available in the online appendix). We calculate average characteristics by gender, age group, and historical period. Sociodemographic data for transgender people are measured in the year of their observed transition to ensure comparability across age groups and periods; however, we note that transition-related factors may influence the traits that we study. For the comparison groups, we include cisgender people in our calculations every year they appear in the data within our period windows.

Our approach has several advantages over analyzing survey data from a single period. All residents of Sweden are included in Swedish administrative data, thus reducing the risk of undercount biases that may arise in surveys if respondents withhold their gender identity for fear of discrimination or other negative consequences (Bates et al. 2022; Festy 2007).⁷ Moreover, the data's longitudinal structure enables us to document how the prevalence of gender transition—as defined by our events—has changed over time. We have consistent numerators and denominators of transgender and cisgender people, facilitating our calculation of rates and prevalence levels by age, period, and cohort.

Context and Interpretation of Our Data

Our results should be interpreted in light of the fact that we define the transgender population through diagnostic codes in administrative health records rather than self-reported data on gender identity (for further discussion, see Collin et al. 2016). This approach provides relevant insights for addressing important policy questions, such as allocating resources across the health service. Nevertheless, we want to stress that receiving a diagnosis of gender incongruence is related to—but not synonymous with—identifying as a different gender from one's assigned sex, taking steps to socially transition (e.g., changing pronouns), or undergoing gender-affirming medical treatments, such as hormone therapy or surgery. This lack of perfect overlap has several implications. First, while we include people who transition medically and legally but might not disclose their transgender status in surveys and censuses, we exclude those who self-identify as transgender or socially transition without pursuing medical steps. Studies suggest that our approach yields a lower estimate of population prevalence than using self-identification to define transgender people (Zhang et al. 2020). In particular, we likely capture a smaller share of nonbinary people relative to what would be captured using self-identification because of the historical focus of gender-affirming care on binary transitions and the fact that Sweden has only two legal gender markers. Although no representative data exist on the self-reported gender identity of those who receive a gender incongruence diagnosis, there is survey evidence that approximately 90% use binary pronouns (Axfors et al. 2023).

Moreover, our transition measures may not coincide with the timing of other milestones in a person's gender transition. While Swedish survey data suggest that nearly everyone who receives a gender incongruence diagnosis has taken steps to socially transition (Axfors et al. 2023), our administrative records do not indicate when their social transitions began. These transitions likely predate the first diagnosis, as many people experience a mismatch with their assigned sex and socially transition before seeking gender-affirming care. Furthermore, because of institutional constraints, legal changes often occur several years after the initial diagnosis. Trends in these measures therefore occur with a lag relative to diagnostic measures.

Our results should also be interpreted in light of the fact that demographic trends in the transgender population are shaped by systemic barriers and other societal processes. In Sweden, these barriers have gradually declined over time (for a detailed discussion, see sections B and C in the online appendix). The diagnostic criteria for gender incongruence have included a less binary perspective of gender since the mid-2010s (National Board of Health and Welfare 2015), and changing legal gender became less burdensome following the removal of divorce, sterilization, and citizenship requirements in 2013. Later-born cohorts have thus had better access to gender-affirming care and legal gender recognition. Although this increase in access has partly been offset by lengthy evaluations and extended wait times,⁸ improved access to care over time is reflected in our results.

Results

Population Trends by Gender, Age, and Birth Cohort

Figure 1 uses two measures to document an increasing trend in the number of transgender people in Sweden over time: the cumulative number of transitions (panel a) and the rate of transition (panel b), both per year and gender. A transition is a nonrepeatable event, defined as the first time a person receives a gender incongruence diagnosis (darker lines) or changes their legal gender (lighter lines).

From 1973 to 2020, we observe a total of 4,140 transgender men and 3,464 transgender women with a gender incongruence diagnosis (out of roughly 10 million people in Sweden). Restricting to the subset who have changed legal gender, we observe 1,537 transgender men and 1,422 transgender women. We find a clear increase in the number of transgender people over time. Moreover, transgender women outnumbered transgender men until 2016 on the basis of diagnoses and until 2018 on the basis of legal gender change.

When defining a transition as the first diagnosis of gender incongruence, we found low and slowly increasing transition rates through 2000. Thereafter, our data cover diagnoses from outpatient visits, leading to a jump in the transition rate for transgender women in 2001 and a less dramatic increase for transgender men. The transition rate for transgender women remained fairly stable over the next decade and then increased from 2011 to 2018. Transgender men had a lower transition rate than transgender women in the early 2000s but their transition rate increased over time and finally surpassed the rate for transgender women in 2013. The transition rates for both groups peaked in 2018, followed by moderate declines in 2019–2020 that were larger for transgender men.

The rates expressed as a probability that a person receives a gender incongruence diagnosis in any given year are low, peaking at about 0.02%. The transition rates when defined as changing legal gender are even lower and display less variation over time; however, we observe an increase in 2013, when the requirements for changing legal gender became less strict, and then again a few years later. Similar to rates based on diagnostic codes, we find a peak in 2018 for the rate of legal gender changes, but the declines in 2019–2020 are quite modest.

Figure 1 hides substantial age variation, which we document in Figure 2. Panels a and b of Figure 2 show that the transition rates based on first diagnosis are much higher for people younger than 30 than for older adults, particularly for transgender men. There was a large increase for transgender adolescent men aged 15–19 from 2013 to 2016; however, this subsequently slowed and reversed after 2018. Additional analyses at a half-year level show that the reversal began in 2019 and was thus unrelated to COVID-19 (see Figure S3, which also shows stable rates of legal gender change in 2021). The reversal instead coincided with increased media attention in Sweden on standards of care for transgender youth, as well as staff shortages at gender clinics.⁹ For transgender women, we find an increase in younger age groups through 2014 and stable rates thereafter. Legal gender changes also increased most among people younger than 30 (see panels c and d of Figure 2).

To understand the prevalence of diagnoses and legal gender changes for people born in different years, we also examine our data from a cohort and life course perspective. In Figure 3, the top yellow line shows the total prevalence of gender transition at the end of 2020 by birth year. The other lines show the prevalence of transition by different ages. Overall, we find a slower increase by birth cohort, compared with the increase over time periods. Nevertheless, each subsequent birth cohort was more likely to transition by a comparable age.

For transgender women, prevalence increased slowly across the 1950–1985 cohorts (Figure 3, panel a). Among those born after 1985, the proportion approximately doubled and then declined as the latest cohorts are observed only at young ages. In younger cohorts, total prevalence increased tenfold—from about 0.015% to about 0.15%—relative to the oldest cohorts. The cohort trends in legal gender changes—shown in Figure 3, panel c—exhibit a more stable pattern.

For transgender men, transition prevalence increased rapidly over cohorts, with diagnoses occurring at younger ages (Figure 3, panel b). In the latest cohort, whom we identify only through age 18, cohort prevalence was still much higher than in earlier cohorts. The cohort prevalence for transgender men increased from approximately 0.01–0.02% for those born in the 1960s–1970s to about 0.5% among those born in 1999. The cohort trend for legal gender changes—shown in Figure 3, panel d—is less marked.

An important implication of these analyses is that prevalence measures of the transgender population vary significantly depending on the event type and age span used to calculate the numerator and denominator. We summarize this point in Table S2, documenting how the prevalence of the transgender population in 2020 changes across event types and a wide range of age definitions (e.g., everyone aged 18 or older with no upper limit, those aged 18–65).

Sociodemographic Characteristics by Age Group and Transition Period

Table 1 documents transgender people's life circumstances at the time of transition and whether these have changed over transition periods. We compare the average sociodemographic characteristics of transgender men and women in the year of their first diagnosis with those of their cisgender counterparts in the same age groups (18–29 and 30–65). The data are also divided into three historical periods:

1. early, 1990–2000, spanning the start of our socioeconomic data through the last year that our data include only hospital stays;

2. pre-reform, 2001–2012, spanning the start of records including visits to specialized care until the end of the sterilization requirement; and

3. post-reform, 2013–2020, spanning the end of the sterilization requirement through the last year for which we have data.

Our approach largely captures an age-balanced comparison population within each period, and Tables S3 and S4 confirm that the results hold if we reweight the cisgender population to produce an exact age-matched sample. Furthermore, Table S5 shows similar findings when measuring characteristics in the year of legal gender change, which typically occurs about three years after the first diagnosis.

Overall, transgender men and women exhibit more similarities to each other than to cisgender men or women for all the sociodemographic characteristics we study. Across age groups and historical periods, transgender men and women had substantially lower earnings and higher proportions with very low earnings compared with cisgender people. For example, in the post-reform period, transgender men and women aged 30–65 earned 40–70% less than cisgender men and women of comparable age, and they were about 2.5 times as likely to have very low earnings. In the younger age group, the difference is even more pronounced. Transgender people typically had slightly lower educational attainment than their cisgender peers, which contrasts with notions of transgender identities as being particularly commonplace in university environments (as also discussed by Lagos (2022)).

Transgender people were much less likely than cisgender people to be in a legal union or have young children living in their households, irrespective of their age and transition period. Transgender women were the least likely to live with children, but both transgender men and women diverge clearly from cisgender men and women. Differences decline somewhat over time. For people who transitioned during the early period, the foreign-born were overrepresented among transgender people relative to the Swedish-born, but this was no longer the case in later periods. The early period also shows a substantial overrepresentation of transgender people living in metropolitan areas, but in later periods, we find smaller differences.

Conclusion

Using population-wide administrative data, we studied the prevalence of gender transitions in Sweden from 1973 to 2020—a period of gradual legal advances, increasing societal acceptance, and a more accommodating health care system for transgender people.

We find that the number of people undergoing medical or legal gender transition is low. By 2020, only 0.083% of Sweden's population aged 18 or older had received a gender incongruence diagnosis and 0.034% had changed legal gender (see Table S2 for different definitions). These numbers are significantly lower than overall prevalence estimates from surveys and censuses—for example, 0.5% in the United States (Herman et al. 2017). Consistent with the age patterns observed in these surveys and censuses, our prevalence measures are highest for younger age groups, peaking at 0.5% for transgender men born around 2000. In general, our estimates are lower than comparable estimates in studies focusing on self-identification (Zhang et al. 2020; Zucker 2017) but higher than in studies focusing on gender-affirming medical procedures or legal gender change in other countries (Arcelus et al. 2015; Collin et al. 2016; Zucker 2017).

We find a sharp increase in the rate of legal gender change in the 2010s, though smaller in magnitude than the increase in the rate of first diagnosis and occurring later in the life course. After 2018, we see a modest decline in overall transition rates and a stabilization for adolescent transgender men. We also find that transgender people—particularly transgender men—receive their first diagnosis at increasingly younger ages. Furthermore, in earlier periods, we find a larger share of transgender women, which then shifted to a larger share of transgender men in the 2010s. Although the trends we find are broadly similar to those found in the United States over the same period, the reversal in the ratio of transgender men to women that we find has not been observed in the United States (Lagos 2022). Additionally, our result that new gender transitions have reached a point where rates are no longer increasing over time and instead are relatively level has, to our knowledge, not been found in other contexts.

In addition to analyzing demographic changes, we examine transgender people's life circumstances in the year of earliest diagnosis and legal gender change. Irrespective of the historical period when they transitioned, transgender people earn substantially less than cisgender people, with larger gaps at younger ages. This finding is consistent with survey evidence from the United States (Carpenter et al. 2020, 2022) and illustrates that Sweden's transgender population is a group with markedly disadvantaged socioeconomic outcomes. Their precarious situation is most likely reinforced by—and a result of—the challenges, negative attitudes, and systemic barriers they face. We also document a lower prevalence of partnership and coresident children, which can be viewed as a social vulnerability. This may partly be explained by previous requirements to be unmarried and to have undergone sterilization before changing legal gender.

We want to emphasize two important points concerning the interpretation of our results. First, our use of medical records to identify transgender people implies that we study a subgroup of the broader gender-diverse population. We focus on the medical and legal aspects of gender transition, as opposed to changes in how people self-identify or express their gender. The applicability of our results may be limited when it comes to understanding other facets of gender transition or gaining insights about gender-diverse people who do not want or receive gender-affirming care. Future research should triangulate our findings with data measuring other aspects of transgender and nonbinary people's experiences.

Second, the demographic changes that we observe do not necessarily indicate that gender incongruence has become more common over time or at younger ages. Our measures are affected by contextual changes, such as increased transgender visibility, improved access to gender-affirming health care, and less restrictive requirements for legal gender recognition. These advances may account for both the upward trend in transition prevalence and the shift toward receiving a diagnosis or changing legal gender earlier in life. Further research is necessary to understand why we find our observed patterns.¹⁰

Despite its limitations, our approach provides valuable insights into the changing composition of the transgender population receiving medical care in Sweden. Our findings can be used to inform policy decisions—for example, on the number of gender identity clinics. Additionally, they highlight the ongoing economic and social vulnerabilities faced by the transgender population, underscoring the need for policy measures that provide support.

Our results also have methodological implications for research on transgender demography. We find large differences by age group and transition event, implying that prevalence measures are sensitive to the definition of both the numerator (who is transgender) and the denominator (whom we compare with, particularly which ages). We therefore encourage researchers to report multiple dimensions of age, period, and cohort trends when studying transgender demography, and to be mindful of how definitional changes can affect their estimates.

Acknowledgments

We are grateful for financial support from the Swedish Research Council (grants 2022-01863 and 2022-02361) and the Royal Swedish Academy of Sciences (grant SO2018-0015). We want to thank Aino-Maija Aalto, Cristina Bratu, Kitt Carpenter, Cecilia Dhejne, Marie Evertsson, Gilbert Gonzales, Stefanie Möllborn, Ely Strömberg, and Maaike van der Vleuten for helpful suggestions and edits. We also received useful feedback from participants of the Economics of LGBTQ+ Individuals Virtual Seminar Series, the 2021 Workshop on Life-Changing Transitions in the LGBTQ+ Community, and the Swedish National Conference on Sociology 2022, as well as members of the SOCPOL and GAINS research groups at SOFI, Stockholm University. Special thanks go to Helena Berglund at FPES for valuable input and to Ana Tramosljanin and Sandra Thiman for excellent research assistance. We are also immensely grateful to everyone at the Swedish gender clinics who generously answered our many questions.

Ethical Approval

Our research has been approved through the regional Ethics Review Board in Stockholm (2018/560-31/5) and the Swedish Ethical Review Authority (2021-03229 and 2022-03462-02).

Notes

References