What the studies show
The rise in diagnoses and the shift in who is presenting are the best-documented findings here, coming from large record-based studies in several countries. In English GP records, the share of under-19s with a recorded gender dysphoria diagnosis rose more than 50-fold between 2011 and 2021, to 8.3 per 10,000, and to 42 per 10,000 among 17- and 18-year-olds; by 2021 rates in girls were about twice those in boys (Jarvis et al., 2025). German insurance data on about 14 million young people showed an eight-fold rise in confirmed diagnoses from 2013 to 2022, highest in girls aged 15–19 (Bachmann et al., 2024). Earlier clinic data captured the start of the shift: in Toronto and Amsterdam, adolescent referrals flipped from a male to a female majority after 2006, while the sex ratio of other mental health referrals barely moved (Aitken et al., 2015). Co-occurring problems are consistently common: 52.7% in English GP records had anxiety, depression or self-harm recorded (Jarvis et al., 2025), 72.4% in Germany had another mental health diagnosis, and only 36.4% of those diagnosed in 2017 still carried the diagnosis five years later (Bachmann et al., 2024).
Why the rise occurred is far less settled. The "rapid-onset gender dysphoria" hypothesis originates in a survey of 256 parents recruited from websites skeptical of youth transition, who described prior mental health diagnoses and friend groups coming out together (Littman, 2018). A study of 173 adolescents under 16 at Canadian clinics found that those who recently realized their gender identity did not have more mental health problems or rely more on online or trans peers (Bauer et al., 2022), but critics argue it measured time since realization rather than onset relative to puberty, and only about 24 adolescents fell in the "recent" group. A narrative review judged the hypothesis unproven but worth serious study (Leonhardt et al., 2025).
Limits of the evidence
Record-based studies count codes and referrals, not true prevalence, and cannot separate a real increase from greater awareness, reduced stigma, recoding or service changes; a dropped code is not necessarily desistance. No study directly tests peer or social media influence with prospective, population-level data, and the evidence on both sides of the social-influence question rests on small, self-selected or clinic-only samples.