Taiwan: Suicide Death Risk 4× Higher in Transgender People — and Still 2.6× Higher Than in Their Own Siblings

Mortality among transgender persons: a Taiwan matched-population and sibling-comparison cohort study

Cohort & registry studies Mental health & suicide Autism, trauma & co-occurring conditions

Authors: Chih-Wei Hsu, Yang-Chieh Brian Chen, Liang-Jen Wang, Mu-Hong Chen, Yao-Hsu Yang, Chih-Sung Liang, Po-Yen Chen, Edward Chia-Cheng Lai
Journal: The Lancet Regional Health – Western Pacific
Year: 2025
Source: https://doi.org/10.1016/j.lanwpc.2025.101715

Summary

Using Taiwan's national health insurance records, researchers followed about 3,900 people diagnosed with gender identity disorder by psychiatrists. They compared their deaths with matched members of the public and with their own non-trans brothers and sisters. Over about nine years, trans people had a 40% higher death rate than the public and about four times the suicide rate. Deaths from accidents and illness were not higher. Compared with siblings, who share family background and genes, overall deaths were no higher, but suicide was still about 2.6 times higher. Depression, anxiety and other mental illnesses were several times more common in the trans group, and taking depression into account shrank the suicide link the most. The study is careful and the sibling comparison is a strength. But it rests on just 35 suicides and has no data on hormones or surgery.

Key findings

The study followed 3,906 people diagnosed at least twice by a psychiatrist with gender identity disorder (average age about 25; 71% legally male) and 15,624 matched people for about 9.5 years on average.
Mental illness was much more common in the trans group: depression 41% vs 6%, anxiety 29% vs 6%, bipolar disorder 7% vs 1%, and psychosis 7% vs 1%.
Compared with the public, trans people were 40% more likely to die from any cause, about twice as likely to die from outside causes such as suicide or accidents, and about 4 times as likely to die by suicide (35 deaths vs 31 in a group four times larger).
Deaths from accidents and from illness were not higher than in the comparison group.
Compared with 4,765 non-trans siblings, overall deaths and deaths from outside causes were no higher, but suicide was still about 2.6 times higher.
The extra suicide risk was largest in people whose legal sex was female (about 6 times) and in teens (about 13 times, but based on very few deaths, so this figure is very uncertain).
Taking depression into account shrank the suicide link the most. Among people with depression, the difference in suicide could easily have been chance.

Abstract

BACKGROUND: Transgender persons have been reported to experience excess mortality, but evidence is dominated by Western cohorts and seldom addresses familial confounding. METHODS: We conducted a nationwide, retrospective cohort study using two prespecified comparison designs: matched population controls (1:4, matched on legal sex and birth date) and within-family cisgender sibling comparisons. Transgender persons were identified by ≥2 psychiatrist-recorded gender identity disorder diagnoses during 2001-2021; cohort entry required age ≥6 years. Participants were followed until death or December 31, 2022. We compared all-cause, external-cause (suicide and accidents), and internal-cause mortality using Cox regression adjusted for sociodemographics and medical comorbidity. FINDINGS: Among 3906 transgender persons (mean age 24.6 years; 70.6% with a male legal sex) and 15,624 controls (mean follow-up 9.5 years), psychiatric conditions were markedly over-represented in the transgender cohort (e.g., depressive disorders 40.8% vs 5.7%). Adjusted hazard ratios were 1.40 (95% confidence interval 1.06-1.84) for all-cause mortality, 2.03 (1.39-2.96) for external causes, and 4.07 (2.52-6.59) for suicides; accidents (0.56, 0.21-1.45) and internal-cause (0.98, 0.65-1.48) mortality did not differ. In sibling comparisons (4765 siblings), excess all-cause and external-cause mortality were not observed (0.96, 0.67-1.36; and 1.14, 0.70-1.85), suicide risk was higher (2.57, 1.32-5.00), and accident mortality lower (0.19, 0.06-0.67). Relative hazards were greatest in those with a female legal sex and in adolescents. Adjustment for psychiatric clusters attenuated-but did not eliminate-the suicide mortality excess. INTERPRETATION: In Taiwan, transgender persons had higher all-cause and suicide mortality than population controls; in within-family comparisons, the suicide mortality excess persisted, indicating influences beyond shared familial factors. Prevention should prioritize targeted mental-health and safety interventions, with particular attention to adolescents and those with a female legal sex. FUNDING: Taiwan National Science and Technology Council.

Conclusion

In Taiwan, transgender persons had higher all-cause and suicide mortality than population controls; in within-family comparisons, the suicide mortality excess persisted, indicating influences beyond shared familial factors. Adjustment for psychiatric clusters attenuated, but did not eliminate, the suicide excess, with depressive disorders producing the greatest attenuation. Prevention should prioritise targeted mental-health and safety interventions, with particular attention to adolescents and those with a female legal sex, including structured suicide-risk assessment at key points of care.

Methodological limitations

Mental illness only partly accounted for: Each type of mental illness was taken into account one at a time rather than all together, and the timing relative to diagnosis was not checked. So the study cannot say how much of the extra suicide risk is explained by mental illness overall, and the authors admit they cannot tell whether mental illness is a separate cause or part of the pathway.
Trans people were checked more often: In Taiwan a psychiatrist must make the diagnosis, so trans people saw psychiatrists far more, and their mental illnesses were more likely to be recorded than other people's. The authors note this may distort how much the results were adjusted for mental illness.
Sibling comparison timing: Siblings were followed from 2001 or from birth, and the trans group's death rate in this comparison was less than half that in the main comparison. This suggests years before diagnosis were counted, when people could not yet have died as diagnosed trans people. That can make the differences look smaller than they are.
Few deaths: The results rest on 35 suicides and 73 deaths in total. Subgroup figures, especially for teens (about 13 times higher, but anywhere from about 1.5 to 110 times), are very uncertain, and many extra analyses were run.
No treatment data: There was no information on hormones or surgery (national insurance does not cover surgery and covers hormones unevenly). So the study cannot say whether transition affects the risk of death.
Legal sex, not birth sex: Sex came from the earliest insurance record from 2001. Surgery before then could not be seen, so some results split by sex may put people in the wrong group.
Many factors not measured: There was no information on smoking, drug or alcohol use beyond diagnoses, family rejection or being victimised. Explanations offered by the authors (minority stress, family rejection, avoiding sports) were not measured.
A selected group: Needing two psychiatrist diagnoses picks out people who sought formal assessment, often to qualify for surgery. The results may not apply to other trans people or to countries with different health systems.

detrans.ai

Summary

Key findings

Abstract

Conclusion