瑞典登記資料：被診斷為性別焦慮者的死亡率幾乎是配對對照組的兩倍

2001–2023年瑞典性別焦慮個體全因及特定原因死亡率的全國估計：一項基於登記的世代研究

Authors: Kristen D. Clark, Richard A. White, Sarah S. Jackson, Alkistis Skalkidou, Thomas Frisell, Fotios C. Papadopoulos
Journal: EClinicalMedicine
Year: 2026
Source: https://doi.org/10.1016/j.eclinm.2026.104088

Summary

Swedish researchers found everyone diagnosed with gender dysphoria in national health records between 2001 and 2023, about 9,000 people who were about 25 on average when diagnosed. They compared deaths with men and women of the same age and county. Over about seven years, 211 people with gender dysphoria died, almost twice the death rate of the comparison groups after allowing for age. Extra deaths came from accidents, violence and suicide (suicide was not counted on its own), from unclear causes, from mental and brain conditions such as drug or alcohol problems, and from cancer in people born female. The authors call for public health action but say the records cannot explain why. Nothing beyond age and county was taken into account, and many causes had very few deaths. So the study shows a gap in deaths but not what causes it.

Key findings

研究涵蓋 2001–2023 年間在瑞典被診斷為性別焦慮的 9,038 人（4,181 人出生時為男性，4,857 人出生時為女性），與 179,816 名年齡和縣市相同的人比較。
共有 211 人死亡；考慮年齡後，死亡率為每年每 1 萬人 183 人對 96 人，約為兩倍。
在出生時為男性者中，死亡率為女性的 2.3 倍、男性的 1.6 倍，意外、暴力和自殺（3.2 倍和 2.5 倍）、原因不明（約 5 倍）以及精神或腦部狀況（3.6 倍和 2.6 倍）造成的死亡較多。
在出生時為女性者中，死亡率為女性的 2.1 倍、男性的 1.7 倍，意外、暴力和自殺（2.2 倍和 1.5 倍）以及癌症（2.4 倍和 1.8 倍）造成的死亡較多。
自殺、謀殺和意外合計造成 211 例死亡中的 76 例（36%），但自殺沒有單獨報告。
平均而言，出生時為男性者每例死亡提早 6 年（與男性相比），出生時為女性者提早 9 年（與女性相比）。
作者表示，這些發現是追蹤死亡的起點，不是關於原因或治療效果的證據。

Abstract

BACKGROUND: Transgender and gender-diverse (TGD) people experience elevated mortality risk, yet population-based estimates of cause-specific mortality remain limited. Prior studies have focused on narrow subgroups or outdated cohorts, leaving gaps in understanding current mortality disparities. Therefore, the purpose of this study is to quantify all-cause and cause-specific mortality among all TGD adults with a diagnosis of gender dysphoria in Sweden. METHODS: This cohort study provides national cause-specific mortality estimates among TGD adults in Sweden using registry data (2001-2023). TGD people were identified using ICD-10 codes for "gender dysphoria" and matched to approximately 10 cisgender controls of each sex by age and county. Age-standardized mortality rates, absolute mortality risk differences, mortality rate ratios (MRR), and years of potential life lost (YPLL) were calculated overall and stratified by sex assigned at birth. FINDINGS: A total of 9038 TGD people were identified, among them, 211 deaths were observed. All-cause mortality risk was almost twice that of the cisgender comparison group (MRR = 1.9, 95% confidence interval [CI] 1.7, 2.2). Stratified analyses showed distinct mortality patterns for transfeminine (assigned male sex at birth) and transmasculine (assigned female sex at birth) people, with higher risk from ill-defined diseases among transfeminine people (cisgender women as reference: MRR = 4.9, 95% CI 2.3, 10.3; cisgender men: MRR = 5.5, 95% CI 2.6, 11.4) and transmasculine people from neoplasms (cisgender women: MRR = 2.4, 95% CI 1.4, 4.1; cisgender men: MRR = 1.8, 95% CI 1.1, 3.1). These disparities resulted in a mean of 6.3 YPLL for transfeminine people compared to cisgender men and 9.0 YPLL for transmasculine people compared to cisgender women. INTERPRETATION: These findings highlight persistent health inequities and underscore the need for targeted public health interventions and an evaluation of data collection processes to improve health monitoring and support TGD populations in Sweden. FUNDING: The present study was funded by a Starting Grant (2023-01397) and a project grant (2021-01968) from the Swedish Research Council for Health, Working Life and Welfare (FORTE).

Conclusion

These findings highlight persistent health inequities and underscore the need for targeted public health interventions and an evaluation of data collection processes to improve health monitoring and support TGD populations in Sweden. The authors note that registry data cannot distinguish whether excess deaths result from biological factors, behavioural risks, systemic barriers or administrative biases. They state that future research linking clinical treatment records to mortality is required, and that the findings serve primarily as a baseline for monitoring mortality trends rather than evidence of specific causal pathways.

Methodological limitations

Almost nothing taken into account: People were only matched on age and county. The study did not account for mental illness, drug use, smoking, income or past self-harm, all of which strongly predict the kinds of deaths behind much of the gap.
No treatment information: There is no data on hormones or surgery, so the study cannot say whether medical transition is linked to more or fewer deaths, as the authors note.
Suicide not counted on its own: Suicide was lumped in with accidents and violence, so the study gives no suicide figure despite how important it is.
Very few deaths for many causes: Many cause-of-death figures rest on fewer than 5 to 10 deaths, so they are very uncertain. For example, deaths from nutrition-related conditions were 22 times higher, but the true figure could be anywhere from 2 to 243 times. Some results look implausible, such as almost no mental or brain-related deaths among people born female.
Age adjustment skews the numbers: Rates were adjusted to match Sweden's whole 2023 population, although the people studied were about 25 on average. This gives a few deaths at older ages a lot of weight, such as a high cancer rate in people born female based on 15 deaths.
Many comparisons, some likely chance: About 15 causes of death were compared against two groups for each sex without correcting for the number of tests, so some "significant" findings may be chance.
Unclear causes on death records: The extra deaths with unclear causes may come from paperwork problems (such as ID numbers changing after a legal gender change) or from accidents and violence not properly recorded, which the authors admit.
Only people with a diagnosis: Only people with a recorded diagnosis were included, which over-represents those in contact with specialist care and leaves out those on waiting lists. Diagnoses from age 10 were counted, although the group is described as adults.
Errors in the tables: The first table has apparent mistakes, such as 411,146 comparison people born outside the Nordic countries in a group of only 179,816, and text calling 18.5% "slightly higher" than 19.5–20.4%. Reported figures should be treated with care.
Explanation not tested: The discussion mainly blames stress from being a minority and barriers to healthcare, but the data did not measure either.

detrans.ai

Summary

Key findings

Abstract

Conclusion