September 11, 2023

Accurate transition regret and detransition rates are unknown

Widespread methodological problems limit the reliability of “low transition regret” claims

Like all medical interventions, “gender-affirming” interventions are associated with a range of physical and mental health outcomes—both positive and negative. Regret and detransition are examples of negative outcomes. Proponents of youth gender transition assert that rates of regret and detransition are extremely low. These assertions are frequently cited in legal proceedings, medical journals, and even treatment recommendations. A new paper by Cohn, “The Rate of Detransition is Unknown,” reviews common limitations of “regret” studies and demonstrates that hormone discontinuation, detransition, and regret rates are largely unknown. It is important that clinicians, law makers, and those contemplating medical interventions understand that frequently cited low rates of regret are based on flawed evidence.

Detransition and regret have varied presentations. Sometimes individuals embrace their ultimately regretted transition as part of the “gender journey” they felt was inevitable for them. Other times, individuals openly express devastating regret. As one detransitioner stated, “Some of us will now never be able to have children and many of us live with great distress and regret every day.” Nearly two thirds of detransitioners in a recent convenience sample survey said they would not have had medical intervention had they known what they know now. For such individuals, medical and surgical “gender-affirming” interventions constituted iatrogenic harm.

Studies that claim low regret commonly suffer from the following methodological limitations, which render the conclusions of “very low regret” at a critical risk of bias:

  • Inadequate follow-up.
    While some individuals report regret shortly after starting treatment (such as postsurgical regret), more typically, regret takes significant time to set in. The median time for surgical regret has been reported to be as much as 8 years. Somewhat shorter average times (3-6 years) to detransition have been reported for groups who had a mix of interventions (including puberty blockers, hormones, and/or surgeries). This is important considering that transition is intended to be a lifelong process.

    The flipside of underestimating the adverse psychosocial findings of regret and detransition due to short follow-up is the corresponding overestimation of positive psychological findings, such as of reduced depression, anxiety, and suicidality. Transition may be associated with a “honeymoon period”, with quality of life and satisfaction rising at 1 year post-transition compared to baseline, but then starting to fall at 3 years and falling even more precipitously at 5 years post-transition. It is therefore alarming that studies extolling the benefits of youth transition often focus on outcomes as short as 3 months and rarely extend beyond 5 years. In the instances when follow-up is longer, studies suffer from other significant methodological limitations outlined below.

  • High rates of loss to follow-up.
    A common limitation of regret studies is reporting only on the individuals who willingly engage in follow-up research. While dropouts can occur randomly due to attrition (people move away, move on with their lives, or simply overlook an invitation to participate in follow-up), at other times dropouts are not random and result in a highly biased sample. For example, one of the most frequently-cited studies asserting a low rate of regret omitted all those who stopped coming to the gender clinic - a remarkably high 36%. It is not known how many of these individuals went on to obtain hormones elsewhere or how many decided to stop using “gender-affirming” hormones altogether.

    Another way in which non-random dropouts may lead to an underestimation of detransition and regret is that those who feel harmed by the treatment may not wish to participate in follow-up research. At least one study showed that fewer than one quarter of detransitioners returned to their clinicians to tell them about their decision to detransition.

    While there is no cut-off for the dropout rate that critically biases a study, methodologists assess the risk by estimating whether the study results would substantively change had the dropouts stayed in the study but reported different outcomes than the subjects or participants who remained. Less than 5% loss to follow-up is often thought to not to critically bias results especially when treatment effect among participants is large. On the other hand, 15%-20% dropout rates lead to “degraded” quality and can pose “serious threats” to the validity of findings, especially when treatment effect is modest.

    While there are several studies that claim low regret rates, such studies routinely lose 20%-60% of the original group to follow-up, rendering the results at a critical risk of bias. This is because patients who still attend the gender clinic and those satisfied with their transitions are likely more willing to participate in follow-up research.

  • Imprecision of the measurement of detransition and regret.
    Conscientious researchers who understand the importance of following up with all of the original cases but are unable to contact many individuals in the original cohort resort to other ways to estimate detransition and regret. Unfortunately, the proxy measures they use, such as analyzing medical or legal records for signs of detransition, likely systematically bias the results toward underreporting detransition and regret.

    Consider, for example, a well-known study that asserted extremely low regret rates by searching records for mention of regret and reversal of hormones, or studies that used similarly weak methodologies to assert low regret of surgery. When medical records do not state that the patient regretted treatment, researchers assume that the patient was happy. The presumption of “no news is good news” is inappropriate for research on detransition and regret since, as previously observed, detransitioners are unlikely to return to the physicians who treated them to share their concerns, so “no news” is as likely to signal “bad news.”

    Another common but problematic methodology of identifying instances of detransition and regret is checking for a legal name change. The problems with reducing the complex phenomenon of regret to a binary action such as requesting a legal name / sex marker change were discussed in another recent study.

    There must be a hierarchy of intensity of regret related to the situations patients ultimately find themselves in. The most extreme form of regret is post-transition suicide and suicide attempts. Individuals who undergo medical detransition to restore the body to its pre-transitioned state are also high on this hierarchy. Lower on this hierarchy are those who regret their transitions but due to the irreversible changes to their bodies’ anatomy and function, adaptively choose to make the best of their lives without detransitioning. Regret and acceptance can co-exist.

  • Reliance on biased samples or samples with poor generalizability.
    The “low regret” narrative stems from older studies that sought legal record changes to identify individuals who detransitioned. However, these more carefully vetted older samples are not generalizable to the population of young gender dysphoric people who have multiple mental health comorbidities, and are transitioning currently under the “informed consent” model of care which requires no psychological evaluations.

    The more recent detransition samples that should be applicable to the current clinical presentations frequently suffer from methodological problems, rendering the results at a high risk of bias. For example, a sample in a well-publicized study, which concluded that most detransitioners did not regret their transition, paradoxically only allowed in the detransitioners who still identified as transgender. This was not disclosed in the published study.

    The fact that detransitioners who realigned with their biological sex were excluded from the study is only apparent once one reads the several-hundred-page report describing the survey methodology of which the study was based. The attitudes of individuals who identify as transgender but who detransitioned (due to medical complications or external pressures) are likely markedly different from the attitudes of detransitioners who no longer identify as transgender.

Cohn illustrates how several frequently quoted detransition and regret studies suffer from one or more of the limitations outlined above. Cohn also details how every study measuring surgical regret rates in a recent systematic review and meta-analysis of surgical regret rates suffers from insufficient follow-up time and/or high dropout rates. There are other previously voiced criticisms and concerns about this study as well, yet it continues to be frequently cited as demonstrating low regret rates for transition surgery.

Cohn concludes that it is important for those considering medical intervention to know that the likelihood of regret, detransition, and discontinuation is unknown and that regret and detransition can be traumatic. Cohn conveys the urgency of this concern by highlighting the rapidly growing numbers of youth pursing gender transition: in the US alone, more than 17,000 children aged 6–17 started puberty blockers or hormones from 2017 to 2021, and there were at least 56 genital surgeries and 776 double mastectomies in the 13–17 age range from 2019 to 2021.

Cohn calls for improved methodologies in studying detransition and regret rates, and advocates for disseminating accurate information about the gap in knowledge of detransition and regret so that young patients and their families can make informed decisions about treatments, rather than being lulled into a false sense of security by the erroneous “low regret” narrative.

SEGM Take-away

While the negative physical health outcomes have increasingly come under scrutiny (including adverse effects on bone and cardiovascular health, sexual dysfunction, and infertility/sterility), less attention has been paid to adverse psychological outcomes. Although the proponents of youth transition assert that detransition should not be thought of as a manifestation of a failed transition, this argument is hard to justify. Hormones and surgery irreversibly change the body and some of its key functions. And since gender transition is a lifelong process required to maintain a masculinized or feminized appearance, instances of medical detransition—reported by one study as reaching 30% within just 4 years of initiating treatment—is an alarming warning signal of high numbers of inappropriate transitions.

The current narrative by gender-affirming clinicians that regret is extremely rare is based on studies that suffer from significant methodological limitations, which critically bias those studies toward underreporting of detransition and regret. A recent study, which claims significantly reduced depression and suicidality following testosterone administration for gender dysphoric females is a case in point: the subjects were followed for a mere 3 months (long before any physiological effects of testosterone—positive or negative—could set in).

A potential “honeymoon period” associated with starting treatment has been observed. Short-term improvements in mood do not provide credible evidence that the highly invasive medical and surgical interventions involved in gender transition will assure a regret-free, high-quality life. Studies that do not extend sufficiently long after transition should explicitly state that they are unable to ascertain true regret rates.

Until reliable measures of regret are available (which will take years to collect, given the recent rise in gender transitions of youth), patients, families, clinicians, policy makers, and the public-at-large need to know that the regret and detransition rates are unknown, and that the evidence in hand does not demonstrate that these rates are very low.

A larger list of study regret rates, without their follow-up times or percentages, can be found here.

This analysis underwent minor modifications after its original publication date.