Regret and the Path Not Taken
What should one make of the claim that transgender medical interventions have extremely low levels of measured regret?
Part of the problem in approaching the subject of regret is that it’s so large and pervasive, part of almost any interesting story. It’s difficult to think of any more particularly human emotion than regret because it requires the ability to reflect on how the past led to the present, and to consider how alternate actions could have led to better outcomes. It combines our highest cognitive abilities with the primal emotional capacity by which all living creatures struggle for advantage or survival within an uncertain and often adverse environment. Raw emotion can be hard to bear on a sustained basis, which is part of how it functions to drive behavior. Intense fear or desire can be relieved quickly by fleeing or pursuing. Regret can foster a similar emotional intensity, but at greater remove from the immediate senses through the application of imagination, comparing oneself with others, and reasoning about counterfactuals. A little capacity for regret must be a good thing, otherwise from an evolutionary point of view that ability wouldn’t be so universal. Perhaps the emotional component stimulates the development of planning skills through imaginary rehearsal and retaining memories particularly important for avoiding trouble in the future. But too much regret is pathological. Aside from how unpleasant it can feel, it can lead one to wallow in a paralyzing whirlpool of self pity that lowers one’s social value and inhibits initiative. So it’s not surprising that we have powerful psychological defenses against regret, in particular rationalization and developed indifference.
An ordinary human life is a minefield of potential regrets. Realistically, no one marries the one most perfect possible partner; no one is in the best possible job or home or neighborhood; one could always have done something more or better for loved ones or friends. Hindsight combined with imagination can provide infinite persuasive opportunities for regret. Yet most of us will admit to few regrets and that’s probably an honest indication of the degree to which we tend to regard almost all of life’s imperfections with a fair degree of equanimity. My guess is that there’s basically two processes that come into play, as needed. The first might be called the Panglossian approach. This is the tendency to actively rationalize that whatever has happened is actually all for the best. Whatever problems one may seem to perceive at the moment, past alternative choices were impossible or would have led to worse outcomes, and the resilience one has developed through adversity will contribute to a better future. The second might be called accommodative or strategic indifference: simply ignoring the possibility of a better alternative or deciding that one doesn’t care. For myriad minor issues, this makes perfect sense, they’re not worth worrying about. For bigger issues that could be worth some worry, eventually that worry needs to be cast aside and one naturally grows indifferent as the causes recede in time and the consequences are normalized. For chronic and acute potential regrets, one route to improved functioning is simply cultivating a superior indifference: I’m stronger than this and I don’t care. I suppose this could be called the Stoic approach.
In line with the above considerations, one factor that can very much affect whether or how regret is explored or actively resisted is a perception of whether there is some utility to dwelling on reconsideration of some decision. I’m reminded of the story of Cortes destroying his ships to forestall mutiny and ensure complete dedication of his men to the conquest of Mexico. Extreme danger requires extreme commitment which can be fostered by passing a point of no return. All transgender medical interventions are largely to entirely irreversible. That fact must weigh heavily on the minds of those who have taken that path and on many of those contemplating it. For those who haven’t yet made the leap, it can be both intimidating and fascinating. The fact encourages a ratcheting of certainty in order to cross a perceived threshold: There must be a true trans nature that can make that passage without risk of regret, the only question is whether one has it. Fear of making a mistake heightens emotion and the urge to ruminate, and those in turn can be read as evidence that one has the necessary extreme ‘true’ nature. For those on the other side of transition, there is no going back to what could have been. Indulging in regret must look like a pointless pit of despair. It would be instinctive to skirt widely of that possibility by nurturing a positive interpretation, first by sustaining belief in one’s threshold satisfying true nature, and second by rationalizing problems in a Panglossian light or as due to the shortcomings of others. It’s also worth pointing out that one psychological tactic for resisting an impulse towards post-transition regret could be embracing an evangelism about the success of one’s own transition and the advisability of others doing the same.
So what ought one to make of the studies claiming to show extremely low regret rates for medical transition? I would argue that an ordinary understanding of regret, as sketched above, should bias the reader towards skepticism as one considers how such studies have been carried out and what they really measure. Where there is no effective reversibility of the medical consequences, and emotional exploration of regret may be seriously painful and counter-productive, one would expect most people in this position to make the best of their situation and move on, regardless of real perceived benefit of the outcome. I’m aware of two kinds of study that are used to argue for low regret rates. The first kind is based on questionnaires, usually given by providers to former patients. The second kind is based on some other type of followup data, where some documented action is taken as a proxy for regret.
It’s difficult to force someone to answer a survey questionnaire and to do so honestly. Some things are very hard to measure without one, but that doesn’t justify placing much confidence in quantitative measurements made by surveys that have not been carefully calibrated against data obtained by other methods. Some well-known systematic weaknesses in surveys include non-representative samples, courtesy bias or a desire to please the researcher, and issues with the framing of questions to encourage particular responses. These seem particularly relevant when patient surveys are designed and administered by clinics with an obvious interest in the outcome. Relevant to the subject being surveyed is choice support bias which would be expected to tilt responses towards expressing satisfaction with an elected treatment. All of the patient survey based studies I’ve seen have significant loss to followup, usually 20% or greater. Some loss is inevitable for various reasons, but there’s good reason to suspect that in the case of gender medicine the responding and non-responding populations differ significantly, and it’s irresponsible not to investigate this possibility. Qualitative interviews with detransitioners (as a subset of overall regretters) suggests that transition regret often co-occurs with a reluctance to engage with the provider, either because of negative emotions towards the provider as a result of feeling injured, or guilt as a result of feeling that one didn’t live up to expectations.
One occasionally sees the claim that gender surgeries have a much lower regret rate than e.g. knee surgeries. I suspect that this has much more to do with the psychological aspects touched on above than the actual quality of life impact and medical quality of care received. Anyone undergoing knee surgery probably knew what it was like once to have healthy knees and wished to recover that state, although no common knee injury addressed by surgery will result in a youthfully healthy knee. (I have had knee injuries and one surgery.) Those passing through gender surgery are entering a novel reality with less to compare to, and are likely conditioned to embrace it with optimism.
A frequently cited study using a proxy measure that claims a low level of regret is Dhejne et al. (2014) https://pubmed.ncbi.nlm.nih.gov/24872188/ . Laws and registries particular to Sweden allowed the authors to perform a non-survey followup on all patients receiving sex reassignment surgery and legal recognition to see how many of them later reapplied to have their birth sex reinstated as their legal sex. This re-application was the proxy measure of regret. Remarkably, the paper contains no mention of the possibility that real regret could occur without applying to reverse legal sex, and that at best this measure is probably a loose lower bound on regret. Instead, the authors encourage conflation of this measure with real regret via statements like “the risk of regretting the procedure was higher if one had been granted a new legal gender before 1990.” However, qualitative (anecdotal) interviews with regretters indicates that some of them feel that there’s no point in attempting to transition back to or be recognized as their birth sex since the changes are irreversible. This Dhejne paper is also remarkable for its claim, made in the abstract, that the data showed “a significant decline of regrets over the time period.” This was true in the way they chose to compute a rate, as the number of known eventual ‘regret’ applications over the number of transitions in each decade, but it’s not necessarily the most accurate interpretation of the data. They admit that the overall median time to regret was 8 years and that the number and demographics of patients changed significantly over time, but they make no attempt to further analyze time distribution and estimate future regret applications and instead are happy to assume no future such applications in publicizing their 2.2% figure.
Any honest study of transition regret needs to look closely at how it plays out over time, that the median time to regret seems to be on the order of a decade, and could be even longer. What's going on with this? Does the quality of life actually decrease post transition, or does it just take that long to understand the consequences and work through the psychological resistance to admitting one made a mistake?
Another proxy for regret is discontinuation of gender transition prescriptions. In the past couple of years there have been several studies published that tried to track cross-sex hormone prescriptions using health system or insurance records in order to identify how often such prescriptions are discontinued by patient choice. This is an interesting approach that has led to some much higher estimates of detransition than previously published, but as yet it’s difficult to guess what unrecognized factors might be contributing to inaccuracy of these estimates.
In an ideal situation, how should regret be measured? At one time I got interested in economics and game theory and read up on auction theory which influenced how I think about this kind of question. The basic insight is that talk is cheap and public talk can be heavily influenced by factors that incentivise dishonesty. The most accurate way to measure how much someone values something is to see what goods of unambiguous value they’ll exchange for it, especially if that can be done privately. In this case the irreversibility of time and medical interventions seem to pose an insurmountable barrier. The most important use case for measuring transition regret is to be able to best inform those contemplating transition the real likelihood of comparative life satisfaction along each choice arm. What we’d ideally like to measure is how much, without being exposed to any social shaming or embarrassment, a person N years post transition would prefer to go back in time and choose not to transition, or somehow magically recover their original body and social standing at the present time. Since neither of those possibilities are available at any price, it seems impossible to truly know and that all proxy or survey measurements are likely to be significant underestimates. Note that a long-term randomized control trial would likely be the best alternative, especially if it looked at lots of objective (other than solicited opinion) measures of social thriving and life satisfaction.