Tavistock’s Experimentation with Puberty Blockers: Scrutinizing the Evidence

blockers

Tavistock’s Experimentation with Puberty Blockers: Scrutinizing the Evidence

by Michael Biggs, Dept of Sociology, University of Oxford

(2 March 2019)

In 2010, Tavistock and Portman NHS Trust’s Gender Identity Development Service (GIDS) launched a trial of puberty blockers for children in their early teens with gender dysphoria. This was—and remains—an experimental treatment. These drugs, Gonadotropin-Releasing Hormone agonists (GnRHa), have not been certified as a safe or effective treatment for gender dysphoria by their manufacturers, nor by the National Institute for Clinical Excellence.

The Director of GIDS, Polly Carmichael, was keenly aware of the controversy over these drugs. ‘The question is, if you halt your own sex hormones so that your brain is not experiencing puberty, are you in some way altering the course of nature?’ (Guardian, 14 August 2008). ‘[T]he debate revolves around the reversibility of this intervention—physical and also psychological, in terms of the possible influence of sex hormones on brain and identity development’ (Carmichael and Davidson 2009). Before 2010, GIDS administered blockers to children only when they reached 16; this is the age at which young people have the presumptive capacity to consent to medical treatment.

This cautious approach was vociferously opposed by two organizations devoted to transgendering of children, Mermaids and the Gender Identity Research and Education Society. As Carmichael later recounted: ‘There was a lot of pressure coming from certain group [sic] to introduce it—families were travelling abroad because they knew it was available in Holland and America. As a service, we didn’t have the evidence one way or the other, so the best way to do it was as part of a research study’ (Vice, 16 November 2016).

Tavistock Trust announced the study on its website in April 2011. It stated that GnRHa treatment ‘is deemed reversible’. This assertion contradicted the study’s own research protocol (which I obtained under Freedom of Information from the NHS Health Research Authority). ‘It is not clear [my emphasis] what the long term effects of early suppression may be on bone development, height, sex organ development, and body shape and their reversibility if treatment is stopped during pubertal development’ (Early Pubertal Suppression in a Carefully Selected Group of Adolescents with Gender Identity Disorder, 4 November 2010, Research Ethics Committee number 10/H0713/79). A paediatrician on the study team, Russell Viner, frankly acknowledged the risks. ‘If you suppress puberty for three years the bones do not get any stronger at a time when they should be, and we really don’t know what suppressing puberty does to your brain development. We are dealing with unknowns’ (Daily Mail, 25 February 2012).

The study received considerable publicity, being reported in the Mirror, the Daily Telegraph, and the Times. As Carmichael observed, ‘as professionals we need to be looking at the long term and making sure this treatment is safe’ (Daily Telegraph, 15 April 2011). The bare outlines of the study can be gleaned from a conference presentation and a half-page published abstract (Gunn et al. 2015a; Gunn et al. 2015b). From May 2010 to July 2014, 61 children were recruited; puberty blockers were administered to 50 of them, aged between 10 and 16.

Even before the final patient was enrolled, Carmichael announced success to the tabloid press. ‘Now we’ve done the study and the results thus far have been positive we’ve decided to continue with it’ (Daily Mail, 17 May 2014). In fact the decision had already been made, at least six months earlier (Daily Mail, 17 November 2013). Tavistock Trust then embraced the drug regime with enthusiasm. Three years later, GIDS (and its satellite operation in Leeds) had prescribed puberty blockers for a total of 800 adolescents under 18, including 230 children under 14 (Daily Mail, 30 July 2017). By 2018, new prescriptions were running at 300 per year (BBC News, 2 July 2018). Freedom of Information requests have failed to elicit more recent figures because GIDS does not collate basic data on this experimental treatment—and neither does the University College London Hospitals NHS Foundation Trust, which provides its endocrinology services.

Over a thousand adolescents have been given puberty blockers on the basis that the 2010–14 study yielded ‘positive’ results. Tavistock is surprisingly reticent to share these results with the scientific community. GIDS has a webpage on the evidence base for puberty blockers. It notes that ‘research evidence for the effectiveness of any particular treatment offered is still limited.’ There is no mention of its own study; it cites only research from the Netherlands. The former director of GIDS stated last year that the ‘project is ongoing and the results are yet to be published’ (De Ceglie 2018).

Diligent searching does, however, uncover some unpublished results. Most revealing is an appendix within a report to Tavistock’s Board of Directors (Carmichael 2015). It tracks the first 44 children on GnRHa, measuring changes after one year of the drug regime. The text is sometimes internally inconsistent and occasionally contradicts the tabulated figures, suggesting that it was prepared in haste. But we can summarize those changes that were reported as statistically significant (p-value < .05). Only one change was positive: ‘according to their parents, the young people experience less internalizing behavioural problems’ (as measured by the Child Behavior Checklist). There were three negative changes. ‘Natal girls showed a significant increase in behavioural and emotional problems’, according to their parents (also from the Child Behavior Checklist, contradicting the only positive result). One dimension of the Health Related Quality of Life scale, completed by parents, ‘showed a significant decrease in Physical well-being of their child’. What is most disturbing is that after a year on blockers, ‘a significant increase was found in the first item “I deliberately try to hurt or kill self”’ (in the Youth Self Report questionnaire). Astonishingly, the increased risk of self-harm attracted no comment in Carmichael’s report. Given that puberty blockers are prescribed to treat gender dysphoria, it is paradoxical that ‘the suppression of puberty does not impact positively on the experience of gender dysphoria’ (measured by the Body Image Scale). When differentiated by sex, the impact was positive for boys on one aspect of body image, but negative for girls on two aspects.

These preliminary results (44 children after one year on GnRHa) were also presented at a symposium at the World Professional Association for Transgender Health (Carmichael et al. 2016). Only the abstract is available. ‘For the children who commenced the blocker, feeling happier and more confident with their gender identity was a dominant theme that emerged during the semi-structured interviews at 6 months. However, the quantitative outcomes for these children at 1 years time suggest that they also continue to report an increase in internalising problems and body dissatisfaction [my emphasis], especially natal girls.’ Why were these negative results never published?

The study apparently contributed data on outcomes to one publication, coauthored by Carmichael (Costa et al. 2015). The abstract proclaims that ‘adolescents receiving also puberty suppression had significantly better psychosocial functioning after 12 months of GnRHa … compared with when they had received only psychological support’. The article is treated in the literature (e.g. Heneghan and Jefferson 2019) as providing evidence in favour of puberty blockers. But the abstract is misleading: the analysis actually failed to detect any difference between children who were given blockers and those who were not. To understand this, we need to scrutinize the article in detail. (Statistically minded readers will recognize the fallacy described by Gelman and Stern 2006.)

The analysis starts with 201 adolescents diagnosed with gender dysphoria. The children were divided into two groups: those deemed eligible for puberty blockers immediately, and those who needed more time due to ‘comorbid psychiatric problems and/or psychological difficulties’. This second group did not receive any physical intervention during the time of this analysis, and so serves as a control group. Both groups received psychological support. The article chooses one outcome, psychosocial functioning as measured by the Children’s Global Assessment Scale (CGAS). This scale was administered at the outset, and then after six, twelve, and eighteen months. It is intriguing that the article omits the outcomes that were negative in the preliminary results: the Child Behavior Checklist, the Youth Self Report Questionnaire, the Health Related Quality of Life scale, and the Body Image Scale.

The authors graph the CGAS results, but without confidence intervals—which indicate the extent of random statistical variation or noise. The smaller the sample, the greater this noise. These samples shrank over time: after eighteen months, the group getting puberty blockers numbered only 35, and the control group 36. The article does not explain why two thirds of the subjects disappeared. Presumably they did not stop the medication, because all 50 children given blockers in the 2010–14 study continued the drug regime for two years (Gunn et al. 2015b).

My graph plots the results with standard 95% confidence intervals. The group given puberty blockers from six months onwards showed improvement at eighteenth months: the average CGAS score had increased from 61 to 67 (coloured red on the graph). This improvement is statistically significant, and it is the one that the authors chose to highlight. However, these children also received psychological support, and so attributing this improvement to medical intervention is unjustified. The crucial comparison is between the group receiving blockers and the control group. The latter’s average CGAS score (coloured blue) after eighteen months was lower, 63 compared to 67. But this difference is not statistically significant; the 95% confidence intervals substantially overlap. (For statistically minded readers, a two-tailed t-test for the difference between group means yields a p-value of .14, far beyond the conventional .05 threshold.) In other words, the samples were so small, and there was such wide variation in scores within each group, that we can draw no conclusions. There is no evidence that puberty blockers improve psychosocial functioning. Presumably this is why GIDS omits the article from its own evidence base.

The abstract describing the baseline characteristics of the children in the 2010–14 study concluded: ‘Assessment of growth, bone health and psychological outcomes will [my emphasis] be important to assess the medium and long-term safety and effectiveness of early intervention’ (Gunn et al. 2015b). However, GIDS apparently failed to collect any data on its experimental subjects after they turned 18. In a startling admission, Carmichael and coauthors blame ‘the frequent change in nominal and legal identity, including NHS number in those referred on to adult services’—‘to date they have not been able to be followed up’ (Butler et al. 2018). (Transgender activists successfully lobbied the NHS to provide new numbers to patients as well as to change the ‘gender’ on their medical records.)

To summarize, GIDS launched a study to administer experimental drugs to children suffering from gender dysphoria. Between 2010 and 2014, puberty blockers were given to 50 children. This study yielded only one published scientific article on outcomes. It showed no evidence for the effectiveness of GnRHa: there was no statistically significant difference in psychosocial functioning between the group given blockers and the group given only psychological support. In addition, there is unpublished evidence that after a year on GnRHa children reported greater self-harm, and that girls experienced more behavioural and emotional problems and expressed greater dissatisfaction with their body—so puberty blockers exacerbated gender dysphoria. Yet the study has been used to justify rolling out this drug regime to several hundred children aged under 16. Almost five years after the last patient was enrolled in the experiment, there is no evidence to substantiate Carmichael’s claim ‘that the results thus far have been positive’.

The Director of GIDS needs to answer these questions about the 2010–14 experimentation with puberty blockers:

  • On what evidence did you claim in 2014 that ‘the results thus far have been positive’?
  • When preliminary results in 2015 showed that children after a year on blockers showed a statistically significant increase in reported self-harm, was this ever investigated?
  • Why did you never publish the negative results reported to Tavistock’s Board of Directors in 2015 and to WPATH in 2016?
  • Why did your only published article (Costa et al. 2015) using data from the study omit all the outcomes that were negative in the preliminary results (Child Behavior Checklist, Youth Self Report questionnaire, Health Related Quality of Life scale, and Body Image Scale)?
  • In your article, why did the abstract and conclusion not report the finding that there was no statistically significant difference between the group given GnRHa and the control group?
  • In your article, what accounts for the reduction in the number of subjects from 201 to 71 over eighteen months?
  • What steps have you taken to monitor the ‘long-term safety and effectiveness of early intervention’, as these experimental subjects become adults?

Note

How many subjects from the 2010–14 study are included in Costa et al. (2015) is unclear. The 2010–14 study gave GnRHa to 50 adolescents, starting at ages from 10 to 16 (Gunn et al. 2015a, 2015b). One would expect all of them to be included in the ‘immediately eligible’ group in Costa et al. (2015), along with some older adolescents to boost the sample size. The article counts 101 children in this group at 6 months when GnRHa commenced (Table 2), starting at ages from 13 to 17 (Table 1). The age range indicates the exclusion of some children from the 2010–14 study: those who commenced GnRHa from ages 10 to 12. Why? Another puzzle is worth noting. When I requested the Research Ethics Committee number from Tavistock and Portland NHS Trust under Freedom of Information, it provided the number 10/H0718/62. According to the NHS Health Research Authority, however, this number refers to a study that was given an ‘unfavourable opinion’ and therefore could not proceed.

References to the 2010–14 GIDS Study

Carmichael, Polly, 2015, ‘Preliminary Results from the Early Intervention Research’, Appendix 7 in ‘Service Line Report: Gender Identity Development Service (GIDS)’, Tavistock and Portman NHS Trust, Board of Directors Part One: Agenda and Papers … 23rd June 2015, pp. 50–55.

Carmichael, Polly, Sally Phillott, Michael Dunsford, Amelia Taylor, and Nastasja de Graaf, 2016, ‘Gender Dysphoria in Younger Children: Support and Care in an Evolving Context’, World Professional Association for Transgender Health 24th Scientific Symposium.

Costa, Rosalia, Michael Dunsford, Elin Skagerberg, Victoria Holt, Polly Carmichael, and Marco Colizzi, 2015, ‘Psychological Support, Puberty Suppression, and Psychosocial Functioning in Adolescents with Gender Dysphoria’, Journal of Sexual Medicine, vol. 12, pp. 2206–14.

Gunn, H.M., C. Goedhart, G. Butler, S.N. Khadr, P.A. Carmichael, R.M. Viner, 2015a, ‘Gender Dysphoria: Baseline Characteristics of a UK Cohort Beginning Early Intervention’, presented to the Youth Health Conference, Australia.

Gunn, H.M., C. Goedhart, G. Butler, S.N. Khadr, P.A. Carmichael, R.M. Viner, 2015b, ‘Early Medical Treatment of Gender Dysphoria: Baseline Characteristics of a UK Cohort Beginning Early Intervention’, Archives of Disease in Childhood, vol. 100, supp. 3, p. A198.

Acknowledgements

Valuable information was provided by Stephanie Davies-Arai, Susan Matthews, Heather Brunskell-Evans, and Elin Lewis.

This Post Has 24 Comments

  1. GILAW

    Thanks so much for doing this analysis. There are SO MANY red flags in the design, administration, and analysis of this work. I hope someone in the government starts asking some hard questions.

  2. Klara Lo

    Dear prof Biggs, thank you for this data. There are methodological problems in the Tavi pub blockers study, and also a severe lack of transparency. Also how come that a study can be performed even if the IRB does not give a favorable opinion? Do you know the sources of financial support to this project? It seems that this study needs an external review.

    1. Margaret

      Thank you Prof Biggs. I too wonder who provided monetary support for this survey. Did they expect to get one result and when that seemed unlikely radically cut down the numbers?

      1. Michael Biggs

        I don’t think there was any specific funding for the study. The kids were coming to the clinic anyway. I’m not sure who pays for the blockers, it will depend on whether the GP agreed to prescribe them (on GIDS recommendation) or whether the GP refuses and so GIDS (via UCL Hospital) has to prescribe. There’s nothing suspicious about the numbers: the research protocol envisaged 45, and they ended up with 50 on blockers.

    2. Michael Biggs

      There was a favourable outcome for the ethics application (Research Ethics Committee number 10/H0713/79). But it seems that there was a previous unfavourable application? I have made an FOI request and we will see.

  3. Bepis Drink

    Why doesn’t this make headlines around the world? The treatment promoted as “saving children’s lives” actually makes children MORE at risk of killing themselves!

  4. Katherine

    Thank you. One wonders if Tavistock or any of its service providers could be receiving payment, funding or other incentives from the pharmaceutical industry that encourage the prescription of hormone treatments absent any authoritative evidence that it a) actually benefits children psychologically and b) does not do significant long-term physical or neurological damage. One also wonders if Tavistock was deliberately hiding the limited data it has collected or attempting to hide its research incompetence. Thank you, Professor Biggs, and I hope this report leads to further investigation.

    1. Heather

      Is this a cover up or incompetence? Both?

  5. Alex

    The CGAS measure is a clinician-rating. It is a single number (1-100), which is supposed to represent the child or young person’s overall psycho-social functioning. There are a couple of issues with this.

    Firstly, the CGAS is a rough and ready snap shot based on a clinician’s overall impression. In practice, clinicians may have a brief look over the brief descriptors for each 10-point score range but scores tend to be recorded as much on gut-feeling as on objective criteria. The scores are not based on a specific interview or a questionnaire. Often clinicians will be deciding these scores for many young people at the same time as they catch up with this task, rather than at the time of seeing the young person.

    Secondly, when the CGAS is used in clinical practice as an outcome measure there can be pressure (either explicitly or implicitly) for the scores to reflect progress. There is likely to be some rater-bias to scores falling to show improvement and that the intervention is working and the clinician is doing their job well. This is on top of the pressure the client may also feel to present as doing better in sessions, both so as not to jeopardise their further transition but also in a perhaps unconscious effort to please their clinician.

    In summary, the measure GIDS has chosen to publish to demonstrate the effectiveness of the blocker is a rough and ready professional guesstimate provided by GIDS own staff on how effective their own intervention is. There are good reasons to believe that (whether unconsciously or consciously) we can expect that clinician-rater bias would lead to lower scores (ie improved psychosocial functioning) at follow-up, even if this wouldn’t accurately reflect reality. That would be a serious criticism and flaw in the study even if it had found significant differences between scores and given the lack of significant differences all the more so.

    1. Michael Biggs

      That’s very interesting information, thanks Alex!

  6. charles lewis

    I am not a statistician, merely a lawyer, but it is no surprise to me to read that the Tavistock people are distorting their results. They are controlled, as indeed are the NHS, by the transgender activists. The other day I heard Dr Bernadette Wren address a meeting at which she assured us that gender was not binary but was fluid — and not of course biological. It was a ludicrous display of nonsense being parrotted as reality. I wrote to tell her so but — surprise surprise– got no response.

    1. Alan Henness

      Charles

      You may remember we were at the same meeting. I’d always assumed the TRAs just didn’t understand what they were talking about and simply spouted doctrine that had been passed down to them – doctrine that stopped them having to think for themselves. It was truly staggering to hear Wren spout the same nonsense the TRAs repeat ad nauseam. They were getting their nonsense from the very top.

      I understand the meeting will be written up for the journal so it’ll be interesting to see how it’s presented.

      1. Heather

        I saw Bernadette Wren speak last year. She is very charming and beguiling speaking beautifully on gender fluidity and creative teens. When the audience asked about homosexuality though she seemed less interested, as though it wasn’t relevant to her work at all. She seemed quite ignorant on LGB issues actually. That concerned me. If she can’t even consider that some of her patients might be gay then how can she screen them out before suggesting medical treatment? Worrying service.

  7. Moyra

    This article ends with questions that you say the Director of GIDS ‘needs to answer’.
    Have these questions actually been put to the Director of GIDS?

    1. Michael Biggs

      I believe that a journalist is putting these questions to her, and we will shortly see how she responds …

  8. charles

    Unbelievable how the monstrous trans cult with its lunatic creed has prospered the other side of the Atlantic. Are these people who support it mentally defective? Thanks for your article, Laurie

  9. David Curtis

    Thanks for this. One point which could be more strongly emphasised is that the “control” group is not really a control group at all in any way that we would normally understand this. As you mention, the report states this about them: “In those specific cases clinicians needed more time to make the decision of starting GnRHa because of possible comorbid psychiatric problems and/or psychological difficulties.” (https://www.jsm.jsexmed.org/article/S1743-6095(15)34443-X/pdf) So the children who don’t receive medical treatment are actually the ones who have more severe comorbidities. Comparing their lack of improvement in CGAS to those that did get medical treatment is therefore nonsensical. The fact that in the difference is not statistically significant is noteworthy but I think the main point is that the groups are not comparable in the first place.

    1. Michael Biggs

      Yes, that’s a good point David!

  10. Alex

    In the Telegraph article the Tavistock is quoted as saying “fewer than half of those referred to the service go on to access physical interventions.”

    This is a stat that needs investigating.

    What would happen if we weren’t talking about those referred to the service but those who actually attended at least one appointment? How many people drop out before they get to the clinic? Presumably there is no way of knowing how many do so because of a change in heart, and how many drop out because they have obtained blockers or hormones privately instead. And what about those who reach 18 whilst on the waiting list and are then transferred or referred on to an adult service without even been seen at GIDS? Are they counted in the numbers of those not going on to access physical interventions too?

    But say we got a new stat that takes out those who don’t even get seen by the service, many of whom might get physical interventions elsewhere (whether privately or from an NHS adult service).

    Does the number counted as not going forward for intervention also include, for example, all the pre-pubertal children who haven’t yet started hormone blockers but many of whom will do once the first signs of puberty begin?

    And does the Tavistock only mean physical interventions within the child and adolescent service? What about many 16 and 17 year olds who have been seen by the service but don’t get hormone blockers and instead get a referral direct to an adult service (sometimes counting as a first opinion meaning they are fast-tacked through the adult service assessment)?

    What about the many older teenagers where GIDS clinicians suggest obtaining the contraceptive pill from their GP as an alternative to the blocker, whilst they then wait to get testosterone from an adult clinic (who unlike GIDS don’t require a year on the blocker first before starting hormones).

    It would be interesting if the Tavistock would provide the figures to back up its claims.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.