Puberty Suppression in Adolescents

Evidence Review · ER-014

Part I — The frame, and what is mechanically firm

1. Two questions, kept apart

Almost every public exchange about puberty blockers runs two different questions together, and the confusion is not accidental — collapsing them is how each side converts a modest evidential claim into a sweeping conclusion. Pulling them apart is the whole method of this review.

The evidence question. What do the available studies show about the benefits and harms of suppressing puberty in adolescents with gender dysphoria, and how much confidence can be placed in those findings? This is an empirical question with a real, if uncomfortable, answer, developed across Parts II to IV: the treatment reliably achieves endocrine suppression, several of its physical effects are reasonably characterised, and the evidence for its psychological benefits — the reason it is principally given — is weak and inconsistent. That last point is not a fringe reading. The systematic review commissioned for the Cass Review found a lack of high-quality studies, and the National Institute for Health and Care Excellence rated the evidence low quality across the outcomes that matter most.^[1,2]

The policy question. Given evidence of that quality, what should be done — permit the treatment routinely, permit it only within research, restrict it, or prohibit it? This is not an empirical question. It is a question about how much weight to give uncertain benefits against uncertain harms, how to treat an evidence gap when the alternative (untreated distress) also carries risk, who should decide, and how much precaution a treatment for minors warrants. Reasonable, well-informed people reach opposite answers from the same evidence, because they weigh the uncertainties differently — and that is a values disagreement, not a factual one.

Why the seam matters. Watch how the two get welded. “The evidence is weak” is an evidence claim; “therefore the treatment should be banned” is a policy claim, and the first does not entail the second — weak evidence can justify caution, more research, restriction or continued provision with consent, depending on values the evidence cannot supply. Run the weld the other way and it fails just as clearly: “young people report relief” is an evidence claim of limited certainty; “therefore access should be unrestricted” is a policy claim it cannot carry. This review answers the evidence question and characterises the policy question without pretending the evidence settles it.

Two symmetric errors, named once so they can be watched. The evidence base is weak enough that both sides are tempted into the same two mistakes, mirrored. The first is treating absence of evidence as evidence of absence: low-certainty evidence of benefit is not proof the treatment fails, and low-certainty evidence of harm is not proof it is safe. The second is the false absolute: “reversible” and “irreversible,” “safe” and “harmful,” “settled” and “experimental” are wielded as though the evidence licensed certainty at either pole, when for most outcomes it licenses neither. Wherever this review reaches for a strong word, it is because a specific finding earns it — not because the surrounding argument wants it.

Confidence: not applicable — this section is the method, not a finding. The claims graded below begin in Section 2.

2. What puberty suppression actually is

The mechanism, which is not in dispute. Puberty begins when the hypothalamus starts releasing gonadotrophin-releasing hormone (GnRH) in pulses; that pulsatile signal is what drives the pituitary to secrete the gonadotrophins luteinising hormone (LH) and follicle-stimulating hormone (FSH), which in turn switch on the gonads. GnRH agonists exploit a quirk of that system: given continuously rather than in pulses, they first over-stimulate and then downregulate and desensitise the pituitary GnRH receptor, sharply reducing the secretion of LH and FSH and, with them, the gonadal production of oestrogen or testosterone — after a brief initial surge, the “flare,” before suppression takes hold.^[3,4] Suppression is not always complete, FSH is not always lowered to the same degree as LH, and — because the drug acts on the gonadal axis — it does not switch off adrenal androgen production, so adrenarche-driven changes such as some pubic and axillary hair, body odour and acne can still proceed. The effect is to hold gonadal sex-hormone levels down and so pause the gonadally-driven progression of puberty. The receptor-level mechanism is firm and well established across indications; what does not automatically transfer is clinical safety, which turns on age, pubertal stage, duration, dose and what follows.

The agents and how they are given. The GnRHa used include leuprolide, triptorelin, goserelin and histrelin — given by depot injection at product-specific intervals of weeks to months, or by a subcutaneous implant replaced roughly annually — and nafarelin, which is a daily intranasal spray; none is a daily tablet, but route and schedule vary by agent.^[3,5] In gender care they are typically started once an adolescent has reached Tanner stage 2 — the earliest objective sign of gonadarche, breast budding in ovarian puberty or testicular enlargement in testicular puberty (pubic hair alone reflects adrenarche and does not confirm the gonadal axis has switched on) — a threshold set by the 2017 Endocrine Society guideline and carried into subsequent practice.^[5,6] The stated logic of starting at that point is that suppression can only prevent secondary sex characteristics it gets ahead of; small observational studies suggest that when treatment is begun later, in Tanner 4–5, biochemical suppression of the gonadotrophins is often less complete — sex-steroid levels fall in nearly all, but gonadotrophins remain elevated — and hypo-gonadal symptoms are more commonly reported, with hot flushes described by around three-quarters of the late-stage group in one single-centre retrospective series; that estimate rests on limited, single-centre data.^[7] Population flag: much of the dosing and timing framework is carried across from the treatment of central precocious puberty in young children — a different indication, discussed below.

Where the “decades of safe use” claim is strong, and where it thins. GnRHa have indeed been used for many years, and used widely — but predominantly for central precocious puberty, in which the drugs halt a puberty that has begun abnormally early in a young child, and are then stopped so that puberty proceeds at a typical age.^[6] That track record supports two specific claims — that the drugs suppress the hypothalamic–pituitary–gonadal axis reliably, and that the axis itself generally reactivates once the drug is stopped. It supports much less than it is often asked to. Even axis recovery is better described as usual than guaranteed, and “reversible” has to be attached to a named endpoint: recovery of gonadotrophin secretion is not the same claim as reversibility of every downstream developmental consequence of suppression at a typical pubertal age. Pausing an abnormally early puberty so it can resume nearer the typical age is not the same intervention, on the same body, at the same developmental moment, as pausing a normatively-timed puberty for an open-ended period — in precocious puberty the treatment moves sex-steroid exposure towards the typical window, whereas here it delays otherwise typically-timed exposure beyond it, and the two populations differ in age, indication and what happens after the drug stops. The safety inheritance is real for the mechanism and partial for everything else. Population flag: precocious-puberty safety data are extrapolated to a different indication, age profile and treatment trajectory.

Confidence: 🟢 Robust (the endocrine mechanism; the agents; suppression of gonadal sex-hormone levels; Tanner-2 starting convention) / 🟡 Moderate (completeness of suppression across pubertal stages; persistence of adrenal androgen effects; how far precocious-puberty safety data transfer to this indication).

3. What it is for — the stated rationale, and the seam to the evidence

The treatment was not introduced as a feminising or masculinising intervention. It was introduced as a pause. The approach that established it — developed by a Dutch team from the mid-1990s and described across a series of publications rather than in one fixed protocol — set out puberty suppression with three broad purposes: to relieve the distress of developing secondary sex characteristics an adolescent may experience as deeply wrong; to widen the window for gender identity to clarify before irreversible bodily changes occur, sometimes framed as “time to think”; and, practically, to make any later transition less medically arduous by heading off characteristics — a deepened voice, a developed chest — that are hard to reverse afterwards.^[8,9]

Those are the rationales. Whether the treatment delivers them — whether the pause reliably reduces distress, whether it genuinely leaves the decision more open or instead tends to settle it, and at what cost to bone, fertility and development — is precisely the evidence question, and it is graded outcome by outcome in Parts II to IV. Two of those rationales contain empirical claims worth flagging now because the rest of the review tests them. The first is that suppression improves psychological wellbeing; the most direct attempt to test that claim, a small uncontrolled UK study using measures similar to the original Dutch work, did not find the expected group-level benefit (Part III).^[10] The second is that the pause is decision-neutral — that it buys time without tilting the outcome; the continuation data, in which almost everyone who reaches gender-affirming hormones stays on them and, in the small selected cohorts that report it, most who start suppression go on to those hormones, are read by some as confirming the pause was used well and by others as evidence that it channels rather than neutrally delays (Part IV). Naming the rationales precisely is what lets the evidence be held against them rather than substituted for them.

Confidence: 🟢 Robust (that these were the stated rationales for the treatment) / 🔴 Limited (that the treatment delivers each rationale — the substance of the outcome grading that follows).

Part II — The physical outcomes, graded

4. Does it do what it is designed to do?

At the level it is meant to work, yes. Once GnRHa suppression is established, the gonadally-driven progression of puberty largely halts: in adolescents assigned female at birth menstruation stops and breast development generally goes no further; in those assigned male at birth further voice deepening and testicular growth are arrested and facial-hair progression slows. Two honest qualifiers travel with that. Established changes do not reverse — a voice that has already broken stays broken, breast tissue already laid down generally remains — so this is prevention of further change, not undoing; and because the drug spares the adrenal axis, adrenarche-driven pubic and axillary hair, acne and body odour can still advance. This is the best-evidenced effect in the field, because it is a fairly direct read-out of the mechanism — assessed through pubertal staging, menstrual history and biochemical markers, each with its own measurement limits — rather than because it is simply easy to measure.^[3,11] Suppression can be incomplete when started late (Tanner 4–5), where gonadotrophin suppression is often partial, but the core claim — that the drugs pause pubertal progression — is firm.^[7]

Two physical effects travel with the suppression and are worth stating because they are sometimes lost in the debate over psychological benefit. First, growth: puberty normally brings a substantial, sex-dependent acceleration in growth through the interaction of sex steroids with growth hormone and IGF-1, so suppression reduces growth velocity during treatment. Final adult height is a separate question, and here the evidence is more reassuring than the velocity data alone would suggest: Amsterdam cohort studies of adolescents treated with GnRHa and subsequent gender-affirming hormones found adult height attainment broadly unaffected — trans girls reached expected adult height, and the effect in trans boys was neutral to positive. Growth is therefore better tracked through height velocity and bone age than expressed as a fixed share of adult stature lost.^[11,12] Second, body composition and tolerability: fat mass tends to rise, and the reported adverse effects — headache, vasomotor symptoms such as hot flushes, fatigue, mood symptoms, and injection-site or implant complications — are generally described as mild in the available cohorts, which also suggest that stopping treatment because of acute side effects is uncommon; but the denominators, duration and reporting behind that are inconsistent, and mood is hard to attribute to the drug in a group with a high baseline psychiatric burden.^[7,11] These are the predictable consequences of removing sex hormones from a body that is developmentally expecting them.

Confidence: 🟢 Robust (that GnRHa halt the progression of secondary sex characteristics and suppress sex-hormone levels; the growth-velocity effect; the common side-effect profile) / 🟡 Moderate (completeness of suppression at later Tanner stages).

5. Bone — the clearest physical signal, and its nuance

Bone is where the physical-harm evidence is strongest, and it repays being read carefully rather than as a slogan. Childhood, puberty and early adulthood are the critical periods for acquiring peak bone mass, and pubertal accrual is strongly sex-steroid-driven — so suppressing puberty slows bone-density accrual. During treatment, bone-mineral-density (BMD) z-scores, which compare a person to age- and sex-matched peers, fall. It is worth being precise about what that means: a falling z-score signals that the individual’s bone is not keeping pace with the reference population, not necessarily that absolute BMD is dropping — in several cohorts absolute density keeps rising, just more slowly. That z-scores fall during suppression is not in dispute; the causal contribution of the drug is biologically plausible and consistent, but interpretation is complicated by often-low baseline BMD, by growth and body size (which distort areal DXA measures), by the reference database chosen, and by the absence of an untreated comparison group.^[13,14] BMD is also not the same as bone strength, geometry or fracture risk. The harder question is what happens afterwards, and the answer differs by direction of care.

The most informative data come from a long-term Amsterdam cohort followed to a median age of 28, more than a decade after starting.^[15] In participants assigned female at birth who went on to testosterone, mean z-scores at the measured sites returned to approximately their pre-treatment values — a return towards baseline, which is not the same as showing that individuals reached their genetically expected peak bone mass or that fracture risk normalised. In those assigned male at birth who went on to oestradiol, hip and femoral-neck density returned to their (already below-average) pre-treatment values, but the lumbar spine did not: the mean lumbar-spine z-score at follow-up was −1.34, roughly 0.87 z-score units below the cohort’s mean pre-treatment value (using the study’s stated reference convention). The authors proposed lower oestradiol exposure as one possible contributor, on an observational association — plausible, but not established as the cause, given confounding by adherence, body composition, nutrition and baseline.^[15] Earlier, shorter follow-up had shown a similar unrecovered dip, so the recovery picture has improved with longer and better-dosed hormone treatment but is not uniformly complete.^[14]

Three honest qualifiers. Baseline BMD in this population is often already low before any treatment — so not all of the deficit is the drug. The long-term cohort lost a large share of its eligible participants to follow-up (75 of 143), which can bias a recovery estimate in either direction. And the outcome that actually matters — whether any of this translates into more fractures in later life — has not been established: no adequately powered long-term study has tested it, and the z-score is a proxy, not a fracture count. Population flag: BMD reference norms and much of the fracture-risk framework derive from cisgender populations; long-term fracture outcomes in this group are unstudied.

The accurate summary is neither “fully reversible” nor “permanent damage.” It is: accrual is reduced during suppression; with adequate subsequent sex-hormone treatment much of the reduction in z-score recovers, with mean z-scores returning close to baseline on testosterone; a lumbar-spine shortfall can persist on oestradiol and may be dose-related; and the downstream fracture consequence is genuinely unknown.

Confidence: 🟢 Robust (that suppression reduces BMD accrual and lowers z-scores during treatment) / 🟡 Moderate (the recovery pattern — mean z-scores return close to baseline on testosterone, with a lumbar-spine shortfall persisting on oestradiol; return to baseline is not evidence of attained peak bone mass) / 🔴 Limited (long-term fracture risk, which is unstudied).

6. Fertility, sexual development, and the surgical seam

This is the outcome most often mis-stated in both directions, because the honest answer depends on distinguishing the blocker from the pathway it usually sits in. The blocker does not ablate gonadal tissue. GnRHa pause gamete development rather than destroying the gonad; in other indications, stopping the drug lets the axis reactivate and maturation resume. Direct, long-term evidence of preserved future fertility after several years of suppression begun in early puberty for gender dysphoria is, however, limited, and “sterilise” is too blunt an endpoint — fertility spans gametogenesis, the ability to retrieve or produce gametes, and eventual live birth, which are not the same thing. Small reports in boys treated for central precocious puberty describe testicular maturation and spermatogenesis resuming after discontinuation, but the evidence base is sparse and may not transfer cleanly to early-pubertal gender care.^[6,16] On its own, the drug is best read as a pause rather than a cause of permanent infertility.

The complication is the trajectory. Suppression is typically begun at Tanner 2, before the gonads have matured — mature gametes are generally not available at early Tanner stages, and spermarche, the ability to give a semen sample, menarche and adequate gamete quality do not arrive at a single uniform stage — and most who continue proceed to gender-affirming hormones without an intervening window of endogenous puberty (Section 9). On that standard path, gametes may never mature: in histology of testicular tissue removed at later gender-affirming surgery, mature spermatozoa were found in only about 5% of specimens — all from people who had begun treatment at Tanner stage 4 or later — while the large majority contained immature germ cells only. That immature germ cells persist is itself relevant: it is why cryopreserving testicular tissue taken at gonadectomy is discussed as a residual option, albeit one with no established route to a live birth.^[16] And the conventional fertility-preservation options largely need a maturity that early suppression prevents — semen banking needs the ability to ejaculate, and oocyte retrieval needs sufficient follicular responsiveness (it does not strictly require prior menarche, and has been done in selected pre-menarchal adolescents, but remains innovative and low-yield in early puberty). Tissue cryopreservation splits by type: ovarian-tissue cryopreservation is an established option with many live births in adult and oncology settings, though its use in pre-pubertal or this specific population is limited; prepubertal testicular-tissue cryopreservation remains experimental, with no established route to a live birth.^[16] A theoretical “off-ramp” — pausing the blocker to let puberty and gametes mature before deciding — exists but is seldom chosen, because it means undergoing the very development the young person sought to avoid.

So the accurate statement is layered: the blocker does not itself destroy the gonad; but starting early and going straight to gender-affirming hormones may leave no window for conventional gamete preservation unless treatment is interrupted, and the tools to preserve fertility at that age are limited or experimental — with the probability and timing of any later gametogenic recovery uncertain and dependent on anatomy, dose and duration. This is why pre-treatment fertility counselling is treated as essential rather than optional. The decisive evidence simply does not exist: no study has yet followed adolescents who began suppression in early puberty through to attempted conception in adulthood, so the outcome that would settle the question has never been measured. Population flag: recovery-of-fertility data after stopping are largely extrapolated from precocious-puberty and animal models; direct long-term fertility outcomes in this population are sparse.

Confidence: 🟢 Robust (that GnRHa alone pause rather than destroy gamete development) / 🟡 Moderate (that the early-start-then-gender-affirming-hormone pathway may leave no window for conventional gamete preservation unless interrupted) / 🔴 Limited (the effectiveness of experimental preservation at early Tanner stages, and long-term fertility outcomes).

Sexual development, largely unstudied. The same suppression that pauses fertility also pauses aspects of sexual development the evidence base has mostly left uncharacterised — libido, spontaneous erections, orgasmic capacity, and the maturation of genital tissue. This matters beyond completeness: these are among the outcomes a claim of “reversibility” has to answer to, and they bear directly on what an adolescent and their family can meaningfully be told when consenting. The honest position is that they are under-studied, not that they are reassuring or alarming.

The surgical seam. One consequence is concrete. In adolescents with testes, suppression begun early can limit penile and scrotal growth, which can reduce the genital tissue available for certain later vaginoplasty techniques; alternative surgical approaches exist but carry different burdens and trade-offs. This is a real downstream effect of the blocker phase on a later decision, and it belongs in the reversibility ledger — not as a reason for or against the treatment, but as something the standard pathway makes materially different and consent should cover.

Confidence: 🔴 Limited (sexual-function and psychosexual outcomes are largely unstudied in this population — the 2025 systematic review found no moderate- or high-quality evidence on fertility, sexual function or cancer risk) / 🟡 Moderate (that early suppression can reduce genital tissue relevant to some later surgical techniques).^[18] Population flag: much of the sexual-development framework is extrapolated; direct data in early-suppressed adolescents are sparse.

Part III — The psychological outcomes, graded

7. The benefit the treatment is principally given for

Puberty suppression is not given to change bone or body composition; those are its costs. It is given, above all, in the expectation that it will relieve psychological distress — that pausing an unwanted puberty will reduce the dysphoria, depression and anxiety that can intensify as the body develops. This is the central claim, and it is among the outcomes for which blocker-specific causal evidence is weakest. Stating that plainly is not a verdict against the treatment; it is the finding.

The Dutch foundation. The claim originates in the Amsterdam cohort. In the key prospective study — an uncontrolled pre-post cohort with no untreated comparator — several behavioural, depressive-symptom and general-functioning scores improved over follow-up, while anxiety and anger did not and, importantly, gender dysphoria itself did not resolve during suppression; body satisfaction and dysphoria improved only later, after gender-affirming hormones and surgery.^[8] Because the design was uncontrolled, it establishes within-group change over time, not that suppression caused it — concurrent psychological care, maturation, regression to the mean and selection could all contribute. A young-adult follow-up reported wellbeing broadly comparable to population norms, but it measured a highly selected group who completed the whole pathway — suppression, then hormones, then surgery — against a population (not matched-clinical) norm, so it cannot isolate the blocker; the one death it recorded followed vaginoplasty, a complication of the surgical pathway rather than of the blocker.^[9] The foundational evidence, then, shows improvement in general functioning but not in dysphoria during the suppression phase, and cannot separate blocker from pathway thereafter.

The replication that did not replicate. The most direct attempt to test the Dutch psychological claim was a UK study using measures similar to the Dutch work, in 44 carefully selected 12-to-15-year-olds with persistent dysphoria. It found no statistically significant group-level change in psychological functioning across the suppression period: on average, young people were neither better nor worse, with some improving and some deteriorating.^[10] A later reanalysis using individual-level reliable-change methods reached the same conclusion — most showed no reliable change. Two cautions keep this in proportion: it was small and uncontrolled, so a null group mean adds uncertainty rather than proving no effect; and it did not reproduce the Dutch population, selection or care context, so it is a similar test, not a strict replication. On the strength of evidence of this kind, NICE rated the psychological-benefit base low quality.^[2,10]

The studies that report benefit — and their limits. More recent cohorts, mostly from the United States, do report improvements: the most-cited reported adjusted odds ratios of about 0.40 for moderate-to-severe depression and 0.27 for suicidality over twelve months.^[17] Those are adjusted odds from a non-randomised cohort — not relative risk reductions, and not the same as “60% fewer” young people affected — and the treated group comprised young people who received puberty blockers, gender-affirming hormones, or both over the follow-up period, analysed together, so the benefit cannot be attributed to suppression specifically. Several such studies carry heavy, unexplained attrition, and part of the apparent effect comes from deterioration in the untreated rather than improvement in the treated.^[17] A 2025 systematic review that graded this literature read the mental-health signal more favourably than the Cass-commissioned review did; the divergence turns on which studies are included, whether uncontrolled pre-post evidence can be rated moderate quality, and whether outcomes measured after sequential blocker-plus-hormone treatment can answer a blocker-specific question at all — so the more favourable read does not settle the point.^[18]

The honest grade. Put together: the foundational study found improved general functioning but not reduced dysphoria during suppression; the best replication found no significant change; the studies reporting clear benefit cannot separate the blocker from the hormones that follow; and the systematic reviews land on low or inconsistent certainty. The defensible conclusion is that a psychological benefit from puberty suppression specifically is not established — the evidence is genuinely mixed and low in certainty.

That conclusion has to be held with both hands, because it is symmetrically easy to abuse. It does not mean the treatment fails or that young people who report relief are mistaken — absence of high-quality evidence of benefit is not evidence of no benefit, individual adolescents do describe real relief, and being unable to obtain wanted gender-related care — and exposure to restrictive policy environments — has been associated with distress and worse mental health, though that observational signal does not cleanly isolate the effect of GnRHa access from stigma, family conflict or loss of other care. It also does not license the opposite claim that benefit is proven — the strongest positive studies measure the combined hormonal pathway, which is the subject of a separate review, not the blocker alone. What can be said is exactly this much, and no more.

The central promise of puberty suppression — relief of psychological distress — is among the major clinical claims the evidence supports least well. That is a statement about the evidence, not a verdict on the treatment.

Confidence: 🔴 Limited (that puberty suppression specifically improves mental-health or dysphoria outcomes at the group level — the direct evidence is mixed, the most direct UK test null in a small uncontrolled sample, and the positive studies confounded by combined treatment; certainty may differ by specific outcome) / 🟡 Moderate (that the wider gender-affirming pathway is associated with improved wellbeing — graded in the forthcoming adolescent hormones review, not here).

Part IV — Neurodevelopment, and the continuation question

8. Neurocognitive development — the thinnest evidence in the field

Adolescence is a period of intense brain development, and sex hormones are among the signals shaping it — so it is biologically reasonable to ask whether suppressing them affects cognition or brain maturation. It is worth naming the other side of that plausibility: distress, disrupted sleep, depression and social adversity — and unwanted pubertal development itself — also affect cognition and its measurement, so the biology cuts more than one way and does not single out drug harm. The honest position is that the direct human evidence is too thin to answer the question in either direction. The most-cited human study used a functional-imaging executive-function task and found no statistically significant group difference in task performance attributable to treatment — but it was small (a few dozen participants) and cross-sectional, which cannot establish “no effect”, and has been read by others as hinting at a subgroup difference confounded by lower baseline IQ.^[19] A second, cross-sectional study points the other way: membership of a long-duration pubertal-suppression group was associated with poorer executive functioning — but the same study found autism and anxiety symptoms independently associated with poorer executive function too, and executive function was never measured before treatment, so a drug effect cannot be separated from a pre-existing difference at baseline.^[19] The two human studies point in opposite directions, and neither design can settle which is right. A separate study found that pre-treatment IQ predicted later educational achievement; that speaks to baseline ability, not to whether treatment altered cognition, so it is not in itself reassurance about the drug.^[20] Animal work raises a caution that can be neither dismissed nor over-read: pre-pubertal suppression in sheep impaired long-term spatial memory that later hormone treatment did not reverse — a hypothesis worth taking seriously, but an animal finding, in a species whose brain and treatment course differ from a human adolescent’s, and not a demonstration of clinical harm.^[21]

A 2024 review of the neuropsychological evidence reached the only defensible summary: critical questions about the nature, extent and permanence of any cognitive effect remain unanswered, and this is an urgent research priority.^[22] Neither “no cognitive impact” nor “irreversible harm to brain development” is supported by the human data — the first overstates the reassurance, the second overstates the alarm. What can be said is that this is the least-studied major outcome, and that the honest word for it is unknown — and the reason it stays unknown is specific: no longitudinal controlled human study has tracked cognitive trajectories from before treatment through into adulthood, which is precisely the design the question requires. Population flag: the clearest adverse signal is from an animal model; the human executive-function and IQ data are minimal and confounded.

Confidence: 🔴 Limited / genuinely uncertain. The direct human evidence is minimal and mixed; the strongest adverse signal is from animals; permanence is unstudied.

9. The continuation question — “pause” or “pathway”?

The treatment’s founding rationale was that it buys time — a reversible pause that leaves the eventual decision open. The most-cited number in this debate is a Dutch figure that is routinely misread, so it is worth stating exactly. In a cohort of 720 people who started GnRHa in adolescence and went on to add gender-affirming hormones, 704 (98%) were still using those hormones at follow-up.^[23] That is a measure of persistence after starting hormones — once someone had reached the hormone stage, discontinuation was rare (about 2%). It is not a measure of how many adolescents who begin the blocker later progress to hormones, because the cohort was defined by having already reached that stage; used as a blocker-to-hormone transition rate, the 98% is a denominator error. The progression rate itself — of unselected blocker-starters, how many go on — is far less cleanly quantified; the small, highly selected cohorts that report it do show most continuing (in one UK series, 43 of 44 later commenced hormones), but those figures are imprecise and specific to particular selection models, not a general law.^[23,10] What is genuinely established is that very few who reach hormones later stop. What that means for the “pause versus pathway” question is unresolved, and splits cleanly.

The two readings. On one account, the high persistence is the pause working as designed: suppression is started only in carefully assessed adolescents with persistent dysphoria, so very few changing course later is what correct selection should produce, and the near-absence of discontinuation into adulthood is reassuring.^[23] On the other, so few reversing course is read as a sign that starting suppression may not be decision-neutral — that by halting an endogenous puberty during which, in some young people, dysphoria might otherwise change, or by its own social and psychological momentum, the blocker could make continuation more likely rather than simply revealing a pre-set path.^[10] Both readings are internally coherent, and the same figure is compatible with each — which is not the same as saying it supports each equally; what it cannot do, on its own, is tell them apart.

Why the evidence cannot currently adjudicate. The two accounts make opposite claims about a counterfactual — what would have happened to these same adolescents unsuppressed — which is not observed, because starters are not compared with closely matched non-starters. The desistance literature is sometimes brought in to settle it, but it cannot. Older studies of pre-pubertal children found a substantial share no longer identifying as trans by adolescence, though the figure is contested and more recent socially-transitioned cohorts report much lower rates.^[24] Two things stop that resolving the question. It is a largely childhood phenomenon, and those historical clinic samples are not interchangeable with contemporary adolescents assessed for treatment, who differ in sex ratio, age, diagnostic criteria and comorbidity — so the childhood non-persistence rate cannot be transported wholesale onto the treated group; and the studies show only that identity changed over a developmental window, not that puberty itself caused the change. So the desistance data neither prove the blocker channels nor prove it is neutral. Nor is discontinuation uniformly rare everywhere: a large non-probability online survey of transgender and gender-diverse people aged 15–29 across Canada and the United States found that about 17% of those who had started gender-affirming medical treatment had discontinued it at least once — with a change in gender identity given as a reason by roughly a third of those who answered that question, alongside health reasons, cost and access. That is a self-selected sample using a different definition and a broader population, not a follow-up of a treated cohort, so it is not directly comparable with the Dutch prescription-continuation figure — but it is a reminder that continuation estimates are partly an artefact of who is counted and how.^[25]

The honest verdict: persistence once hormones are reached is very high, and progression from blocker to hormones appears high in the selected cohorts that report it; whether that reflects accurate selection of young people who would have persisted regardless, or a channelling effect of suppression itself, is unresolved and may be unresolvable without a matched comparison group. This is the empirical core of the “pause versus pathway” dispute, and it is stated here as an open question rather than answered in either side’s favour.

Confidence: 🟢 Robust (that persistence on gender-affirming hormones, once started, is very high in the studied cohorts) / 🟡 Moderate (that progression from blocker to hormones is high — shown in small selected cohorts, not cleanly quantified for unselected starters) / 🔴 Limited (that either reading — correct selection or a channelling effect — is the right causal interpretation).

Part V — The policy divergence, characterised

10. Why comparable systems reach opposite conclusions

The most revealing fact about the policy landscape is that well-resourced health systems, looking at substantially the same evidence, have reached opposite conclusions — and that this is a feature of the two-questions structure, not a puzzle. There is broad international agreement on the core of the evidence question: for the outcomes that matter most, certainty is low. The divergence is largely — though not purely — on the policy question. Part of it is genuinely empirical and simply unresolved: the outcomes of alternative care pathways, diagnostic reliability, service quality, and the consequences of delay. But the larger part is how much precaution a low-certainty treatment for minors warrants, weighed against the risk that withholding it also carries — and that is a values judgement, which is why the same systematic reviews can sit beneath opposite policies.

The restrictive turn. Several European systems moved toward caution after appraising the evidence. In England, NHS England stopped routine commissioning of puberty blockers for gender dysphoria in March 2024; a separate statutory instrument then restricted the sale and supply of these medicines to under-18s for gender incongruence or dysphoria — chiefly via private and overseas prescriptions, with exceptions for existing patients, other indications and authorised trials — coming into force on 1 January 2025 and due for a first review report by October 2027. (These are two distinct actions — an NHS commissioning decision and a supply restriction — not a single blanket prescription ban.) The trial intended to generate better evidence, the PATHWAYS trial (ISRCTN12491684, King’s College London), had its preparations paused in February 2026, before recruiting, after the MHRA raised safety and wellbeing concerns. Following scientific dialogue, a modified protocol was agreed in June 2026 and approved by the Health Research Authority: it adds minimum entry ages (11 for birth-registered females, 12 for birth-registered males), clearer discontinuation criteria for bone health, cognition and vaginal bleeding, and fuller fertility-preservation information. Recruitment is not due to begin before 1 August 2026 and remains subject to a judicial review of the regulators’ processes.^[26] Sweden restricted suppression to exceptional cases (2022), and Finland, Norway, Denmark and others moved to tightly controlled or research-first settings.^[27] In the United States the picture is different in kind — more legislative and litigated: in United States v. Skrmetti (18 June 2025) the Supreme Court held that Tennessee’s ban did not violate the Equal Protection Clause and was subject only to rational-basis review — it did not rule on the treatment’s clinical merits or resolve whether transgender status is a protected class; a federal executive order (January 2025) defines “child” as under 19 and directs federal agencies to withdraw support for the treatment, though it operates through funding levers and has been preliminarily enjoined; and, as of writing, more than twenty states have enacted bans, several contested in court.^[28]

What Cass did and did not recommend. Because the Cass Review is invoked by both camps, it is worth stating what it actually concluded. It found the evidence insufficient to support routine clinical use, and recommended that puberty suppression for new patients be provided in future only within a formal research protocol — with consent sought for long-term follow-up into adulthood — rather than recommending a permanent prohibition, and separately advised “extreme caution” for masculinising or feminising hormones under 18. The statutory supply restriction that followed was a separate governmental act, not the Review’s recommendation. That gap between the two is the evidence/policy seam of this review, appearing in the primary document itself: an appraisal of evidence quality did not, on its own, dictate the policy that came after it.^[26]

The affirmative position. Against that, major professional bodies maintain that puberty suppression is appropriate care within a multidisciplinary pathway: the leading international standards of care and the 2017 endocrine guidance support puberty suppression for appropriately assessed adolescents after pubertal onset, within specialist multidisciplinary care — neither recommends it for every adolescent simply because Tanner stage 2 has been reached, and both turn on individualised assessment.^[6,29] Bodies including national paediatric societies and the clinic that pioneered the treatment have defended continued access and contested the restrictive reviews’ reading of the evidence — arguing that low-certainty evidence is the norm across much of paediatric medicine, that restriction carries its own harms, and that some appraisals set an unreachable bar.^[26] This is a live, good-faith disagreement among serious clinicians, not a contest between evidence and its absence.

The neutral reading. The divergence should not be mistaken for one side following the evidence and the other ignoring it. Both are responding to the same weak evidence with different tolerances for two different risks — the risk of intervening on the basis of uncertain benefit, and the risk of withholding care from adolescents in distress. Where a system lands depends on which error it more fears, how it weighs irreversibility against delay, and its broader legal, institutional and policy context. This review takes no position on that question; it notes only that it is the question, and that the evidence does not answer it. This section describes a fast-moving policy landscape as it stood at the time of writing and will date faster than the evidence sections; positions and laws are changing across jurisdictions.

Confidence: 🟢 Robust (that systems diverge, and that the divergence tracks the policy question rather than disagreement about evidence quality) / not graded (the policy question itself is a values judgement outside the remit of an evidence review).

Part VI — Evidence quality and bottom line

11. Evidence quality, gathered

The grading in one place. The pattern is consistent: what the drugs do is firm, several of their physical costs are reasonably characterised, and the psychological benefit for which they are principally given — along with the neurocognitive question and the meaning of near-universal continuation — is where the evidence is weakest. Ratings use the series scheme — 🟢 reasonably robust, 🟡 moderate, 🔴 limited — as editorial judgements, not formal GRADE. See Reading the Evidence.

12. Frequently asked questions

Do puberty blockers work? At what they physically do — pausing puberty — yes, reliably. At the psychological benefit they are mainly given for, the evidence is weak and mixed, and does not establish that the blocker itself improves mental health. Both halves of that answer are true at once. (Sections 4, 7)

Are they reversible? The pause is: stop the drug and endogenous puberty usually resumes. But “reversible” is not “without lasting consequence” — bone, fertility and neurodevelopmental outcomes depend on how long treatment lasts and what follows, and are not fully characterised. “Completely reversible” overstates it; “irreversible damage” overstates it the other way. (Sections 5, 6, 8)

Do they cause infertility? The blocker alone does not destroy the gonad. But starting early and going straight to gender-affirming hormones may leave no window for conventional gamete preservation unless treatment is interrupted, and the options at early puberty are limited or experimental — which is why counselling beforehand matters. (Section 6)

Is it true that almost everyone continues to hormones? The most-quoted 98% is narrower than it sounds: it means that, among those who reached gender-affirming hormones, almost all stayed on them — discontinuation after starting hormones is rare. How many blocker-starters progress to hormones in the first place is less cleanly measured (high in the small selected cohorts that report it). Whether high continuation reflects good selection or a channelling effect of the treatment is genuinely unresolved. (Section 9)

Why do countries disagree so sharply? Mostly not about the evidence — nearly everyone agrees it is low-certainty — but about how much caution a low-certainty treatment for minors warrants against the harm of withholding it. That is a values question the evidence cannot settle. (Section 10)

13. The bottom line

1. Keep the two questions apart: what the evidence shows, and what should follow from it. The first is answerable and, for the outcomes that matter most, converges on “low-certainty”; the second is a values judgement this review does not make.
2. What is firm: GnRHa reliably suppress gonadal puberty; they reduce bone accrual during treatment; the blocker alone does not ablate the gonad and the axis usually reactivates on stopping.^[3,13,16]
3. What is genuinely uncertain: the psychological benefit of suppression specifically, the neurocognitive effect, long-term fracture and fertility outcomes, and whether near-universal continuation reflects selection or channelling.^[8,10,15,22]
4. What is mis-stated in both directions: “completely reversible / fully safe” and “irreversible damage / chemical castration” both claim a certainty the evidence withholds.
5. The honest posture for anyone weighing this — clinician, family, policymaker — is to hold uncertain benefit against uncertain harm openly, counsel fertility and bone explicitly, and resist the false comfort of a confident answer in either direction.