The Tools Were Built for Boys

What the research found

This is a narrative review, published in Neuropsychiatric Disease and Treatment in 2024 by Julia Cook, Laura Hull and Will Mandy of University College London. A narrative review is not a new experiment. The authors did not recruit participants or run a trial. They read the existing literature and built an argument from it. So there is no sample size here, and no new numbers of their own. What they offer instead is a synthesis: why the diagnostic process misses so many girls and women, and how it might be adjusted.

The starting point is a ratio. For years, autism has been described as roughly four times more common in boys than girls. The authors report that this figure is now doubted. Once you account for girls who are diagnosed late or missed entirely, they write, the true male-to-female ratio “may be close to three-to-one or even two-to-one.” Autism overall is estimated at 1 to 2 percent of the population, though in some UK groups, such as boys aged 5 to 9, reported diagnosis rates reach around 4 percent. The gap between the sexes, in other words, may be smaller than the referral numbers suggest.

The review names three problems and pairs each with a recommendation. The first is that the standard tools were built around a male picture of autism. The most widely used instruments, including the “gold standard” ADOS-2 and ADI-R, “were created at a time when a male-centric understanding of autism predominated, and were evaluated with predominantly male samples.” A girl whose interests and social behaviour look different from that template can score below the threshold while still being autistic. The recommendation is to widen the range of behaviours an assessment counts as evidence.

The second problem is camouflaging. The authors define it as “hiding or compensation for autistic characteristics” — learning and rehearsing neurotypical social behaviour, suppressing visible autistic behaviours such as hand-flapping, and mimicking others. Camouflaging, they write, “may account for the later age of diagnosis for many girls and women.” A person who has spent years masking may present, in a one-hour clinical appointment, as unremarkable. The recommendation is to ask about masking directly and to use structured measures of it, such as the CAT-Q questionnaire.

The third problem is diagnostic overshadowing. Autistic girls and women frequently arrive at services with a mental-health condition already attached. The review cites research in which 70 percent of autistic young people had at least one mental-health diagnosis and 40 percent had two or more, and notes that 20 to 30 percent of women in treatment for anorexia nervosa are autistic. When anxiety, depression or an eating disorder is the presenting concern, clinicians may attribute the underlying autistic traits to that condition and stop looking. The recommendation is careful, comprehensive assessment so that a co-occurring condition does not hide the autism beneath it.

What it means for you and your child

If you are a parent who has felt that your daughter does not match the description of autism you were given, this review explains, in the authors’ words, why that mismatch happens. The clinical picture most of us hold in our heads was drawn from boys. A girl can meet the criteria and still be waved through, because the assessment is looking for a version of autism she does not display on the surface.

That is the useful part. But it is worth being precise about what this paper can and cannot do. It is a review, not a trial. It gathers evidence and reasons from it. The authors are honest about the limit this imposes. They point to “the lack of research evaluating any adaptations to diagnosis through experimental paradigms” and say their recommendations “would benefit from thorough empirical evaluation.” In plain terms: nobody has yet run the studies that would show, with hard data, that asking about camouflaging or widening the behavioural criteria produces more accurate diagnoses. The case is reasoned and plausible. It is not settled.

The numbers here carry the same caution. The “two-to-one or three-to-one” ratio is an estimate about a population, not a claim about your child. It tells you that girls are under-identified as a group. It cannot tell you whether any particular girl is autistic. Only a proper assessment can do that, and this review is an argument for making that assessment better, not a substitute for it.

So what is practical? If you are seeking an assessment, it is reasonable to ask the clinician how their process accounts for masking, and whether they will look past a mental-health label that is already in place. A daughter who “seems fine at school” and “falls apart at home” is displaying exactly the pattern this review describes — effortful masking in public, exhaustion in private. That contrast is worth naming to whoever is doing the assessment.

Keep your expectations calibrated. A good assessment considers how a child behaves across settings, over time, and in her own account of her inner effort — not just how she presents for one hour in an unfamiliar room. This review does not offer a new test or a shortcut. It offers a well-argued reason to look more carefully, and an honest admission that the tools to confirm its own recommendations are still to be built. Both halves of that sentence matter.