Sex-Related Measurement Bias in Autism Spectrum Disorder Symptoms in the Baby Siblings Research Consortium.

Researchers studied over 4,500 young children to see if autism assessment tools work differently for boys and girls. They found that girls with autism often show milder signs than boys, particularly in eye contact and social interactions. This means current autism tests might miss girls or diagnose them later because they don't show symptoms the same way boys do. The study suggests we need better tools and training to recognize autism in girls earlier.

This large-scale cohort study examined sex-related measurement bias in autism diagnostic tools using data from 4,550 children (56% male) in the Baby Siblings Research Consortium. Researchers analyzed ADOS assessments across multiple visits between ages 20-40 months, comparing children with and without autistic siblings. Key findings revealed systematic measurement bias favoring males in autism symptom detection. Females showed milder presentations in eye contact, joint attention response, and social overtures compared to males with equivalent underlying autism traits.

The study suggests current diagnostic thresholds may inadequately account for sex differences in autism presentation, potentially contributing to delayed or missed diagnoses in females.

Clinicians should be aware that autism may present more subtly in females, particularly in eye contact and social behaviors. Diagnostic training and tools may need modification to better identify autism in girls. Earlier identification could improve outcomes for autistic females.

Study limited to children aged 20-40 months from families already involved in autism research. Findings may not generalize to broader population or older children. Long-term diagnostic outcomes not fully examined.

Disparities exist in age of diagnosis and prevalence of autism spectrum disorder (ASD) for female compared with male children. Correcting for sources of bias is critical for improving equitable ASD identification. To determine whether sex differences exist in measurement of ASD symptoms using the Autism Diagnostic Observation Schedule (ADOS) among young children at high familial likelihood (HFL) and low familial likelihood (LFL) of ASD. This cohort study collected longitudinal, prospective data from the Baby Siblings Research Consortium between January 1, 2003, and December 31, 2021.

Participants included 3106 children who had an older sibling with ASD (HFL group) and 1444 without (LFL group). Data from as many as 3 visits when participants were aged 20 to 40 months were included. Analysis occurred between March 1, 2023, and May 29, 2025. Child sex and age and ASD diagnosis.

Measurement invariance by sex and age was examined across item-level ADOS data. Diagnostic group and sex differences were then examined using mixed-effect models on corrected scores. Repeated visits (n = 7557) from 4550 participants (2548 [56.0%] male) were included, of whom 1444 (31.7%) were in the LFL and 3016 (68.3%) in the HFL groups. Confirmatory factor analysis indicated social communication and restricted and repetitive behaviors models fit the data well in the HFL group but poorly in the LFL group.

In the HFL group, females were rated as less impaired in eye contact (differential item functioning estimate [SE] = 0.088 [0.033]; P = .01), and their response to joint attention (differential item functioning estimate [SE] = 0.290 [0.105]; P = .01) and quality of social overtures (differential item functioning estimate [SE] = 0.053 [0.019]; P = .005) was associated with less underlying social communication difficulties compared with males. Adjusting for differential item functioning by age and sex resulted in moderate levels of measurement differences. Females showed milder autistic traits than males, although this gap was smaller in the participants diagnosed with ASD. Sex differences exist in the general population in many social communication traits, yet ASD diagnostic thresholds do not account for these sex differences.

Future instrument development, as well as clinician training, should acknowledge milder presentation (fewer difficulties with eye contact or quality of social impairments) in many females. This may help identify developmental differences earlier and improve outcomes for autistic females (estimate [SE] = -0.160 [0.061]; P = .009).