Classifying Autism Spectrum Disorders by ADI-R: Subtypes or Severity Gradient?

The team fed 463 ADI-R records into a cluster analysis. Ages ranged from 3 to 21. The goal was to see if autism sorts into neat boxes or a sliding scale.

Three severity groups popped out: high, medium, and low. The groups sat on a continuum, not in separate bins. This favors a gradient view of ASD rather than rigid subtypes.

Sajith et al. (2008) used taxometric math and claimed discrete subgroups based on social-communication and IQ. Cholemkery et al. (2016) disagree, finding smooth severity clusters instead. The clash is only skin-deep: G sampled mostly clinic-plus-ID cases, while Hannah cast a wider net, letting milder kids blur the lines.

Morales-Hidalgo et al. (2018) echoed the continuum idea in mainstream schools. Kamensek et al. (2023) saw the same smooth spread for face-processing deficits. Three different tools, three different samples, same story: autism traits slope upward, they don’t jump categories.

Shuster et al. (2014) swept 36 factor papers and still landed on two core domains: social/communication and restricted/repetitive. Hannah’s three severity clusters nest neatly inside those two domains, showing you can have both broad factors and fine-grained levels.

Stop forcing kids into ‘mild’ or ‘severe’ buckets. Think slider, not switch. Use ADI-R totals to place a child along the gradient, then match teaching pace, prompt level, and sensory breaks to that spot. Update the slider each reassessment; severity can move.