Skip navigation

Calls to Action

Can autism be measured in a sniff?

Perspective on some splashy headlines about a sniff test for autism - and the real importance of research on autism’s early warning signs

By Lucia Murillo, Autism Speaks assistant director of education research and developmental pediatrician Paul Wang, vice president and head of medical research.


This Sniffing Test Can Detect Autism in Children With 81 Percent Accuracy”

New autism study shows you can sniff it out!”

These are just a couple of more than a score of headlines we’ve seen in the past week – following the publication of a study that looked at autism-related differences in how long children smell pleasant versus unpleasant odors. In fact, we were both interviewed for some of these stories, though in some cases our precautionary warnings didn’t make it into the news report.

Why caution?

This study was small – just 36 children, ages 4 and older, half of whom had been previously diagnosed with autism. So while the “sniff test” correctly classified 12 of the 18 kids who had autism, it missed a third of them. It also correctly identified 17 of 18 of the “typical” kids. So technically, yes, it was 81 percent accurate, correctly classifying 29 out of the 36 kids.

But how meaningful was this result? And are we anywhere near ready to install sniff machines in doctor offices to diagnose autism?

The answer to the second question is definitely “no.” Autism is complex, and many signs and symptoms need to be considered in making a diagnosis. And as mentioned, the sniff test missed a third of the kids who had been diagnosed with autism using traditional methods.

Still, it would be interesting to see others repeat this study with a larger – and ideally younger – sample of children.

Why pursue a larger study if it doesn’t make sense to diagnose autism based on something as one-dimensional as sniffing behavior?

On one thing, we agree: The earlier a child is diagnosed with autism the better. This is because evidence-based early intervention services can often reduce autism’s core symptoms and improve both learning abilities and overall daily functioning. Early intervention also offers parents support and information that will help them better understand their child’s behavior and needs. A good early intervention program also teaches parents strategies for working and playing with their children in ways that support their development.

So it would be wonderful if a larger study with very young children confirmed that this smell test gave us another indicator that a child is at risk for developing autism.

What’s more, studies on early behavioral differences give us insights into how autism develops and how we might better target early interventions to improve children’s outcomes.

Reliably identifying more very early warning signs can help us do that. Currently, autism can be most- reliably diagnosed after 18 months. Before then, we have but a limited number of red flags that can help us identify children who should be closely followed and considered for early intervention. Some of these include:

A child not meeting their developmental milestones (See Learn the Signs.)

A lack of gestures such as pointing, waving (See the Modified Checklist for Autism in Toddlers.)

Loss of skills (regression)

Although under and over responsiveness to sensory information (like smell) can be a characteristic of autism, it’s only one of multiple symptoms we consider before making a diagnosis. A reliable, gold-standard diagnosis requires evaluating key behaviors involving social communication, social interaction and restricted or repetitive patterns of behavior. 

The Autism Speaks blog features opinions from people throughout the autism community. Each blog represents the point of view of the author and does not necessarily reflect Autism Speaks' beliefs or point of view.