Strengthening the evidence for early childhood interventions for autism

Early childhood interventions have become a cornerstone of treatment for toddlers and young children on the autism spectrum. But which children are likely to benefit, which interventions are most effective, and could some cause unintended harms?

To answer such questions, Micheal Sandbank, PhD, assistant professor of health sciences at the University of North Carolina at Chapel Hill School of Medicine, is turning a discerning eye toward the research literature. Sandbank leads Project aim (Autism Intervention Meta-Analysis), a comprehensive meta-analysis of research on early interventions for autistic children.

Early in her career, Sandbank worked as a special education teacher and clinician trained in Applied Behavior Analysis (ABA) therapy, a widely used intervention considered a best practice for treating autism and other developmental disorders. But her understanding of ABA evolved as she moved into research. During her master’s and PhD in the renowned special education program at Vanderbilt University’s Peabody College, Sandbank trained in meta-analytic approaches and learned to evaluate research methodology through a rigorous lens. Now, she has circled back to take a fresh look at the various interventions designed to support young autistic children. We asked Sandbank to sum up what Project aim has revealed so far.

With Project aim, we include any nonpharmacological intervention studied with autistic children under age 8, regardless of what outcomes measures the study looked at.

A typical meta-analysis starts by looking at an intervention and its effect on some specific outcome. We’re not expectant in that way. Instead, we’re seeking to describe and summarize the body of research on this topic. We also rate studies on measures of quality and reliability so we can start to assess the evidence from high-quality studies.

When I started, I thought it was going to be one paper—I didn’t realize it would take over my entire research program.

When we peel things back, we see this progression. When you include all the studies, you get a very rosy picture suggesting there are all sorts of interventions that are effective in producing a range of positive outcomes. But if you limit the analysis to only include, say, randomized controlled trials, things don’t look so rosy.

There’s a stark difference between what’s endorsed by all studies versus what’s endorsed by the high-quality studies. In my opinion, it tells a story that certain interventions have been oversold as being more evidence-based than they actually are.

In the last 10 or 20 years, researchers began doing randomized controlled trials of more developmentally informed interventions. That work suggests, for example, that naturalistic developmental behavioral interventions can promote early social communication. In a young child, that may be something like learning to follow an interaction partner’s gaze.

These kinds of early social communication skills are theorized to be foundational to other kinds of communication skills, such as language. There’s also evidence that child-led, play-based developmental interventions are likely to benefit early social communication skills that may trigger developmental cascades on other learning domains.

Then there are some interventions, such as horseback riding therapy, where there really aren’t many studies, so we can’t yet say whether the intervention is helpful. And we shouldn’t assume that untested interventions are benign. In some cases, an intervention might even be harmful.

We also see a group of interventions that are heavily studied, but often with quasiexperimental research and far fewer randomized controlled trials. One example is intensive ABA therapy, which is the primary intervention type provided to autistic children in the United States. These intensive therapies recommend really high numbers of hours, up to 40 hours per week. Across the United States, ABA therapy clinics have become extremely prominent, yet there are only maybe two or three randomized controlled trials of that kind of intervention—and they are small and have mixed results.

I came up in the ABA world and was always told it was the gold-standard evidence-based intervention, so I was surprised to find we don’t actually have a lot of rigorous tests to guide our recommendations for these high-intensity protocols.

There’s a lot of work that suggests that children benefit from really high amounts of intervention, lasting from 25 to 40 hours a week. But when we peel back the layers, we find holes. Some studies use stronger research methodology than others, and when we control for the quality of the research, we don’t see that relationship between more therapy and better outcomes.

To be clear, that doesn’t mean more intervention is not better. But the truth is, we can’t say so with certainty. It’s likely that there’s some amount at which children do benefit from more therapy. But at some point, you may not get any more bang for your buck—and that point probably differs based on the developmental profile of the child. We need more research so we can recommend interventions—and intervention amounts—that are most likely to benefit a given child.

There are definitely kids who have experienced significant developmental progress in intensive interventions, and there are families who say intensive therapy has changed their lives. But the studies also find that some participants don’t seem to benefit from these interventions. One mother told me she has a lot of heartache about it now, because they did the 40 hours a week for years and it didn’t help.

Forty hours is a lot. We’re basically recommending that toddlers have full-time jobs. The implication is that being autistic is so bad that any possible improvement is worth it to be withdrawn from school, to work at this up to 40 hours per week, to treat it. I think we must face the ableism inherent in that recommendation. If we really value autistic children, we’ve got to consider that they deserve a childhood.

Of course, we do sometimes recommend treatments that have costs or side effects, because we decide the cost is worth the benefit. But we shouldn’t make those recommendations without taking a lot of care to weigh the costs.

Parents are sometimes told this is the thing you have to do, and you have to do it as soon as possible and move mountains to do it. But there is not one optimal intervention for every autistic child, and there is not one optimal intervention amount for every autistic child.

We need to make sure families are fully informed about the fact that not every child will benefit from a given intervention, and to not feel pressured into an intervention—especially if it’s something that requires them to withdraw their child from school, for example, or to take on significant financial debt.

We also want families to be informed about outcomes they would hope to see as the result of an intervention, as well as potential side effects we might be concerned about. Might an intervention interfere with sleep, for example?

We’ve been recommending interventions based on the assumption that they can only be helpful, but we shouldn’t assume. One finding from our analyses is that there’s very little monitoring of adverse effects from these interventions. That doesn’t mean that they’re harmful, but it means we don’t always know enough to fully weigh those risks versus benefits.

It’s tricky because there’s a desire to boil things down into clinical recommendations, but it’s nuanced. My overall takeaway is that there are fewer high-quality studies than we’d like to see. That doesn’t mean we can’t or shouldn’t provide interventions and supports to children, but it does mean we need to be cautious in our recommendations and not present anything as a silver bullet. Clinicians are sometimes led to believe that intensive ABA therapy is always the best approach. But in fact, we don’t yet know the optimal intensity of treatment.

Clinicians can help families understand that while there’s some good evidence for a variety of different approaches, we’re not sure that they are helpful for everyone, and there’s also a potential for adverse effects. It’s helpful for clinicians and families to talk about the skills they want their child to learn, and to think about what they would consider a marker of progress, and what might be a sign of concerning outcomes that could arise.

I would also advise clinicians to get familiar with the networks for autism support in their communities. Ideally, interventions should be interdisciplinary, with children getting support from a variety of professionals (such as speech therapists, occupational therapists, and behavioral therapists) working together.

I’m now trying to get funding to start a large randomized study looking at the amount of intervention that children get every week. I want to have enough participants that we can determine if the benefit differs, not only by the intervention amount but also by their developmental profile. That kind of evidence would help us give more individualized intervention recommendations that are rooted in high-quality evidence.

In the meantime, I’m definitely not suggesting children shouldn’t get the supports they need. The point is to make sure that families are fully informed when they’re choosing those supports.