Landmark Meta-Analysis Finds Promising but Immature Evidence for Autism Behavior Therapy

Challenging behaviors such as aggression, self-injury, and destructive outbursts can upend the lives of autistic children and their families, often forcing schools and care services into restrictive or exclusionary responses. For more than a decade, a behavioral intervention known as skill-based treatment, or SBT, has been promoted as an efficient, effective, and safe way to address these behaviors by teaching children to communicate their needs instead. Now, a systematic review and meta-analysis published in the Journal of Autism and Developmental Disorders has delivered the most rigorous accounting yet of the evidence behind the approach, and the verdict is a nuanced one: the treatment shows genuinely encouraging effects, but the research base supporting it remains surprisingly thin.

The study, conducted by Nathan Bunyan, Clarissa Feuerer, and Athanasios Vostanis of the Tizard Centre at the University of Kent, was pre-registered on PROSPERO and followed PRISMA reporting guidelines. The team searched seven databases, hand-searched thirteen behavior-analytic journals, contacted key authors, and combed grey literature in the form of dissertations. After screening more than thirteen thousand records, they identified thirty studies, twenty-seven published and three unpublished, representing 111 individual applications of SBT across 107 participants. That scale matters, because until now only a single scoping review had examined SBT itself, and it reported no formal effect sizes and no assessment of study quality or risk of bias.

To understand what the reviewers were evaluating, it helps to know how SBT works. The intervention package rests on two pillars. The first is functional communication training, one of the most extensively researched behavioral interventions, in which a child is taught an alternative communication response that serves the same function as the challenging behavior, such as requesting a break or asking for attention. The second pillar is delay and denial tolerance training. Because the new communication response is initially reinforced every single time, the reinforcement schedule must gradually be thinned so the skill can survive in the real world, where requests are sometimes refused or delayed. In SBT, this thinning is contingency-based: children are asked to engage in meaningful tasks or activities during the delay period before earning reinforcement, an approach research suggests produces better tolerance and lower rates of problem behavior than simply making children wait.

Before treatment begins, practitioners conduct a functional analysis to identify the environmental events that evoke and reinforce the challenging behavior. Most SBT implementations in the reviewed literature relied on a variant called the interview-informed synthesized contingency analysis, in which caregivers are interviewed openly and the resulting contingencies, often combinations such as escape from demands paired with access to preferred items, are tested together in a single synthesized condition. In the treatment phase, many applications taught an omnibus communication response, a single phrase such as “my way” that accesses all identified reinforcers, before progressively shaping more complex requests. A specific tolerance response, such as taking a breath and saying “okay,” is then trained to help children cope with denial of reinforcement without relapsing into problem behavior.

The meta-analysis quantified outcomes using Baseline-Corrected Tau, a non-overlap effect size metric that adjusts for baseline trends and aligns well with visual analysis of single-case data. Across the 59 participants for whom values could be computed, the mean Tau-BC was 0.47, falling squarely in the moderate range, with individual values spanning from minus 0.35 to a perfect 1.00. Twenty-five participants showed moderate change, fifteen showed large change, and eight showed very large change, while eleven showed only small change. The reviewers caution, however, that forty-two participants had wide 95 percent confidence intervals, indicating substantial uncertainty around many of the estimates. When the analysis was restricted to participants without wide intervals, the picture brightened considerably: eight very large changes, two large, six moderate, and just one small.

Quality assessment told a less flattering story. Using the Evaluative Method, the reviewers rated only two studies as borderline-strong and four as adequate, while thirteen were borderline-adequate and eleven were weak. In the abstract’s summary figures, 37 percent of the literature scored weak and just 7 percent reached borderline-strong. An evidence-based practice Z score of 46, calculated from the six studies rated adequate or above, suggests SBT qualifies as a probable evidence-based practice on paper. Yet the authors urge caution. They note that the Evaluative Method can be overly sensitive, penalizing studies that publish unflattering data from some participants, which may actually indicate a lower-risk design such as a consecutive case series. The risk of bias tool flagged further concerns: every study was rated high risk for blinding of participants, personnel, and outcome assessors, and in half the studies participants received pre-session teaching that inflated the apparent immediacy of treatment effects.

One clear strength of the SBT literature is its attention to social validity. Twenty-three of the thirty studies, covering 73 percent of participants, reported social validity data, almost always via questionnaires completed by caregivers, home care teams, or school staff. Wherever procedures and outcomes were rated, responses were positive, and two individuals who received the treatment themselves also reported positive experiences. Some studies went further, involving parents in goal setting, allowing parents to observe or terminate sessions at will, and letting children choose whether to participate. Notably, parents given the option to end sessions chose not to. Still, the reviewers point out that social validity was rarely measured from the perspective of the children receiving treatment, and they call for qualitative research with autistic adults, parents, and recipients of support to broaden the evidence base.

The review also exposes striking gaps in who has been studied and where. Around three-quarters of participants were male, 96 percent were under fourteen years old, and 80 percent were autistic, with most able to speak in full sentences or multi-word phrases. Only two participants were over eighteen. Nearly all studies originated in the United States, most were conducted in outpatient clinics by therapists, graduate students, or board-certified behavior analysts, and most research teams were connected to the originators of the approach, with only eleven studies coming from independent groups. Vocal verbal communication dominated both the initial and terminal communication responses, with aided augmentative and alternative communication appearing in roughly a fifth of cases and unaided AAC almost never. For practitioners hoping to apply SBT in schools, homes, or with minimally verbal individuals, the reviewers warn that procedures may need adaptation, timelines may differ, and effectiveness is not yet guaranteed outside the well-studied profile.

Practical details emerged that will interest clinicians and families weighing the costs of care. Where data were available, treatment averaged 68.78 total hours, with a range of 23 to 167 hours, and an average estimated cost of just under 6,000 dollars, suggesting SBT may be comparatively cost-effective. Most applications allocated fewer than twenty hours per week. Implementers also adapted procedures in ways that appear to reduce distress: some achieved reductions in challenging behavior without extinction at all, others applied extinction only to less dangerous behavior or used differential reinforcement, and the majority reinforced precursors or milder forms of problem behavior. These choices align with growing movements toward trauma-assumed, assent-based, and compassionate care in behavior analysis, though the reviewers note that some schedule-thinning practices, such as resetting reinforcement criteria, could still be experienced as aversive and deserve closer study.

The authors’ bottom line is measured rather than triumphant. SBT emerges as an emerging, potentially evidence-based practice with moderate average effects, strong stakeholder endorsement, and a flexible structure that clinicians individualize extensively. But with most studies rated weak, blinding absent across the board, and the evidence concentrated in a narrow slice of the population, the field needs higher-quality, independently conducted research in more diverse settings before confident, widespread recommendations can be made. The reviewers recommend concrete fixes: clearer operational definitions, reported inclusion criteria, demographic transparency, fidelity measurement, consideration of blinding, maintenance and generalization testing, and including pre-session teaching data in outcome measurement. For now, they advise practitioners to treat SBT as a general template, weighing the evidence for a particular population or context and tailoring the intervention to each individual’s needs and preferences, rather than as a one-size-fits-all prescription.

Subject of Research: Systematic review and meta-analysis of skill-based treatment for challenging behavior in autism and developmental disabilities

Article Title: Evaluating the Effects of Skill-Based Treatment: A Systematic Review and Meta-Analysis of Single-Case Studies

Article References: Bunyan, N., Feuerer, C., & Vostanis, A. (2026). Evaluating the Effects of Skill-Based Treatment: A Systematic Review and Meta-Analysis of Single-Case Studies. Journal of Autism and Developmental Disorders. https://doi.org/10.1007/s10803-026-07547-2

Image Credits: AI Generated

DOI: 10.1007/s10803-026-07547-2

Keywords: skill-based treatment, challenging behavior, autism, functional communication training, meta-analysis, single-case design, schedule thinning, functional analysis, social validity, evidence-based practice, behavior analysis, Tau-BC