Spectrum Connect · Research

Do telehealth therapies actually help autistic children?

20 systematic reviews Ray Kawai, Co-Founder June 2026
⚠ Test run — not for publication This is a draft evidence summary built from an internal methodological audit trail. Its conclusions are pending independent sign-off (the external review bench, "Gate D," is not yet in place) and the review was not pre-registered on PROSPERO/OSF. It is shared here for review only — not medical advice, and not a finished, citable Spectrum Connect publication.

Section 01

Where we looked

PubMed Cochrane Library Epistemonikos ERIC PsycINFO Embase CINAHL

Search query (runnable on PubMed)

("autism spectrum disorder"[MeSH] OR autistic[tiab] OR ASD[tiab]) AND (telehealth[tiab] OR telepractice[tiab] OR telemedicine[tiab] OR telerehabilitation[tiab] OR videoconferencing[tiab]) AND (ABA[tiab] OR "applied behavior analysis"[tiab] OR speech[tiab] OR "occupational therapy"[tiab] OR "parent training"[tiab] OR "caregiver coaching"[tiab]) AND (systematic[sb] OR "meta-analysis"[pt])

Live PubMed counts captured 24 June 2026: 516 records unfiltered, 45 once limited to systematic reviews. Paywalled databases (PsycINFO, Embase, CINAHL) were proxied through Epistemonikos, which independent testing found covers ~92% of reviews; Cochrane, Epistemonikos, and ERIC counts are pending a reproducible re-run.

Section 02

What we did with what we found

516 records found in the PubMed search (unfiltered, live count)
↓
45 systematic reviews read in full after screening
↓
20 reviews included in this overview

The 25 reviews set aside were narrative or assessment-only reviews, protocols with no results yet, or mixed-population reviews with no extractable autism data — each one itemised with a reason. The funnel is anchored on the live PubMed count; the full multi-database total is pending a human re-run of the JavaScript-rendered databases.

Section 03

Where the 20 reviews landed

Supportive
12
Mixed / sparse
6
Insufficient
2
Pan 2023 — fidelity↑, self-efficacy↑, parent stress↓
Meta-analysis · 17 trials, 902 parents
Neely 2021 — supports remote assessment + intervention
54 studies (44 single-case)
Ferguson 2019 — improvement on ≥1 outcome in every study
28 studies, mostly single-case
Hao 2023 — pragmatic (social) language improved
Meta-analysis · 21 studies
Ellison 2021 — equal to or better than in-person
Critical/systematic review
de Nocker & Toolan 2021 — comparable to face-to-face
16 group-design studies
Sutherland 2018 — positive, equivalent to in-person
14 studies · speech
Clarke 2023 — daily-living skills improved
17 studies (few)
Unholz-Bowden 2020 — caregiver training works
~30 single-case · thin outside N. America
Sivaraman & Fahmie 2020 — adapts across cultures
9 studies, 9 countries
Knutsen 2016 — early telemedicine-in-autism review
Mixed designs
Boisvert 2010 — foundational telepractice review
8 studies (now dated)
Tomlinson 2018 — some null results
Implementer training · mixed
Parsons 2017 — only one RCT found
Systematic review · sparse
Tan-MacNeill 2021 — some null results
Mixed neurodevelopmental
Parent-training SR 2025 — autism subset only
SR + meta-analysis · mixed-DD
Bundy 2023 — COVID service disruption
45 sources · scoping
Kane & DeBar 2023 — descriptive, not systematic
Descriptive review
Dehghani 2023 — tele-OT, but mixed disability
RCTs · not autism-specific
Cochrane CD014793 — protocol only, no results yet
Cochrane protocol

Section 04

Strongest evidence (and how sure we are)

What was measured Certainty What it means
Behavioural coaching (FCT) reducing challenging behaviour — vs no treatment Low Large, repeated drops in problem behaviour — but mostly small single-case studies from one research group
Parent fidelity, confidence, and stress Low–Moderate Controlled trials and a meta-analysis show parents coached over video improve and feel less stressed
Speech — pragmatic (social) language Low A meta-analysis shows improvement, but only for this one language domain
Is telehealth as good as in-person? (any therapy) Very low Only a few tiny head-to-head studies found "no difference" — no properly powered trial exists
Occupational therapy — any outcome Insufficient No autism-specific review evidence; a single 15-child pilot is all that exists

Section 05

What this means for you

If therapy over video is what you can access, the evidence is genuinely encouraging — especially when a therapist coaches you, the parent, to work with your child. Coaching parents in behavioural strategies remotely produces large, repeated drops in challenging behaviour, and parents gain skill and confidence while stress goes down. Two honest cautions: "it works" is not the same as "as good as in-person" — that head-to-head question hasn't been properly tested for any therapy yet. And for occupational therapy, there simply isn't enough autism research to judge.
This summary is for information only and is not medical advice. Talk to your child's doctor or therapy team before starting or changing any treatment. Telehealth also depends on reliable internet, a device, and some comfort using it. Status: test run — conclusions pending independent sign-off and not yet a finished Spectrum Connect publication.
Full methodology record
01 Question & scope — Four telehealth modalities (behavioural/ABA, speech-language, occupational therapy, parent training) were kept separate, and two claims were held rigorously apart: "the therapy works" vs "telehealth equals in-person."
02 Three-layer search — A reproducible PubMed Boolean backbone (the system of record), a logged discovery booster, and a deterministic de-duplicated merge. Search could order what to read first, but never decide what counted as evidence.
03 Per-database provenance — Every named database has either a reproducible live count or a documented reason none was available; no funnel number is an estimate.
04 Funnel & screening — 516 → 45 → 20, anchored on live PubMed counts; the 25 excluded reviews are itemised with reasons and resolving identifiers.
05 Appraisal — Each key review was rated with AMSTAR 2, adapted so single-case-design reviews aren't unfairly under-credited. Ratings are drafted and await human duplicate review.
06 Overlap check — Apparent agreement is partly an artefact of two recurring primary-study clusters (the Iowa FCT lineage and the Ingersoll/Vismara parent-coaching lineage); shared trials were resolved at the study level.
07 Synthesis & GRADE — Findings were summarised narratively by outcome to avoid double-counting shared trials, and certainty was graded per outcome. No modality reaches "Settled" — the correct result given the evidence.

Prepared by: Research agent (Claude) under the accountability of Ray Kawai, Co-Founder, Spectrum Connect. Independent sign-off (Gate D): PENDING · Registration (PROSPERO/OSF): PENDING. This is a test-run audit artefact, not a published clinical recommendation. Source verification complete — every cited identifier confirmed live and non-retracted as of 24 June 2026.

Think we got something wrong?

We publish the whole record so it can be checked — and that only counts if we act on what you find. If a number looks wrong, a study is missing or has been retracted, or we’ve read a finding in a way the evidence doesn’t support, tell us.

You don’t need a research background to file one. “This doesn’t match what our doctor told us” is a useful report. Every one reaches a person: we reply within seven days, and within thirty we have either corrected the page or told you when we will. Substantive reports send the affected steps back through the protocol and need fresh sign-off before anything here changes.

Report a factual error