Insomnia has a first-line treatment that is not a drug: cognitive behavioural therapy for insomnia, known as CBT-I, has held that position in guidelines for years. What it has never had is enough clinicians to deliver it. A course means weeks of structured sessions, and in most health systems the waiting list is the treatment’s defining feature.

That arithmetic is what makes a trial published on 10 November in the journal Sleep worth reading. Researchers in the Netherlands took an existing internet-delivered CBT-I programme, rebuilt it with a panel of autistic adults, and tested the result in 163 people with moderate clinical insomnia. It worked, and it worked specifically on sleep.

Why this population needed its own programme

Estimates cited by the authors put insomnia symptoms in about 60% of autistic adults, with broader sleep problems reported across 79% to 89%. The paper lists sleep among the priorities the autistic community itself identified for research, and connects poor sleep to depression, suicidality and reduced day-to-day functioning.

Standard CBT-I is first-line in European insomnia guidelines, and it works in people with co-occurring depression, PTSD and alcohol dependence. But no randomised trial had tested it in autistic adults before this one, and the adaptations are not cosmetic. Face-to-face clinical contact can be taxing for people who find eye contact, body language and simultaneous sensory input hard to process. Sleep restriction — a core CBT-I ingredient — asks people to change routines, which collides with a preference for predictability.

The digital format answers both problems: self-paced, predictable, and reachable without a waiting room. It also scales, which matters when the alternative is a trained therapist. Every participant here had a coach who, the authors report, spent under 30 minutes per module.

To adapt the course, the team worked with seven autistic adults in two rounds of written feedback and focus groups. Instructions became informative rather than directive, and the generic case examples were replaced with three autistic people describing their own experience. Sleep restriction, stimulus control and cognitive restructuring survived intact.

What the trial actually compared

The 163 participants came through the Netherlands Autism Register, all scored at least 10 on the Insomnia Severity Index, and averaged 48.5 years old. Roughly 60% were women, more than half reported another condition alongside autism, and the average reported duration of insomnia was close to 19 years. More than a third had used sleep medication in the previous two weeks.

The comparison group was not a do-nothing group. It received online psychoeducation and sleep hygiene advice, also adapted for autism, without exercises, videos or coach contact. Both arms were followed for six weeks, and the treated arm was re-assessed at six months.

On the primary outcome, mean insomnia severity fell from 15.54 to 10.05 in the treated group, against 16.21 to 14.95 in controls. As a group-by-time interaction that is −3.97 points (95% CI −5.52 to −2.42), a baseline-adjusted effect size of d = 1.02 (95% CI 0.65 to 1.40). That is a large effect on the authors’ own scale, and close to the d = 1.05 an umbrella review cited in the paper reports for guided online CBT-I in the general population.

The categorical results say it more plainly. Clinically meaningful improvement, at least a six-point drop, occurred in 37.9% of the treated group against 11.1% of controls (odds ratio 7.02, 95% CI 2.93 to 16.81). Remission, a score of 9 or below, was reached by 35.4% versus 11.1% (odds ratio 4.54, 95% CI 1.94 to 10.66). Nobody in either arm reliably got worse. Pre-sleep arousal — the racing mind and physical tension at bedtime — fell from 42.5 to 32.4, against 44.3 to 40.9 in controls (d = 0.73), and the insomnia effect was already visible at three weeks (d = 0.49).

Where the effect stopped

Three of the five secondary outcomes beyond sleep did not move. Anxiety drifted in the right direction but not reliably (d = 0.23, 95% CI −0.12 to 0.59). Depression likewise (d = 0.29, 95% CI −0.06 to 0.65). Work and social adjustment, which measures how much insomnia interferes with daily life, was essentially unchanged (d = 0.17, 95% CI −0.18 to 0.53). Quality of life did reach significance (d = 0.39), but mostly because the control group’s scores fell over the six weeks rather than the treated group’s rising — a distinction the authors make themselves.

That runs against the usual case for CBT-I, which is that fixing sleep improves mood as a bonus. The authors offer two explanations and pick neither: either the spillover genuinely does not happen in autistic adults, or the trial was too small to detect modest effects. A per-protocol look at the 46 participants who completed the four core modules did find a medium effect on depression (d = 0.52), but that is a selected subset and is presented as such.

The engagement problem

Half of the treated group completed all five modules. Just over 81% started at least one, 18.3% never began, and post-test data were missing for 29.3%. The paper cites 69% full adherence and 18% dropout for guided online CBT-I in patients with co-occurring conditions — better figures than these.

The authors do not soften it. They call feasibility lower in this population and name the friction: a platform and sleep diary some found hard to use, the burden of the assignments, sleep restriction experienced as particularly difficult. Some participants applied the exercises very rigidly, which the authors suggest may have added stress instead of removing it. Those who finished took about 41 days rather than the intended 35.

Two findings cut the other way. Effects held at six months within the treated group — controls had been offered the programme by then, so there is no comparison at that point — with only pre-sleep arousal creeping back up. And 70.7% of respondents preferred the online format to face-to-face treatment, citing flexibility and pace. The most-used components were the behavioural ones; cognitive restructuring, the “C” of CBT-I, was the least used. Benefit was stronger in younger participants and in those reporting a stronger alliance with their coach.

What the trial could not answer

The control condition was a waiting list, and the authors say directly that waitlist designs can inflate effect sizes; some of the benefit may come from guidance and structured engagement rather than CBT-I’s specific ingredients. There was no in-person CBT-I arm, so relative effectiveness is unknown.

All outcomes were self-reported, with no actigraphy or other objective sleep measure, and the authors note that the insomnia, functioning and quality-of-life scales have not been validated specifically in autistic people. Other sleep disorders were not assessed, so undiagnosed apnoea may have been present. The sample was mainly female, highly educated, Dutch, and drawn from an existing research register — far from the full autistic population, including people with intellectual disability or high psychological distress, for whom the authors say the programme may be less suitable. Suicidality, despite its links to insomnia, was not measured.

What this means outside the trial

The components tested here are the ones general guidance already describes: a regular schedule, stimulus control, relaxation work and cognitive techniques. MedlinePlus notes that cognitive behavioural therapy can relieve the anxiety linked to chronic insomnia, and NHLBI lists CBT-I alongside healthy sleep habits and medicines. None of that is an argument for self-treating a decades-long sleep problem with a downloaded worksheet.

Two details from the trial are worth keeping. Sleep hygiene advice alone was the control condition, and it produced a modest improvement at best — matching the paper’s own note that sleep hygiene has relatively weak empirical support. And sleep medication use was similar in both arms before and after, so the result was not simply a proxy for people who stopped their tablets.

Chronic insomnia means difficulty on three or more nights a week for more than three months, and it raises the risk of high blood pressure, coronary heart disease, diabetes and cancer, according to NHLBI. MedlinePlus points out that most chronic cases are secondary — a symptom of another condition, a medicine, or another sleep disorder. That is why persistent sleep problems belong with a clinician. CDC puts the daily need at seven or more hours for adults aged 18 to 60, and its guidance is blunt about the destination: talk to your healthcare provider if you have problems sleeping.

What we did not find

No trial comparing guided online CBT-I with in-person CBT-I in autistic adults. No objective sleep data. No cost-effectiveness analysis, which the authors call a prerequisite before blended-care use. No test of whether better sleep reduces suicidal ideation in this population. And no evidence that treating insomnia alone lifts anxiety or depression in autistic adults — the trial looked and did not find it.

This article describes research findings, not medical advice. If you are considering treatment for insomnia, or you are taking sleep medication and thinking about changing it, talk to a healthcare professional. Our medical disclaimer explains the limits of what we publish.

For more on the sleep evidence base, our earlier reporting covers why sleep regularity got its own research agenda and the evidence behind permanent time policy. The archive piece on habits to avoid before bed covers the sleep-hygiene basics this trial deliberately went beyond.