Most digital mental health trials compare their app against a waitlist, which is to say against nothing, and then report a large effect. A trial published in The International journal of eating disorders in 2026 did not do that, and the difference is worth dwelling on because it is what makes the result believable. [pubmed-eating-disorders-2026-source]
What the researchers did
One hundred and thirty people who had been discharged from intensive eating disorder treatment within the previous four months were randomly assigned to one of two self-guided digital conditions:
- Five sessions of written imaginal exposure, in which participants wrote repeatedly and in detail about feared outcomes such as rapid weight gain
- Online journaling, as an active control
The study was pre-registered, with fear of weight gain named in advance as both the primary outcome and the mechanism the treatment was supposed to work through.
What they found
Imaginal exposure beat journaling on fear of weight gain, the primary outcome.
It also beat it on overall eating disorder symptoms, fear of food, food avoidance, clinical impairment, and post-traumatic stress symptoms.
Why the control group is the story
Compare this design with the typical digital trial.
A waitlist control receives nothing: no activity, no attention, no expectation of improvement. Beating that tells you the package did something, and leaves entirely open which part.
Online journaling, by contrast, shares nearly everything with the intervention. Both groups wrote regularly. Both spent time on their eating disorder. Both were doing something structured with the expectation it would help. The single systematic difference is that one group was deliberately approaching feared content and the other was not.
When a trial with that structure finds a difference, the difference is much more plausibly the active ingredient. The researchers designed it to isolate exposure specifically, and the result supports the mechanism rather than just the product.
The other thing it gets right
It targets the gap. Discharge from intensive eating disorder treatment removes structure, supervision and daily contact simultaneously, usually before outpatient care has started. Relapse risk is high in exactly that window, and it is a window in which very little is currently offered.
A five-session self-guided digital programme is unglamorous and it fits there. It does not compete with specialist treatment; it occupies a period when specialist treatment has stopped.
What it does not settle
The trial reports which measures moved and, in the abstract, not how much. Whether the improvements are large enough to change someone’s clinical course is a separate question from whether they are statistically detectable.
Nor does it tell you about durability. Improvement over a five-session programme in the months after discharge is meaningful; whether it reduces relapse over a year is the question that matters most for this population, and it is not one this design answers.
And 130 participants recently discharged from intensive care is a specific group. Whether the same approach helps someone earlier in the illness, or someone who has never had intensive treatment, is untested here.
Is the gap after discharge actually covered?
For anyone leaving, or who has recently left, intensive treatment. This is about the structure around you, not about symptoms.
0 of 6 ticked
This is what a well-managed discharge looks like. If any single item is the weak one, it is usually the written plan for hard situations, which is worth putting on paper while things are stable rather than during a difficult week.
Look at which items you could not tick and take exactly those back to the discharging team. A missing contact route or an unbooked appointment are concrete requests, and they are much more likely to be actioned than a general worry about coping.
What you have described is the drop in support this research is about. Ask specifically for a named contact, a booked follow-up date, and arrangements for physical monitoring. You are entitled to all three, and this is the period in which they matter most.
The general point
This is a good template for reading any digital mental health study. The two questions that separate a strong trial from a weak one are almost always the same:
What was the comparison? If it was nothing, the effect size tells you about receiving an intervention in general, not about that intervention.
Was the primary outcome declared in advance? If not, the reported headline may be whichever of many measures happened to move.
This trial answers both well, which is rarer than it should be.
If you are recovering from an eating disorder
The period after intensive treatment is genuinely hard and the drop in support is a known problem rather than something you are experiencing wrongly. If there is a gap before outpatient care starts, it is worth asking the discharging team what they suggest for it specifically, including whether structured exposure work is appropriate for you.
Do not improvise exposure work alone. Eating disorders carry physical risk that needs monitoring, and unstructured writing about feared outcomes can turn into rumination, which makes things worse.
The source
These findings are drawn from “A Randomized Controlled Trial of Self-Guided Digital Written Imaginal Exposure Therapy for Eating Disorders in Patients Recently Discharged From Higher Level Specialty Eating Disorder Care” (Levinson CA, Penwell T, Sandoval-Araujo LE, et al., 2026), published in The International journal of eating disorders. Read the full study on PubMed.