The Department for Education has added a new review of behavioural and social communication interventions to its evidence series for mainstream settings. Its central message is useful, but not comfortable: several approaches show promise, yet no single intervention works for every autistic pupil or every pupil with social, emotional and mental health needs.
For schools, that makes the quality of the decision and the record as important as the name of the intervention.
The DfE evidence review collection was updated on 30 July 2026 with a 93-page review focused on autism and SEMH. It brings together 73 systematic reviews. Sixty-one focused on autistic pupils and 12 on children and young people with SEMH needs.
In brief: the review finds positive but variable effects across several approaches. Impact depends heavily on matching the support to the pupil, delivering it well, training staff and checking whether any improvement lasts or transfers beyond the intervention session. Schools therefore need a clear baseline, a delivery record and a review that compares like with like.
What did the new DfE review find?
The review examined strategies that mainstream educators can use to improve behavioural and social communication outcomes.
It reports promising findings for approaches including:
- peer-mediated interventions for social engagement and communication, particularly in early years and primary settings
- structured behavioural supports such as activity schedules, Check In Check Out and Daily Behaviour Report Cards
- physical activity programmes, especially where delivery is sustained and sufficiently intensive
- naturalistic developmental behavioural interventions and play-based approaches
- LEGO-based therapy and structured social skills programmes
- video-based interventions and some technology-supported tools
That list should not become a shopping catalogue. Effects were often short term and measured through observable behaviour. Evidence about maintenance, generalisation and broader outcomes was limited. Implementation quality, dosage, staff training and individual profile all affect results.
A promising approach badly matched or inconsistently delivered is not evidence-based practice. It is an intervention name attached to an uncertain process.
Why does "no single intervention" matter in school?
Schools are under pressure to act. A pupil is distressed, finding social communication difficult or struggling to stay engaged. Staff want something practical, and families reasonably want to know what will change.
The temptation is to choose a recognised programme and treat selection as the decision. The DfE review points to a more demanding standard.
The school must be able to explain:
- which specific need or barrier the support is intended to address
- why this approach fits this pupil in this setting
- what delivery will look like in practice
- how staff will know whether it helped
- when the decision will be reviewed
Those questions turn an intervention from a hopeful activity into a testable part of SEN support.
The SEND Code of Practice places this within the graduated approach: assess need, plan with the pupil and parents, deliver the provision and review its effect.
What should the baseline record contain?
A baseline should describe the behaviour, communication difficulty or participation barrier in terms that staff can recognise again.
"Has poor social skills" is too broad. It says little about context and gives the review nothing reliable to compare.
A stronger baseline might read:
During unstructured lunchtime periods, Maya usually stays within two metres of the supervising adult and does not respond when peers invite her into a game. In a familiar paired classroom task, she responds to a peer's question in four out of five observed opportunities.
This does two useful things. It identifies where the difficulty appears and where the pupil is already more successful. That difference may shape the intervention far better than a general label.
Do not turn a pupil's distress into a behaviour score without context. Do not infer a diagnosis from a pattern. The record should help staff understand the pupil, not flatten them into a target.
How should schools match an intervention to need?
Start with the intended change, not the programme name.
If the aim is to increase peer interaction in a defined setting, a peer-mediated approach may be relevant. If the immediate barrier is difficulty moving through a classroom routine, a structured visual schedule may be a better fit. If regulation and participation improve after movement, a planned physical activity approach may be worth testing.
The DfE review does not say that any one of these will work automatically. It says the evidence is conditional.
This is also where pupil and parent views matter. An approach can produce more observable compliance while making the pupil feel less safe. A narrow measure may miss that entirely.
Why is delivery evidence essential?
A review cannot judge an intervention unless the school knows what was delivered.
Suppose a plan specifies four short sessions each week for six weeks. Staff absence, timetable changes and pupil absence reduce that to nine sessions. The review shows little change.
That result does not prove the approach failed. It shows that the planned dose was not tested.
Delivery records do not need daily essays. Capture the information needed to interpret the result:
- date and duration
- whether the session happened
- material changes from the agreed approach
- who delivered it
- any immediate response relevant to safety or access
- a brief reason when delivery was missed
This distinction protects pupils and staff. Without it, schools may continue an ineffective intervention because "it has not had long enough", or abandon a useful one that was never delivered consistently.
How should the review measure change?
Compare the same outcome with the baseline. If the starting measure concerned peer responses at lunchtime, a review based only on completion of adult-led tasks does not answer the original question.
Use more than one perspective where the outcome warrants it. Staff observation may show increased participation. The pupil may report that the setting remains exhausting. A parent may notice a rise in distress after school. Those accounts can coexist and should influence the next decision.
A useful review should state:
- what was delivered and how closely it matched the plan
- what changed against the baseline
- what did not change
- whether improvement appeared beyond the intervention session
- the pupil's and parent's view
- the decision to continue, adapt, stop or seek further advice
- the reason for that decision
The review also warns that long-term maintenance and generalisation are often poorly assessed in the research. Schools can respond by planning a later check. If progress disappears as soon as the intervention stops, that matters. If a skill appears in the session but nowhere else, that matters too.
A practical intervention record for SENCOs
Use one connected record rather than a plan in one file, attendance in another and the review in someone's inbox.
For each intervention, keep:
Decision
- current need or barrier
- baseline and sources
- intended outcome
- reason for choosing the approach
Delivery
- responsible staff
- planned frequency and duration
- training or instructions required
- actual delivery and material changes
Review
- evidence collected
- comparison with baseline
- pupil and parent views
- signs of generalisation or maintenance
- decision, owner and next review date
MeritDocs supports this kind of traceable workflow by linking review versions to an earlier baseline and keeping review context with the document history. Staff can see what the new version is building on rather than treating each review as a fresh form.
That is the practical answer to variable evidence. Software cannot decide which intervention is right for a pupil. It can make the school's reasoning, delivery and review much harder to lose.
Senior leaders should not ask only which programmes the school uses. Ask whether staff can show that support was matched, delivered and reviewed. If the answer depends on one adult's memory, the process is fragile.
MeritDocs keeps SEND documents, versions, assignments and review context in one governed workflow. It does not promise that every intervention will work. It keeps the account of what the school tried, what happened and why staff chose the next step.
Frequently asked questions
Does the DfE review recommend one best intervention for autism or SEMH?
No. It finds positive but variable evidence across several approaches and concludes that no single intervention is universally effective. Matching, implementation, staff training and outcome monitoring all affect results.
Is a behaviour change enough to show success?
Not always. Consider communication, participation, wellbeing, distress and whether any change generalises beyond the intervention context. The chosen measures should fit the pupil and the original aim.
The takeaway
The new DfE review gives schools a useful evidence base, not a universal prescription.
Its strongest operational lesson is simple: choose support against a specific need, record whether it happened, compare the result with a clear baseline and preserve the reason for the next decision. When evidence is variable, a trustworthy review trail is not extra paperwork. It is how a school learns what works for this pupil.