"A new form of therapy has for the first time been shown to improve the symptoms and behaviour of autistic children," The Guardian reports.
A new trial looked at the impact of early intervention in children with severe autism. This programme of treatment aimed to mainly focus on the parents, who were trained to pick up on communication cues from their child, which are usually much more subtle than in other children.
For example, a small shift in head movement could be a sign that the child wanted to communicate.
The hope is that once parents receive sufficient training they can then provide "around the clock" therapy to their child rather than one-off sessions provided by external therapy.
The programme, Parent-mediated social communication therapy for young children with autism (known as PACT), showed early promise. The children of parents who took the course showed improvements in symptoms such as communication and repetitive behaviour after one year. This study tested the children again, five to six years after the end of treatment, to see if the effects had lasted.
The children in the PACT group had on average lower symptoms scores for autism when compared to those who had normal care. But the difference was small enough that it could have been down to chance (it wasn't statistically significant). That doesn't mean the treatment didn't work, but arguably suggests the PACT programme should be now tested in larger groups of families affected by autism.
The study was carried out by researchers from Kings College London, Newcastle University, the University of Manchester and Guys and St Thomas University NHS Trust and was funded by the Medical Research Council, Department of Health and National Institute of Health Research.
Most of the UK media reports were enthusiastic. The Daily Telegraph called it "the first successful treatment for autism" while The Guardian reported a "potential breakthrough".
But at the risk of sounding like a kill-joy, none of the media sources mentioned that the main finding of the study was not statistically significant.
Many made the effort to include some useful feedback and commentary from independent experts. For example, Dr James Cusack, the director of science at the charity Autistica, was quoted widely, saying: "Too often, parents fall victim to the false claims of charlatans who prey on desperate families.
"These results look promising for the many thousands of parents who want to find early interventions for their children based on solid science,"
This was a long-term follow up of a randomised controlled trial. These types of studies are usually good ways to assess the effects of treatments.
Researchers randomly divided parents of pre-school children with autism into two groups.
One group of 75 had normal care, while in the other group, 77 parents were coached in how to communicate better with their children, using videos of play sessions to spot opportunities for communication.
The programme, called PACT, lasted for a year.
Five to six years later, researchers contacted the families again and asked them to have follow-up tests of autism symptoms.
They compared the results from the group who'd had normal care with those who'd had PACT.
In the PACT training, parents were coached to recognise what might be very subtle clues that their children wanted to engage with them and then respond appropriately, in a way that was intended to help children learn social interaction and language.
They had 12 two-hour coaching sessions over six months, then a further six support sessions over six months.
Unlike many treatments for autism, therapists worked with the parents rather than directly with the children. The aim was to produce long-lasting improvements, by changing the children's home environment.
Children were aged between two and four years 11 months when they started in the study. The average age at follow-up was 10 and a half.
Follow up assessments for the main results were done by researchers who didn't know which treatment group the children had been in.
The researchers also asked parents about their children's symptoms and behaviour.
They analysed the data in different ways, but the main result was a change in autism symptom score severity.
They looked at children's scores at the start of the study, after the 12 month treatment period, and at follow up.
The original study showed that children in the PACT group had a bigger improvement in symptom scores after treatment, compared to those who'd had usual care.
At follow-up, both groups of children had worse scores than immediately after the study.
There was still a difference between the groups, but it was no longer statistically significant.
That means we cannot be sure that, five years on from treatment, PACT improved average symptom scores more than normal care.
Average scores (1 to 10, higher being more severe) were:
Fewer children in the PACT group had severe symptom scores at follow up (46%) than those who'd had usual care (63%).
The difference between the groups was too small to be sure this was not simply a chance finding (group difference 17.2%, 95% confidence interval [CI] -2.9 to 37.3).
However, looking at the combined mean symptoms scores from immediately after the study and at follow-up compared to baseline (before treatment), the results show a statistically-significant moderate effect in favour of PACT treatment (effect size 0.55, 95% CI 0.14 to 0.91).
The researchers say their results "are encouraging and provide evidence that sustained changes in autism symptoms can be possible after early intervention," in a way that has not been shown before.
They go on to say that "on the basis of these results, we are now able to support the use of the PACT intervention for reducing symptoms of autism in young children".
They caution that the treatment has not yet been tested in older children, or in children with autism spectrum disorder, rather than "core" autism.
This study seems to provide some much-needed good news for parents of children with autism, and has been welcomed by experts and campaigners. However, the lack of statistical significance of some of the results mean we can't be sure the findings are reliable.
Statistical significance is a way of including a margin of error in calculations and allowing for chance. So the "true result" for PACT could be between 6.3 and 8.3, and the true results for usual care could be between 6.9 and 9.6. As these results overlap, we can't be sure that PACT treatment led to better scores.
One expert said that the main outcome measure of the autism symptom scale is "insensitive to change" meaning that it may not be the best way to show improvements. Another said the effects of PACT in the study were "not dramatic" and "very variable" across the group of children.
However, most seem to think that the results are promising, especially for an intervention that does not require the intensive time and commitment of some other autism treatments.
The National Institute for Health and Care Excellence says "social communication interventions" should be considered for children with autism, although it doesn't mention PACT specifically.
Hopefully further trials of PACT in larger groups of parents will point to a significant improvement in autism symptoms.