Recommendations for research
The guideline committee has made the following recommendations for research
Does training professionals to recognise features suggesting possible autism lead to earlier assessment of needs and earlier diagnosis (and by implication reduce morbidity/improve health outcomes) among children and young people compared with no training?
Successful training of healthcare professionals in the Netherlands has been shown to improve their ability, confidence and skills in identifying children or young people who need an autism diagnostic assessment. A fully trained workforce can identify the number of autistic children or young people and provide accurate information both for planning individual care and at a strategic level for planning appropriate service provision.
If training improves earlier recognition and referral, this could be of particular benefit to at-risk groups for which there is evidence that autism is currently under-diagnosed, such as girls, and children and young people:
with parents of lower educational attainment
with English as an additional language
with sensory impairments
with a learning (intellectual) disability.
Before extending training to a wider population, it is important to better understand its effectiveness in terms of age, number of children and young people at referral, and time between parents' concerns and autism diagnosis. 
Does routine additional information from educational settings (such as nursery or school) improve accuracy in diagnosing autism among children or young people up to the age of 19 compared with signs and symptoms alone?
Autism is primarily characterised by difficulties in social reciprocity, social communication and social understanding, along with rigid and repetitive ways of thinking and behaving. Diagnostic accuracy may be improved by interpreting information about how the child or young person presents in social settings away from the home and immediate family.
Nurseries or schools are the most obvious settings from which such information may be collected. However, the degree to which information from teachers and schools helps in accurate diagnosis has not been well tested. 
Do additional assessments (for IQ, language ability and motor ability) improve accuracy in diagnosing autism among preschool children (younger than 5 years) compared with signs and symptoms alone?
Current NHS practice varies widely with regard to the proportion of children having an autism diagnostic assessment who also routinely undergo assessments of IQ, language and motor abilities.
As a consequence we do not know whether such assessments aid more accurate diagnosis of autism. This is particularly important if a differential or coexisting diagnostic decision is called for and/or if there may be specific management implications.
Studies may prove valuable to parents in terms of explaining some of the child's behaviours, leading to more targeted and informed support for the child, parents and the wider family. 
What is the effectiveness and acceptability of comparative genomic hybridisation (CGH) array compared with current genetic testing in autistic children and young people?
Recent scientific advances have led to the detection of genetic abnormalities that may partly or wholly explain why a child or young person has autism. As the tests become increasingly sophisticated (for example, using methods such as CGH array that detect more subtle variations), more genetic abnormalities are being identified, although their causal role in autism is not always clear. Improved detection of genetic causes of autism could increase the precision of genetic counselling for parents of an autistic child or young person and also for the wider family. At present, the yield of abnormal genetic results using CGH array is known to be higher in those with dysmorphic features and/or a learning (intellectual) disability, but this may extend to the wider autism population with increasing test sophistication. Before extending CGH array testing to a wider population, it is important to have a better understanding of its diagnostic yield. It is also essential to identify any negative consequences that may result from routine testing.