Motivation
Autism spectrum disorder is a condition that currently is estimated to effect 1-2% of the population (Maenner et al., 2020). Due to the complexity of symptom presentations, identifying autism is not straightforward and the condition may go undiagnosed for many years. However, identification is important as it allows parents to access supports (and, critically, funding) for their children, which are generally locked behind receiving a diagnosis.
Consequently, the sooner that potential signs of autism are identified, the earlier the diagnosis process can begin, and the sooner that support for children who need it can be provided.
One factor that accounts for delayed (or missed) diagnoses, is that there are currently no free, scientifically backed screeners for children that assess autistic traits as described by current diagnostic criteria. Existing screeners are either costly, and/or based on older diagnostic criteria that does not reflect the current understanding of autism. For example, sensory sensitivity was only added to the diagnostic criteria relatively recently (American Psychiatric Association, 2013), and older screening questionnaires lack this aspect (e.g. Autism Spectrum Quotient: Children’s Version; (Auyeung, Baron-Cohen, Wheelwright, & Allison, 2008).
Our goal
The proposed project seeks to overcome this issue by developing a new screening tool that reflects current diagnostic criteria for autism in children, is accurate, and is freely available to anyone to use . This tool could be used by clinicians and parents alike to help identify traits and behaviours in children that may be associated with autism, and set in motion the path to potential diagnosis as early as possible. The new tool could also be used more broadly by researchers interested in measuring autistic traits in children.
Public interest
We believe there is large public interest in the development and distribution of free, accurate tools like the one we are proposing based on the reception of a similar tool we developed for adults, the Comprehensive Autistic Trait Inventory (English et al., 2021). The publication associated with that measure is in Altmetrics top 5% of research publications , and the website dedicated to it received over 5000 hits in the month after it was released.
Methodology
Study 1: Developing the screener
Approximately 80 pilot items have been developed that span the seven main diagnostic criteria for autism spectrum disorder as defined in the DSM-5. To develop the screener, we will need to administer these pilot items to a large number of parents of children aged 4-12 in the general population, where parents of autistic children reflect current population estimates. These participants will be recruited through Prolific Academic using the appropriate eligibility filters.
When the data has been collected, exploratory factor analysis will be used to identify the different autistic trait dimensions. Ideally, the factor structure will mimic the current diagnostic criteria.
Study 2: Testing the screener
The next step would be to develop useful cut-offs or indicators for at-risk children. To achieve this, we will need to administer the screener to a large number of parents of autistic and non-autistic children (aged 4-12). To recruit enough parents of autistic children, an additional ‘screener’ study may need to be administered to identify such parents, and then invite them into the main study (this will be funded independently).
Logistic regression will be conducted to help identify useful thresholds for identifying ‘at risk’ children by using the previously identified trait dimensions to predict diagnosis (or lack thereof) of autism. Depending on the outcomes, risk ‘bands’ may be established rather than a single threshold that is met/not met. Developing similar thresholds for each autistic trait dimension will also help enable greater predictive accuracy, as clinical diagnoses also require multiple criteria to be met to be given a diagnosis.
Sample size
Our target sample for Study 1 would be 1200 participants. Our previous work in developing an autistic trait measure for adults (English et al., 2021) has indicated that this is an appropriate amount to conduct a meaningful factor analysis of the data, and that the factor structure only begins to settle around 700-800 participants. Study 2 would require 400 participants, with 200 in each group (parents with vs without autistic children). If 200 parents of autistic children cannot be recruited via Prolific, we have access to a University-affiliated autism clinic where participant numbers can be supplemented.
Costs
It is estimated that in both studies that it will take about 20 minutes to complete the questionnaires administered. As a buffer, we will set the estimated completion time on Prolific to be 30 minutes. As the study is not too demanding of participants, and to maximise use of funding, we will pay participants at the recommended minimum rate (£5.00/hour), meaning that each participant will receive £2.50.
| Sample size | Participant payments @ £2.50 / participant | Service fee @ 33% | Total Cost | |
|---|---|---|---|---|
| Study 1 | 1200 | £3000 | £990 | £3990 |
| Study 2 | 400 | £1000 | £333 | £1333 |
| Total | 1600 | £4000 | £1323 | £5323 |
Based on the required sample, payments to participants, and Prolific fees, we are seeking £5323 to complete both studies .
Preregistration
We are currently in the process of pre-registering this study, and the link will be added here soon!
Open Science
We will commit to the analyses, data, and materials to being made publicly available following the completion of the study. Participants will be informed at the beginning of the study of the intention to make their anonymised data available at the conclusion of the study, and we will explain the reasoning for data sharing in order to maximise ‘consent to share’. The study materials will be hosted in several places (e.g. Open Science Foundation, University Research Repository) to ensure that they are accessible for years to come.