The Autism, Tics–ADHD and other Comorbidities inventory (A–TAC), Validity, reliabilty and the measurement of autism in males and females
We meet up with GNCs Caroline Mårland who discusses her newly published thesis on the A-TAC.
First of all, I wonder if you could tell me a bit about yourself?
I trained as a psychologist at the University of Gothenburg and started my career as a psychologist in the forensic psychiatric inpatient unit in Gothenburg. After some time, I got the opportunity to pursue PhD studies (50%) at CELAM (Centre for Ethics, Law and Mental Health) which is a centre at the Department of Neuroscience and Physiology, University of Gothenburg. I defended my PhD in May 2022 on a thesis entitled "The Autism, Tics-ADHD and other Comorbidities inventory (A-TAC), Validity, reliability and the measurement of autism in males and females". After my PhD I wanted to deepen my clinical knowledge in the field of ESSENCE, and I am now working at the Neuropsychiatric Clinic for Children and Adolescents (BNK) at Sahlgrenska University Hospital.
You have just completed your PhD. Congratulations! What was your project about?
Thank you! My PhD project consisted of 4 substudies that evaluated the A-TAC screening instrument in different ways. The first substudy focused on the validity of several different diagnostic domains of the A-TAC while the other studies had a specific focus on autism. In the second study, a short scale for autism was created, while the third and fourth studies focused on the measurement of autism in boys and girls. All studies were based on data from the CATSS, i.e., the Children and Adolescent Study in Sweden. CATSS is a large twin study based on the Swedish twin registry at Karolinska Institutet. This study has included parental assessment with A-TAC at age nine, but during the first three years of CATSS, twins who turned 12 were also included. In my PhD project, we have had access to these scorings, which currently consist of data from over 30,000 twins. By linking the participants in CATSS with the Swedish Patient Registry, we have also had access to information on registered diagnoses.
Can you tell us a bit about A-TAC as a screening tool?
Absolutely, the A-TAC was developed at the Department of Child and Adolescent Psychiatry at the University of Gothenburg, (Gillberg Neuropsychiatry Centre’s predecessor), as there was a need for a broad screening instrument that could be used in CATSS. The A-TAC was created to capture neuropsychiatric conditions such as autism, ADHD and learning difficulties but also other common diagnoses in child and adolescent psychiatry such as oppositional defiant disorder or OCD. It is important to remember that A-TAC is not a diagnostic instrument, but it can be used to get an indication of the difficulties that exist. No training is required to use the A-TC and it is freely available on the GNC website.
Why did you choose this particular topic for your doctoral studies?
My PhD project was started when my main supervisor Sebastian Lundström suggested that I could conduct a validation study of A-TAC using data from CATSS. Over time, my PhD project came to have an increasingly pronounced focus on psychometrics, and using various statistical methods, we have been able to look more closely at the strengths and limitations of A-TAC as a screening instrument.
Your first study examined the validity of A-TAC. Can you describe the study and the results?
In this study, we examined several diagnostic domains in the A-TAC by looking at whether, for example, a positive screening result for autism and ADHD could identify or predict a registered diagnosis in the Swedish patient registry. We examined whether the diagnoses were registered before or after the A-TAC interview was conducted at the time participants turned 9 or 12 years old. The results indicated that the A-TAC has particular strength as a screening instrument for autism, ADHD, learning disabilities and oppositional defiant disorder. Participants' age at diagnosis mattered; the A-TAC was able to identify a recorded diagnosis more accurately before the age of 9 or 12. Predictive validity varied more between the different domains, but the results pointed to good predictive power in areas such as learning disabilities, autism and ADHD.
Development of a brief screening for autism using item response theory was your second study. Please tell us about it!
Item response theory (IRT) is a psychometric method that can be used, among other things, to reduce the number of questions in a form. In IRT, each question is studied individually, which creates the opportunity to identify the questions that best measure the characteristic that an instrument is designed to measure. Shorter instruments can thus be created by selecting the questions with the best precision. In the second study, we used IRT to study all 17 questions in the autism domain of the A-TAC with the goal of creating a shortened version. Four questions with good precision were identified and included in the short version whose psychometric properties were similar to the results we found for the entire autism domain of 17 questions in the first study. A short version, for example, may be useful in large-scale research studies where prevalence is to be examined and when there is no need to examine the broader picture of autism.
Your third article was about the measurement of autism in girls and the importance of a gender-specific comparison. Can you summarise the study?
In the third study, we looked at the approximately 300 boys and 120 girls from CATSS who had a registered autism diagnosis in the Swedish patient registry. We used A-TAC scorings and looked at the domains of autism, ADHD, learning disabilities and oppositional defiant disorder. In the first step, we looked at differences in mean scores for each gender, i.e., we compared boys and girls. This comparison showed that boys with autism in particular scored higher on the autism and ADHD domains. In the next step, a statistical adjustment of the scales was carried out using standard scores, which in practice means that girls are compared with girls and boys with boys. Through this analysis, we could instead see that girls with an autism diagnosis scored higher on all scales compared to boys with an autism diagnosis. This suggests that girls with an autism diagnosis may have greater difficulties than boys, but this is something that only becomes apparent when girls with autism are compared to girls rather than boys.
In the fourth and final study, a method called Differential Item functioning was used to measure autism in boys and girls, tell us a little about it.
The study is based on the whole population of the CATSS and here we looked at whether one or a few items in the autism domain of the A-TAC had a better ability to capture autism in boys or girls. The statistical analysis showed that 3 questions favoured boys, such as the question about whether the child gets so caught up in their interest that it becomes tedious or too intense. We also found 3 questions that favoured girls, including the question whether the child speaks too loudly or too quietly. In the next step, we used these results to create gender-specific scales of the existing questions in the autism domain. Overall, the results indicated that the use of shorter and gender-specific scales did not improve accuracy in screening. The fact that there was an equal number of questions that favoured boys and girls suggests that the entire autism domain of the A-TAC with 17 questions can measure autism equally in boys and girls.
What are the implications of the results for practice and thus for clinical work?
Evaluating an instrument and studying its validity and reliability is a fundamental and important step in interpreting the results we obtain. The A-TAC is used both in research and in clinical practice and, as a screening instrument, has its strength in providing a broader picture of the difficulties commonly seen in child and adolescent psychiatry. The fact that girls with autism scored lower on the autism domain of the A-TAC highlights the importance of focusing not only on the number of symptoms but also on the quality, i.e., the impairment that a symptom causes in everyday life. Finally, sub-study four showed that the autism domain of the A-TAC provides an equivalent measure for boys and girls, which in practice means that the differences we see in mean scores represent meaningful variation within autism and are not the result of systematic measurement error.
What has been the highlight of your research career?
Obviously, obtaining a PhD was a major goal and a highlight of my academic training. For me, the whole PhD period has been incredibly rewarding and challenging. I have been fortunate to have both fantastic supervisors and colleagues who have supported me all the way to the finish line. In everyday life, the big highlight has been being able to collaborate with other researchers who take the time to comment on and review the various studies, which has also helped each study to develop and improve over time.
Looking ahead, what is the next research topic for you?
With each passing year of my doctoral studies, I realized more and more that I need to complement my research with working clinically with children as a psychologist, which I have now had the opportunity to do. By working clinically, I hope to gain new thoughts and ideas about research questions that are clinically relevant. I also hope to devote a few hours a week to research and be part of ongoing research projects at the Gillberg Neurospychiatry Centre.