Service audit: Does adding the Autism Quotient as a screening measure increase cost-effectiveness of a local adult autism diagnostic clinic?

Background and objectives / introduction Diagnosing autism spectrum disorders (ASD) in adults is time-consuming and referral sources often provide limited prior information. We added the Autism Quotient (AQ) to assess its... [ view full abstract ]

Background and objectives / introduction

Diagnosing autism spectrum disorders (ASD) in adults is time-consuming and referral sources often provide limited prior information. We added the Autism Quotient (AQ) to assess its usefulness for screening out inappropriate referrals.

Methodology

This was a naturally occurring ABAB design where the Autism Quotient (AQ) was introduced as part of a referral screening protocol during two separate time-intervals (n=53/149). All new referrals scoring below 28 (n=19) were subjected to additional scrutiny and 2 were closed prior to assessment.

Results

In total 149 patients (110/39 male/female, mean age=34.6, SD=10.9) were referred to a community-based inner-city assessment clinic between June 2011 and December 2014. Four patients opted out before the assessment, and 3 assessments were inconclusive.

A ROC-curve analysis of the AQ showed fair predictability (AUC=0.716, 95%CI=.602-.829) with limited specificity (69.2%) and sensitivity (51.9%) using the recommended cut off score of 32 (Baron-Cohen, et al., 2001). We recommend using a cut off score of 25 (sensitivity=95.4%, specificity=47%) and combine the AQ score with additional patient information for patients scoring between 25 and 36 (specificity=77.8%). Overall, using the AQ together with clinical information nearly tripled the accuracy of patients receiving a diagnosis of ASD in the clinic (OR=2.86, 95%CI=1.09-7.56).

Conclusions/Perspectives
Overall, the AQ provides some limited value for screening out clearly unsuitable referrals but has poor specificity in a clinical population. Combining the AQ score with additional clinical information is essential for scores between 26 and 36. Improving the proportion of suitable referrals will benefit patients by decreasing waiting times for diagnostic assessments.

MJ, GL and SS are NHS employees and no external funding was used. VG worked in an honorary capacity on a voluntary basis. We have adhered to professional standards and codes of conduct for the research. The audit and the data... [ view full abstract ]

MJ, GL and SS are NHS employees and no external funding was used. VG worked in an honorary capacity on a voluntary basis.

We have adhered to professional standards and codes of conduct for the research. The audit and the data analysis has been peer reviewed by independent colleagues. The project was registered as an audit with the employing trust and patient information was not used outside of the normal clinical practice, hence additional informed consent was not sought. Data was anonymised prior to analysis to protect the confidentiality of the patients.