A Validation Study of the Observation for Autism Screening (OERA), a New and Brief Low-Cost Instrument to Screening for ASD

Background:

The ADOS is the gold standard observational schedule for diagnosis of ASD. However, it is lengthy, time-consuming, costly and requires a high level of training, which makes it rarely feasible in non-research settings and large-scale proposals. This is particularly problematic in most Low-Middle-Income-Countries/LMIC. For that reason, our team developed a new tool called Structured Observation for Autism Screening/OERA.

Objectives:

(1) to describe the OERA instrument, (2) to assess its construct validity and invariance regarding gender and IQ, and (3) to calculate its sensitivity and specificity.

Methods:  Local: São Paulo, Brazil. Participants: 99 children: N=79 with ASD, and n=23 children without ASD (11 with intellectual disability and 22 with typical development). Instruments: ASD diagnosis based on the ADI-R; IQ based on SON-R 2½-7 or WISC-III. Statistical analysis: (1) Evaluation of the agreement between two independent examiners via kappa coefficient, percentage of agreement and expected agreement. (2) Construct validity and invariance via confirmatory factor analysis/CFA. (3) Elaboration of ROC curve and its cutoff for sensitivity and specificity confronting OERA total score with clinical judgment by psychiatrist allowing to estimate the area under the curve.

Results:

OERA is an 8-item semi-structured assessment tool to screening ASD in children 3–10 years old of age. After a short training, non-specialists can administer the OERA in approximately 10 minutes, based on a standardized low-cost set of objects and toys. After that, an expert in ASD watches the record to code the 8 semi-structured items, and a list of 5 additional behaviors, summing 13 items. The unidimensional model for OERA showed:

(1) High agreement between the observers (all items above 80%);

(2) Good fitness of indices, high factor loadings (on average, factor loading=0.878), no item showed differential functioning item regarding gender and IQ, and;

(3) Area under the ROC curve=0.9443. Scores equal of greater than five correspond to a sensitivity=92.75% and specificity=90.91%.

Conclusions:  The OERA was useful to identify ASD children with good psychometric features (construct validity and invariance regarding gender and IQ), discriminating children with ASD and typical development with good sensibility and specificity. This new instrument is promising to identify ASD, particularly in LMIC.