A Training Model for the Diagnosis of Autism in Community Pediatric Practice

Background:

Early screening and diagnosis of autism spectrum disorders (ASD) currently represents a critical public health and clinical practice issue.  Historically, waits for diagnostic services are quite lengthy and hinder the start of early intervention services thought to be crucial for optimizing functional developmental outcomes for children and their families. 

Objectives:

In this study we developed and evaluated a training program designed to help pediatricians identify and diagnose young children with ASD in the context of traditional community practice settings within a very time-limited framework (1 hour). 

Methods:

Five community pediatricians participated in an intensive training (i.e., use of MCHAT, Screening Tool for Autism in Two-year-olds (STAT), and focal diagnostic interviewing), conducted specialized 1-hour ASD diagnostic evaluations within their own practices, and then referred a consecutive series of children to a traditional medical center diagnostic clinic for an independent assessment of ASD. 

Results:

Of the five community physicians attending the START-ED training, four referred patients for subsequent independent evaluation.  21 or 25 referrals completed the independent evaluation process (child age M = 30.48, SD = 3.74).  Of the four families who declined invitation to participate in the evaluation, only one had received an ASD-risk classification from their referring pediatrician.   Of children seen for independent evaluation, 19 (90%) received diagnoses of ASD-risk from their pediatrician.  An ASD diagnosis was confirmed based on independent evaluation in 14 out of these 19 cases (74%). Of the 2 children (10%) referred without an ASD-risk classification, one child received a diagnosis of ASD subsequent to independent evaluation and one did not (50% agreement).  Overall independent diagnostic evaluation was in agreement with initial pediatrician classification in 15 out of 21 cases (71%).  Agreement varied greatly between the four referring pediatricians: 1/1 = 100%; 6/7 = 86%; 4/6 = 67%; 4/7 = 57%.  

In all cases (n=6) where there were diagnostic disagreement clinically significant developmental concerns were confirmed during the independent evaluation (i.e., global developmental delays or speech/language delays).  Clinical diagnostic certainty ratings from the independent evaluation process were significantly lower for children not receiving an ASD diagnosis (ASD mean = 4.27; Non-ASD mean = 2.41; t = 3.72, p < .05). 

Conclusions:

The development of training methods for the classification of ASD within traditional community based pediatric practice holds promise.  Specifically, introduction of basic interactive screening tools into a pediatric consultative model may be able to successfully classify young children in a timely fashion to appropriate categories of risk.  Such models must take into account the reality that significant revision and condensation of gold-standard assessment methodologies will undoubtedly contribute to more errors in definitive classification.  However, if we are able to train community providers to identify risk and initiate intervention services based on such assessed risk status, long waits for more definitive diagnostic assessment services will not hold deleterious effects for the child, nor will clinicians be asked to provide definitive diagnoses within a timeframe that it is unrealistic to expect clear-cut meaningful classifications.