Adding ESCS Measures of Initiating and Responding to Joint Attention to the M-CHAT

Background: The American Academy of Pediatrics recommends universal autism-specific screening. Differentiating autism spectrum disorders (ASD) from language or global developmental delay (DD) is challenging due to symptom overlap in young children. The Modified-Checklist for Autism in Toddlers (M-CHAT; Robins et al., 1999) identifies children at risk for ASD, but has poor to moderate positive predictive value, indicating that some cases who screen positive on the M-CHAT have a non-ASD DD. Studies show that children with ASD show more impairment in initiation of joint attention (IJA) and in response to joint attention bids from others (RJA) than children with DD.

Objectives: The current study examines whether the combination of M-CHAT and Early Social Communication Scales (ESCS; Mundy et al., 2003) IJA and RJA scores differentiate ASD from DD in a sample of toddlers who screened positive on the M-CHAT better than M-CHAT alone.

Methods: Thirty-one children who screened positive on the M-CHAT during a well-child checkup were recruited from a larger study. Of these children, 11 (7 males; mean age=23.67 months, SD=4.40) were subsequently diagnosed with DD and 20 (20 males; mean age=24.83 months, SD=3.37) were diagnosed with an ASD. Mean age was not significantly different between groups, t(29)=-.823, p=.417. Those who screened positive on the M-CHAT and subsequent follow-up interview completed the ESCS and a diagnostic evaluation, which included the Mullen Scales of Early Learning (MSEL; Mullen, 1995) to measure cognitive development. Analyses using M-CHAT critical score were similar to results using M-CHAT total score, so only total score analyses are reported.

Results: M-CHAT scores were not significantly different between the ASD group (M=7.90, SD=2.75) and the DD group (M=6.27, SD=4.10), t(29)=-1.32, p=.197. However, the ASD group demonstrated less IJA (M=6.45, SD=8.80) than the DD group (M=18.36, SD=11.11), t(29)=3.29, p=.003, and similarly less RJA (M=12.30, SD=19.44) than the DD group (M=57.05, SD=23.10), t(29)=5.74, p=.000. Binary logistic regression analyses were conducted to classify participants as ASD or DD. M-CHAT alone correctly classified 71% of the participants.  The addition of ESCS IJA and higher level RJA scores classified 87.1% of the participants, a significant improvement over M-CHAT alone, χ²(3)=19.95, p=.000. Neither IJA nor M-CHAT total scores were significant independent predictors of diagnostic group membership, though higher level RJA score was significant, Exp(B)=.916, SE=.04, p=.02. When level of cognitive development was controlled using the Mullen early learning composite standard score, 90.3% of the sample was correctly classified, χ²(4)=20.51, p=.000; neither M-CHAT nor ESCS scores were significant independent predictors once cognitive level was included.

Conclusions: Adding ESCS JA measures to the successfully improved differentiation between ASD and DD. Given that ESCS JA measures help differentiate ASD DD, a second level screen after a positive M-CHAT may be useful to determine which cases should be referred to ASD specialists. A combination of M-CHAT and ESCS may reduce over-referrals to ASD specialists, while still maintaining very successful early detection of autism spectrum disorders.