Screening Profile on M-CHAT(-R/F) for Toddlers Who Screened Positive for ASD and Received “No Diagnosis”

Background: Previous findings indicate that children who fail an autism screener, the Modified Checklist for Autism in Toddlers with Follow-Up (M-CHAT/F) and its revision (M-CHAT-R/F) are highly likely to have significant developmental concerns. 97.7% (Chlebowski, Robins, Barton, & Fein, 2013) and 94.6% (Robins et al., 2014), respectively of screen-positive cases were diagnosed with ASD or other developmental delay or concern following comprehensive evaluations. Research has investigated which items of the M-CHAT(-R/F) best discriminate toddlers with and without ASD (Kamio et al., 2015; Robins et al., 2010). However, we have yet to specifically characterize a group of “No Diagnosis” children, or those who screen positive for risk for autism, but present with sub-clinical developmental concerns at evaluation.

Objectives: To compare items failed on the M-CHAT(-R/F) of children with “No Diagnosis” (ND) with typically developing children (TD; i.e., children who screened positive on the screener, but were within normal limits on all areas of development) and children diagnosed with ASD, global developmental delay (DD), and developmental language delay (DLD).

Methods: 169 toddlers (mean age at evaluation = 23.7 months, range: 16-41 months) who screened positive on the M-CHAT(-R/F) completed a diagnostic and developmental evaluation. 28 TD children, 39 children with ASD, 33 with DD, and 27 with DLD were matched on sex, age, and race/ethnicity to 42 children with ND. Total items failed on the M-CHAT(-R/F) was regressed onto diagnostic group in separate comparisons between ND and each other group. For each M-CHAT(-R/F) item, Fisher’s Exact Tests compared the proportion of children in the ND group who failed the item to that in each other group. Data from M-CHAT items were combined with data from its corresponding M-CHAT-R/F item, with data from three items from the M-CHAT (i.e., peek-a-boo, playing with toys, and wandering without purpose) excluded. Due to the exploratory nature of these analyses, alpha levels were not corrected.

Results: Groups did not differ significantly on level of maternal education. The ND group had significantly fewer average total M-CHAT(-R/F) items failed compared to the ASD group (p=.001), but did not differ from the TD (p=.076), DD (p=.144), or DLD (p=.958) groups. Results from item-level analyses are presented in Table 1, and graphically represented in Figure 1 (attached).

Conclusions: At the first stage of autism screening, toddlers with sub-clinical developmental delays are distinguishable from toddlers with ASD but not other diagnostic groups, highlighting the role of the M-CHAT(-R/F) as an autism-specific screener. Items about social communication (e.g., response to name and language, response to and initiation of pointing) and play (e.g., pretend play, imitation) best distinguish ASD from ND, relative to other items (e.g., unusual motor activity). A subset of the same items also distinguish ND from DD. At screening, children with subclinical presentations are indistinguishable from children who have a DLD. Furthermore, the ND group is highly similar to the TD group at screening, other than being more likely to have poor language comprehension. These findings underline the importance of conducting further diagnostic evaluation for children who fail screening.