Minimally Verbal Children with ASD and Cognitive Impairment: Do Definitions Matter?
Guidelines for defining “minimally verbal” (MV) have been proposed (e.g., Kasari et al., 2013); however, to date, studies have used a variety of methods to classify samples. It is unknown how different parent-report or direct assessments used to classify MV samples compare. Some instruments may yield a very narrowly defined group of MV children, which may limit variability and make it difficult to identify dimensions that affect later language acquisition or other outcomes. This has implications for answering basic questions, such as the extent to which cognitive and language skills overlap in children with nonverbal cognitive impairment (i.e., NVIQ<70). A better understanding of methods used to define MV children is important to advancing understanding of this understudied group.
To examine how using different instruments to define MV cohorts affects sample characteristics (demographics; ASD symptom, cognitive and adaptive profiles) and the overlap between cognitive and verbal abilities in children with NVIQ<70.
Children with ASD between 6 and 18 years old (N=2044) were drawn from the Simons Simplex Collection. All children had the ADI-R, Vineland, ADOS, cognitive assessment and parent questionnaires. Children were divided using the following MV definitions: ADOS Module 1 (No words/Some words), ADI-R language level (<5 words, 5+ words/no flexible phrases), Vineland Expressive age equivalent (<18 months), Social Communication Questionnaire (item 1 indicating no phrases) and parent estimate of vocabulary (none, 1-5, 5-25 words). Within children with NVIQ <70, MV children (ADOS Module 1, some words vs. no words) and verbal children (ADOS Module 2-4) were compared.
328 children (16% of the sample) were classified as MV. Of the 328, 24% were MV on two instruments and 38% on three or more. Most children (93%) were administered an ADOS Module 1 and 61% were reportedly not using daily, flexible phrases (ADI-R). Defining MV on the basis of the ADOS yielded the largest group with the most discrepancies across measures (e.g., 62% MV on ADI-R). Across MV definitions, 44-60% of children had VIQ<NVIQ profiles (i.e., VIQ 1+ SD below NVIQ), with as many as 15% of children with NVIQ>70.
Within the NVIQ<70 group (n=507), MV children were younger (F(2,504)=12.63, p<.001) and had lower VIQ (F(2,504)=222.97, p<.001) and NVIQ (F(2,504)=122.95, p<.001) than verbal children. A higher proportion of MV children (45%) had VIQ<NVIQ profiles than verbal children (20%; X2=78.82, p<.001). MV children had less severe current social-communication impairments on the ADOS (F(2,504)=24.38, p<.001), but more past symptoms on the ADI-R (F(2,504)=34.59, p<.001).
While no single approach is obviously “best,” these results demonstrate how using different instruments to define MV samples affects sample composition. Results will inform comparisons across existing samples and design of studies ascertaining new samples of MV children. Within children with NVIQ<70, verbal and MV children differed on several characteristics. Nearly half of MV children exhibited VIQ<NVIQ profiles, suggesting a distinct subgroup of MV children whose language impairments are not attributable to general cognitive impairment. This highlights the need for future research to understand the unique strengths and challenges of MV children with ASD.