Objectives: To determine how severity estimates vary depending on age, cognitive skills, and adaptive scores in a large sample of children diagnosed with autism
Methods: Participants included 726 participants diagnosed with ASD, ages 15 months through 17 years, drawn from a university based clinical research database. Examined measures included the Vineland Adaptive Behavior Scales – II (Adaptive Behavior Composite, Communication, and Socialization scores), Autism Diagnostic Observation Schedule Comparison Scores (CS; also known as Calibrated Severity Score; Gotham et al., 2009) and several different measures of cognitive ability, including the Mullen Scales of Early Learning and Differential Ability Scales-II. Mild/No Impairment, Moderate, and Severe Impairment groupings were created in two ways. First, standard deviation splits were created by splitting data by standard deviation cutoffs (<70, 70-85, >85) for each measure. Next, tertile splits were created by separating the sample’s scores into equivalent thirds for each measure. Grouping labels were compared for consistency across the three measures used to define level of impairment.
Results: Discrepancies emerged between all groups such that participants with Mild, Moderate, and Severe CS demonstrated varying levels of adaptive, communicative, social, and cognitive impairment. A large proportion of the Mild CS group fell into Moderate-to-Severe groups on Vineland and Cognitive variables (52%-85% based on standard deviation splits, 59-84% based on tertile splits). Many participants in the Mild/No Impairment IQ group were also defined as Moderate-to-Severe based on the CS and Vineland variables (51%-90% for standard deviation, 35-72% for tertile). Discrepancies in severity classification for all variables were observed between standard deviation and tertile groupings as well as between age groups.
Conclusions: Discrepancies were found in the distribution of severity categorizations across adaptive, communicative, social, and cognitive functioning. Greater variability emerged when using tertile splits, suggesting that basing severity categorizations on comparisons with other diagnosed individuals within a group may lead to provider or site-specific biases in severity assessment. The differences between groups and age levels highlight the need for a more clearly elucidated method of classifying ASD diagnoses as mild, moderate, or severe according to proposed diagnostic labels, and further study of how those designations may change with development.
See more of: Clinical Phenotype
See more of: Symptoms, Diagnosis & Phenotype