Sensitivity and specificity of the low verbal investigatory survey in a large, diverse, American sample

The acquisition of productive language is a key predictor of quality of life in autism; however, this developmental process yields widely varying outcomes, from fluent verbal speech to an inability to communicate verbally. Currently available measures fail to characterize autism-specific communication strategies, examine only broad outcomes, or require significant time and expertise. This preregistered study rigorously tests the psychometrics and utility of a low-cost and efficient measure in a large, diverse, sample of caregivers of autistic children (including a minimally and low verbal, MLV, group), and establishes its utility for characterizing language abilities in that sample.

A 30-item yes/no questionnaire, the Low-Verbal Investigatory Survey (LVIS), was administered to 796 caregivers of autistic individuals ages 6 to 16 years, along with surveys of sociodemographics, adaptive and cognitive functioning, autistic symptomatology, and caregiver familiarity with autism. Group differences by race and ethnicity were assessed using ANOVA and chi-square analyses; the factor structure of the LVIS was assessed using confirmatory and exploratory factor analysis; convergent and divergent validity were assessed using partial correlation and regression models; and optimal LVIS cut-point for evaluating MLV status based on caregiver report was assessed via Receiver Operator Curve (ROC) analyses.

The LVIS yielded three factors: Verbal Ability, Atypical Communication, and Nonverbal Communication. Results indicated that Verbal Ability was strongly and significantly correlated with other communication measures, and less strongly with measures of other abilities (suggesting both convergent and divergent validity). The Atypical Communication factor converged with broader autistic characteristics (Social Communication Questionnaire, and Repetitive Behaviors Scale - Revised scores). LVIS scores were consistent across race and ethnicity, income levels, and levels of autism knowledge; educational attainment was associated with higher LVIS scores. ROC results indicated a clear cut-point (Verbal Ability = 12) for predicting MLV status with good sensitivity and specificity.

Study participants may be better resourced and more connected to services than the broader population.

The LVIS offers a low-cost, quick, and informative means of characterizing autism-specific communication skills in individuals across the autism spectrum. It may be used to track the development of language skills in large samples and longitudinal studies.