A Cross-Validated Approach Towards Identifying the Unique and Cumulative Contributions of Child and Family Factors Predictive of Speech-Language Therapy Start Time

Abstract

Language is an important milestone in early childhood and paves the way for later achievement and outcomes. Early identification and communication-focused interventions can provide additional environmental supports to foster language development for children who face challenges in acquiring language. To accurately identify and predict who may be likely to receive speech/language services, it is crucial to understand the unique contribution and the cumulative effect of factors that best predict speech-language service status and service receipt start time. Using data from the Early Childhood Longitudinal Study – Birth Cohort (ELCS-B), the current study sought to identify, within a bio-ecobehavioral framework, the unique and combined contribution of child and family factors predictive of speech-language therapy (SLT) start time. Multinomial logistic regression was used to evaluate child and family factors predictive of SLT start time. In addition, k-fold cross-validation was performed to assess generalizability. The final model accurately predicted SLT start time 62% of the time and only included child-related factors. The presence of an intellectual and developmental disability (IDD), prelinguistic performance, cognitive growth, and the number of words said at 24-months all provided a unique contribution in predicting SLT start time.