Senior Research Scientist Children's Hospital of Philadelphia Philadelphia, Pennsylvania, United States
Background: It is well recognized that impairment in executive function (EF) capacities in clinical adolescent populations (e.g. ADHD ASD) is related to poorer driving outcomes and performance. However, there is documented wide variability in EF capabilities even in typically developing adolescents and young adults also. Some subdomains of executive function have been associated with crash risk (e.g. primarily working memory, and inhibition). However, studies have often examined the subdomains of EF in isolation, precluding ability to comparatively differentiate which poses the greatest risk. Objective: To systematically test the association between individual differences in different subdomains of executive function and crash risk in typically developing young drivers. Design/Methods: A cross-sectional analysis of n=93 typically developing young drivers from an age-balanced sample of 16-24 years (Mean Age = 19.86 years) recruited from the community in Philadelphia, USA. We measured their performance in 3 subdomains of EF (working memory, inhibition, task switching), as well as speed of processing. Participants also completed a validated virtual driving assessment (VDA) that measures crash risk by exposing drivers to common serios crash scenarios, and classifies drivers as having No Issues, Minor Issues, Major Issues, or Major Issues with Dangerous Behavior. We collapsed the classes to establish a binary indicator of VDA performance as our primary outcome: safer classes (No issues, Minor issues: n=54) and the less safe classes (Major issues, Major issues with Aggressive Driving: n=38). We ran univariate models including all identified covariates and EF measures, and conservatively retained variables with p< 0.2 for a forward step-wise model selection process (choosing the best model by AIC). Results: Model results indicate that variation in working memory alone, as measured by the 3-back task was the most consistent factor significantly associated with VDA performance. Better 3-back task scores were associated with being in the safer class on the VDA (OR 2.60, p=.01, 95% CI: 1.31, 5.85). This association remained significant when covariates of crash risk were added to the model.
Conclusion(s): These results extend results from previous studies of clinical populations with impaired EF to typically developing young drivers. Variability in working memory was related to objective driving performance on the VDA, positioning working memory as a candidate factor for clinical screening, and possible preventative interventions.
