Predictive validity of the Addenbrooke’s cognitive Examination III for on-road driving outcomes in a clinical fitness-to-drive population: a prospective cohort study

Abstract

Background Evidence-based fitness-to-drive assessment must balance public safety and fairness. Cognitive screens are commonly used, but must be accurate in high-stakes contexts. The predictive utility of the Addenbrooke’s Cognitive Examination III (ACE-III) for driving outcomes remains unclear. Methods This prospective cohort study analysed data from drivers referred to a specialist fitness-to-drive clinic. Participants completed the ACE-III and a standardised occupational therapy on-road driving assessment (clinical reference standard). Associations between ACE-III scores/classification, demographic variables, and pass/fail on-road outcome were examined. Predictive validity was evaluated using ROC analysis and accuracy indices, applying a pre-specified two-cut-off model with an indeterminate grey zone. Results Between August 2022 and June 2024, 133 drivers (33–97 years) were included. Participants who passed the on-road assessment had higher ACE-III total and subdomain scores. Overall discrimination was good (AUC = 0.81). Using the two-cut off model (≤81 predicting failure; ≥88 predicting pass; 82–87 indeterminate), sensitivity was 81%, specificity 75%, PPV 82%, NPV 73%, and overall accuracy 79%. Misclassification remained: 27% of drivers classified as Not Impaired failed the on-road assessment, 18% classified as Definitely Impaired passed, and 26% were indeterminate. Years of education and communication limitations further constrained validity. Conclusions ACE-III is associated with on-road outcomes but is not accurate enough for stand alone fitness-to-drive decisions. It may assist screening within a broader functional assessment framework, with on-road evaluation remaining the reference standard.