The Concurrent Criterion Validity of the 32-Item Fitness-to-Drive Screening Measure

• Main Authors: Shabnam Medhizadah, Sherrilene Classen, Andrew M. Johnson
• Format: Article
• Language: English
• Published: Frontiers Media S.A. 2019-02-01
• Series: Frontiers in Psychology
• Subjects: psychometric; automobile driving; proxy raters; ROC (Receiver Operating Characteristic) curve; sensitivity and specificity (MeSH)
• Online Access: https://www.frontiersin.org/article/10.3389/fpsyg.2019.00253/full

Background: The Fitness-to-Drive Screening Measure© (FTDS) is a free online screening tool that identifies at-risk older drivers. This tool screens for at-risk drivers using proxy rater responses (family, friends, and caregivers) to 54 driving-related items. Consumer usage analysis of the FTDS determined that reducing the time commitment to complete the 54-item FTDS might increase usability and uptake of the tool. To address this need, we used classical test theory and exploratory factor analysis to construct a 32-item version of the FTDS. This study aims to establish the concurrent criterion validity of the 32-item FTDS.Method: Two hundred older driver on-road assessments and Two hundred caregiver FTDS responses were used to generate a receiver operating characteristic (ROC) curve, in which we plotted the rate of true positives against the rate of false positives, calculated the area under the curve (AUC), and used Youden's index to identify the optimal cut-point for the 32-item FTDS. In this study, the true positive rate was the 32-item FTDS' ability to predict a fail when the older driver actually failed the on-road assessment, and the false positive rate was the the 32-item FTDS' ability to predict a pass when the older driver actually passed the on-road assessment. We computed the sensitivity, specificity, positive predictive value, negative predictive value and total number of misclassifications for the optimal cut-point.Results: The ROC curve results indicated an acceptable AUC, with a magnitude of 0.75, p < 0.05. At the optimal cut-point of 4.87, the 32-item FTDS had a sensitivity of 0.74, specificity of 0.69, positive predictive value of 0.30, negative predictive value of 0.93 and 61 (of 200) misclassifications.Conclusion: Although the 32-item FTDS met the criterion (AUC 0.75, p < 0.05.) for good concurrent criterion validity in predicting older driver on-road outcomes, it also misclassified 30% of the drivers and as such may be overly sensitive.