Related Experiment Video
Updated: May 12, 2026
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
Validation of the Candrive Older Driver Risk Stratification Tool for Assessing Medical Fitness-to-Drive in Older
Judith L Charlton1, Sjaan Koppel1, Amanda Stephens1
1Monash University Accident Research Centre, Monash University, Clayton, Australia.
Background:
Assessing older drivers' fitness-to-drive is challenging, with decisions impacting mobility and health. This study aimed to validate the Candrive older driver risk stratification tool for screening medical fitness-to-drive in an independent cohort of older adults from the Ozcandrive 8-year prospective study.
Methods:
A convenience sample of drivers aged 75 and older residing in Melbourne, Australia completed the Candrive assessments. Their vehicles were instrumented to collect vehicle and global positioning system data, including trip distance. The first 4 years of Ozcandrive data were analyzed. The primary outcome measure was self-reported at-fault collisions, adjusted per 10 000 km driven. Collision risk status was modeled using Generalized Estimating Equations with Poisson regression using predetermined Candrive risk stratification tool predictor variables.
Results:
A total of 257 older drivers (70.8% male) were recruited with an average age at study enrollment of 79.7 years (standard deviation = 3.5). Of the 755 adjusted person-years of driving, 74.1% were in the Low risk category (vs original sample, Candrive: 74.8%) and 10.5% were in the Low-Medium risk category (Candrive: 9.3%). Only 15.4% were in the Medium-High risk category (Candrive: 15.9%), where the relative risk for self-reported at-fault collisions was 1.79 (95% confidence interval = 1.06-3.03) compared to the Low risk category.
Conclusions:
This study demonstrates an association between self-reported at-fault collisions and Candrive risk stratification tool scores. This result is promising given the primary outcome measure differed from the original Candrive study that used police-reported, at-fault collisions, and supports Candrive risk stratification tool's use by healthcare providers when initiating fitness-to-drive conversations.
