Development and validation of a prediction model for severe traffic accident injury caused by drivers with disabilities in Korea

Research Objectives: This study aimed to develop a prediction model for severe traffic injuries caused by drivers with disabilities in Korea.

Design: A retrospective cohort design.

Setting: General community

Participants: We obtained occasional medical checkup and traffic accident investigation reports of drivers with disabilities who caused traffic accidents between 2014 and 2019 (n=1939). Only participants who caused traffic accidents after being registered as drivers with disabilities were selected as the study population. All participants were randomly assigned to the training and validation sets at a ratio of 7:3 (training set, n=1291; validation set, n=553).

Interventions: This was an observational study.

Main Outcome Measures: The primary outcome of this study was the occurrence of traffic accidents in which victims suffered severe injuries, including death. Death and injury severity were recorded in the traffic accident investigation reports. Severe injuries were defined as hospitalization for >3 weeks due to a traffic accident.

Results: Among traffic accidents caused by drivers with disabilities, the proportion of traffic accidents in which victims suffered severe injuries was 21.15% and 20.80% in the training and validation sets, respectively. Using a multivariate logistic model with stepwise elimination, independent risk factors associated with severe traffic accidents were identified: type of traffic accident, victims’ protective gear, type of vehicle, driver’s driving behavior, and vehicle speed before the traffic accident. A nomogram was used to construct a risk prediction model. The areas under the curves of the model were 0.781 and 0.842 for the training and validation sets, respectively. The calibration curve indicated that the model exhibited proper discrimination and good calibration. Decision curve analysis showed that this prediction model was clinically useful when the risk threshold probability was 7–81%.

Conclusions: The prediction model developed in this study can contribute to the prevention of severe traffic accidents caused by drivers with disabilities and help them drive safely.

Author(s) Disclosures: This study was supported by a grant from the Ministry of Land, Infrastructure and Transport (MOLIT) Research Fund (NTRH RF-2023001) and the Korean National Police Agency (PR09-02-000-22).