Crash Risk of Cell Phone Use While Driving: A Case – Crossover Analysis of Naturalistic Driving Data
This report investigates the relationship between using a cell phone while driving and the risk of being involved in a crash by comparing cell phone use immediately prior to crashes versus during ordinary driving by the same drivers using in-vehicle video from a large study.
For media inquiries, contact:
Cellular telephone use while driving is a risk factor, but how much of one is a hotly debated issue – particularly as more people use smartphones that are essentially hand-held internet-accessible computers. Numerous studies conducted with driving simulators and on-road driving suggest that using a cell phone while driving, particularly visual-manual interaction, can significantly impair driving performance.
This study investigated the relationship between cell phone use and crash risk using data from the Second Strategic Highway Research Program Naturalistic Driving Study, which included data from a sample of 3,593 drivers whose driving was monitored using in-vehicle video and other data collection equipment for a period of several months between October 2010 and December 2013. The relationship between driver cell phone use and crash involvement was quantified using a case-crossover study design in which a driver’s cell phone use in the six seconds immediately prior to the crash was compared with the same driver’s cell phone use in up to four six-second segments of ordinary driving under similar conditions (time of day, weather, locality, lighting, and speed) within the three months prior to the crash. Cell phone use, crash involvement, and traffic and environmental conditions were assessed using in-vehicle video. The final study sample included 566 severe, moderate, and minor crashes matched to 1,749 segments of ordinary driving.
Odds ratios for the association of cell phone use with crash involvement were estimated using conditional logistic regression. Odds ratios were calculated for overall cell phone use, conversation, overall visual-manual cell phone use, and several specific visual-manual tasks including texting, dialing, browsing, and reaching for or answering the phone; the reference condition was driving without performing any observable secondary task. Results were also stratified by traffic density, crash severity, and crash type.
Visual-manual tasks overall and texting in particular were associated with significantly elevated incidence of crash involvement relative to driving without performing any observable secondary tasks (visual-manual interaction overall: Odds Ratio [OR] 1.83, 95% Confidence Interval [CI] 1.03 – 3.25; texting: OR 2.22, CI 1.07 – 4.63). The increase in the incidence of crash involvement associated with visual-manual tasks was greater for crashes in free-flow traffic conditions (OR 2.46, CI 1.10 – 5.51) and in types of crashes in which the subject driver generally played a clear role (run-off-road crashes: OR 3.15, CI 1.30 – 7.67; rear-end crashes: OR 7.77, CI 1.65 – 36.56) than for all crash types taken together. The incidence of crash involvement was elevated slightly during hand-held cell phone conversation; however, the estimate was very imprecise and was not statistically significant (OR 1.16, CI 0.50 – 2.70). The relationship between hands-free cell phone conversation and crash involvement could not be assessed meaningfully because there were very few crashes or baseline epochs in which hands-free cell phone conversation was observed.
In general, results reflected similar patterns to previous studies, with visual-manual tasks (particularly texting) associated with significantly increased crash risk. Estimated risks were somewhat lower than in previous studies, likely due to the careful matching of crashes to baseline epochs in which the same drivers were driving under similar traffic and environmental conditions, thereby inherently controlling for many individual driver-specific and situational factors that may be related to both cell phone use and crash risk.
For media inquiries, contact: