Validating lane drifts as a predictive measure of drug or sleepiness induced driving impairment

Background: Standard deviation of lateral position (SDLP) has been accepted as a reliable parameter for measuring driving impairment due to lowered vigilance caused by sleepiness or the use of sedating drugs. Recently, lane drifts were proposed as an additional outcome measure quantifying momentary lapses of attention. The purpose of this study was to validate lane drifts as outcome measure of driver impairment in a large data pool from two independent research centers. Methods: Data from 11 placebo-controlled studies that assessed the impact of alcohol, hypnotics, and sleep deprivation on actual driving performance were pooled. In total, 717 on-the-road tests performed by 315 drivers were subjected to an automated algorithm to detect occurrences of lane drifts. Lane drifts were defined as deviations > 100 cm from the mean (LDmlp) and from the absolute lateral position (LDalp) for 8 s. Results: The number of LDmlp was low and did not differ between treatments and baseline, i.e., 14 vs. 3 events, respectively. LDalp were frequent and significantly higher during treatment relative to baseline, i.e., 1646 vs. 470 events. The correlation between LDalp and SDLP in the treatment conditions was very high (rs = 0.77). The frequency of the occurrence of treatment-induced lane drifts however depended on baseline SDLP of drivers, whereas treatment-induced changes in SDLP occurred independent of baseline SDLP. Conclusion: LDmlp is not useful as an outcome measure of driver impairment due to its rare occurrence, even when treatment-induced increments in SDLP are evident. Treatment effects on LDalp and SDLP are closely related.