Lapses during on road driving: Determining a cut-off value for clinically relevant driving impairment

Purpose: Traditionally, the Standard Deviation of Lateral Position (SDLP), i.e. the weaving of the car, is used as primary outcome measure of the Dutch on-road driving test [1]. Recently, lapses have been introduced as outcome measure of the on road driving test. Lapses are defined as short periods of inattention, which can be identified by a continuous change in lateral position of greater than 100 cm, lasting for at least 8 seconds [2]. The purpose of the current analyses is to determine a cut off point for clinically relevant driving impairment for the number of lapses. Methods: Data from two driving studies (150 driving tests) was re-analyzed, examining SDLP and number of lapses for the hypnotic drugs zolpidem (10 and 20 mg), zaleplon (10 and 20 mg), ramelteon (8 mg), zopiclone (7.5 mg), and placebo [3,4]. The increase in number of lapses relative to placebo corresponding to a DSDLP of +2.4 cm was determined to establish an appropriate cut-off value for a lapses criterion. This SDLP increment was chosen as it corresponds to an increase seen with a blood alcohol concentration of 0.05%, i.e. the legal limit for driving in many countries. Using DSDLP >+2.4 cm as criterion for driving impairment, sensitivity (true positive) and specificity (true negative) were computed for different number of lapses. Results: Of the 150 driving tests in drug conditions, 59 had a DSDLP >+2.4 cm. An increment of SDLP with 2.4 cm corresponds with and increment of one lapse. Hence, at a group level analysis, a clinically relevant increase in number of lapses, relative to placebo, is defined as an increase of more than one lapse. According to the best fitting trend line, a DSDLP of +2.4 cm corresponds to an increase of 1 lapse. Hence, a criterion for clinically impaired driving can be defined as having an increase of more than 1 lapse relative to placebo. When using 1 lapse as criterion for clinically relevant impairment, most unimpaired drivers (DSDLP +2.4 cm are identified based on the 1 lapse criterion (57.6%). Conclusions: Taking into account the outcomes of the investigated hypnotic drugs, at a group level an average increase of 1.5 lapses relative to placebo should be regarded as corresponding to clinically relevant driving impairment. While at the group level the number of lapses can be treated as a continuous variable, for individual drivers this is a discrete variable (i.e. half a lapse does not exist). As a result, for individual drivers a clinically relevant increase in number of lapses is defined as having an increase of 2 or more lapses relative to placebo.