Short-term reliability and validation of a night-time hazard visibility test

Abstract

Purpose : The ability to respond to hazards is a fundamental driving skill. Hazard perception can be measured using computer-based hazard perception tests, that include daytime driving scenes (DayHPT) in currently available tests and night-time scenes in our novel night-time hazard visibility test (NHVT) that was developed to assess visual performance for night driving. One aspect of these tests that has not been well explored is their short-term reliability; the association between the NHVT and the DayHPT is also unknown.

Methods : Participants included 35 young drivers (20 female; M=22.5±3.8years) tested across 2 sessions 4-5 weeks apart. Vision testing at Visit 1 included binocular visual acuity (VA) and Pelli-Robson letter contrast sensitivity (CS). At Visits 1 and 2, participants completed the DayHPT (20 videos of daytime scenes) and NHVT (26 videos of night-time scenes), filmed from the driver’s perspective, with a range of hazards (pedestrians, bicyclists, vehicles). Participants responded when they first recognized potential hazards in each video that would require them to take evasive action to avoid a collision. Raw response times were converted to z scores relative to the whole sample for each of the DayHPT and NHVTs over the 2 visits; more positive scores represented longer (delayed) response times. Test-retest differences were compared and the association between the DayHPT and NHVT explored.

Results : Participants had normal levels of VA (M=-0.24±0.05logMAR) and CS (M=2.01±0.10logCS). There were no significant differences between visits for NHVT (p=0.127) or DayHPT scores (p=0.738), and Bland-Altman plots demonstrated good agreement between visits (mean NHVT difference of 0.08; 95% CI[-0.57, +0.57]; mean DayHPT difference of -0.03; 95% CI[-0.85, +0.85]). There was also a significant positive association between DayHPT and NHVT scores at Visit 1 (r=0.73, R²=0.53; p<0.001).

Conclusions : Short-term reliability of DayHPT and NHVT performance was confirmed in a population whose performance was expected to be stable. The moderate correlation of NHVT and DayHPT scores confirms the validity of the NHVT, but suggests further variance may be explained by processes unique to night-time driving. Future research should investigate whether these results are repeatable in different cohorts and explore NHVT performance in those with complaints of night-driving, as well as current and emerging ocular corrections and eye diseases.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.