Hazard Perception in Visually Impaired Drivers Who Use Bioptic Telescopes

Abstract

Purpose: Bioptic telescopic spectacles can allow individuals with central vision impairment to obtain or maintain driving privileges. The purpose of this study was to (1) compare hazard perception ability among bioptic drivers and traditionally licensed controls, (2) assess the impact of bioptic telescopic spectacles on hazard perception in drivers with vision impairment, and (3) analyze the relationships among vision and hazard detection in bioptic drivers

Methods: Visual acuity, contrast sensitivity, and visual field were measured for each participant. All drivers completed the Driving Habits Questionnaire. Hazard perception testing was conducted using commercially available first-person video driving clips. Subjects signaled when they could first identify a traffic hazard requiring a change of speed or direction. Bioptic drivers were tested with and without their bioptic telescopes in alternating blocks. Hazard detection times for each clip were converted to z-scores, converted back to seconds using the average response time across all videos, and then compared among conditions.

Results: Twenty-one bioptic drivers and 21 normally sighted controls participated in the study. The hazard response time of bioptic drivers was improved when able to use the telescope (5.4 ± 1.4 seconds vs 6.3 ± 1.8 seconds without telescope); however, it remained significantly longer than for controls (4.0 ± 1.4 seconds). Poorer visual acuity, contrast sensitivity, and superior visual field sensitivity loss were related to longer hazard response times.

Conclusions: Drivers with central vision loss had improved hazard response times with the use of bioptic telescopic spectacles, although their responses were still slower than normally sighted control drivers.

Translational Relevance: The use of a bioptic telescope by licensed, visually impaired drivers improves their hazard detection speed on a video-based task, lending support to their use on the road.

Introduction

Driving is an important method of transportation for many adults living in the United States and around the world, and individuals who lose or are unable to obtain driving privileges report decreased quality of life.¹ Many states and jurisdictions have vision requirements for driving, such as a minimum level of central visual acuity or visual field. For drivers with central vision impairment, bioptic telescopic spectacles provide an avenue for licensure in at least three countries and more than 40 U.S. states.² A bioptic telescope is a small telescope mounted in the upper portion of a pair of spectacles with prescription lenses (referred to as carrier lenses, see Fig. 1). Bioptic drivers rely mostly on their reduced central vision through these carrier lenses, but approximately 2% of the time they use the telescope for brief periods to provide a magnified view of the road ahead, traffic signals, road signs, or other objects of interest.³ 

Figure 1.

 

The large majority of the time, the driver looks underneath the telescope through prescription carrier lenses (left). When spotting a distant target, the driver slightly tilts his head and lifts his eyes to use the telescope for a magnified view (right).

There has long been debate about the safety of bioptic drivers, with retrospective studies of state records frequently, although not always, finding higher collision rates for bioptic drivers when compared with controls.^4–9 However, a study using on-road assessments by backseat raters found similar safety ratings for bioptic drivers and normally sighted controls,¹⁰ and a study using naturalistic recording found that the near-collision rate for bioptic drivers was not statistically significantly elevated compared with that of a control group.¹¹ We have previously found that visual acuity, contrast sensitivity, and peripheral visual field measures are associated with the per-mile collision rate in bioptic drivers,⁹ but there remains a need to identify other tests that may be indicators of their driving safety. Additionally, it is not clear how useful the bioptic telescope actually is for improving driving performance or safety. 

Hazard perception is a key skill for driving with two main components: the perceived hazardousness of a situation, and the time required to react to that perceived hazard.^12,13 It is the ability to anticipate potential road hazards to avoid a collision,¹⁴ and can be considered to be situational awareness for roadway danger.¹⁵ Depending on the location of objects and obstacles in the field of view, drivers must gather information from both the central visual field and the periphery, and often must shift their eyes and attention to hazards in the scene before responding to them.¹⁶ Especially when driving, the time needed to acquire the critical visual information is very important. In a hazard perception task, the participant must notice the hazard, determine its hazardousness, and then respond. Depending on the study, the subjects may be asked to respond by pressing a button, describing the hazard verbally, or clicking with a mouse when the cursor is over the hazard being displayed on a screen. All these components taken together represent the hazard response time. Hazard perception is a component of licensure testing in the UK (www.gov.uk/theory-test/hazard-perception-test),^17,18 where prospective drivers must identify 15 hazards within 14 one-minute video clips. Scores are assigned based on accuracy and speed of responses, where the highest scores are awarded for responses within the first quintile of the time the hazard appears on screen. A test taker must score 44 out of 75 possible points on the hazard perception test to be eligible for licensure. 

Hazard perception ability has been shown to correlate with collision records,^19,20 and simulated cataracts,²¹ optical blur,²² and simulated visual field loss²³ have been shown to result in increased hazard perception time. Another study found the ring scotoma of a monocular bioptic telescope decreased hazard detection for subjects with simulated vision loss performing a sign reading task in a driving simulator.²⁴ More recent studies have assessed the hazard perception skills of drivers with eye disease. Using a computer-based task, Wood et al.²⁵ reported on hazard perception of drivers with cataract, glaucoma, macular degeneration, diabetic retinopathy, and stroke, and associated vision impairment, finding a 0.73-second delay in hazard perception for subjects with eye disease compared with normally sighted controls. A study of drivers with glaucoma demonstrated a 0.42-second delay in response to hazards compared with normally sighted controls.²⁶ 

Although these studies included drivers with real, not simulated, eye disease, all subjects were able to meet standards for traditional licensure. Less is known about the hazard perception ability of drivers with central vision impairment who use bioptic telescopes for licensure. One study of bioptic telescopes focused on the effect of the ring scotoma on hazard detection in bioptic drivers and drivers with normal vision,²⁷ but the drivers were required to continually view through the telescope to ensure driving hazards developed in the ring scotoma. 

Little is known about the hazard detection abilities of visually impaired drivers using bioptic telescopes. Because hazard perception test results have been shown to be associated with on-road driving performance and safety^19,20 for traditionally licensed drivers and because there is evidence of increased collision risk in bioptic drivers,^4–9 it is possible that hazard perception plays a role in bioptic driver safety. Thus, the purpose of this study was to compare the hazard perception of bioptic drivers with that of control drivers, as well as to assess whether the use of bioptic telescopic spectacles, used briefly for magnification as they would be on the roadway, improved the hazard perception of these drivers. 

Methods

Participants

Drivers with central vision impairment licensed to drive through the Ohio bioptic driving program and normally sighted, age-similar licensed control drivers with no history of eye disease were recruited from The Ohio State University College of Optometry. In Ohio, drivers with central vision impairment may use a bioptic telescope to meet the visual acuity standard for licensure, and they also must meet the same visual field standard (≥115° of horizontal visual field) as all traditionally licensed drivers. We excluded non-English speakers and those who reported not driving despite valid licensure. This study was conducted in accordance with and approved by the Ohio State University Institutional Review Board. 

Vision Measurements

Visual acuity was measured using Early Treatment Diabetic Retinopathy Study charts²⁸ with by-letter scoring for each eye individually through habitual distance correction or bioptic carrier lenses, and through the participant's bioptic telescope when indicated. All drivers using bioptic telescopic spectacles had monocular, focusable telescopes. Binocular contrast sensitivity was measured using the Mars chart²⁹ at 50 cm. Visual field testing was completed using the Humphrey Field Analyzer 3 with 24-2C pattern and SITA Faster strategy for each eye with appropriate refractive correction. A binocular visual field was constructed using the better total deviation value from corresponding monocular points,^30–33 a method that has been compared previously with standard measures for driving fitness.³⁴ The average deviation for the superior, inferior, right, and left hemifields were also calculated from these binocular visual fields. 

Driving History, Difficulty, and Exposure

Each participant self-reported the total number of years they had been driving, and bioptic drivers also reported their years of bioptic driving experience in cases where they were licensed without a bioptic telescope initially and later needed to enter the bioptic driving program to maintain licensure. All participants verbally completed the Driving Habits Questionnaire³⁵ with a trained member of the research team. Average weekly mileage was calculated based on responses to the survey. Summary scores based on a 5-point scale (5 = no difficulty at all, 4 = a little difficulty, 3 = moderate difficulty, 2 = extreme difficulty, 1 = does not do due to visual reasons) were calculated for questions related to difficulty with the following driving situations: rain, driving alone, unprotected left turns, interstates or expressways, high-traffic roads, rush-hour traffic, and night driving. All participants reported whether they were usually the driver (score = 1), usually the passenger (score = 3), or split the driving (score = 2) with the people with whom they regularly traveled, and an average dependency score was calculated. 

The Hazard Perception Test

Hazard perception testing required participants to respond to a series of commercially available videos (Imagitech Ltd, Swansea, UK) that were chosen because of their previous use in studies of the impact of vision and divided attention on driving.^27,36 The clips are first-person recordings of real roadway driving scenes that are available for training for the hazard perception component of the licensure examination in the UK. From a total of 150 videos, a subset of 30 unique 1-minute test videos, each containing a single hazard, was selected. Videos containing multiple hazards, a high number of roadway signs, significant roadway markings, or settings that would not commonly occur in the United States were eliminated, resulting in a subset of 30 one-minute test videos, each containing a single hazard. The video clips were mirrored using Adobe Premiere so that driving scenes appeared as right-hand traffic, as participants would encounter on roads in the United States. Subjects were seated 1.25 m from a screen and clips were projected onto a screen with a field of view of 52° horizontal by 33° vertical. 

Participants were instructed to signal, by pressing a button on a keyboard in front of them, as soon as they identified a developing traffic hazard that would require the driver to change speed or direction to avoid a collision (Fig. 2). They were instructed that each video would contain exactly one hazard. Each participant completed three practice runs. The first included full narration of the video and identification of the hazard by a member of the study team. The next two were used as practice runs, with discussion about the target hazard and any other questions after completion of the run. All participants reported understanding of the task before testing. The hazards included were pedestrians, animals, cyclists, or another vehicle. The 30 selected hazard videos included 4 animals, 7 cyclists, 10 pedestrians, and 9 other vehicles. Control drivers completed a subset of fifteen trials, including a selection of one-half of the videos in each category. Bioptic drivers completed 30 trials, separated into 5-video blocks, alternating between being able to use their telescope (focused for the test distance) or with their telescope occluded so that they were restricted to carrier lens use only. No videos were shown to any subject more than once; one-half from each category were used for testing with the bioptic available and one-half were used with carrier lenses only. The first condition for testing (with or without telescope available) was randomized for participants. Test videos were selected such that each clip was shown an equal number of times across the entire study set. MATLAB (2021b, MathWorks Inc., Natick, MA) was used for all video presentation and trial recording. 

Figure 2.

 

Stills representing various possible response times during a sample hazard perception task. The driver would signal as soon as the hazard in the video was identified (circled in yellow for demonstration).

Figure 3.

 

Mean standardized hazard response time as a function of group or condition. The center line in each box is the median, with boxes representing the interquartile range. Individual data points are jittered. BTS, bioptic telescope available during testing; No BTS, telescope occluded.

Visually impaired drivers were instructed to use their bioptic as they normally would while driving. It has been shown previously that bioptic drivers use their telescopes only briefly to access magnified views of the road ahead.³ Bioptic drivers were observed during testing and demonstrated the ability to properly use the telescope; however, we did not quantify telescope use formally. 

Data Analysis

Hazard perception time was recorded as the time in seconds between the first appearance of a hazard and the time the subject responded. The response times were converted to z-scores using the average response for each hazard to account for differences among the videos and hazards presented. Then, z-scores were converted back to seconds using the mean and standard deviation of all responses to all clips for ease of interpretation.^{12,25,26,37,38} Hazards that were not identified within the hazard window were assigned the time of the entire window during which the hazard was visible, as in previous work.³⁹ Responses were also scored based on methods from the UK licensure examination (www.gov.uk/theory-test/pass-mark-and-result), where hazard perception testing is required for licensure. For this scoring, the total time each hazard is visible in the video is divided into quintiles. For a response within the first quintile, a participant would receive a score of five; for the second quintile, a score of four; and so on. If the hazard is missed, a score of zero is assigned for that video. The scores are summed, and a total of 44 out of a possible 75 points is required to pass. 

Characteristics including age, vision, and driving habits were compared among bioptic drivers and controls using independent t tests. Hazard response time for bioptic drivers (with ability to use their telescopes) was compared with that of control drivers using independent t tests. To assess the potential effect of the bioptic telescope on hazard perception, the average hazard response time for bioptic drivers with and without the ability to use their telescopes was compared using a paired t test. Summary driving difficulty scores (excluding parallel parking due to a lack of participant response to the item), dependence on others for transportation, and weekly mileage were calculated using responses to the Driving Habits Questionnaire. Vision, driving habits, and demographic variables were assessed for relationships with hazard perception, adjusting for age, using linear regression. Linear regression model assumptions (independence of observations, homoscedasticity, and approximate normal distribution of residuals) were confirmed. P values of <0.05 were considered evidence of significance for all tests. Statistical testing was performed using IBM SPSS version 29 (IBM, Armonk, NY). 

Results

Vision and Demographics

Twenty-one drivers with central vision impairment licensed to drive through the Ohio bioptic driving program (81% male) and 21 normally sighted control drivers (29% male) completed the study. Table 1 details demographic and vision data of participants. Bioptic and control driver groups were not different in age (P = 0.926). Eye conditions reported by bioptic drivers included maculopathies (n = 7), albinism (n = 7), diabetic retinopathy (n = 2), optic atrophy (n = 2), history of stroke (n = 1), history of congenital cataract (n = 1), and retinopathy of prematurity (n = 1). 

Table 1.

 

Demographic and Basic Vision Data of Participants

 

The mean better-eye visual acuity of bioptic drivers was approximately 20/100 through their carrier lenses, which was significantly worse than that of control drivers (P < 0.001). When viewing through their telescopes, mean visual acuity improved to about 20/30, although this was still significantly worse than that of control drivers (P < 0.001). Contrast sensitivity was also significantly poorer for bioptic drivers than for control drivers (P < 0.001). Bioptic drivers also had worse visual field mean deviation (P < 0.001 for all comparisons). 

Hazard Perception

The mean standardized hazard detection time for bioptic drivers when able to use their bioptic telescopic spectacles was 5.4 ± 1.4 seconds, which was significantly slower than that of control drivers’ mean response time of 4.0 ± 1.7 seconds (independent t-test, P = 0.006). This response time by bioptic drivers when able to use their telescopes was faster than the response time of 6.3 ± 1.8 seconds when their telescope was occluded. (paired t test; P = 0.005) (Fig. 3). Overall, few hazards were missed by participants (n = 3 misses [1%] during 315 telescope available trials; n = 14 misses [4%] during 315 telescope occluded trials; and n = 12 misses [4%] during 315 control trials). For individual participants, the median number of hazards missed out of 15 was zero for all three conditions (bioptic telescope available, bioptic telescope occluded, and control). 

When using UK scoring standards, 86% of control drivers (n = 18) would pass the hazard perception test (mean score, 51 ± 8). Bioptic drivers met the UK test passing criterion 66.7% of the time in trials in which the telescope was available versus 33.3% for trials in which the telescope was covered. One-half of bioptic drivers who had a failing score with the bioptic telescope covered had a passing score for trials when they were allowed to use the telescope, and all drivers with a failing score for trials with the telescope also failed for trials in which the telescope was covered. We tested the null hypothesis that the probability of passing with the telescope available is the same as when the telescope is covered. McNemar's test for paired comparisons rejected this null hypothesis (P = 0.016). 

Driving Exposure and Difficulty

The two groups of drivers reported similar total years of driving experience (30 ± 14 for bioptic drivers vs 34 ± 12 for controls; P = 0.282). They also reported driving similar weekly mileage (274 ± 262 for bioptic drivers vs 224 ± 210 for control drivers; P = 0.540). When surveyed about various challenging driving scenarios with the Driving Habits Questionnaire, bioptic drivers reported more total driving difficulty than controls (P < 0.001) and more dependence on others (maximum score of 3, where the participant is usually the passenger and a minimum score of 1, where the participant is usually the driver) to serve as the driver (2.50 ± 0.56 vs 1.87 ± 0.54; P < 0.001). 

Vision and Hazard Perception

Relationships among vision measurements and standardized hazard response time in regression models controlling for age are shown in Table 2. For drivers with vision impairment, hazard response time with their bioptic telescopes available was used. Visual acuity (P < 0.001), contrast sensitivity (P=0.008), and binocular superior visual field deviation (P = 0.001) were found to be the best statistically significant predictors in univariate regression models. Slower hazard perception time was associated with greater reported difficulty with road scenarios (P = 0.024) and greater dependence on other drivers (P = 0.003) among all participants, controlling for age. There was no relationship between hazard perception time and reported weekly mileage (P = 0.752) or years of driving experience (P = 0.957), controlling for age, nor was sex related to standardized hazard response time (P = 0.512). 

Table 2.

 

Results of Regressions Assessing Relationships Among Hazard Perception and Vision Measures, Adjusted for Age

 

There were relatively few self-reported motor vehicle collisions among participants. Two bioptic drivers each reported one collision where police were called to the scene, and one bioptic driver reported two collisions with no police involvement. Three normally sighted control drivers each reported one collision where police were not called. 

Discussion

Visually impaired drivers who use bioptic telescopes responded more quickly to hazards in a video-based test when able to use their bioptic devices compared with their telescopes occluded, but still responded more slowly than normally sighted control drivers of similar age, even when able to use the telescope. The finding that visually impaired drivers respond more slowly to hazards than controls is consistent with previous research documenting longer hazard response times for participants with eye disease, including cataracts, glaucoma, and age-related macular degeneration,²⁵ as well as for those with simulated cataract.²¹ Bioptic drivers responded to hazards 1.3 seconds slower than normally sighted controls. This means a bioptic driver would travel about 95 feet (30 m) farther at 50 miles per hour (80 km per hour) than a control driver before identifying and responding to a roadway hazard. This delay could translate to a greater risk of adverse on-road outcomes, such as hard braking, evasive maneuvers, or collisions. 

Previously, naturalistic recording of bioptic drivers demonstrated that bioptic drivers mainly view the road with their impaired central vision, accessing the telescope magnification for spotting only about 2% of the time,³ and that the most common telescope spotting task was previewing the road ahead. In the present video-based study, the mean time to hazard detection when drivers were able to access the magnification provided by the bioptic telescope was 0.8 seconds faster than when the telescope was occluded. Additionally, when participants were able to spot through their bioptic telescope, they were more likely to obtain a score that would qualify as passing on the hazard perception portion of the UK licensure examination, which has been shown to predict non–low-speed reported collisions on public roads.¹⁸ Because drivers were able to detect hazards more quickly in trials where they were able to use the bioptic telescope than when it was occluded, this finding supports a potential practical role for the telescope in detecting obstacles in the driver's path and potentially leading to decreased collision risk. 

Studies of traditionally licensed drivers have shown correlations between hazard perception testing results and state crash records.^19,20 It has also been demonstrated previously that bioptic drivers have higher collision rates than normally sighted drivers,^4,5 and that worse visual acuity and contrast sensitivity are related to higher per-mile collision rate.⁹ It is logical that there may be a relationship between the on-road performance of bioptic drivers and their laboratory-based hazard perception skills, and that visual function is a component of this relationship. This is an important area for future study. 

However, the use of recorded driving videos has limitations. Participants were informed that there would be a hazard in each video, whereas the appearance of hazards is unpredictable in real-world driving. The present study included other vehicles, animals, pedestrians, and cyclists as hazards. These items were chosen for their easy understanding as roadway hazards, their commonality across rural and suburban driving areas, and their availability in videos obtained for this testing. Future studies might include other types of hazards, such as roadway debris or construction markers and materials, or use newer video sets ranked for level of hazardousness, such as those recently published by Song et al.⁴⁰ (which were not available during our study development). Owing to the expected difficulty viewing a cursor and operating a mouse while using a bioptic telescope, participants pressed a button on a keyboard to signal detection of a hazard, which limits the ability to determine whether the subject was responding to the selected hazard in each video. A lack of videos including major roadway markings or prominent signage is a limitation of this work; in some scenarios, signs may indicate impending hazards in a meaningful way for drivers. The exclusion of these videos was necessary owing to mirroring of the roadway scenes, which would make the contents of the signs unreadable. Future work could include recording of new, U.S.-based videos that could include important warning signs and road markings and could also assist with decreasing the potential confusion experienced by participants owing to video mirroring. 

Additionally, participants in this study did not need to complete other tasks to control the vehicle (steering, braking, etc.) as they would on the road (or in a driving simulator), and this factor could affect their ability to detect hazards, for instance, looked but failed to see errors occur when someone fails to recognize an object that is clearly visible.⁴⁷ For example, a driver may describe a collision in which they report not having seen the object with which they collided, even though they also report having looked in the direction of that object. These errors are a well-known source of collisions that are related to attention rather than vision loss.⁴⁷ They may be less common in video-based hazard perception testing, like in the present study, than they are in real-world driving because the cognitive and physical demands of controlling a vehicle or doing other common in-vehicle activities (like conversing with a passenger) are not present.^41–43 

Videos like the ones used in this study are a useful way to assess hazard perception owing to the ethical dilemma of deliberately exposing subjects to driving hazards, as well as the ability to standardize the presentation of hazards throughout the study. Future studies using different methods could help to elucidate an important question—what role does actual use of the bioptic telescope play in hazard detection when driving? There are a number of ways in which this question could be pursued. Driving simulator scenarios in which hazards are programmed to appear more unpredictably and in which the driver must control the simulated vehicle could address some of the limitations inherent with videos like the ones used in the present study.^24,44 Objective measures of bioptic telescope use and head and eye tracking could also be of value, as the road locations attended to by drivers, both through the carrier lenses and through the bioptic telescope may be an important component of their hazard detection abilities and the role of the bioptic device.^3,10 Naturalistic recording studies in which both the road scene and driver are recorded during real hazardous situations, which can be analyzed retrospectively, could also play a useful role.³ 

We also found that both a summary score for self-reported difficulty in challenging driving situations from the Driving Habits Questionnaire and reported dependence on others to drive were associated with hazard perception testing results, with those reporting greater difficulty having slower hazard perception times. It is possible that those with poor hazard perception skills may perceive driving as more difficult and, thus, rely on others to drive instead. This finding would be consistent with previous studies that have found relationships among poorer vision and reduced mileage or driving cessation for older adults,^45,46 as well as other work that has shown poorer visual acuity, contrast sensitivity, and visual fields that predict lower yearly mileage for bioptic drivers.⁹ Interestingly, however, in the current study there was no relationship between either self-reported driving difficulty or hazard perception testing results and self-reported mileage. It has been shown that hazard perception is a trainable skill.^47–49 It is possible that training focused on hazard perception may be beneficial for bioptic drivers, especially because little research exists on the best ways to train these individuals. A training protocol aimed at the use of the bioptic for more efficient hazard detection, such as the best areas of the roadway to look for hazards or best practices for spotting techniques, could improve a driver's hazard perception. Perceived hazardousness can affect hazard detection, and training on the ways in which particular roadway obstacles are (or become) hazardous may play a role in improving drivers’ performance. These are an important areas for future study, especially if coupled with on-road studies of driver performance. 

It has been reported previously that visual field sensitivity is related to hazard perception, but the literature is mixed on the location with the greatest impact on performance. One study demonstrated simulated visual field loss to the right and left of fixation negatively impacts hazard detection time,²³ and another study has shown that simulated superior defects have more impact that inferior defects.⁵⁰ In the present study, true right, left, superior, and overall binocular visual field deficits were found to be related to hazard perception time. We found superior binocular visual field defects to have the strongest relationship with hazard perception time, similar to Glen et al.’s⁵⁰ comparison between superior and inferior fields. All included hazards had looming qualities, and as the videos included driving down the roadway, hazards often appeared in relatively more superior positions within the visual field. It is important to note that in the present study only defects in the central 24° of visual field were measured with standard automated perimetry, and licensure in Ohio requires drivers to have ≥115° of total horizontal visual field. Thus, the effects of more peripheral visual field deficits on hazard perception could not be assessed with these data. Additionally, it is worth noting that central vision impairment can make fixation during visual field testing difficult, and thus these results should be viewed with some caution. 

Conclusions

Bioptic telescopic spectacles allow some individuals with central vision impairment the ability to obtain or maintain licensure. Although these drivers exhibited slower detection of roadway hazards than drivers without vision loss, the use of the telescope improved their response time in a video-based hazard perception test. Delayed hazard response times were associated with decreased visual acuity, poorer contrast sensitivity, and worse binocular superior visual field mean deviation, as well as with self-report of more difficulty with challenging driving situations. 

Acknowledgments

Ohio Lions Eye Research Foundation Fellowship; Prevent Blindness Ohio Young Female Investigator Award; NIH T35 EY007151; NIH T32 EY034824. 

Disclosure: R.A. Deffler, None; S.-S.L. Cooley, None; H.A. Kohl, None; T.W. Raasch, None; B.E. Dougherty, None