Printer Friendly

Relative effects of age and compromised vision in driving performance.

INTRODUCTION

Impairments of visual, perceptual, and cognitive functions have been shown to occur with age (Baltes, Cornelius, Spiro, Nesselroade, and Willis, 1980), and higher-order perceptual/cognitive decline has been suggested to be a contributing factor in motor vehicle accidents among older drivers (Ball, Owsley, Sloane, Roenker, and Bruni, 1993; Coyne, Fiens, Powell, and Joslin, 1990; Lucas-Blaustein, Filipp, Dungan, and Tune, 1988; Odenheimer et al., 1991; Sekuler, Kline, and Dismukes, 1982). Besides the contribution of perceptual/cognitive decline to deficits in driving skill, we have demonstrated that sensory changes can also affect driving abilities (Szlyk, Alexander, Severing, and Fishman, 1992; Szlyk, Fishman, Severing, Alexander, and Viana, 1993).

In the present study our aim was to determine how the normal aging process, with its concomitant perceptual/cognitive changes and visuosensory changes, interacts with disease-based sensory loss to affect driving performance. To do this we compared the real-world accidents and driving simulator performance of four groups of subjects: (1) a younger, normally sighted group; (2) an older, normally sighted group; (3) a younger, visually compromised group; and (4) an older, visually compromised group.

METHOD

Subjects

A total of 107 paid volunteers who drove at least 1000 miles per year served as subjects in this study. The younger and older groups did not differ significantly in sex or the mean number of miles driven per year (assessed by a self-report questionnaire). Neither was there a significant difference in mean age or mean years of driving experience between the two groups.

Control subjects. Of the subjects, 47 were normally sighted, had corrected binocular visual acuity of 20/20 or better (as measured with the Lighthouse Charts, Long Island City, NY), clear ocular media (i.e., the fluid within the eye and the lens were clear), and normal fundi (i.e., the vasculature of the retina was normal with no lesions or leaky blood vessels that would suggest disease). Of that 47, 27 subjects (13 women and 14 men) ranged in age from 19 to 49 yrs, and 20 subjects (7 women and 13 men) ranged from 50 to 83 yrs.

Patients with compromised vision. The remaining 60 subjects had compromised vision. Of that number, 37 (18 women and 19 men) ranged in age from 22 to 49 yrs, and 23 subjects (7 women and 16 men) ranged from 50 to 80 yrs. Each had a diagnosis of one of the following disorders: age-related macular degeneration (N = 3: older subjects), retinitis pigmentosa (N = 22: 17 younger subjects, 5 older subjects), cone-rod dystrophy (N = 12: 11 younger subjects, 1 older subject), Stargardt's disease (N = 11: 7 younger subjects, 4 older subjects), or hemianopsia resulting from a cerebral vascular accident affecting primarily the occipital cortex (N = 12: 2 younger subjects, 10 older subjects). Diagnosis was determined from the clinical files of the patients and verified by an ophthalmologist of University of Illinois at Chicago Eye Center.

In the older group with compromised vision, only two subjects had a binocular visual acuity worse than 20/80, one subject had a visual acuity of 20/100, and the other had a visual acuity of 20/200. Likewise, in the younger group with compromised vision, only two subjects had binocular visual acuities that were worse than 20/ 80, one had 20/100, and another had 20/200 visual acuity.

Test Measures

Visual field measures. The visual fields of all subjects with compromised vision were measured with a Goldmann perimeter using the II-4-e target. We used two indexes of binocular visual fields: the total extent of the visual field in degrees along the horizontal meridian to the II-4-e target, and the extent of central visual field loss in degrees to the Goldmann II-4-e target. The total extent of the visual field in degrees along the horizontal meridian to the II-4-e target ranged from 10 deg to 150 deg across the entire patient sample, with a mean [+ or -] SD of 114.0 deg [+ or -] 51.0 deg. The extent of central visual field loss along the horizontal meridian to the II-4-e Goldmann target ranged from 0 deg to 30 deg across the entire patient sample, with a mean [+ or -] SD of 3.00 deg [+ or -] 5.95 deg.

Driving assessment system (interactive simulator). All subjects underwent testing on an interactive simulator, which has been described in detail in Szlyk et al. (1992, 1993). Subjects were instructed to operate the simulator as they would normally drive their own car and to obey all traffic signs and signals along the roadway. Even though all subjects reported feeling fully comfortable with the equipment after only 5 min of practice, there may have been some age-related differences in the speed of learning the task. Therefore, all subjects were given 15 min of practice. Data were collected for subjects' responses during an 8-min session of driving the test course.

Simulator performance indexes. The simulator indexes analyzed included the following: (1) mean speed (in miles per hour, or mph); (2) braking response time to a stop sign along the simulator course (defined as the elapsed time between the objective presentation of a stop sign and the initiation of a brake pedal response); (3) number of lane boundary crossings (defined as any tire crossing over any of the lane's boundaries); (4) variability of braking pressure (in arbitrary units; calculated as the standard deviation of brake pedal pressure during the session); and (5) number of simulator accidents.

Eye and head movement analysis. A camera (Hitachi model AP-130U, Hitachi Denshi America, Ltd., Woodbury, NY) mounted above the simulator display captured the image of each subject's face while driving. The image was stored on tape via a video cassette recorder (U-Matic model NV-9200, Panasonic Corp., Secaucus, NJ). The pixel location of the center of the pupil of each eye, the location of the center of a cross on a head marker, and the time code were recorded. The variability of eye and head movements in the sequence of images was the index of movement.

Real-world accident measures. We analyzed information obtained from the subjects' self-reports about past accidents within a previous 5-yr period. We had found in our previous studies (Szlyk et al., 1992, 1993) that self-reported accidents were correlated with state records of accidents, and that we could obtain information about more accidents through self-reports, especially with subjects who have compromised vision. State of Illinois records of automobile accidents and convictions for traffic violations (e.g., failure to stop at a stop sign) were obtained for 66 subjects with their written permission.

Assessment of risk-taking in driving. All subjects were asked to respond "true" or "false" to questions intended to assess their perceived level of risk-taking during real-world driving. A complete list of the questions may be found in Szlyk (1993). The summed total of scores given for each question was defined as the subjective risk analysis score. Scores ranged from a minimum of 0 (least risk) to a maximum of 9 (most risk).

RESULTS

Performance on the Interactive Driving Simulator

The driving simulator proved to be a sensitive assay for age-related changes in driving-related skills. A series of two-way analyses of variance (ANOVAs), Age Group (younger, older) x Visual Status (normally sighted, visually compromised), were performed to test the relative effects of age and compromised vision.

There was a systematic decrease in speed as a function of increasing age for both the normally sighted group and for the visually compromised group. Younger, normally sighted subjects had an average speed of 29.2 [+ or -] 7.6 mph, and the older, normally sighted subjects had an average of 24.2 [+ or -] 5.9 mph. The average speed for the group of younger, visually compromised subjects was 28.0 [+ or -] 7.7 mph; for the group of older, visually compromised subjects, it was 21.4 [+ or -] 7.6 mph. A two-way ANOVA found a significant main effect for age, F(1,99) = 15.4, p [less than] 0.001, but not for visual status, F(1,99) = 1.6, p = 0.22. There was no significant interaction between age and visual status, F(1,99) = 0.27, p = 0.61.

The mean braking response time for the younger, normally sighted group was 5.3 [+ or -] 2.0 s, and for the older, normally sighted group, it was 13.36 [+ or -] 12.4 s. For the visually compromised groups, braking response time averaged 7.3 [+ or -] 2.3 s for the younger group and 9.27 [+ or -] 7.0 s for the older group. A two-way ANOVA found a significant main effect of age, F(1,92) = 12.2, p [less than] 0.001, but not for visual status, F(1,92) = 0.06, p = 0.82. There was a significant interaction between age and visual status, F(1,92) = 5.2, p = 0.03. Because the oldest normally sighted subject (age 83) had a braking response time that was much greater than that of the rest of the older, normally sighted subjects, we reanalyzed the data omitting this point. The mean for the older, normally sighted subjects without this point was 10.44 [+ or -] 5.1 s; a two-way ANOVA once again revealed a similar pattern. There was a significant main effect for age, F(1,91) = 13.7, p [less than] 0.001, but not for visual status, F(1,91) = 0.89, p = 0.35. However, there was no significant interaction between age and visual status F(1,91) = 3.06, p = 0.08.

The mean number of lane boundary crossings for the younger, normally sighted group was 0.41 [+ or -] 0.92, and for the older normally sighted group it was 5.47 [+ or -] 11.9. For the visually compromised groups, lane boundary crossings averaged 1.14 [+ or -] 1.9 for the younger group and 13.6 [+ or -] 32.4 for the older group. There was a significant main effect for age, F(1,99) = 8.0, p [less than] 0.006, but not for visual status, F(1,99) = 1.3, p = 0.25. However, there was no significant interaction between age and visual status, F(1,99) = 1.3, p = 0.25.

The mean variability of braking pressure for the younger, normally sighted groups was 21.7 [+ or -] 5.9 units of pressure, and for the older normally sighted group it was 24.6 [+ or -] 10.6 units of pressure. For the visually compromised groups, variability averaged 23.4 [+ or -] 7.9 units for the younger group and 24.0 [+ or -] 11.6 units for the older group. There was no significant main effect either for age, F(1,99) = 0.77, p = 0.38, or for visual status, F(1,99) = 0.91, p = 0.66. In addition, there was no significant interaction between age and visual status, F(1,99) = 0.4, p = 0.53.

None of the younger, normally sighted group had an accident on the simulator course, whereas the mean number of accidents for the [TABULAR DATA FOR TABLE 1 OMITTED] older, normally sighted group was 0.89 [+ or -] 2.0. For the visually compromised groups, the number of accidents averaged 0.17 [+ or -] 0.61 for the younger group and 0.43 [+ or -] 0.81 for the older group. A two-way ANOVA found a significant main effect for age, F(1,98) = 7.1, p [less than] 0.009, but not for visual status, F(1,98) = 0.14, p = 0.71. There was no significant interaction between age and visual status, F(1,98) = 2.3, p = 0.13.

Relationships between Visual Status Indexes and Simulator Indexes

The correlations between the visual status indexes (degrees of central field loss measured with the Goldmann II-4-e target, the extent of binocular field along the horizontal meridian in degrees to the II-4-e target, decimal visual acuity, and age) and the simulator indexes are presented in Table 1. Age was significantly correlated with four of the five indexes examined (all but variability of braking pressure), and central field loss was also significantly correlated with speed. Visual acuity, however, was not significantly correlated with any simulator indexes.

Real-World Driving

Because there were no statistical differences among the groups in the number of miles driven per year (for age, F(1,103) = 0.87, p = 0.35; for visual status, F(1,103) = 0.39, p = 0.34), we analyzed the number of self-reported accidents, state-recorded accidents, and state-recorded traffic violations.

The younger, normally sighted group averaged 0.63 [+ or -] 0.8 self-reported accidents, and the older, normally sighted group had an average of 0.42 [+ or -] 0.6. The averaged number of accidents for the group of younger, visually compromised subjects was 0.73 [+ or -] 0.8; for the group of older, visually compromised subjects it was 0.48 [+ or -] 0.7. There was no significant main effect either for age, F(1,102) = 2.1, p = 0.15, or for visual status, F(1,102) = 0.28, p = 0.60. In addition, there was no significant interaction between age and visual status, F(1,102) = 0.02, p = 0.89.

Again, the younger groups had higher averaged numbers of accidents than the older groups. The younger, normally sighted group averaged 0.39 [+ or -] 0.5 state-recorded accidents, and the older, normally sighted group had an average of 0.27 [+ or -] 0.4. The averaged number of accidents for the group of younger, visually compromised subjects was 0.40 [+ or -] 0.6; for the group of older, visually compromised subjects, it was 0.18 [+ or -] 0.5. There was no significant main effect either for age, F(1,62) = 1.6, p = 0.20, or for visual status, F(1,62) = 0.06, p = 0.80. In addition, there was no significant interaction between age and visual status, F(1,62) = 0.14, p = 0.71.

For the index state-recorded convictions, the younger, normally sighted group averaged fewer convictions (0.28 [+ or -] 0.4) than the older, normally sighted group (1.91 [+ or -] 3.0). For the visually compromised groups, the averaged number of convictions for the younger subjects was also lower (0.35 [+ or -] 0.6) than for the older visually compromised subjects (0.65 [+ or -] 2.2). There was no significant main effect either for age, F(1,62) = 1.4, p = 0.20, or for visual status, F(1,62) = 0.93, p = 0.34. However, there was a significant interaction between age and visual status, F(1,62) = 4.6, p = 0.04.

Compensation Factors

There was a discrepancy between the negative effects of age on simulator performance and the lack of significant effects of age on real-world driving. For the simulator, both age and visual status (for visual fields, see Table 1) were significantly correlated with simulator performance indexes. We examined measures of compensation such as speed, risk-taking while driving in the real world, and eye movement while driving the simulator to determine if they could account for our results.

Speed. A significant main effect for age on the speed index was reported, as described earlier. To determine how speed on the simulator relates to real-world driving, we performed a Speed x Self-Reported Accidents ANOVA. A significant main effect for accidents showed that drivers who had more accidents had higher average speeds, F(3,99) = 3.00, p [less than] 0.05. Subjects (averaged across group) who had no accidents had an average speed of 24.9 [+ or -] 7.5 mph; those having one accident had an average speed of 27.7 [+ or -] 8.4 mph; those having two accidents had an average speed of 26.9 [+ or -] 6.4 mph; and those having three or more accidents had an average speed of 34.8 [+ or -] 5.3 mph.

On-road risk-taking. The younger, normally sighted group had a mean risk-taking score of 5.22 [+ or -] 1.5; the older, normally sighted group had an average score of 4.1 [+ or -] 1.6 (a lower score indicates a lower level of risk-taking). For the younger, visually compromised group, the average score was 3.4 [+ or -] 1.8; for the older, visually compromised, it was 3.3 [+ or -] 1.7. A two-way ANOVA found a marginally significant main effect for age, F(1,97) = 3.6, p = 0.06, and a significant effect for visual status, F(1,97) = 13.1, p [less than] 0.001. There was no significant interaction between age and visual status, F(1,97) = 1.6, p = 0.21. Interestingly, both groups of visually compromised subjects reported taking less risk than did the older, normally sighted group.

Eye movements. For vertical eye movements, the mean for the younger, normally sighted group was 19.23 [+ or -] 7.9; for the older, normally sighted group, it was 52.7 [+ or -] 51.2. For the visually compromised groups, vertical eye movements averaged 18.6 [+ or -] 10.5 for the younger group and 38.2 [+ or -] 40.7 for the older group. A two-way ANOVA found a significant main effect for age, F(1,65) = 6.5, p [less than] 0.003, but no significant effect for visual status, F(1,65) = 0.64, p = 0.43. There was no significant interaction between age and visual status, F(1,65) = 0.89, p = 0.35.

Horizontal eye movements also increased as a function of age. The mean for the younger, normally sighted group was 24.3 [+ or -] 12.2 and 41.3 [+ or -] 32.3 for the older, normally sighted group. For the visually compromised groups, horizontal eye movements averaged 26.9 [+ or -] 13.0 for the younger group and 31.7 [+ or -] 16.1 for the older group. A two-way ANOVA found a significant main effect for age, F(1,65) = 5.09, p [less than] 0.03, but no significant effect for visual status, F(1,65) = 0.19, p = 0.66. There was no significant interaction between age and visual status, F(1,65) = 1.8, p = 0.18.

Relationship of Age, Vision, Simulator Performance, and Risk-Taking to Accidents

We performed a series of logistic regression analyses intended to determine the combination of indexes that provided the most significant prediction of real-world driving (i.e., self-reported accidents and state-recorded convictions for traffic violations). Because the self-reported accidents and state-recorded accidents were significantly correlated, r(65) = 0.25, p [less than] 0.05, and because we were able to obtain information from more subjects through self-report (Szlyk et al., 1993), the following regression analyses used the self-reported accident results.

A combination of vision group status, central field loss, simulator accidents, and speed produced a marginally predictive model of the dichotomous variable accident group membership (0 for those with no accidents, 1 for those with one or more accidents) with a [[Chi].sup.2] (4) = 8.0, p [less than] 0.09; removing speed from the regression produced a [[Chi].sup.2] (3) = 7.0, p [less than] 0.09; and removing both speed and simulator accidents produced a statistically significant [[Chi].sup.2] (2) = 6.0, p [less than] 0.05. A combination of braking response time and risk-taking was marginally predictive of the dichotomous variable state conviction group membership ([[Chi].sup.2] (2) = 5.24, p [less than] 0.09); however, risk-taking alone significantly predicted membership ([[Chi].sup.2] (1) = 4.77, p [less than] 0.05).

CONCLUSIONS

The major finding of this study was that there were significant differences between the younger and older groups on driving-related skills as measured with our interactive driving simulator. The older subjects, whether visually compromised or not, performed more poorly on the simulator. Although our data show that visuocognitive/motor skills, as measured with the simulator, decline with age and are affected by compromised vision, these deficits do not necessarily place these drivers at a greater risk for accidents in the real world. We have found that the older drivers (50 yrs and above) and all drivers with compromised vision have reduced risk-taking (e.g., not passing or not changing lanes in traffic). In addition, all older drivers, regardless of vision status, have increased eye movements and drive more slowly compared with younger drivers. These behaviors may allow these drivers to compensate for deficiencies in their visuocognitive/motor skills and ultimately reduce their accident risk.

Our results also show that good driving skills are only one factor that contributes to successful real-world driving. Interestingly, the regression analyses for a younger, normally sighted group published previously (Szlyk et al., 1993) showed that models including only simulator indexes (braking response time to a stop sign, variability of braking pressure, and braking response time to a traffic signal) were predictive of accident involvement. However, we also found in these previous reports that in the visually compromised younger groups, other factors were more closely related to accident involvement than the simulator indexes, including increased risk-taking (e.g., driving at night; see Szlyk et al., 1993) and extent of visual field loss (Szlyk et al., 1992).

In essence, those individuals (the younger, normally sighted group) who did not restrict their driving behavior showed the purest relationship between raw driving-related skills as measured with the driving simulator and accident involvement (Szlyk et al., 1993). When all the groups are combined for analysis in the present study, the importance of the vision variables and behavioral indexes (i.e., risk-taking) is highlighted, muddying the relationship between simulator performance and real-world driving. When including older drivers, it appears that losses in driving-related skills (i.e., more lane boundary crossings, prolonged braking response times) can be compensated for by such factors as decreased risk-taking, decreased speed, and increased eye movements. However, our data also show that compromised vision - specifically visual field loss - is not as easily compensated for as these age-related losses of driving skills, given that compromised vision (visual status group and degrees of central visual field loss) predicts accident involvement.

In conclusion, age alone should not be a criterion for restricted licensure or for more frequent road tests. Vision should be assessed more thoroughly by state licensing facilities, and individuals of any age with compromised vision should be monitored for progressive changes in their visual status and for how this status might affect their on-road performance. In addition, we suggest that level of risk-taking and compensation strategies should be considered when granting licenses to drive. These factors may be measured through direct observation and questionnaires to predict whether or not drivers present themselves as high-risk drivers on the road.

ACKNOWLEDGMENTS

We thank Andrew M. Glenn FRCS, FCOphth., for verifying the diagnoses of the patients with compromised vision. This research was supported by a grant from the AARP Andrus Foundation; a grant from the U.S. Department of Veterans Affairs; grants from the National RP Foundation Fighting Blindness, Inc.; unrestricted grants from Research to Prevent Blindness, Inc.; and grant EY01792 from the National Eye Institute.

REFERENCES

Ball, K., Owsley, C., Sloane, M., Roenker, D., and Bruni, J. (1993). Visual attention problems as a predictor of vehicle accidents among older drivers. Investigative Ophthalmology and Visual Science, 34, 3110-3123.

Baltes, P. B., Cornelius, S. W., Spiro, A., Nesselroade, J. R., and Willis, S. L. (1980). Integration versus differentiation of fluid/crystallized intelligence in old age. Developmental Psychology, 16, 625-635.

Coyne, A. C., Fiens, L. C., Powell, A. L., and Joslin, B. L. (1990, March). The relationship between driving and dementia. Paper presented at the Third Cognitive Aging Conference, Atlanta, GA.

Lucas-Blaustein, M. J., Filipp, L., Dungan, C., and Tune, L. (1988). Driving in patients with dementia. Journal of the American Geriatric Society, 36, 1087-1091.

Odenheimer, G., Albert, M., Jette, A., Beaudet, M., Grande, L., and Minaker, K. (1991). Cognitive skills are related to driving performance in the elderly. Journal of the American Geriatric Society Supplement, 39, A9.

Sekuler, R., Kline, D., and Dismukes, K. (1982). Aging and human visual function. New York: Alan R. Liss.

Szlyk, J. P., Alexander, K. R., Severing, K., and Fishman, G. A. (1992). Assessment of driving performance in patients with retinitis pigmentosa. Archives of Ophthalmology, 110, 1709-1713.

Szlyk, J. P., Fishman, G.A., Severing, K., Alexander, K. R., and Viana, M. (1993). Evaluation of driving performance in patients with juvenile macular dystrophies. Archives of Ophthalmology, 111, 207-212.

JANET P. SZLYK, WILLIAM SEIPLE, and MARLOS VIANA (Relative Effects of Age and Compromised Vision on Driving Performance)

JANET P. SZLYK is an assistant professor in the Department of Ophthalmology and Visual Sciences at the University of Illinois at Chicago Eye Center, and a research health scientist at the West Side Veterans Administration Medical Center. She received her B.A. degree from Brown University in 1983 and her Ph.D. degree in experimental psychology from Fordham University in 1988, where she specialized in perceptual psychology. She conducted postdoctoral research at the Vision Research Laboratory of the New York Lighthouse, and also completed a fellowship in the area of psychophysics applied to the study of inherited retinal disease at the UIC Eye Center. She is currently director of the Laboratory for the Assessment of Functional Vision at the UIC Eye Center. Her research focuses on determining how eye disease affects a patient's ability to perform tasks such as driving.

WILLIAM SEIPLE is a research professor in the Department of Ophthalmology at New York University Medical Center. He received a B.A. from Albright College, an M.A. from the University of North Carolina, and a Ph.D. in zoology from the University of Illinois. His current research efforts include electrophysiological assessment of cone photoreceptor function in humans, examination of the effects of alterations of sensory input on higher-order visual processing, and ecology of decapod crustaceans.

MARLOS VIANA is an assistant professor of biostatistics in the Department of Ophthalmology and Visual Sciences and in the School of Public Health at the University of Illinois at Chicago College of Medicine. He received a Ph.D. from Stanford University in 1978 and currently directs the Laboratory for Statistical Assessment of Diagnostic Tests. His research interests include multivariate statistical analysis, combined inferences, and Bayesian inference.
COPYRIGHT 1995 Human Factors and Ergonomics Society
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1995 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Szlyk, Janet P.; Seiple, William; Viana, Marlos
Publication:Human Factors
Date:Jun 1, 1995
Words:4327
Previous Article:Information-processing slope scores can be reliable!
Next Article:Taxonomy of scheduling systems as a basis for the study of strategic behavior.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters