Table of Contents
Next Section


II.A. VISION


1. Static Acuity

(a) MultiCAD
(b) Rosenbaum Card
(c) Snellen E Chart
(d) Snellen E (Computerized Presentation)
(e) Snellen Letter Chart (Modified)

2. Dynamic Acuity

(a) MultiCAD
(b) Snellen E (Computerized Presentation)

3. Static Contrast Sensitivity

(a) MultiCAD
(b) Pelli-Robson Test of Low Contrast Acuity
(c) Smith-Kettlewell Low Luminance Card (SKILL)
(d) Vistech Contrast Sensitivity Gratings/Optec 1000

4. Static Contrast Sensitivty/Glare

(a) Berkeley Glare Tester

5. Dynamic Contrast Sensitivity

(a) MultiCAD

6. Peripheral Visual Fields

(a) Goldman Perimeter
(b) Manually Operated Perimeter

7. Eye Disease

(a) Cataracts
(b) Diabetic Retinopathy
(c) Glaucoma

8. Multiple Visual Capabilities

(a) Keystone Telebinocular Testing Device
(b) Sight Screener II


FUNCTIONAL TEST

SUBJECTS

PROCEDURE/TEST DESCRIPTION

WHERE APPLIED

FINDINGS

RESEARCHER(S)

VISION

Static Acuity: MultiCAD

82 "referred" subjects aged 60-91 (26 of which were identified as probably being cognitively impaired to some degree). The drivers were referred to the DMV for reexamination due to a medical condition (by physician, optometrist, ophthalmologist), a series of licensing test failures, a flagrant driving error (police referral), or some other indicator of driving impairment.

This test used MultiCAD to measure drivers' ability to resolve fine detail on a stationary target under high contrast conditions. The subject was shown a driver's eye view of travel along a suburban arterial, approaching and then stopping at an intersection with a traffic signal in plain view. The image centered and then zoomed on the signal until it filled the screen, while the subject was instructed to use the 3-button response pad to identify which face on the (conventional, 3-face) signal looked different than the other two. Instead of solid red, yellow, and green circles, however, the signal faces contained acuity test stimuli. Square wave gratings with vertical bars were used, such that one signal face contained a high contrast test stimulus (90% contrast) and the other two faces showed a uniform luminance (without bars). The ability to discriminate which two signal faces are were"blank" versus which one contained the vertical bars defined the subject's static acuity level. Three levels of testing were conducted--20/40 (15 cycles per degree), 20/80 (7.5 cycles per degree), and 20/200 (3 cycles per degree)--with a pass/fail score assigned at each level. A passing score was defined as at least 2 correct responses out of the 3 presentations for each level tested. Mean response time was also calculated for correct responses at each level. Three replications of each measurement were performed.

Scoring was also conducted on a "gross" level across all stimulus characteristics. Average response time and average error score were calculated across all 9 trials.

Multiple linear regressions were conducted to arrive at the best linear combination of variables for predicting performance (weighted error score) on a standard DMV road test (see On-road Performance Measures of Driving Safety: California MDPE at the end of this Compendium), and comparisons were made between cognitively impaired and cognitively non-impaired referral drivers to determine whether there were differences in performance.

California DMV Field Office

* Correlation between gross static acuity errors and weighted errors on driving test was not significant (r=.0983, p< .395)

*Correlation between gross static acuity time and weighted error score on road test was significant (r=.3519, p< .002)

* Correlations between static acuity score (20/20, 20/80, and 20/200) and weighted errors on driving test were not significant.

* Correlations between static acuity time at each level of acuity and weighted error scores on driving exam were as follows:

20/40 time: r=.3395 (p<.004)

20/80 time: r=.4230 (p<.000)

20/200 time: r=.1970 (p<.090)

*Neither gross or precise scoring of static acuity accuracy or time differentiated between cognitively impaired and cognitively unimpaired referral subjects.

*Using the precise MultiCAD measures a multiple linear regression model using knowledge test score, Auto Trails time, Doron Cue Recognition 2 score, MultiCAD Static Contrast Sensitivity time with the high contrast 20/80 target, and MultiCAD Static Acuity time for correct responses at 20/80 accounted for 56.4% of the variance in performance on the road test (weighted road test error score).

*Using gross MultiCAD measures, a model including knowledge test score, Auto Trails time, Doron Cue Recognition 2 score, MultiCAD static acuity time, and MultiCAD static contrast sensitivity time accounted for 47.7% of the variance in performance on the road test (weighted road test error score).

Janke & Eberhard (1998)

Staplin, Gish, Decina, Lococo, and McKnight (in press)

VISION

Static Acuity:

Rosenbaum Card

283 community-dwelling individuals age 72 to 92 (mean age = 77.8) from the Project Safety cohort living in New Haven, CT who drove between 1990 and 1991. 57% were males.

Rosenbaum Card was used to measure corrected static near visual acuity. Performance was measured as 20/40 or better vs worse than 20/40

The outcome variable was self-reported involvement in automobile crashes, moving violations, or being stopped by police in the year following administration of the test battery.

New Haven, CT. Subjects were interviewed and given the assessments in their homes by a trained research nurse.

The occurrence of adverse events did not substantially differ between persons with better than 20/40 (13% had adverse events) or worse than 20/40 (14% had adverse events) near static visual acuity.

Marottoli, Cooney, Wagner, Doucette, and Tinetti (1994)

VISION

Static Acuity:

*Snellen E Chart

*Snellen E (Computerized Presentation)

Matched pair case-control study, with close (1 year) age matching conducted in Sweden

* 37 drivers age 65+ (mean age 75.5) with temporarily-suspended licenses due to crashes (23 drivers) or other moving violations (14 drivers). Moving violations were: speeding (2), running stop sign (4), running red light (4) run off the road (4). Mean distance driven/yr = 12000 km; # males = 30, # females = 7

* 37 matched controls age 65+ (mean age 74.8) with no license suspensions within the past 5 yrs; mean # miles driven = 9200 km; # males = 30, # females = 7

Static Visual Acuity was measured with a standard letter chart (SVA:L) at a distance of 4 m, and was measured as the smallest row of 10 letters read binocularly with no errors. Static Visual Acuity was also measured using Snellen E=s (SVA:E), using a (PC) computerized system including a slide projector which projected the optotypes on a white screen using first-surface mirrors. Luminance of E=s was 85 cd/m2, background was 195 cd/m2, and contrast was 0.39. Snellen E=s were shown in any of 4 possible directions (up right, up left, down right, down left), and subject responded by pressing corresponding orientation printed on a button on the response box. Each E was shown for 6 s. A logarithmic scale was used with the different object sizes (equivalent to Snellen acuity) of 0.10 (10 min of arc), 0.13, 0.16, 0.20, 0.25, 0.32, 0.50, 0.63, 0.79, 1.0, 1.3, and 1.6. SVA:E was measured with 3 randomly chosen directions on each size of the target, starting at 0.10, and increasing stepwise. A pass was a correct response on all 3 readings of a particular size. Two trials on each acuity size were allowed. The SVA:E was measured with non-moving optotypes (mirror remained still).

Subjects were also given the Trail Making Test (Part A), the Mini Mental Status Examination, and a cube copying task .

________________________________________________

FINDINGS (Cont=d)

* Correlation between drop in acuity from SVA:L to SVA:E and scores on the cube copy test were significant (p<0.002).

* Using a decrease in acuity of more than 2 steps from SVA:L to SVA:E as a cut-off limit, the ability to detect drivers with crashes had a low sensitivity (21%) but a specificity of 98%.

* Drivers with conspicuously low results on the SVA:E were cognitively impaired, which implies, according to the authors, that acuity testing with Snellen E=s (or Landolt C=s perhaps), might also be a simple test for identifying persons with moderate cognitive impairment and subsequent increased crash risk.

Hospital clinic

(Unit of Traffic Medicine, Section of Geriatric Medicine, Department of Clinical Neuroscience & Family Medicine, Karolinska Institutet, Stockholm, Sweden)

* No differences between cases and controls with respect to Static Visual Acuity measured with the standard letter chart (SVA:L). Mean SVA:L for case group = 0.79, and for controls = 1.0. Also, no differences between cases with crashes and their controls or cases with violations and their controls.

* No significant differences between all cases (across crashes and violations) and controls in visual acuity measured with automated Snellen E test (SVA:E).

* 3 drivers with crashes performed well on SVA:L (acuities of 0.60, 0.80, and 1.0) but performed very poorly (_0.10) on SVA:E. These drivers (ages 72, 76, 78) showed cognitive impairment on the Mini Mental Status Exam (scores of 24, 27, and 20), and clinical dementia ratings (CDR) of questionable dementia (2 drivers) and mild (1 driver) dementia. None of them were able to copy a cube design. Psychomotor speed (Trails A) was low (79, 90, and 388 s) compared to controls. One fulfilled DSM IV criteria for dementia, the other 2 were suspected of having a dementing disease.

* Visual performance of drivers with crashes measured with SVA:E was significantly lower than that of controls (p=0.017), but removal of above 3 drivers reduced the difference to a tendency (p=0.087). In violation group, there was no difference in SVA:E between cases and controls).

Johansson, Seideman, Kristoffersson, Lundberg, Lennerstrand, Hedin, and Viitanen (submitted)

Johansson (1997)

VISION

Static Acuity:

Snellen Letter Chart (Modified)

* 102 "referred" subjects aged 60-91 (34 of which were identified as probably being cognitively impaired to some degree). 47% of the noncognitively impaired referred drivers had visual impairment noted on their record, and 24% of the cognitively impaired had a visual disability noted). The drivers were referred to the DMV for reexamination due to a medical condition (by physician, optometrist, ophthalmologist), a series of licensing test failures, a flagrant driving error (police referral), or some other indicator of driving impairment.

* 33 paid "volunteers" aged 56-85, recruited through signs posted at study site or word of mouth.

Chart contained 5 lines of letters at 20/40 size, viewed at a distance of 6 m. Snellen errors and Snellen failures were measured.

Three tiers of analyses were conducted in this research: (1) logistic regressions to determine what combination of tests, observations, or survey variables, with what weightings, would best predict whether a subject was a volunteer or referral; (2) multiple linear regressions were conducted to arrive at the best linear combination of variables for predicting performance on road tests; and (3) comparisons were made between cognitively impaired and cognitively non-impaired referral drivers to determine whether there were differences in performance on nondriving tests and driving tests.

(see On-road Performance Measures of Driving Safety: California MDPE at the end of this Compendium)

California DMV Field Office

The referral group performed significantly worse than the volunteer group. Average Snellen error score for referrals = 2.42, for volunteers = 0.09. Average Snellen failure (0=pass, 1=fail) for referrals = 0.57, for volunteers = 0.03.

There was no significant difference in performance on Snellen errors or Snellen failures as a function of cognitive impairment

Correlations between Snellen performance and weighted error score on the test performance were significant when combining referrals and volunteers (n=135). Correlations between weighted error score and Snellen errors = .3360 (p<.000), between weighted error score on road test and Snellen failure = .3553 (p<.000).

Correlations between Snellen performance and weighted error score on drive test were lower (.1704, and .1846) for Snellen errors and failure respectively and not significant, when considering performance of the referrals only

This variable was also significantly correlated with age: correlation of age with snellen errors = .402; age and snellen failure = .401)

Janke & Eberhard (1998)

VISION

Dynamic Acuity:

MultiCAD

82 "referred" subjects aged 60-91 (26 of which were identified as probably being cognitively impaired to some degree). The drivers were referred to the DMV for reexamination due to a medical condition (by physician, optometrist, ophthalmologist), a series of licensing test failures, a flagrant driving error (police referral), or some other indicator of driving impairment.

This test used MultiCAD to measure drivers' visual acuity, for a target that was moving relative to the observer, under high contrast conditions. The subject was shown a driver's eye view of travel along a suburban arterial, approaching and then stopping at an intersection with a traffic signal in plain view. The image centered and then zoomed on the signal until it filled the screen, while the subject was instructed to use the 3-button response pad to identify which face on the (conventional, 3-face) signal looked different than the other two. Instead of solid red, yellow, and green circles, however, the signal faces contained acuity test stimuli. Square wave gratings with vertical bars were used, such that one signal face contained a high contrast test stimulus (90% contrast) and the other two faces showed a uniform luminance (without bars). The ability to discriminate which two signal faces are were"blank" versus which one contained the vertical bars defined the subject's dynamic acuity level. The rate of movement across the screen (12 degrees per second) corresponded to a driver trying to read a street sign posted at roadside while passing by at a moderate (25-40 mi/h) rate of speed. Three levels of testing were conducted--20/40 (15 cycles per degree), 20/80 (7.5 cycles per degree), and 20/200 (3 cycles per degree)--with a pass/fail score assigned at each level. A passing score was defined as at least 2 correct responses out of the 3 presentations for each level tested. Mean response time was also calculated for correct responses at each level. Three replications of each measurement were performed.

Scoring was also conducted on a "gross" level across all stimulus characteristics. Average response time and average error score were calculated across all 9 trials.

Multiple linear regressions were conducted to arrive at the best linear combination of variables for predicting performance (weighted error score) on a standard DMV road test, (see On-road Performance Measures of Driving Safety: California MDPE at the end of this Compendium), and comparisons were made between cognitively impaired and cognitively non-impaired referral drivers to determine whether there were differences in performance.

California DMV Field Office

*Correlation between gross dynamic acuity errors and weighted errors on driving test was significant (r=.2346, p<.040).

*Correlation between gross dynamic acuity response time and weighted errors on driving test was significant (r=.3346, p<.003).

*Correlations between dynamic acuity score (20/20, 20/80, and 20/200) and weighted errors on driving test were not significant.

*Correlations between dynamic acuity time at each level of acuity and weighted error scores on driving exam were as follows:

20/40 time: r=.3092 (p<.010)

20/80 time: r=.3256 (p<.005)

20/200 time: r=.3297 (p<.004)

*Neither gross or precise scoring of dynamic acuity accuracy or time differentiated between cognitively impaired and cognitively unimpaired referral subjects.

 

Janke & Eberhard (1998)

Staplin, Gish, Decina, Lococo, and McKnight (in press)

VISION

Dynamic Acuity:

Snellen E (Computerized Presentation)

Matched pair case-control study, with close (1 year) age matching conducted in Sweden

* 37 drivers age 65+ (mean age 75.5) with temporarily-suspended licenses due to crashes (23 drivers) or other moving violations (14 drivers). Moving violations were: speeding (2), running stop sign (4), running red light (4) run off the road (4). Mean distance driven/yr = 12000 km; # males = 30, # females = 7

* 37 matched controls age 65+ (mean age 74.8) with no license suspensions within the past 5 yrs; mean # miles driven = 9200 km; # males = 30, # females = 7

Dynamic visual acuity (DVA) was measured using the apparatus described for Johansson et al. (submitted) for Static Visual Acuity (SVA), using a (PC) computerized system including a slide projector which projected the optotypes on a white screen using first-surface mirrors. Luminance of E=s was 85 cd/m2, background was 195 cd/m2, and contrast was 0.39. Snellen E=s were shown in any of 4 possible directions (up right, up left, down right, down left), and subject responded by pressing corresponding orientation printed on a button on the response box. Each E was shown for 6 s. A logarithmic scale was used with the different object sizes (equivalent to Snellen acuity) of 0.10 (10 min of arc), 0.13, 0.16, 0.20, 0.25, 0.32, 0.50, 0.63, 0.79, 1.0, 1.3, and 1.6. DVA was measured with 3 randomly chosen directions on each size of the target, starting with 0.10 and increasing stepwise. A pass was a correct response on all 3 readings of a particular size. Two trials on each acuity size were allowed. Using a system of mirrors, where one was rotating and angled in relation to the optic axis, the Snellen= s E described a circular movement on the screen; however, the orientation of the E was not influenced by the circular movement. The test was repeated with 3 different angle velocities: 10 /s, 30 /s, and 50/s, performed after each other. The circular movement used a diameter of 0.8 m and an observation distance of 3 m. The dependent measures included dynamic visual acuity and response time.

Subjects were also given the Trail Making Test (Part A), the Mini Mental Status Examination, and a cube copying task .

 

Hospital clinic

(Unit of Traffic Medicine, Section of Geriatric Medicine, Department of Clinical neuroscience & Family Medicine, Karolinska Institutet, Stockholm, Sweden)

* There was a significant difference in DVA for the case drivers as a group compared to the controls at an angular velocity of 30/s. This difference was eliminated when the 3 case drivers with dementia were eliminated (see SVA:E study by Johansson et al.). There was no significant difference at 10/s or 50/s.

* Comparing only case drivers with crashes to their matched controls showed significantly lower DVA performance among drivers with crashes at 30/s, even when the 3 drivers with dementia were eliminated.

* Comparing only drivers with violations to their matched controls revealed no difference in DVA performance at any velocity.

* Drivers with crashes took significantly longer to respond to the Snellen E=s than drivers with only violations and drivers in the control group.

* Drivers with conspicuously low results on the DVA were cognitively impaired, which implies, according to the authors, that acuity testing with Snellen E=s (or Landolt C=s perhaps), might also be a simple test for identifying persons with moderate cognitive impairment and subsequent increased crash risk.

Johansson, Seideman, Kristoffersson, Lundberg, Lennerstrand, Hedin, and Viitanen (submitted)

Johansson (1997)

VISION

Static Contrast Sensitivity:

MultiCAD

82 "referred" subjects aged 60-91 (26 of which were identified as probably being cognitively impaired to some degree). The drivers were referred to the DMV for reexamination due to a medical condition (by physician, optometrist, ophthalmologist), a series of licensing test failures, a flagrant driving error (police referral), or some other indicator of driving impairment.

This test used MultiCAD to measure drivers' sensitivity to differences in brightness, as required to detect edges between adjacent lighter and darker areas in the roadway environment. The subject was asked to use the 3-button response pad to indicate which of three signal faces contained a test pattern. The traffic signal image remained stationary during this test. The test patterns were the same as used for the static acuity test for 20/40 (15 cycles per degree) and 20/80 (7.5 cycles per degree), and were presented at 2 contrast levels (medium contrast=20.6%; low contrast = 4.9%). Three replications of each measurement were performed with a pass/fail score assigned at each level. A passing score was defined as at least 2 correct responses out of the 3 presentations for each level tested. Mean response time was also calculated for correct responses at each level.

Scoring was also conducted on a "gross" level across all stimulus characteristics. Average response time and average error score were calculated across all 12 trials.

Multiple linear regressions were conducted to arrive at the best linear combination of variables for predicting performance (weighted error score) on a standard DMV road test, (see On-road Performance Measures of Driving Safety: California MDPE at the end of this Compendium), and comparisons were made between cognitively impaired and cognitively non-impaired referral drivers to determine whether there were differences in performance

California DMV Field Office

*Neither gross static contrast sensitivity errors nor response time was significantly correlated with on-road weighted error score

*Static contrast sensitivity response time for the high contrast 20/80 target was significantly correlated with weighted error score on the driving test (r = .3884 p< .001)

*Neither gross or precise scoring of static contrast sensitivity accuracy or time differentiated between cognitively impaired and cognitively unimpaired referral subjects.

*Using the precise MultiCAD measures a multiple linear regression model using knowledge test score, Auto Trails time, Doron Cue Recognition 2 score, MultiCAD Static Contrast Sensitivity time with the high contrast 20/80 target, and MultiCAD Static Acuity time for correct responses at 20/80 accounted for 56.4% of the variance in performance on the road test (weighted road test error score).

*Using gross MultiCAD measures, a model including knowledge test score, Auto Trails time, Doron Cue Recognition 2 score, MultiCAD static acuity time, and MultiCAD static contrast sensitivity time accounted for 47.7% of the variance in performance on the road test (weighted road test error score).

Janke & Eberhard (1998)

Staplin, Gish, Decina, Lococo, and McKnight (in press)

VISION

Static Contrast Sensitivity:

Pelli-Robson Test

* 102 "referred" subjects aged 60-91 (34 of which were identified as probably being cognitively impaired to some degree). 47% of the noncognitively impaired referred drivers had visual impairment noted on their record, and 24% of the cognitively impaired had a visual disability noted). The drivers were referred to the DMV for reexamination due to a medical condition (by physician, optometrist, ophthalmologist), a series of licensing test failures, a flagrant driving error (police referral), or some other indicator of driving impairment.

* 33 paid "volunteers" aged 56-85, recruited through signs posted at study site or word of mouth.

48-letter test designed by Pelli, Robson, and Wilkins, 1988, of contrast sensitivity at one spatial frequency. The contrast between letters and background decreases as one moves down and toward the right of wall-mounted chart, viewed at dist. of 2 m under normal room illumination. The letters from left to right and from top to bottom progressively fade out as if they must be read in thicker and thicker fog. Letters (in groups of 3) range from 90% contrast (upper left) to 0.5% contrast (lower right). Testing requires no more than 3 minutes

Rather than standard scoring (number correct), errors were counted to conform with scoring of other tests.

Three tiers of analyses were conducted in this research: (1) logistic regressions to determine what combination of tests, observations, or survey variables, with what weightings, would best predict whether a subject was a volunteer or referral; (2) multiple linear regressions were conducted to arrive at the best linear combination of variables for predicting performance on road tests; and (3) comparisons were made between cognitively impaired and cognitively non-impaired referral drivers to determine whether there were differences in performance on nondriving tests and driving tests.

(see On-road Performance Measures of Driving Safety: California MDPE at the end of this Compendium).

California DMV Field Office

Referral group performed significantly worse than the volunteer group (correlation with group = .484).

Pelli-Robson error score for referrals = 15.87; for volunteers = 10.33

Note: this variable was also significantly correlated with age (correlation = .436).

There was no significant difference in performance on Pelli-Robson test as a function of cognitive impairment (mean Pelli-Robson errors for cognitively impaired group = 16.64; for cognitively unimpaired = 15.48)

Correlation between Pelli-Robson errors and weighted error score on road test was significant (r=.4009, p,.000) for combined referrals and volunteers (n=135). For referral group only (n=102), correlation between Pelli-Robson errors and weighted error score on road test was also significant (r= .2069, p,.044).

A model using number of observed problems plus Pelli-Robson errors with a cut-point of p=.80 of being a referral, gave specificity of 97 percent (32 of 33 volunteers classified correctly) with sensitivity of 71.4 percent (70 of 98 referrals in the model correctly classified).

Janke & Eberhard (1998)

VISION

Static Contrast Sensitivity:

Pelli-Robson Test

3,669 randomly-selected Class C license renewal applicants, licensed in California for at least 12 years, and unable to renew by mail. Four driver age groups were studied:

26-39, 40-51, 52-69, and 70+.

48-letter test designed by Pelli, Robson, and Wilkins, 1988, of contrast sensitivity at one spatial frequency. The contrast between letters and background decreases as one moves down and toward the right of wall-mounted chart, viewed at dist. of 2 m under normal room illumination. The letters from left to right and from top to bottom progressively fade out as if they must be read in thicker and thicker fog. Letters (in groups of 3) range from 90% contrast (upper left) to 0.5% contrast (lower right). Testing requires no more than 3 minutes

5 experimental vision tests were employed:

* Pelli-Robson Low-Contrast Acuity Test (measures loss in low contrast acuity; ability to see objects and borders)

* Smith-Kettlewell Low-Luminance Card (measures high-contrast near-acuity loss and low-contrast near-acuity loss)

* Berkeley Glare Tester (measures low-contrast near acuity loss, and low-contrast near-acuity loss in the presence of glare)

* Modified Synemen Perimeter (measures standard visual field-integrity loss and attentional visual field-integrity loss

* Visual Attention Analyzer (measures loss in UFOV, the area of the visual field in which useful information can be rapidly extracted from a complex visual display)

The dependent measure was the crash frequency during the previous 3-year period, extracted from the DMV database.

Drivers also completed a Driving Habits Survey measuring level of restriction (never, sometimes, often or always) for night driving, rain or fog, sunrise or sunset, driving alone, left turns, and heavy traffic.

California DMV Field Offices:

Carmichael

El Cerrito

Roseville

Study subjects rated test as face valid (clear instructions, safety-related, and fair in requiring driver license applicants to pass similar sensory tests to get full driving privileges).

For all age groups combined, test score was not significantly associated with total prior 3-year crash involvement when considered in isolation.

There was a very small percentage of drivers age 70+ with good low-contrast acuity.

Using a pass-fail criterion of 36 or more correctly identified letters as pass and less than 36 letters fail, Pelli-Robson specificity=53%, sensitivity=29% in predicting citations for age 70+ drivers, and accuracy of predicting citation occurrence=6.5%. For S=s age 52-69, specificity=65%, sensitivity=19%, and positive prediction=7%.

Approximately 5% of the variation in reported level of self-restriction was explained by test performance or age (the worse the visual performance or the older the driver, the more restriction). 5.3% of the variation in crash involvement for S=s age 70+ was explained by low Pelli-Robson scores and the avoidance of heavy traffic.

Hennessy (1995)

VISION

Static Contrast Sensitivity:

Pelli-Robson Test

1,475 ITT Hartford Insurance Co. policyholders for whom past driving histories were available through insurance records, divided into two groups based on the presence or absence of recent at-fault accidents. Driver age ranged between 50 and 80+ and was distributed as follows:

* 26 percent of the sample were between 50-64,

* 54 percent were between 65-74,

* 20 percent were over 75.

Participants were active drivers who had (generally) been pre-screened for risk in the insurance underwriting process. Also, participants who came in for testing appeared confident in their driving abilities.

Subjects participated in a 2-hour testing session consisting of visual, perceptual, and cognitive performance tests, and completed a self-report questionnaire. Contrast Sensitivity was measured using the Pelli-Robson Test, which is a 48-letter test designed by Pelli, Robson, and Wilkins (1988), of contrast sensitivity at one spatial frequency. The contrast between letters and background decreases as one moves down and toward the right of wall-mounted chart, viewed at dist. of 2 m under normal room illumination. The letters from left to right and from top to bottom progressively fade out as if they must be read in thicker and thicker fog. Letters (in groups of 3) range from 90% contrast (upper left) to 0.5% contrast (lower right). Testing requires no more than 3 minutes

Insurance and motor vehicle department records provided information about the following variables: at-fault accidents, non-fault accidents, non-accident claims, violations and convictions, miles driven, age, gender and marital status.

Testing rooms in hotels in 15 cities throughout Connecticut, Florida, and Illinois

Results showed that 42 percent of the sample had an at-fault accident between 1989-1991. Univariate correlations and multiple regression analyses were computed to determine the relationships between the variables and accidents.

The Pelli-Robson Letter Sensitivity Chart consistently yielded the highest correlation to accidents in the sample during 1989-1991 (r=-0.11, p<0.05).

Results indicated the following relationship between Pelli-Robson test scores and accident involvement:

45% of the drivers with scores of 1.95 were involved in accidents; 50% of drivers with scores of 1.80 were involved in accidents; 55% of drivers with scores of 1.65 were involved in accidents; 65% of drivers with scores of 1.50 were involved in accidents, and 70% of drivers with scores of 1.35 were involved in accidents.

The Pelli-Robson was relatively highly correlated to age, and thus the observed correlation between test performance and accidents is likely to be understated.

Brown, Greaney, Mitchel, and Lee (1993)

VISION

Static Contrast Sensitivity:

Smith-Kettlewell Low-Luminance (SKILL) Card

(high-contrast near-acuity loss and low-contrast near-acuity loss)

3,669 randomly-selected Class C license renewal applicants, licensed in California for at least 12 years, and unable to renew by mail. Four driver age groups were studied:

26-39, 40-51, 52-69, and 70+.

A letter chart viewed at a distance of 40 cm (16 in). From the top of the chart to the bottom, each line of letters is smaller than the line preceding it. One of the SKILL Card charts shows black letters on a white background (high-contrast letters); the other card shows black letters on a dark gray background (low contrast letters on a low-luminance background). The low-contrast SKILL Card chart is likened to viewing the worn-darkened lane striping at a busy intersection.

5 experimental vision tests were employed:

* Pelli-Robson Low-Contrast Acuity Test (measures loss in low contrast acuity; ability to see objects and borders)

* Smith-Kettlewell Low-Luminance Card (measures high-contrast near-acuity loss and low-contrast near-acuity loss)

* Berkeley Glare Tester (measures low-contrast near acuity loss, and low-contrast near-acuity loss in the presence of glare)

* Modified Synemen Perimeter (measures standard visual field-integrity loss and attentional visual field-integrity loss

* Visual Attention Analyzer (measures loss in UFOV, the area of the visual field in which useful information can be rapidly extracted from a complex visual display)

The dependent measure was the crash frequency during the previous 3-year period, extracted from the DMV database.

Drivers also completed a Driving Habits Survey measuring level of restriction (never, sometimes, often or always) for night driving, rain or fog, sunrise or sunset, driving alone, left turns, and heavy traffic.

California DMV Field Offices:

Carmichael

El Cerrito

Roseville

Study subjects rated test as face valid (clear instructions, safety-related, and fair in requiring driver license applicants to pass similar sensory tests to get full driving privileges).

For all age groups combined, test score was not significantly associated with total prior 3-year crash involvement when considered in isolation.

Best-corrected near acuity of drivers age 70+ differs on average only 3 letters for that of drivers ages 40-51, when tested under optimal conditions (well-illuminated, high-contrast text). When contrast was reduced by making the black letters gray, 70+ year old drivers read 2-3 lines (5 letters per line) less than 40-51 year old drivers (a marked reduction).

There was a very small percentage of drivers age 70+ with good low-contrast acuity.

Approximately 5% of the variation in reported level of self-restriction was explained by test performance or age (the worse the visual performance or the older the driver, the more restriction).

Hennessy (1995)

VISION

Static Contrast Sensitivity:

Vistech Contrast Sensitivity Gratings/Optec 1000

12,400 drivers in Pennsylvania, who came to Photo ID centers for license renewal, who were unaware that their vision would be tested when they arrived at the photo license facility. Ages ranged from 16 to 76+.

An Optec 1000 (Stereo Optical Company, Inc., Chicago, Il) vision screener was used to test contrast sensitivity at spatial frequencies of 6, 12, and 18 cycles per degree. Also tested with this device were visual acuity (Sloane letters at 20/20, 20/30, 20/40, 20/50, 20/70, 20/100, and 20/200 acuity ranges) and horizontal visual field (mini-lamps set at the horizontal peripherals of 85, 75, 55, and 45 [nasal] degrees on each side of the nasal region). Conduct of these 3 tests required 3 to 5 minutes per driver.

In the contrast sensitivity test, drivers were required to choose between one of three orientations of a test patch with line gratings that pointed either diagonally up to the left, to the right, or straight up and down.

Contrast sensitivity measurements show that the ability to see targets of low spatial frequency is statistically independent of the ability to see high spatial frequency targets, such as those presented in routine vision tests.

Statistical analyses of the relationship between visual performance at the time of screening and prior (3.67-year) crash experience were performed.

Three PennDOT Photo ID Centers

(Northeast Philadelphia/

urban area; Schuykill County/ rural area; and Delaware County/

suburban area)

* No significant relationships were found between binocular visual acuity, horizontal visual field scores, or contrast sensitivity at any particular spatial frequency and crash frequency (Chi square).

* Chi square analysis comparing observed vs expected crash counts for drivers who failed the vision test (static visual acuity worse than 20/40 and/or horizontal visual field less than 140 degrees), was significant: relative overinvolvement in accidents was found for drivers with "good" vision in age groups 16-20 and 21-25, and by drivers with "poor" vision in age groups 66-75 and 76+.

* Failure on the combined criteria that incorporates the current PennDOT standard (binocular acuity of 20/40 and horizontal visual field of 140 degrees) and a broadly defined contrast sensitivity criterion (scores below normal for 1 or more of the 3 spatial frequencies tested) produced the strongest relationship linking poor vision and high crash involvement, especially for 66-75 and 76+ driver age groups.

* Using the current PA standard, there is a modest upturn in crash experience for drivers age 76+ who pass, and a larger increase for drivers who fail; however, using the combined criteria (PA standard plus contrast sensitivity), the increase in crash rates for drivers age 76+ rises steeply, and the biggest difference in rates between passed and failed drivers in this group is found using combined criteria. No increase in crash rate with age was found for drivers who passed according to the combined criterion.

Decina and Staplin (1993)

VISION

Static Contrast Sensitivity/Glare:

Berkeley Glare Tester (BGT Chart)

(low-contrast near acuity loss, and low-contrast near-acuity loss in the presence of glare)

3,669 randomly-selected Class C license renewal applicants, licensed in California for at least 12 years, and unable to renew by mail. Four driver age groups were studied:

26-39, 40-51, 52-69, and 70+.

A letter chart viewed at a distance of 40 cm (16 in). From the top of the chart to the bottom, each line of letters is smaller than the line preceding it. The letters on the BGT chart are gray on a white background (low-contrast letters). The chart is mounted on a translucent screen behind which are lights; the chart is read in the presence and in the absence of glare. Testing requires no more than 3 minutes, and is administered in a dark or very dimly-lit room. Reading the chart is likened to viewing a white car in a fog.

5 experimental vision tests were employed:

* Pelli-Robson Low-Contrast Acuity Test (measures loss in low contrast acuity; ability to see objects and borders)

* Smith-Kettlewell Low-Luminance Card (measures high-contrast near-acuity loss and low-contrast near-acuity loss)

* Berkeley Glare Tester (measures low-contrast near acuity loss, and low-contrast near-acuity loss in the presence of glare)

* Modified Synemen Perimeter (measures standard visual field-integrity loss and attentional visual field-integrity loss

* Visual Attention Analyzer (measures loss in UFOV, the area of the visual field in which useful information can be rapidly extracted from a complex visual display)

The dependent measure was the crash frequency during the previous 3-year period, extracted from the DMV database.

Drivers also completed a Driving Habits Survey measuring level of restriction (never, sometimes, often or always) for night driving, rain or fog, sunrise or sunset, driving alone, left turns, and heavy traffic.

California DMV Field Offices:

Carmichael

El Cerrito

Roseville

Study subjects rated test as face valid (clear instructions, safety-related, and fair in requiring driver license applicants to pass similar sensory tests to get full driving privileges).

For all age groups combined, test score was not significantly associated with total prior 3-year crash involvement when considered in isolation.

Best-corrected near acuity of drivers age 70+ differs on average only 3 letters for that of drivers ages 40-51, when tested under optimal conditions (well-illuminated, high-contrast text). When luminance and contrast were reduced by adding glare, 70+ year old drivers read 2-3 lines (5 letters per line) less than 40-51 year old drivers (a marked reduction).

There was a very small percentage of drivers age 70+ with good low-contrast acuity.

Approximately 5% of the variation in reported level of self-restriction was explained by test performance or age (the worse the visual performance or the older the driver, the more restriction).

Hennessy (1995)

VISION

Dynamic Contrast Sensitivity:

MultiCAD

82 Areferred" subjects aged 60-91 (26 of which were identified as probably being cognitively impaired to some degree). The drivers were referred to the DMV for reexamination due to a medical condition (by physician, optometrist, ophthalmologist), a series of licensing test failures, a flagrant driving error (police referral), or some other indicator of driving impairment.

This test used MultiCAD to measure drivers' contrast sensitivity for a target that is moving relative to the observer. Immediately following the MultiCAD static contrast sensitivity test, exactly the same type of stimuli were shown while moving at a predetermined rate (12 degrees per second) from one side of the screen to the other. The same "which signal face is different?" discrimination was required of the subject, using the 3-button response pad. Three replications of each measurement were performed with a pass/fail score assigned at each level. A passing score was defined as at least 2 correct responses out of the 3 presentations for each level tested. Mean response time was also calculated for correct responses at each level.

Scoring was also conducted on a "gross" level across all stimulus characteristics. Average response time and average error score were calculated across all 12 trials.

Multiple linear regressions were conducted to arrive at the best linear combination of variables for predicting performance (weighted error score) on a standard DMV road test, (see On-road Performance Measures of Driving Safety: California MDPE at the end of this Compendium), and comparisons were made between cognitively impaired and cognitively non-impaired referral drivers to determine whether there were differences in performance

California DMV Field Office

*Correlation between gross dynamic contrast sensitivity errors and weighted error score on road test was significant (r=.2420, p<.034)

*Correlation between gross dynamic contrast sensitivity time and weighted error score on road test was not significant.

*Correlation between dynamic contrast sensitivity time for the high contrast 20/80 target and weighted errors on the road test was significant (r=.2466, p<.049).

*Neither gross nor precise scoring of dynamic contrast sensitivity errors or response time differentiated between cognitively impaired and cognitively unimpaired referral subjects.

Janke & Eberhard (1998)

Staplin, Gish, Decina, Lococo, and McKnight (in press)

VISION

Peripheral Visual Fields:

Goldman Perimeter (II/4e, III/4e and V/4e targets)

 

21 Retinitis Pigmentosa Patients age 29-67 (mean age 42 yrs) 9F, 12M

4 major peripheral field loss profiles were represented:

* partial concentric restriction (n=5)
* residual temporal islands (n=5)
* ring scotoma (n=7)

severe peripheral restriction (n=4)

31 Controls with normal vision age 21-64 (mean age 40 yrs) 16F, 15M

All S=s held unrestricted driving licenses and drove at least 1000 miles/year.

Visual function measures:

Peripheral Visual Field Loss - Visual field tested using Goldman Perimeter to produce a binocular, three dimensional map of the visual field. Additionally, the total scotomatous area of the binocular visual field and total horizontal field extent were calculated.

[S=s were required to have visual acuity of 0.2 LogMar (Snellen equivalent of 20/40) or better in at least 1 eye tested with Bailey-Lovie (ETDRS) charts.]

Visuocognitive and motor performance measures:

S=s performed in driving simulator as described in Szlyk et al. (1993) to collect data on simulator accidents, RT to stop sign, reaction distance, brake and gas pressure, out-of-lane events, etc.

Psychosocial factors:

Risk-Taking Questionnaire
State-Trait Anxiety Inventory.

Dependent measures:

* Self-reported accident involvement in the past 5 years
* State-recorded accident involvement in the past 5 years
* Driving simulator accidents

________________________________________________

FINDINGS (cont=d)

* In simulator measures, RP patients slow down at greater dist. than controls (p<.02) to peripheral landmarks (stop signs). Smaller horiz. visual field extent related to longer reaction dist. in RP's (r=-0.51). Reaction dist marginally related to state-recorded accidents [r(30)=.31, p=.07].

* Smaller horiz. visual field extent to II/4e target related to longer reaction dist {r(19)=-.52, p<.02} but not signif. for III/4e or V/4e targets.

* RP subjects were also strayed out of lane more often than controls (p <.02). Out-of-lane events signif. related to state-recorded accidents + violations.

* RP patients report less risk-taking behavior than controls (p<.001). Trait anxiety signif. related to self-reported accidents (r=0.28), while state anxiety was not. No difference in trait or state anxiety found between groups.

Mult. regress. anal.: for RP=s- reaction dist, deviation in lane position, out-of-lane events, braking pressure, simulator accidents, acuity, and residual visual field accounted for 71% of var in self-reported accidents; RP=s + controls: gas press., out-of-lane, horiz. eye mvmt., acuity, residual visual field account for 46% of variance. (Visual function alone not a signif. predictor of self-reported accidents)

Univ. Illinois at Chicago

Eye Center

* RP patients self-report more accidents in general (p<.02) than normals; and more peripheral accidents--not detecting other cars when changing lanes, pulling out of driveways, in parking lots-- (p <.001) than normals.

* RP patients self-report more accidents under low contrast/adverse weather conditions--rainy, snowy, nighttime, dusk-- (p <.06) than normals.

* No difference between groups in state reported accidents.

* RP patients had more simulator accidents than controls, but this difference was nonsignificant.

* Significant correlation found between total remaining horizontal field extent and self-reported accidents for RP group (r= -0.58, p<.01; r= -0.50, p<.05; r= -0.40, p<.05 ) for the three targets (V/4e, III/4e, and II/4e). A significant negative correlation reflects increased accident risk as remaining intact visual field extent decreases.

* Field extents measured with all three targets also significantly (and negatively) correlated with self-reported peripheral accidents [V/4e, r(20)=-.58, p<.01; III/4e, r(20)=-.61, p<.01; II/4e, r(20)=-.50, p<.05].

* Measure of binocular scotomatas area ro V/4e target (in sq in) significantly related to self-reported accidents and peripheral accidents

* Significant correlation also found between field profile and self-reported accidents in RP group [r (50)=0.42, p<.01)] between field profile and peripheral accidents [r(50)=.66, p<.01]. S=s with severe field restriction (profile 4) are at greater crash risk than those with partial restriction (profile 1)

Szlyk, Severing, and Fishman (1991)

VISION

Peripheral Visual Fields:

Manually Operated Perimeter

97 drivers age 55+recruited from CA DMV driving records

 

In this pilot study, a manually operated perimeter (not described) was used to measure peripheral visual fields with and without attentional demand, along with several other Smith-Kettlewell vision tests, to discriminate between accident-free and accident-involved drivers. Subjects were divided into 2 groups: no accidents on record vs 2 or more accidents on record, within the preceding 3 years.

An accident-proneness index was used to account for whether the subject was at fault in the accident. For each accident, a score was assigned as follows: 4=driver primarily at fault; 3= subject contributed to the fault; 2 =fault undetermined; 1=subject not at fault; 0=no accidents.

Drivers also completed questionnaires to evaluate driving habits, difficulties while driving, and standard ocular and medical history.

Smith-Kettlewell Eye Research Institute, San Francisco, CA

* The relationship between overall decreased performance on the vision tests and the increased accident-proneness index was significant (p=.04).

* The skills with the strongest statistical relationship to accident involvement included low contrast low luminance visual acuity; disability glare; the extent of the standard visual field directly to the right, and down and to the right; and the extent of the attentional visual field directly down and to the right, or to the left.

* A comparison of 10 people with the worst vision scores with 10 people having the best vision scores, showed that people with the worst scores on the attentional visual field, standard visual field, and bright glare tests were several times more likely to be in the accident group than in the nonaccident group.

* Although the accident-involved drivers had more difficulty on vision tests, they were unaware of any problems with their vision.

Brabyn (1990)

VISION

Eye Disease:

Cataracts

279 drivers with cataract

mean age=71

53% male

86% White/13% African-American

105 drivers with no cataract

mean age = 67

48% male

84% White/16% African-American

The project is an intervention evaluation study to determine how improvement in vision impacts crashes and driving habits.

Other Objectives:

To determine the natural progression of crash frequency and driving habits in a group of older adults who, at the outset of the project, are in good eye health.

To determine whether certain factors serve as effect modifiers, thus altering the relationship between improvement in visual sensory function and crash frequency/driving habits.

Prospective study where all subjects are assessed once annually, with the 1st visit before cataract surgery, then annually after surgery for 2 years. Crash data from 5 years prior to enrollment and 3 years following enrollment are obtained from Alabama Dept. of Public Safety.

Visual functional status was measured as follows:

Distance Acuity - ETDRS Chart

Contrast Sensitivity - Pelli-Robson Chart

Visual Field - Humphrey Field Analyzer 81-point screening program for the central 60 degrees

Cataract was the only diagnosed eye condition (other than refractive error) in 75% of subjects in the cataract group.

For this report, crash data for the previous 5 years were obtained from the Alabama Department of Public Safety, and at-fault crashes were used as the dependent measure.

________________________________________________

FINDINGS (Cont=d)

*When adjusted for impaired health, the association between cataract and crash involvement remained significant (relative risk = 2.49, 95% CI = 1.0-6.27).

University of Alabama, Birmingham

*Subjects in the cataract group averaged 20/60 and 20/40 in the worst and best eye respectively, compared to the no cataract group who averaged 20/25 and 20/20 respectively. This difference was significant (p<.001).

*Contrast sensitivity was significantly worse in both eyes for subjects with cataracts (p<.001). Age adjusted log CS for cataract group was 1.39 (best eye) and 1.19 (worst eye) compared to 1.61 (best eye) and 1.52 (worst eye) for no cataract group.

*Cataract subjects detected fewer points in their visual field than the no cataract subjects.

*Proportionately more cataract subjects preferred to have someone else drive when they travelled in a car, drove slower than the general traffic flow, and received advice that they limit or stop driving (self-reports on driving habits questionnaire).

*Cataract was associated with reduced number of days driving per week and a reduced number of destinations. (Cataract drivers 2 X more likely to reduce driving)

*Subjects with cataracts were (2 times) less likely to drive beyond their neighboring towns than subjects without cataracts.

*Cataract was significantly associated with driving difficulty in the rain, driving alone, making left turns across traffic, driving on interstates, in high traffic, in rush hour, and at night (Cataract drivers 4X more likely to report these difficulties).

*After adjusting for driving exposure, the association between cataract and at-fault crash involvement was significant (relative risk = 2.48, 95% CI = 1.0-6.14).

Owsley, Stalvey, Wells, and Sloane (1999)

VISION

Eye Disease:

*Diabetic Retinopathy

*Glaucoma

294 older drivers, ages 56-90 years at enrollment, drawn from the population of licensed drivers in Jefferson County over age 55.

33% had 0 crashes on record

49% had 1 to 3 crashes over the prior 5 year period

18% had 4 or more crashes over the prior 5 year period

Objective: To identify measures of visual processing associated with crash involvement by older drivers, in a prospective follow-up study.

*Subjects received the following sensory tests:

Letter Acuity - ETDRS chart

Contrast Sensitivity - Pelli-Robson chart

Stereoacuity - TNO Test

Disability Glare - MCT-8000 (VisTech)

Visual Field Sensitivity - Humphrey Field Analyzer 120-point program for central 60 degree radius field

*Subjects received comprehensive eye exam resulting in a primary diagnosis (cataract, age-related maculopathy, glaucoma, diabetic retinopathy)

*Mental status was assesses using the MOMSSE

*Visual Attention was measured with the Vision Attention Analyzer:

*AOn the road" exposure was estimated using questionnaire data on number of days/week subjects drove and annual number of miles driven. Subjects were asked if anyone had ever suggested they limit or stop driving.

Dependent variable: Motor vehicle crash occurrence during the 3 years following clinic assessment, obtained from Alabama Department of Public Safety. Person-years to first crash was calculated from enrollment date; Person-miles of travel was calculated by multiplying person-years times reported annual mileage.

University of Alabama, Birmingham

Ophthal-

mology clinic

*56 S=s had at least 1 crash in the 3-year follow-up period, and 11 of these had 2 or more.

*Estimated annual crash rate was 7.4 per 100 person-years of driving and 7.1 per million person-miles of travel.

*Crash involvement in prior 5-year period was significantly associated with increased crash risk (Risk ratio = 2.0)

*Significant, independent associations with crash risk in 3-year follow-up were found only for:

*UFOV reduction of > 40%: RR=2.3; 95% CI = 1.27 - 4.29)

*Driving < 7 days/week: 48% decreased crash risk (95% CI = 0.27 - 1.01).

*Dx of diabetic retinopathy (5X greater risk, 95% CI = 1.13 - 21.8).

*Dx of glaucoma: (RR=5.20, 95% CI = 1.19-22.72). Relationship for glaucoma and crashes stronger for males (RR=9.81) than females (RR=5.14).

Owsley, Ball, McGwin, Sloane, Roenker, White, and Overley (1998).

VISION

Multiple Visual Capabilities:

Keystone Telebinocular Testing Device

105 drivers licensed in Nebraska, aged 65-88 (mean age = 71.4). 54 were females (mean age = 70.5 years); 51 were males (mean age = 72.2 years). All subjects were volunteers, and were paid *25.00 for participating. 36 had taken a driver education course in the past 10 years.

Vision was assessed using a Keystone telebinocular testing device to measure near acuity, depth perception, left and right peripheral vision, color vision, and lateral and vertical phoria.

The driving performance of the subjects was evaluated using the on-street driving performance measurement (DPM) technique developed by Vanosdall and Rudisill (1979). The subjects were evaluated by a driver education expert trained in the use of the DPM technique, while they drove in their own cars. The DPM route was a 19-km circuit designed to evaluate the subjects in the situations that are most often involved in the accidents of older drivers. Therefore, their performance was evaluated at 7 intersections where they were required to make left turns at 5 intersections and right turns at the other 2 intersections. Four of the left turns were made from left-turn lanes onto four-lane divided arterial streets in suburban areas, and one was made from a left turn lane onto a two-lane one-way street in an outlying business district. Performance on the DPM was evaluated as follows. Each of the 7 turning maneuvers was divided into 3 segments; (1) the approach to the intersection, (2) the turning maneuver itself, and (3) the departure form the intersection. Performance in each segment was evaluated as being either satisfactory or unsatisfactory; 1 point was given for satisfactory performance and 0 points were given for unsatisfactory performance. The criteria for determining satisfactory or unsatisfactory performance were in terms of the subject's search pattern and control of the vehicle's speed and direction. Since subjects made 2 trips around the route, the maximum score was 42. The measure of driving performance used in the analysis was a percentage of 42.

Cognitive measures: University laboratory.

Driving measures:

business district and residential street networks

Results of the correlational analysis showed that among the vision factors, only depth perception and right visual field correlated significantly with driving performance (p<0.05); the correlational coefficients for these factors were .35 and .22 respectively.

Tarawneh, McCoy, Bishu, and Ballard (1993)

VISION

Multiple Visual Capabilities:

Sight Screener II

(AO Safety Products)

1,475 ITT Hartford Insurance Co. policyholders for whom past driving histories were available through insurance records, divided into two groups based on the presence or absence of recent at-fault accidents. Driver age ranged between 50 and 80+ and was distributed as follows:

* 26 percent of the sample were between 50-64,

* 54 percent were between 65-74,

* 20 percent were over 75.

Participants were active drivers who had (generally) been pre-screened for risk in the insurance underwriting process. Also, participants who came in for testing appeared confident in their driving abilities.

Subjects participated in a 2-hour testing session consisting of visual, perceptual, and cognitive performance tests, and completed a self-report questionnaire.

Tests of visual function were performed using a Sight Screener II (AO Safety Products) and included:

* Acuity (far and near)

left eye, right eye and binocularly

* Stereopsis (far and near)

* Color perception (severe and mild)

Insurance and motor vehicle department records provided information about the following variables: at-fault accidents, non-fault accidents, non-accident claims, violations and convictions, miles driven, age, gender and marital status.

Testing rooms in hotels in 15 cities throughout Connecticut, Florida, and Illinois

Results showed that 42 percent of the sample had an at-fault accident between 1989-1991. Univariate correlations and multiple regression analyses were computed to determine the relationships between the variables and accidents.

The visual acuity measures were among the predictors which did not reach significance.

Brown, Greaney, Mitchel, and Lee (1993)