Research Notes- Winter 95/96
Crashes Predicted when Compensation for Visual Impairment is Inadequate
David Hennessy
Two of five experimental vision tests have been identified as promising for further validation in a large-scale statewide study.
From February 5 through October 29, 1992, specially trained Motor Vehicle Field Representatives administered five experimental vision tests and a driving habits survey to 3,669 randomly selected renewal applicants in the Carmichael, El Cerrito, and Roseville field offices.
One of the promising tests, the Pelli-Robson low-contrast acuity test, indicates the examinee's ability to see objects and borders. For example, good low-contrast acuity is critical to discerning whether there is an object lying in the roadway or seeing faded lane-boundary markings. The Pelli-Robson low-contrast acuity test requires the examinee to read from a distance of 2 meters (a little more than 6 feet) a letter chart on which the letters from left to right and from top to bottom progressively fade out as if they have to be read in thicker and thicker fog.
The other promising test, perceptual reaction time (PRT) assessment, estimates the speed with which the examinee can process visual information. Good PRT is critical to seeing hazards and recognizing/reading warning signs in a timely manner. In the study it was estimated what the shortest duration was that an examinee could accurately identify a computer-generated silhouette of a car or a truck 75% of the time. PRT estimates were obtained with a machine designed to measure visual attention, the Visual Attention Analyzer (manufactured by Visual Resources Inc.).
The results of statistical analyses supported the hypothesized relationship between vision test performance (VTP) and crashes. The crash predictive value of performance on the two promising vision tests depended, as indicated below, on the applicant's age, general visual ability (Vision), and reported level of self-restriction (Compensation).
Compensation for impaired visual ability by self-restricting is perhaps the most common reason suggested for why one should expect to find only a weak relationship between VTP and crashes. If drivers with poor vision tend to avoid driving under those visually demanding conditions that increase their exposure to crash risk, then the association of poor VTP with crashes would be expected to be weak if the drivers adequately compensate. Adequate compensation for impaired visual abilities may be constrained by a variety of factors such as the driver not being aware of impaired visual abilities, having to cope with multiple functional impairments, and the purpose(s) in driving, e.g., getting to work on time may keep drivers from avoiding driving in heavy traffic.
When the drivers' reported level of self-restriction was not considered, poor Pelli-Robson low-contrast acuity was not found to be predictive of crashes either for specific age groups or when all ages were combined. However, drivers aged 26-39, 40-51, and 70+ who have poor Pelli-Robson low-contrast acuity and who reported never avoiding heavy traffic tended to have an elevated crash rate relative to drivers with good Pelli-Robson low-contrast acuity (see figure below for drivers aged 26-39). The reverse (very low crash risk) was true for drivers who have poor Pelli-Robson low-contrast acuity and who reported often avoiding heavy traffic. Avoiding heavy traffic appears to compensate at least in part for poor Pelli-Robson low-contrast acuity; however, not all drivers with poor Pelli-Robson low-contrast acuity adequately compensate by avoiding heavy traffic.
| Note: Predicted values are based on a small sample size (n=335), and therefore should be regarded only as rough approximations. |
| Predicted number of crashes over a 3-year period for drivers aged 26-39 by Pelli-Robson low-contrast acuity and level of avoidance of heavy traffic. |
Further analyses showed that when the drivers' reported level of heavy traffic avoidance was considered and calculations were confined to applicants who failed the department's 20/40 Snellen test, the crash predictive value of Pelli-Robson low-contrast acuity was more than doubled. Failing the department's Snellen test is indicative of generally poor visual ability, especially among 70+ year old drivers. About 65% of the Snellen fails were 70+ year old drivers. Older drivers whose best-corrected visual acuity is not sufficient to read a line of 20/40 letters most likely have one of the following age-related vision disorders: cataracts, macular degeneration, other retinal pathology such as diabetic retinopathy, or glaucoma. If so, in addition to reduced visual acuity, contrast sensitivity, glare resistance, light sensitivity, and visual field sensitivity may all be impaired too. Therefore, among Snellen fails one might expect that poor VTP would be more strongly associated with crashes than it would be among Snellen passes.
Unlike poor Pelli-Robson low-contrast acuity, poor PRT was somewhat predictive of crashes for 70+ year old drivers. This predictive value doubled for the 70+ year old drivers after adjusting for the extent to which drivers avoid left-turns. As was the case with poor Pelli-Robson low-contrast acuity, this latter predictive value of having a slow PRT doubled when calculations (adjusted for various forms of self-restriction) were confined to applicants who failed the department's Snellen test.
As indicated earlier, the objective of this study was to identify the experimental vision tests offering the most potential for a large-scale demonstration project. It seems clear, however, from the nature of the relationship between VTP and crashes that the use of vision testing to identify and properly restrict high risk drivers will involve more complex scoring procedure than are currently used. In determining whether countermeasures need to be applied to a renewal applicant after having measured their PRT for instance, one might adjust the PRT score in accordance with the applicant's age or DMV Snellen test result. A potential advantage of the latter is that it avoids the direct use of age in generating test norms. There may also be a need to assess a driver's level of compensation as part of the renewal process in order to better capture the crash prediction power of the VTP measures.
Note: A copy of the complete report Vision Testing of Renewal Applicants: Crashes Predicted when Compensation for Impairment is Inadequate (Report 152) can be obtained from the DMV Research and Development Section at 2415 First Avenue, Mail Station F-126, Sacramento, CA, 95818.