Reasearch Notes-Summer 1997
Mary K. Janke
The dementia/frailty study, funded under a cooperative agreement between the California Department of Motor Vehicles (DMV) and the National Highway Traffic Safety Administration, has the goal of developing a "model" battery of tests for use by licensing agencies in identifying impaired older drivers and evaluating their competency to drive. We tested 135 drivers at the Santa Teresa field office in San Jose. Then we moved the study to the Buck Center for Research in Aging in Marin County-a very different kind of environment!-and tried some old tests and some new ones on 101 volunteer subjects there. So far the results look promising.
Our subjects at Santa Teresa were mainly drivers aged 70 or older, and most of them were reexamination referrals, although we collected data on a small sample of volunteers as well. Subjects took a battery of laboratory tests of such things as perceptual speed, reaction time, visual scanning and other tests of visual functioning, as well as DMV tests of knowledge and high-contrast static acuity. Then they took two road tests, one in the neighborhood of the field office, a relatively unfamiliar area, and the other in the neighborhood of their home. Both of these tests were based on DMV's Driver Performance Evaluation or DPE--one was called the MDPE, where M stands for Modified, and the other was called the ADPE, where A stands for Area. Of course the area test couldn't have a standardized route, so this limited its reliability.
There are two main goals of the research. One is to identify tests that distinguish between volunteers and referrals, and the other is to identify tests that predict number of errors on the MDPE. These two sets of tests will be considered as forming a two-tier testing scheme in which applicants take short first-tier tests to identify people with possible age-related impairments, and longer second-tier tests are given to applicants who do poorly in first-tier testing and also to referrals, in order to predict how well they will actually do on the road.
There has been some success in achieving both of those goals. Most nondriving tests, as well as the road tests, differentiated referrals from volunteers in our sample, according to preliminary findings. Considering only the shorter, simpler tests, those showing the most promise for identifying impaired applicants are 1) the Pelli-Robson test of low-contrast static acuity (similar to a test of vision in fog or glare); 2) a visual scanning test called Auto-Trails, in which people must touch, in numerical order and as fast and accurately as possible, 14 numbers randomly arranged against the background of a driving scene; and 3) an observational measure made by the test administrator, who notes the number of observable "problems" the driver has-things like tremor, impaired balance, inability to understand test instructions, etc. (Our laboratory test administrator was Sandra Winter, whose article also appears in this edition.) Using the best score cut-off point to minimize identification of volunteers as referrals, I found early on that a combination of these three measures correctly identified all volunteers but one as volunteers (that one had suffered a stroke and was still somewhat impaired, so the fact that the person was identified is a positive thing) and 63% of the referrals as referrals.
For predicting road test errors of referrals and volunteers combined, I have found for the first 100 or so subjects that a combination of a simulator exercise measuring subjects' recognition speed of two different kinds of cue and their accuracy in making a different response to each, and measures of speed and accuracy on an automated static visual acuity test in which subjects have to push a button to identify the correct stimulus, allowed about 60% of the variation in road test errors to be predicted. These tests, or something like them, may turn out to be suitable for a second tier of testing.
Although the study concerns dementia, as late as November 1996 few subjects had been referred for that. But 26 drivers were identified in preliminary results as showing either dementia or some lesser degree of cognitive impairment which may have been early undiagnosed dementia. These 26 drivers did more poorly than the other referrals on several of the laboratory tests, but the only difference they showed from other referrals on real-world driving measures was that they were more likely to become confused when asked to find their way back to the field office after being directed to drive a short distance past it.
A final report on the Santa Teresa and Buck Center data should be completed by the end of August 1997 and available in the Fall.
Many staff in the Santa Teresa office played critical roles in the successful execution of this project. I want to thank them all, and am particularly indebted to Marilyn Patterson, office manager; Peter Marquez, DL supervisor; and the two expert examiners, Bernard Beckwith and James Nelson, who conducted the road tests. I'm also very appreciative of the tolerance everyone showed for our disruption of normal office procedures.