Banner -- Identifying Strategies to Collect Drug Usage and Driving Functioning Among Older Drivers


Level II: Interactive, Computer Graphic Visuals, Restricted or No Motion

This level of simulation is exemplified by numerous commercial platforms offering driving scene displays on one, two, or three screens that typically provide a field of view (FOV) of roughly 60° on each screen; and a “cockpit” including driver’s seat plus brake and steering wheel/column assembly, sometimes from an actual vehicle. Motion cues, if any, are likely to be limited to “force feedback” on the vehicle controls and a belt tensioner system that constricts during simulated deceleration and is loosened during simulated acceleration. A high frequency “road vibration” element may be added, too. As for the realism of the visual displays – the same limitations apply as noted above for the high-level simulators. In addition, with these more modestly priced (less than $100,000) systems, problems with aliasing and with “pop-ups”—where the level of detail in a scene is such that not all elements can be redrawn without a perceptible delay, and objects, especially at the horizon, abruptly pop into the scene—are more likely to occur. An argument can be made that an equally wide range of driver behaviors at the tactical level could be examined in these simulators as can be examined in the high-level simulators, though of course advanced vehicle maneuvering skills at the operational level (e.g., recovering from a skid) cannot similarly be evaluated. Unfortunately, the increasingly wide-spread use of this class of simulators, for driver training as well as research applications, has indicated that simulator adaptation syndrome (“simulator sickness”) affects a sizeable minority of subjects/trainees, with older people being especially vulnerable.

Documented research studies pertinent to NHTSA’s interest in polypharmacy that are accessible to searches in the open scientific literature appear to have most often been conducted using “Level II” simulation methods.

As one example of such research, Szlyk et al. (2004) used a driving simulator to research visual deficits resulting from diabetic retinopathy. This was an interactive device that included a seat, steering wheel, gas and brake pedals. The visual display consisted of three 62.5-cm (24-inch) color monitors displaying a total 160° horizontal viewing field and a 35° vertical viewing field of a computer-generated environment to a driver sitting 57.5 cm (22 inches) from the center screen. As previously described by Szlyk, Brigell, and Seiple (1993), stimuli are computer-generated images of a simulated roadway with traffic, signs, and painted roadway lines. The visual scene is updated 20 times per second (the minimum rate to avoid a perceived “choppiness” in the visual display). Simulator performance measures of effectiveness in the research by Szlyk et al. (2004) included:

  • Speed (mean speed in miles per hour)
  • Gas-pedal pressure (mean force applied to gas pedal)
  • Gas-pedal pressure SD (SD of the force applied to gas pedal)
  • Acceleration (mean of 5 speed points after a complete stop at a stop sign)
  • Brake-pedal pressure (mean force applied to brake pedal in arbitrary units)
  • Brake-pedal pressure SD (SD of the force applied to brake pedal)
  • Brake-response time (mean time in seconds elapsed between when a stop sign is displayed and when force is applied to the brake pedal)
  • Response time (mean time in seconds elapsed between when a stop sign is objectively displayed and when no force is applied to the gas pedal)
  • Brake-response slope (mean deceleration calculated as the ratio of change in speed to change in time before a complete stop at a stop sign)
  • Brake duration (mean time in seconds that force is applied to the brake pedal).
  • Ran stop sign (number of stop signs ran)
  • Ran red light (number of red lights ran)
  • Off-lane time (total time in seconds spent over the left yellow line during the course)
  • Off-road time (total time in seconds spent off road to the right onto the road’s shoulder)
  • Near accidents (number of situations in which an accident is narrowly averted, as determined by an experienced observer)
  • Accidents (number of collisions with other cars or objects)

In the Szlyk et al. (2004) study, 25 licensed drivers with diabetic retinopathy with a mean age of 53 (range = 34 to 72) practiced on the simulator for 15 minutes on a training course before completing the 8-minute evaluation course. They were instructed to “drive” as they normally would in their own cars, and to obey traffic rules. During the evaluation, subjects were required to respond to stop signs, traffic lights, and road hazards. While the particular results of the study are not relevant to the current research topic, the Szlyk et al. (2004) study illustrates the use of a simulator to measure aspects of driving performance that likely would also be of interest in NHTSA studies of the effects of multiple medications on driver behavior.

Lee, Drake, and Cameron (2002) employed a driving simulator with similar capabilities to assess the driving performance of 53 community dwelling older people ages 65 to 85. Senior drivers were classified as those 65 to 74, and advanced-age senior drivers were classified as those 75 to 89. The self-reported medical conditions and percentage of subjects self-reporting each condition were as follows: arthritis (28%), diabetes (13%), high blood pressure (47%), visual problems (34%), hearing problems (23%), heart diseases (17%), and respiratory disease (4%).

The assessment conducted by Lee et al. (2002) involved three segments: a 30-minute initial screening; a 45-minute driving simulation; and a 25-minute post-session feedback. One-fourth of the participants understood how to use the controls with minimum instruction upon initial introduction of the simulator. The majority indicated that they felt positive about the simulator and confident in their ability to operate it, following the introduction. Four of the 53 subjects (7%) reported some degree of simulator sickness, stating that they experienced mild dizziness after completing the simulated driving. However, they did not feel a need to withdraw from the assessment. The simulated driving test was conducted over a total distance of 15 km (9.3 mi) over three speed-zone segments of 60 km/h (37 mi/h), 70 km/h (43 mi/h), and 100 km/h (62 mi/h). Driving scenarios included eight possible events (e.g., overtaking a car ahead, driving along a curved road) paired with a speed zone. Assessment criteria included “performance indicators” and “operational parameters.” Performance indicators included total run length, speed violation, proper signaling, divided attention task, and off-road crash—effective performance required participants to exercise cognitive function in making interactive judgments and rapid decisions to modify their driving behavior according to the demands of the traffic scenarios. Operational parameters reflected the automatic responses of drivers to maneuver their vehicles when driving, and included curvature error, heading angle error, steering-wheel rate and lane position—effective performance required subjects to call on their “permanent” skills in driving acquired over years of driving.

Lee et al. (2002) found a strong association between age and performance on the “performance indicators.” The difference between the performance of the senior drivers and the advanced-age senior drivers was significant for all five performance indicators. The senior driver group was involved in fewer crashes,” used their signals more often when changing lanes, and committed fewer speed violations than the older senior drivers. The senior drivers drove faster than the older senior drivers, at speeds appropriate to traffic conditions. A weak association was found between the operational parameters and age of the participants. There was no significant difference in operational parameter performance as a function of age group. The authors concluded that in formulating assessment criteria for simulator studies with older people, that driving skills at the “controlled” processing level should be the focus, rather than “automatized” behaviors that do not deteriorate significantly with age.

In another study conducted by Lee and colleagues (Lee, Lee, Cameron, and Li-Tsang, 2003), the driving performance of 129 volunteer subjects 60 and older was examined in the simulator to determine if any of 10 performance measures were associated with self-reported crashes. Each participant received a 30-minute interview to gather data on self-reported crashes in the past year, medical conditions, and driving habits. This was followed by a 45-minute simulated driving session containing 10 performance tasks:

  • Rules Compliance – Lane changing in a double-lane road, where the participant’s car was in the right lane. Keep Lane signs displayed every 55 yards prompted subjects to go back to the inner lane.
  • Traffic Sign Compliance – Drive through Stop, Give Way, and pedestrian crossings safely
  • Driving Speed – Drive 1.5 miles along the road according to the designated speed of the double-lane straight road (40 mph speed limit)
  • Use of Indicator – Drive around “road work” obstacles blocking the road and return to the inner lane as soon as possible
  • Road Use Obligation – Observe traffic conditions and drive safely through T-junctions leading to main road with Stop signs
  • Decision and Judgment – Avoid crashing into pedestrians 30 yards ahead running across the road hastily, car parked on the roadside moving out without signaling, and car in front suddenly slowing down
  • Working Memory – Recall five street names and five maneuvers (turn right or left) after 10 minutes’ simulated driving. Subjects were given 5 minutes to memorize the route to a fictitious park marked on a road map, followed by 10 minutes of unrelated driving, and then asked to recall the maneuvers and street names on the route.
  • Simultaneous Tasks – Starting from 100, take away 5 every time the “SUBTRACT” billboard is seen. (15 billboards with “SUBTRACT” signs were posted along the road).
  • Speed Compliance – Observe and maintain a speed close to the posted speed limits (40, 45, and 70 mph), which vary according to traffic conditions.
  • Divided Attention Tasks – Signal the traffic indicator when the “diamond” shapes on the monitor screen change to “triangle” randomly and stay for 15 seconds.

Subjects in the Lee et al. (2003) study ranged in age from 60 to 88, with a mean age of 72.9, and 22 percent of the subjects were female. The self-reported medical conditions and percentages of subjects reporting each were as follows: high blood pressure (38%), visual problem (36%), arthritis (26%), hearing problem (25%), heart diseases (15%), and diabetes (10%). Over 70 percent of the subjects reported that they were suffering from multiple medical conditions and took daily medications including analgesic, anticoagulant, antihypertensive, and anti-inflammatory agents, but they perceived that such medications did not interfere with their driving abilities. Seventy-nine subjects (61.2%) reported having at least one crash in the prior year, but none of the crashes involved personal injury.

Lee et al. (2003) found a significant negative correlation between each simulated driving criterion and age of the participants, indicating that simulated driving performance worsens with increasing age. A stepwise logistic regression to determine the association between crashes and simulator criteria found that performance on driving tasks involving working memory, decision and judgment, and speed compliance was negatively associated with the occurrence of a crash. According to the fitted model, each added point on the working memory scale was associated with a 45 percent decrease in risk, on the decision and judgment scale with a 61 percent decrease in risk, and on the speed compliance scale with a 17 percent decrease in risk. The model also indicated that an increase of one year in age could elevate the crash risk by 13 percent.

Overall, 87.7 percent of the 129 participants in the Lee et al. (2003) study were correctly classified on the occurrence/nonoccurrence of a (self-reported) crash. The study finding that cognitive skills such as working memory, ability to make rapid decisions, judgment under time pressure, and confidence in driving at high speed were significantly (negatively) associated with a crash event, together with high sensitivity (82%) and specificity (91%) of the regression model, suggests that driving simulators at this level could be useful in investigations of older drivers using multiple medications. Concerns persist about elevated levels of simulator sickness among older people under the influence of medications, however. Results indicating that age, per se, predicted performance in the simulator, while simultaneously predicting crash involvement better than 7 of 10 simulated driving measures, also raises some fundamental questions about this methodology.

Finally, Freund, Gravenstein, Ferris, and Shaheen (2002) conducted a study using four cognitively impaired older adults and five healthy older adults (ages 67 to 78) to examine the degree to which performance in a simulator compared to performance on an on-road test. The on-road testing was conducted by an occupational therapist, who was certified both as a driving rehabilitation specialist and a commercial driving instructor, using a dual-brake equipped vehicle. Performance measures for the simulator task and the on-road task included hazardous or potentially catastrophic errors, traffic violations, and rule violations. The mean scores for the simulator and on-road tests were significantly correlated at -.670. The lower the score on the simulator (identifying performance with few errors), the higher the score on the road test (identifying performance with competency ratings greater than 90 percent). There was also a strong association between hazardous and lethal errors committed on the simulator and failing the road test for hazardous errors. Subjects who failed the on-road test committed an average of 5.4 hazardous errors and an average of 4 lethal errors on the driving simulator. Subjects who passed the on-road test committed no hazardous or lethal errors in the simulator test. The authors conclude that although the sample size was small, study results support the use of driving simulation as a method to objectively evaluate driving performance of cognitively impaired and healthy older adults.