2.1 Participants: Twenty stroke survivors (64.35 ± 14.83 years) and 20 healthy older adults (67.45 ± 8.39 years) participated in the current study. Inclusion criteria for the stroke participants were as follows: (1) diagnosed with a unilateral cerebrovascular accident at least 6 months prior to testing, (2) current or past drivers, (3) a minimum active range of motion of 15 degrees of ankle dorsiflexion and 5 degrees of active plantarflexion against gravity, and (4) have the ability to understand and follow a three step command (e.g., “Take this piece of paper in your right hand. Fold it in half. Put the paper on the floor.”). Exclusion criteria were (1) presence of any other neurological or musculoskeletal disorder, (2) pain or injury affecting limb movements, (3) spatial neglect, vision and hearing impairments, (4) psychiatric illness (such as clinical depression or anxiety) or untreated sleep disorder, and (5) history of simulator sickness. The self-reports on these impairments were used to screen the participants. Prior to participation, all individuals read and signed an informed consent approved by the University of Florida’s Institutional Review Board.
2.2 Experimental procedures: The experimental session lasted ~3 hours. Participants completed clinical, cognitive, motor, and driving assessments.
2.2.1 Clinical assessments: We examined the global cognitive status with the Montreal Cognitive Assessment (MoCA), a widely used cognitive screening measure where a lower score indicates impaired cognitive status (23). In the stroke group, we assessed the severity of motor impairments using lower extremity subsection of the Fugl-Meyer Assessment (FMA), such that a lower score indicates poor motor function. We determined the self-reported driving behavior by current driving status, driving exposure, space, avoidance and citations (24). We obtained a self-reported driving score by combining the above factors, with a higher score (maximum 15) indicating superior driving behavior.
2.2.2 Cognitive assessments: We used the Useful Field of View (UFOV) test to assess processing speed, divided and selective attention (25). Extensive research suggests that UFOV is a strong predictor of safe driving in older adults (26-28) and stroke survivors(9). Poor performance on the UFOV test is linked with increased crash risk and decline in driving mobility among older adults (29). Experimental set up: Participants sat in an upright position in front of a 32-inch monitor (Sync Master™ 275t+, Samsung Electronics America, NJ, USA) placed 1.25 m away at eye level. The monitor displayed the UFOV assessment (Figure 1A).
UFOV task: For processing speed, we asked participants to identify a briefly presented stimulus (a car or a truck) in the center of the computer screen. For divided attention, we instructed the participants to identify a central stimulus and simultaneously localize a peripheral stimulus. The selective attention task was identical to the divided attention except that the peripheral target was embedded within several distractors. Data measurement and analysis:Processing Speed: Processing speed measured the time needed to accurately identify a central stimulus. A longer time to correctly identify to central stimuli indicated slower speed of processing. Divided attention: Divided attention measured the ability to attend to central and peripheral stimuli simultaneously. A longer time to accurately respond to both stimuli indicated poorer divided attention. Selective attention: Selective attention measured the ability to direct attentional processes to two specific stimuli while voluntarily suppressing attention to distractors. A longer time to accurately respond indicated poorer selective attention.
2.2.3 Motor assessments: We assessed the maximal force produced with the maximum voluntary contraction (MVC) task and the motor accuracy with the visuomotor tracking task. The motor assessments were performed on the paretic leg in stroke and the non-dominant leg in the control group.
Experimental set up: Participants were seated comfortably in an upright position in front of a 32-inch monitor (Sync Master™ 275t+, Samsung Electronics America, NJ, USA) placed 1.25 m away at eye level. The monitor displayed the visual feedback of participants’ performance and the target trajectory (only in visuomotor tracking task). Participants confirmed that they could see the visual display. The hip joint was at ~90° flexion and 10° abduction, the knee at ~90° flexion, and the ankle in a neutral position. Participants maintained a stable posture and avoided extraneous movements at the hip, knee or trunk. The experimenter monitored the participant’s posture to ascertain compliance.
MVC task: We asked participants to exert maximum isometric force for 3s during ankle dorsiflexion and plantarflexion. Each participant completed three to five MVC trials until three MVC trials were within 5% of one another. A 60s rest period was provided between trials to minimize fatigue. The task order for ankle plantarflexion and dorsiflexion was randomized between participants. Force measurement and analysis: A force transducer (Model 41BN, Honeywell, Morristown, NJ, USA) located parallel to the force direction on a customized foot device measured the MVC force. Force signals were sampled at 1000 Hz (NI-DAQ card, Model USB6210, National Instruments, Austin, TX, USA), band-pass filtered from 0.03 to 20 Hz, and amplified by a gain factor of 50 (Bridge-8 world precision instrument Inc., FL, USA). The data were stored on a research workstation for offline analysis. Strength: We determined the maximum force for each trial as the average of 10 samples around the peak force. We quantified strength as the highest force obtained among 3-5 MVC trials.
Visuomotor tracking task:Figure 1B shows the placement of the participant’s foot on an adjustable foot plate secured with Velcro straps to ensure simultaneous movement between the foot plate and participant’s foot. We asked participants to track a sinusoidal target (red line), as accurately as possible using ankle dorsiflexion and plantarflexion movements. Participants received real-time visual feedback of their performance via a blue line that was superimposed on the target. The target frequency was 0.3 Hz. Ankle joint movement ranged from 5° ankle plantarflexion to 15° ankle dorsiflexion. Participants performed 2-3 practice trials and 5 test trials. Each trial lasted ~35s with a 30s rest period provided between successive trials to minimize fatigue. Ankle position measurement and analysis: A low-friction potentiometer (SP22G-5K, Mouser Electronics, Mansfield, TX, USA) located laterally to the fibular malleolus enabled the measurement of ankle position during the visuomotor tracking task. The ankle position signals were sampled at 1000 Hz (NI-DAQ card, Model USB6210, National Instruments, Austin, TX, USA). Visual presentation of each trial was controlled via a custom routine written in Matlab® (Math Works™ Inc., Natick, Massachusetts, USA). The position signal was band-pass filtered between 0.2 and 0.4 Hz to remove the task-related frequency (sinusoidal target at 0.3 Hz). Data were stored and analyzed offline using a custom routine written in Matlab® program. Motor Accuracy: We measured the accuracy of the ankle position as the root mean squared error (RMSE). RMSE quantifies the distance between the target and participant’s position. To account for initial and final position adjustments, we eliminated the first 10s and the last 5s of position data from the analysis. We computed the mean RMSE as the average of the 5 trials.
Driving assessment: We used a commercial driving simulator (AplusB software, Myrtle Beach, South Carolina, 7 USA) to conduct the driving assessment.
Experimental set-up: Participants sat in a professional driving simulator seat with a gas pedal and a brake pedal. Figure 1C shows placement of the participant’s ankle during the driving task. The simulated driving task was performed with the paretic leg in stroke and the non-dominant leg in the control group. The simulated driving environment was displayed on three 24-inch computer monitors.
Simulated driving task: The simulated driving environment included driving a Toyota Yaris on a winding road in clear and sunny weather. We instructed participants to drive in the center of the driving lane at 30 km/h for 3 minutes. At random times during the driving course, a STOP stimulus would randomly appear. We asked participants to respond to the STOP sign as quickly as possible by releasing the gas pedal and pressing the brake pedal. Prior to testing, participants practiced 2 short driving trials. Braking Time measurement and analysis: We measured braking time as the time between the presentation of the STOP stimulus and the application of the brake pedal, averaged over 10 trials. One stroke participant could not complete the simulated driving task and was excluded from the analysis.
Statistical Analysis
We tested the normality of data using the Shapiro-Wilk test. Given that our data were normally distributed, we compared the stroke and control groups using independent samples t-test on i) cognitive assessments: processing speed, divided attention, and selective attention, ii) motor assessments: plantarflexion and dorsiflexion strength and accuracy of ankle position, and iii) braking time. Effect sizes were reported with Cohen’s d. To determine the relationship between cognitive, motor, and driving performance, we performed Pearson’s bivariate correlation. To assess whether motor and cognitive abilities contribute to braking time (criterion variable), we performed a separate hierarchical multiple regression analysis with selective attention as predictor variable in model 1, adding accuracy of ankle position as predictor variable in model 2. The squared multiple correlation coefficient (R2) and the adjusted squared multiple correlation coefficient (adjusted R2) determined the goodness-of-fit of the model. Statistical analysis was conducted with the alpha level set at 0.05 using the IBM SPSS 24.0 (IBM, Armonk, NY).