Overload From Anxiety: A Non-Motor Cause for Gait Impairments in Parkinson’s Disease

Anxiety is one of the most under-recognized non-motor symptoms of Parkinson’s disease (PD) and has been shown to influence gait (e.g., slower velocity, smaller step length, and increased variability) when walking in threatening situations.^1–3 Interestingly, these gait deficits are remarkably similar to behaviors observed when performing a dual-task (DT) paradigm in PD.^4–7 Furthermore, gait impairments in threatening environments have been shown to be exacerbated in PD patients who are highly anxious (HA-PD) (compared with PD patients with low levels of anxiety [LA-PD]).^1,3 If anxiety imposed a processing demand similar to a DT (i.e., cognitive load) via limbic circuitry through the basal ganglia, then one might expect individuals with PD who are highly anxious to show greater impairments during normal, self-paced gait (similar to the effect of a DT in less anxious PD participants and healthy controls). Furthermore, if this was a general effect of anxiety, then the greater processing demands imposed by DT walking might lead to an equivalent detrimental influence on gait as seen in healthy older adults and typical PD patients. Thus, the present study examined whether detectable differences in gait exist between HA-PD and LA-PD during normal self-paced walking and whether these gait impairments persist while walking during a DT.

Methods

Participants

A total of 70 individuals (HA-PD participants, N=26; LA-PD participants, N=26; HC subjects, N=18) participated in this study. PD participants were subdivided into low or high trait anxiety groups based on their trait score from the State Trait Anxiety Inventory using a median split (protocol taken from⁸). PD participants who scored above 34 were classified as HA-PD, and those who scored ≤34 were categorized as LA-PD. All participants performed the Montreal Cognitive Assessment in order to rule out differences between groups due to cognitive impairment, and all PD participants were clinically assessed using the Unified Parkinson’s Disease Rating Scale motor subsection (UPDRS-III). A postural instability and gait disorder score was calculated based on the sum of UPDRS-III items 28–30. In order to rule out differences between groups due to motor symptom severity, the HA-PD and LA-PD groups were actively matched in UPDRS-III, which necessitated the removal of three participants from the LA-PD group who had the most mild symptoms according to the UPDRS-III assessment (score <13), as well as removal of three participants from the HA-PD group who had the most severe symptoms (score >40). Participants who had a clinical diagnosis of dementia were also excluded from data analysis (N=3). Thus, there were 18 HC subjects, 22 LA-PD participants, and 21 HA-PD participants remaining for inclusion in the analysis for this study. PD participants completed this study on their regular dopaminergic medication. Ethical approval was obtained from the research ethics boards at Wilfrid Laurier University and the University of Waterloo. Written informed consent was obtained from all participants, prior to their participation, in accordance with the Declaration of Helsinki.

Protocol

Participants completed six walking trials in random order: three trials of ST walking, in which they simply walked along the mat at a normal pace, and three trials of DT walking, in which they simultaneously performed a digit-monitoring secondary task. The digit-monitoring task consisted of counting the number of times a participant heard two pre-established digits (e.g., 3 and 4) spoken on an audio track. The numbers on the audio track ranged from 1 to 9, and the auditory interstimulus intervals varied from 100 ms to 1,000 ms. Additionally, participants were asked to equally prioritize gait and digit-counting tasks. In order to assess baseline ability to perform the secondary task, participants were asked to count how many times they heard the assigned digits while sitting (three randomized trials). Gait parameters such as velocity, step length, step time, step width, and each of their respective coefficients of variation (CVs) were collected and calculated using a 10-m long ProtoKinetics Zeno Walkway System (ProtoKinetics, Havertown, Pa.). All PD participants were tested in their optimal dopaminergic state.

One-way analyses of variance (ANOVAs) were used to compare demographic variables (see Table 1), and mixed repeated measures ANOVAs (group-by-task) were used to examine digit-monitoring errors and gait variables. Tukey’s post hoc analyses were used to investigate significant main effects. An alpha level of 0.05 was used to detect significant findings. A priori comparisons were also carried out to test our hypothesis regarding whether gait behavior during dual-tasking in the LA-PD and HC groups would be similar to (i.e., not significantly different) gait during normal self-paced gait in the HA-PD group.

Results

Demographic characteristics of the study participants are summarized in Table 1.

Effect of DT (Digit-Monitoring Errors)

A main effect of task was found for digit-monitoring errors (F_{[1, 58]}=6.32, p=0.015), which showed that all participants made more errors when digit monitoring while walking (DT condition) compared with seated digit monitoring. It is also noteworthy that HA-PD participants made more digit-monitoring errors overall compared with LA-PD participants and HC subjects; however, this was not statistically different between the groups (see Table 1).

Effect of Anxiety on Gait

There was a main effect of group for step length (F_{[2, 58]}=6.11, p=0.004); step length CV (F_{[2, 58]}=3.85, p=0.027); step time CV (F_{[2, 58]}=5.43, p=0.007); and velocity (F_{[2, 58]}=7.43, p=0.001). Tukey’s post hoc confirmed that HA-PD participants had significantly greater step length CV and step time CV compared with the HC group (p=0.025; p=0.002). However, the HA-PD group did not differ from the LA-PD group on step length CV or step time CV (p=0.15; p=0.16) (Figure 1). The HA-PD group also walked significantly slower and with a smaller step length compared with both the HC (p=0.006; p=0.007) and LA-PD (p=0.003; p=0.015) groups (Figure 2). Notably, the HC and LA-PD groups were not statistically different across any of the outcome gait variables. Additionally, there were no significant group effects for step width, step width variability, or step time.

A priori comparisons confirmed that there were no statistical differences between groups when comparing the HA-PD group during self-paced normal gait with HC subjects and the LA-PD group during dual-tasking gait (step length: F_{[2, 58]}=2.15, p=0.13; velocity: F_{[2, 58]}=1.65, p=0.2; step length CV: F_{[2, 58]}=2.39, p=0.10; step time CV: F_{[2, 58]}=2.71, p=0.08). This finding confirmed that the HA-PD group’s gait performance during ST walking resembled that of the LA-PD and HC performance during DT walking.

Effect of DT on Gait

As expected, there were also main effects of task across a number of gait variables (step length: F_{[1, 58]}=108.8, p<0.0001; step length CV: F_{[1, 58]}=9.76, p=0.003; step width: F_{[1, 58]}=10.02, p=0.002; step width CV: F_{[1, 58]}=4.69, p=0.035; step time: F_{[2, 58]}=39.41, p<0.0001; step time CV: F_{[1, 58]}=43.66, p<0.0001; velocity: F_{[1, 58]}=110.24, p<0.0001). These effects revealed that all participants reduced their velocity, step length, and step width CV while increasing their step length CV, step time, step time CV, and step width when walking during a DT compared with simply walking (Figures 1 and 2).

Discussion

Overall, PD participants with high trait anxiety walked with slower gait, reduced step length, and greater variability compared with HC subjects, whereas the PD group with low trait anxiety walked similarly to HC subjects in both normal and DT walking conditions. Gait velocity and step length were also worse in the HA-PD group compared with the LA-PD group across both walking conditions, despite being matched in motor symptom severity, Hoehn and Yahr staging, and postural instability and gait disorder scores. These findings support previous research suggesting that gait impairments are exacerbated in PD patients who are highly anxious.^1,3

More importantly, however, highly trait anxious PD participants, during normal self-paced walking, demonstrated similar gait deficits to those observed during DT walking in both the LA-PD and HC groups. These findings support our hypothesis that anxiety may impose processing demands (i.e., limbic load) similar to a DT (i.e., cognitive load) by consuming and potentially overloading processing resources in individuals with PD, thereby interfering with gait in a similar way that has been shown with DT interference. Furthermore, increased step length variability and step time variability have been shown to be the primary gait measures that reflect DT interference^5,6,9 and were found to be significantly worse in HA-PD participants at baseline. Additionally, the typical effects of dual-tasking on gait behavior that were observed in the HC and LA-PD groups (i.e., reduced velocity and step length) also persisted for the HA-PD group, as expected.

It is noteworthy that although performance on the secondary digit-monitoring task was not different between groups, it was significantly worse when all participants simultaneously performed the walking task (compared with seated), demonstrating that the task was effective in dividing attention away from gait. It is also important to point out that throughout the duration of this study, PD participants with low anxiety did not differ from the HC group. This is likely because PD participants were tested while on their regular dopaminergic medication, which is known to improve and normalize gait behavior. If PD participants were to be tested while off their dopaminergic replacement therapy, one might expect the findings from this study to be exaggerated, if indeed limbic circuitry does impose a load on basal ganglia processing. Future research should consider manipulating dopaminergic state while performing DT gait, in order to better understand the influence of dopamine on processing limbic and cognitive loads.

In conclusion, this study provides support for previous findings that anxiety can influence spatiotemporal aspects of gait in PD. More specifically, this study demonstrates that PD patients with high trait anxiety have worse gait at baseline and during DT walking compared with their low-anxiety counterparts, which may suggest that anxiety could be demanding shared resources and thus interfering with other processes in a similar way that a DT condition does in LA-PD patients. Therefore, those PD patients with high anxiety levels may have less capacity to cope with the increased demands of the environment and may be more prone to falling or developing freezing of gait.