Post-concussion exercise volume associations with depression, anxiety, and dizziness symptoms, and postural stability: preliminary findings

Abstract

Objective:

To examine the association between post-concussion exercise volume and changes in depression, anxiety, dizziness, and postural stability.

Design:

Secondary analysis of a single-site prospective clinical trial.

Setting:

Cerebrovascular research laboratory

Participants:

Participants completed questionnaires and underwent tests of gait and balance within two weeks of a concussion (mean=11±3 days post-concussion) and approximately one month later (mean=41±7 days post-concussion). Exercise volume was tracked by weekly exercise logs.

Interventions:

Based on previous work classifying exercise volume following concussion, we grouped participants based on self-reported exercise volume between visits as high exercise volume (≥150 min/week) or low exercise volume (<150 min/week).

Main outcome measures:

Participants completed assessments evaluating anxiety and depression (Hospital Anxiety and Depression Scale), dizziness (Dizziness Handicap Inventory), and postural stability (tandem gait and modified Balance Error Scoring System).

Results:

Thirty-eight participants completed the study: 22 in the high exercise volume group (mean=71±40 min/week; 16.8±2.1 years; 59% female) and 16 in the low exercise volume group (mean=379±187 min/week; 17.5±2.1 years; 31% female). Although depression symptoms were not significantly different initially (mean difference=1.5; 95% CI=−0.68, 3.68; p=0.24), the high exercise volume group had significantly lower depression symptom scores at follow-up (mean difference=3.0; 95% CI=1.40, 4.47; p<0.001). Anxiety symptoms (mean difference=2.8; 95%CI=0.3, 5.4; p=0.03), dizziness symptoms (mean difference=10.9; 95%CI=0.2, 21.5; p=0.047), single-task tandem gait (mean difference=3.1 seconds; 95%CI=0.2, 6.0; p=0.04), and dual-task tandem gait (mean difference=4.2 seconds; 95%CI=0.2, 8.2; p=0.04) were significantly better among high exercise volume group.

Conclusion:

Greater exercise volumes were associated with lower depression, anxiety, and dizziness symptoms, and faster tandem gait performance. These preliminary findings suggest a potentially beneficial role for exercise within several different domains commonly affected by concussion.

Introduction

Concussions are complex and medically challenging injuries to manage due to the multidimensional nature of post-injury presentation and the relatively few evidence-based treatments available to reduce symptom burden and improve behavioral function.¹ Among the different domains affected by concussions are increases in depression,² anxiety³, dizziness,⁴ and postural instability.⁵ However, relatively few studies have examined how an intervention can affect these domains, focusing instead on how physical activity or exercise is associated with overall concussion symptom severity.^6,7

Aerobic exercise has emerged as a feasible and effective approach to reducing overall symptom burden after a concussion.^7–10 While promising, few studies have examined how exercise volume impacts recovery. The majority of prior studies have explored an exercise volume of 100 minutes/week compared to placebo (e.g. stretching), and relied on prescription without an explicit consideration of compliance or actual exercise volume.^7,11 Our recent work suggested low adherence to an aerobic exercise recommendation upon randomization, but that higher post-concussion exercise volumes (>160 mins/week) than conventionally prescribed (>100 mins/week) were associated with symptom burden improvement within the first 6 weeks of a concussion.¹² However, this was based on overall concussion symptom inventories. The association between high post-concussion exercise volumes with other domain-specific measures (e.g. anxiety, depression) remains unknown. Given the numerous psychosocial benefits of aerobic exercise among uninjured individuals,¹³ examining these effects following concussion is warranted.

Therefore, we examined how exercise volume following concussion was associated with changes in depression, anxiety, dizziness, and postural stability. Based on our previous work, ≥160 min/week of exercise is associated with a decrease in concussion symptom burden;¹² and optimal exercise recommendations for cardiovascular health are ≥150 minutes/week.¹⁴ Given this, we compared participants who reported high exercise volume (≥150 minutes/week) vs. low exercise volume (<150 minutes/week) during a four week period, beginning within 2 weeks of sustaining a concussion. We hypothesized that the high exercise volume group would report less severe depression, anxiety, and dizziness symptoms, and would perform better on measures of postural stability, compared to the low exercise volume group during the study period.

Methods

This study is a secondary analysis of a randomized clinical trial comparing an aerobic exercise intervention within two weeks (individualized target heart rate based exercise recommendation) of a concussion to standard-of-care (symptom limited physical activity).¹² Participants were not blinded to group assignment, as they had to know the exercise volume they were assigned to meet recommended exercise guidelines. Given our previous finding that the optimal exercise volume to distinguish concussion symptom resolution after one month of study was ≥160 minutes/week¹² and optimal exercise recommendations for cardiovascular health are ≥150 minutes/week,¹⁴ we grouped participants based on whether they reported aerobic exercise ≥150 minutes/week (high exercise volume group) vs. <150 minutes (low exercise volume group). This poor compliance led us to include all participants based on their reported exercise volume (per protocol analysis). Participants were enrolled and initially tested within two weeks of concussion (average=11, SD=3 days post-concussion), completed weekly exercise logs, and returned for a follow-up visit approximately one month after the initial visit (average= 41, SD=7 days post-concussion). Most participants (79%) were no longer experiencing concussion symptoms at their 2-month visit. Given this, data collected at the 2-month assessment were not included in this secondary analysis. In-person assessments were performed in a cerebrovascular research laboratory. In addition, we selected a one-month follow-up time within the scope of the larger study, as we did not expect any cerebrovascular changes to be present at a two-week follow-up time period, and only collected patient-reported outcome data during laboratory visits.

All participants were recruited between September 25, 2018 and March 5, 2020 from a primary care sports medicine center or emergency department associated with a tertiary care regional children’s hospital. Concussion diagnosis was confirmed by the treating physician, consistent with the definition provided within the 5^th International Consensus Statement on Concussion in Sport criteria.¹⁵ We included participants who reported a post-concussion symptom inventory (PCSI) score >9 at the initial assessment to ensure that symptom recovery had not occurred prior to study enrollment. Exclusion criteria consisted of aerobic exercise contraindications, electrical implants, or a concomitant lower extremity injury at the time of initial assessment. The study protocol was reviewed and approved by the institutional review board and pre-registered (Clinicaltrials.gov ID: NCT03170856). All participants, and parents/guardians, if under the age of 18, provided written informed consent/assent.

To construct our grouping variable, participants completed weekly exercise logs each week via an online reporting platform (REDCap). We defined exercise as any physical activity they participated in with the intention to increase their heart rate. The exercise log asked participants to report, on a daily basis: if they exercised (yes/no), the time they spent exercising (in minutes if they reported any exercise) for each day of the week, and the type of exercise they participated in (mode). We grouped exercise mode responses as walking/running on a treadmill, elliptical, or outside, stationary exercise bike, strength training, stretching, or sport-specific exercise (e.g., shooting hoops in the backyard). We calculated the total exercise average minutes per week over the four weeks of study that they reported participating in exercise and constructed two groups based on these calculations (≥150 min/week=high exercise volume; <150 min/week=low exercise volume). We did not collect exercise intensity data due to the inability to have participants perform exercise under constant supervision. While we required participants to wear an activity monitoring device to minimize this shortcoming, not all participants complied with this requirement over the one-month study period.

At each visit, participants completed assessments evaluating anxiety and depression (Hospital Anxiety and Depression Scale; HADS), dizziness (Dizziness Handicap Inventory; DHI), and postural stability (tandem gait and modified Balance Error Scoring System [mBESS] tests). Within the HADS,^16,17 participants answered questions related to depression and anxiety symptoms by providing a response (ranging from 0 to 3, where 3 indicated they experienced symptoms a great deal of the time) on 14 different questions. Seven items were specific to depression symptoms, and 7 were specific to anxiety symptoms. For each scale (HADS-anxiety and HADS-depression) scores ranged from 0 (no symptoms) to 21 (maximum symptom severity). A score between 0–7 is considered “normal”, 8–10 is “suggestive of the presence of the respective state”, and a score of ≥11 indicates a probable case.¹⁸

We used the DHI to examine how dizziness affected everyday life. Participants answered 25 different questions, and points were assigned based on the response provided. Responses included they were not experiencing (0 points), sometimes experiencing (2 points), or always experiencing (4 points) the symptom described. Scores ranged from 0–100 in 2-point increments. Thus, higher scores reflect a higher level of dizziness-related impairment.¹⁹ Previous work suggests a score >10 following concussion indicates the need for specialist referral.²⁰

To assess postural stability, participants completed a single-task and dual-task tandem gait test, and the mBESS. The tandem gait test is a reliable assessment within a session and across time, and can distinguish between concussion and control patients.^21,22 During the test, participants walked without shoes using an alternate heel-to-toe gait pattern along a 3m strip of tape, while a test administrator timed how long it took them to complete. Participants were instructed to approximate their heel and toe, walk beyond the end of the 3m line, make a 180 degree turn and return to the original starting point while walking in the same heel-to-toe fashion, completing three trials in single-task and dual-task conditions. We calculated the average time from each condition for use in further analyses. During dual-task trials, participants complete one of three different cognitive test continuously throughout the trial. The cognitive test forms included spelling a five-letter word backwards, serial subtraction by 6s or 7s from a randomly presented 2-digit number, or reciting the months in reverse order starting from a randomly selected month, as described previously.^21,22

During the mBESS, participants stood on a solid surface for 20 seconds with their eyes closed. They completed one trial in three conditions: double-leg stance, single-leg stance, and tandem stance.²³ We instructed participants to remain in the test position throughout the trial, and counted an error if they opened their eyes, lifted their hands off their hips, took a step, fell out of the testing position, lifted any portion the foot off the ground, abducted the hip greater than 30 degrees, or stayed out of the test position for 5 seconds or more. The outcome variable was the number of errors committed, with the maximum being 30 errors (10 errors per stance condition). While we recognize the mBESS has limited sensitivity to concussion effects after the initial post-injury week, we included it to provide a clinically relevant multimodal assessment of postural stability.

Statistical Analysis

Data are presented as means (standard deviation) or median (interquartile range) for continuous variables, and the number included and corresponding percentage within group for categorical variables. We first compared participant characteristics between high/low exercise volume groups using independent samples t-tests, Chi square analyses, or Fisher’s exact tests. To examine differences across time and between groups, we used 2×2 repeated measures ANOVAs. Our outcome variables included anxiety and depression symptoms (HADS), dizziness (DHI), and postural stability (single/dual-task tandem gait and mBESS). Our independent variables were time and group. For omnibus ANOVA tests, statistical significance was set at p = 0.05. If there was a significant interaction, we examined differences between groups at each time point using a Bonferroni correction (p = 0.025). In addition, we compared exercise mode between groups using Fisher’s exact tests.

To examine the association between exercise volume (as a continuous variable, rather than a categorical variable as with our primary analysis) on our selected outcomes, we used univariable regression models fit separately for each dependent variable. Dependent variables consisted of anxiety symptoms (HADS-anxiety), depression symptoms (HADS-depression), dizziness symptoms (DHI), and postural stability (tandem gait and mBESS), the predictor variable was average exercise volume (in hours/week) during the month of study. Any missing data were treated as such, and no imputations were performed. All statistical analyses were two-sided, and performed with using Stata version 15 (StataCorp, College Station, TX, USA).

Results

Data were available at both time points for N=38 individuals, who were included in this analysis. For complete recruitment data, including allocation and follow-up, please refer to our original work.¹² One individual not included in the primary analysis (dropped out before completion of the 2-month period) was included in the current analysis, as she completed exercise logs for the first four months of the study. Of the 38 participants in the current study, n=22 reported were in the low exercise volume group (mean=71, SD=40 minutes/week) and n=16 were in the high exercise volume group (mean=379, SD=187 minutes/week). The two groups had similar demographic characteristics, physical characteristics, and medical history (Table 1). On average, participants reported symptom resolution 32 (SD=18.5) days post-concussion and were cleared to return to unrestricted sports 38.4 (SD=21.3) days post-concussion. There was a higher proportion of female participants in the low exercise volume group compared to the high exercise volume group, although this difference was not statistically significant (Table 1). A higher proportion of the high exercise volume group reported participating in strength training and sport-specific exercise compared to the low exercise volume group (Table 2).

Table 1.

Participant characteristics for each group.

Variable	Low Exercise Volume (n=22)	High Exercise Volume (n=16)	P value
Age (years)	16.8 (2.2)	17.5 (2.1)	0.32
Sex (female)	13 (59%)	5 (31%)	0.11
Race	White: 18 (82%) Black/African American: 0 (0%) Native Hawaiian/Other Pacific Islander: 0 (0%) Asian: 0 (0%) More than one race: 4 (18%) Unknown: 0 (0%)	White: 11 (69%) Black/African American: 1 (6%) Native Hawaiian/Other Pacific Islander: 1 (6%) Asian: 1 (6%) More than one race: 1 (6%) Unknown: 1 (6%)	0.14
Ethnicity	Hispanic/Latino: 2 (9%) Not Hispanic/Latino: 20 (91%)	Hispanic/Latino: 0 (0%) Not Hispanic/Latino: 16 (100%)	0.51
Initial visit (days post-injury)	10.9 (3.2)	11.1 (2.8)	0.79
Follow-up visit (days post-injury)	39.8 (7.3)	41.8 (7.3)	0.42
Height (cm)	168.2 (7.0)	174.2 (10.8)	0.05
Weight (kg)	64.4 (8.9)	69.5 (14.2)	0.20
LOC at time of injury	7 (32%)	3 (19%)	0.47
Memory loss at time of injury	2 (9%)	1 (6%)	> 0.99
History of ADD or ADHD	1 (5%)	2 (13%)	0.56
History of diagnosed anxiety	5 (23%)	1 (6%)	0.37
History of depression	3 (14%)	1 (6%)	0.62
History of concussion	13 (59%)	10 (63%)	0.83
Resting heart rate (bpm)	64.9 (11.1)	66.8 (11.9)	0.67
Resting systolic blood pressure (mm/hg)	113.7 (8.2)	112.9 (12.0)	0.82
Resting diastolic blood pressure (mm/hg)	65.3 (8.9)	60.8 (12.7)	0.25
Sport at time of injury	Soccer: 5 Basketball: 3 Activity of daily living: 3 Football: 2 Skiing: 2 Cheerleading: 1 Ice Hockey: 1 Ice Skating: 1 Sledding: 1 Ultimate Frisbee: 1 Volleyball: 1 Wrestling: 1	Basketball: 3 Ice Hockey: 3 Football: 2 Soccer: 2 Activity of daily living: 1 Equestrian: 1 Rugby: 1 Skiing: 1 Softball: 1 Volleyball: 1	-

Table 2.

The type of exercise that participants reported completing during the study period. As participants could select more than one option, the percentages in each group add up to more than 100.

Variable	Low Exercise Volume (n=22)	High Exercise Volume (n=16)	P value
Walking or running (treadmill, elliptical, or over ground)	13 (59%)	12 (75%)	0.49
Stationary exercise bike	10 (45%)	11 (69%)	0.20
Strength training	2 (9%)	10 (63%)	0.001
Stretching	0 (0%)	1 (6%)	0.42
Sport-specific exercise	2 (9%)	11 (69%)	<0.001

No significant between group differences were observed at the initial visit for HADS depression scores. However, at follow-up, the high exercise volume group reported significantly lower HADS depression scores than the low exercise volume group (Figure 1A). Significant main effects of group indicated that HADS anxiety (Figure 1B) and DHI (Figure 1C) scores were significantly lower for the high exercise volume group compared to the low exercise volume group across both time points. DHI scores significantly decreased across time among both groups.

Figure 1.

Patient-reported outcomes for both groups at the initial and follow-up (four weeks later) time points. Violin plots describe the distribution of measurements in each group, and data are presented as median (center dot, with corresponding number included) and interquartile range (box around the median). The shaded area represents the probability density of data at each level of the scale, smoothed using a kernel density estimator. Follow-up pairwise analyses were conducted for significant interaction effects.

For both single-task (Figure 2A) and dual-task (Figure 2B) tandem gait, main effects of group indicated that the high exercise volume group had faster (better) tandem gait performance compared to the low exercise volume group. No significant interactions or main effects were identified for the mBESS (Figure 2C).

Figure 2.

Postural stability outcomes for both groups at the initial and follow-up (four weeks later) time points. Violin plots describe the distribution of measurements in each group, and data are presented as median (center dot, with corresponding value included) and interquartile range (box around the median). The shaded area represents the probability density of data at each level of the scale, smoothed using a kernel density estimator.

Within our regression models, we found that greater exercise volumes during the month of study (hours/week) were significantly associated with lower HADS depression (β= −0.38; 95% CI = −0.63, −0.13; p=0.004; r²=0.22), lower DHI scores (β= −2.1; 95% CI = −3.8, −0.3; p=0.022; r²=0.15), faster single-task tandem gait time (β= −0.58; 95% CI = −0.95, −0.20; p=0.004; r²=0.23), and faster dual-task tandem gait time (β= −0.90; 95% CI= −1.43, −0.37; p= 0.001; r²=0.27) at the follow-up visit.

Discussion

This preliminary study indicates despite similar depression symptom severity initially, the high exercise volume group had significantly lower depression symptom severity at the end of the month of study than the low exercise volume group. Furthermore, greater exercise volumes were significantly associated with less severe anxiety symptoms, less dizziness, and better tandem gait performance throughout the study period, even though these dependent variables were evaluated through a mixture of subjective and objective measurement approaches. Although we cannot infer causality given our study design, the results extend prior observations that exercise after a concussion is associated with reduced overall concussion symptom severity^6,7,24 to include specific patient reported and clinician obtained functional domains.

Among the many documented benefits of exercise, using exercise to induce positive mood effects such as decreasing depression symptoms has been an established approach.²⁵ Our data indicate greater post-concussion exercise volume throughout the month of study was associated with lower depression symptom severity. This supports prior studies in similar populations suggesting exercise can result in lower depression symptoms: adolescent athletes who reported more hours participating in sports also reported lower depressive symptoms,²⁶ and among those with TBI, those who exercised were less depressed than those who did not exercise.²⁷ However, as a HADS score >7 is considered “abnormal”,¹⁸ and initial visit median values for both groups were below this level, basement effects with this questionnaire may have been present that mitigated our ability to understand exercise-depression associations. Furthermore, the mechanisms for changes in depression may have been due to other factors, given the subjective and screening nature of the HADS assessment. Given that we began monitoring within two weeks of injury, these results provide support for clinicians who encourage early physical activity and sub-symptom aerobic exercise as a part of the initial concussion management approach.

While anxiety and dizziness symptoms were different between groups across the testing timeline, the lack of any interaction effect for these measures suggests that the high exercise volume group were less symptomatic initially after injury than the low exercise volume group. As with our depression symptoms, we cannot determine the directionality of these results, given that those who were less dizzy and anxious may have been more willing to begin early post-concussion exercise, or, there may have been some therapeutic effect of high exercise volumes on these measures. Using the DHI, greater dizziness has been observed beyond return-to-play clearance after a concussion, suggesting persistent post-injury effects not detectable using more general concussion symptom inventories.²⁸ One review paper suggests a DHI score >10 warrants referral to a specialist.²⁰ Our data indicate the median DHI for the high exercise volume group at the follow-up visit was under this value, while the low exercise group was not (Figure 1C). However, the close relationship between anxiety and dizziness should also be considered when interpreting these results, as any sort of intervention may have an overlapping effect on these two domains.²⁹ Collectively, these results support a potential beneficial role of including exercise within a post-concussion recovery management plan, although the subjective nature of this assessment reduces ability to understand the role of exercise volume on dizziness recovery.

Among our postural control measures, both single-task and dual-task tandem gait times were significantly faster among the high exercise volume group compared to those who exercised below that level, while mBESS scores did not differ. The tandem gait test appears to be a more sensitive measurement to detect concussion-related effects beyond the acute post-injury phase.^22,30 This may be due to some previously reported limitations of the mBESS, including basement effects and poor reliability.³¹ As such, the mBESS may be ineffective as an end-point measurement to identify treatment-related effects in future concussion intervention studies. Furthermore, dual-task effects have been noted previously in relation to post-concussion exercise: those who began supervised exercise within the first week of a concussion performed better on a dual-task gait evaluation than those who did not begin exercising soon after injury.³² Thus, given the potential of exercise to influence both cognitive and motor outcomes, dual-task evaluations may be useful to determine recovery in the context of exercise treatment programs following a concussion.

Study Limitations

Our study was not without limitations, and our findings should be interpreted while considering them. Our analysis is likely to be underpowered, as the original study was not primarily designed to conduct these analyses, so our findings should be considered preliminary until further replication work is performed. Furthermore, our exercise data only contained self-reported volume information; we had no objective measurement to confirm the accuracy of the self-reported exercise times. Although we collected the data prospectively (i.e., weekly during the study), we were unable to capture exercise intensity information, which is a critical element of any exercise-based training program. Furthermore, self-reported information is subject to bias. Our study was limited to adolescents and young adults in a single geographic region seen for care at one hospital and, therefore, our results may not be generalizable to other populations. Finally, several other factors such as dizziness, gender, history of motion sickness, migraine, and family history of mental disorder^29,33,34 were not included as covariates in our analyses, and as such may have affected our observed results. In the same manner, underlying vestibular dysfunction was not formally assessed, and may have accounted for some persisting symptoms of dizziness, and the relationship between anxiety and vestibular system function was not measured.

Conclusion

Data from this preliminary investigation support past work suggesting that post-concussion exercise is associated with improvement in symptoms. Specifically, we found significant improvements in depressive symptoms among those who reported high exercise volumes (≥150 min/week during the month of study), as well as lower anxiety symptoms, lower dizziness symptoms, and faster tandem gait performance across the month of study for this group. Although our design was not equipped to identify causal effects, clinicians should continue to emphasize sub-symptom exercise as an effective treatment method during concussion recovery.

Abbreviations

ANOVA

Analysis of variance

DHI

Dizziness Handicap Inventory

HADS

Hospital Anxiety and Depression Scale

mBESS

Modified Balance Error Scoring System

PCSI

Post-concussion symptom inventory