Autonomic Dysfunction and Exercise Intolerance in Concussion – A Scoping Review

Abstract

Purpose:

Concussion commonly results in exercise intolerance, often limiting return to activities. Improved understanding of the underlying mechanisms of post-concussive exercise intolerance could help guide mechanism-directed rehabilitation approaches. Signs of altered cardiovascular autonomic regulation – a potential contributor to exercise intolerance – have been reported following concussion, though it is not clear how these findings inform underlying mechanisms of post-concussive symptoms. Systematic summarization and synthesis of prior work is needed to best understand current evidence, allowing identification of common themes and gaps requiring further study. The purpose of this review was to (1) summarize published data linking exercise intolerance to autonomic dysfunction, and (2) summarize key findings, highlighting opportunities for future investigation.

Methods:

The protocol was developed a priori, and conducted in five stages; results collated, summarized, and reported according to PRISMA guidelines. Studies including injuries classified as mTBI/concussion, regardless of mechanism of injury, were included. Studies were required to include both autonomic and exercise intolerance testing. Exclusion criteria included confounding central or peripheral nervous system dysfunction beyond those stemming from the concussion, animal model studies, and case reports.

Results:

A total of 3,116 publications were screened; 17 were included in the final review.

Conclusion:

There was wide variability in approach to autonomic/exercise tolerance testing, as well as inclusion criteria/testing timelines, which limited comparisons across studies. The reviewed studies support current clinical suspicion of autonomic dysfunction as an important component of exercise intolerance. However, the specific mechanisms of impairment and relationship to symptoms and recovery require additional investigation.

INTRODUCTION

Exercise intolerance, characterized by the reproduction and/or exacerbation of post-concussive symptoms with activity, is a common consequence of concussion often limiting return to usual activities. In combination with nearly 30 other features used to identify concussion, post-concussive exercise intolerance is commonly considered an indirect indicator of autonomic dysfunction in the course of clinical evaluation [1–3]. Investigatively, an array of clinical phenotypes (or subtypes) have been described, with potential value in guiding therapy, predicting prognosis, and/or defining common mechanisms [2–4]; among these, exercise intolerance and autonomic dysfunction are often considered together, along with cervical musculoskeletal, vestibular/oculomotor, balance/motor function, mental/behavioral health, cognitive functioning and sleep/wake changes.

Post-concussive exercise intolerance has been linked to a reduction in cerebral blood flow, theoretically prolonging effects of metabolic energy crisis associated with the injury [5], and has been attributed to post-injury changes in autonomic function. In current clinical practice, testing for post-concussion exercise intolerance typically involves graded bike or treadmill exercise while recording a patient’s symptoms and/or heart rate (HR) [6–10], and is used as an aid to guide return to play decisions. For example, individuals experiencing intolerance of maximal exertion during exercise testing may be restricted from return to sport-related activities, which approximates that level of exertion, and instead directed toward a submaximal or graded exercise program [11]. The major challenge with current exercise intolerance testing protocols is that they provide little localizing information as to the origin of the exercise-provoked symptoms, and/or as to whether they are due to autonomic dysfunction, or another activity-provoked symptom-driver (e.g., vestibular, oculomotor, cervicogenic). In contrast, cardiovascular autonomic reflex testing, involving discrete measurement of neuroanatomically well-defined reflex arcs [12–20], has the advantage of being mechanistically localizing, while also providing a physiological correlate to subjectively reported symptoms.

The purpose of this review was to 1) summarize the available literature in concussion populations that utilize assessment of both cardiovascular autonomic function and exercise intolerance testing, and 2) map the evidence for autonomic dysfunction as a contributor to exercise intolerance in people who have recently sustained a concussion.

METHODS

The protocol was developed a priori and registered to Open Science Framework [21]. This review was conducted as a scoping review, with guidance from the latest version of the JBI Manual for Evidence Synthesis [22]. Utilizing the framework as outlined by Arksey and O’Malley [23], and expanded by Peters et. al. [24], the review was conducted in five stages: 1) identify research question, 2) identify relevant studies, 3) study selection, 4) chart data and 5) collate findings, summarizing and reporting the results [22, 25, 26]. For transparency and reproducibility, we adhered to the PRISMA reporting guidelines for scoping reviews and literature searches [27–29].

Literature search.

The literature search was conducted by an information specialist (MMM) with assistance from a medical librarian for the initial Medline strategy, our primary database, then translated for the other databases using a combination of keywords and controlled subject headings unique to each database. Concepts included “post-concussion” or “concussion” and “exercise testing” or “exercise tolerance” or “exercise intolerance” and “autonomic function” or “autonomic dysfunction.” No date nor language limits were applied, although searches specified human participants due to the large number of animal studies. Peer review of the strategy was conducted by a library colleague using the PRESS guidelines [30].

Information sources.

Electronic databases included Medline (Ovid) 1946–2022, Embase (embase.com) 1974 – 2022, CINAHL Complete (Ebscohost) 1937–2022, Cochrane Library (wiley.com) 1898 – 2022 including CENTRAL (wiley.com) 1898–2022, APA PsycINFO (Ebscohost) 1872–2022, Dissertations & Theses Global (ProQuest) 1861–2022, Scopus (scopus.com) 1970–2022. References of included studies were checked to discover other relevant studies. No grey literature was searched. Citation management and duplicate detection and removal was accomplished with EndNote (Clarivate Analytics, Philadelphia, PA, USA). See supplemental file for detailed search strategies for all databases.

Study Selection:

Inclusion criteria:

Individuals diagnosed with mTBI/concussion. No restriction on stage of recovery (inclusive of: immediate, 0–7 days; acute, 1–6 weeks; post-acute, 7–12 weeks; chronic, >12 to 52 weeks) [3], sex, or age. Studies containing cardiovascular autonomic assessment were required to include measurement of HR or blood pressure (BP) metrics during standardized, quantitative autonomic testing (tilt table, supine-to-sit/stand, Valsalva maneuver, deep breathing, etc.) or while performing exercise/physical activity requiring movement of the lower extremities (excluding those restricted to contraction of a single muscle group). Studies including exercise intolerance testing were required to utilize physical exertion while measuring HR, BP, and/or symptom provocation.

Exclusion criteria:

Confounding population factors: any central or peripheral nervous system dysfunction (e.g., vestibular/somatosensory pathology, Parkinson’s Disease, etc.), pregnancy, orthopedic injuries that affect balance and gait. Moderate or Severe TBI[3]: loss of consciousness > 30 min, Glasgow Coma Scale <13, posttraumatic amnesia (PTA) > 24 h. Animal models. Review articles, clinical trials, dissertations, expert opinion, case studies, and conference abstracts.

Two reviewers (RP, ES) independently screened titles and abstracts, then reviewed full text articles. When no consensus was reached, a third reviewer (MC) was available to resolve conflicts. Covidence (Veritas Health Innovation, Melbourne, Australia), an online systematic reviewing platform, to screen and select studies. Details on the screening process are shown in Figure 1.

Fig. 1.

PRISMA flow diagram [27]

Quality assessment

In line with scoping review methodology, no quality assessment of included studies was conducted, as our goal is to rapidly map and summarize existing literature [22].

Data Charting:

One reviewer (RP) charted data and a second reviewer (ES) verified the data. Microsoft Excel (Microsoft Corporation, Redmond, WA, USA) was used to extract and chart our data. The team developed and piloted a charting form before finalizing our protocol and identified the following data: author, year, population, testing timeline, cardiovascular/autonomic analysis measures, group comparisons, autonomic testing results, symptoms related to exercise testing.

RESULTS

Study selection.

Review and article selection is summarized in Figure 1 and Table 1. The searches resulted in a total of 8,105 reports; 4,989 duplicates were removed, resulting in 3,116 results screened. Title and abstract screening were conducted to evaluate relevance to primary aim, which removed an additional 2960 articles. Of the remaining 156 reports evaluated for eligibility at full-text review, 17 were included in the final review; summarized in Figure 1. Of those excluded, 55 did not include a cardiovascular autonomic assessment or an exercise intolerance test that measured HR, BP, or symptom provocation during physical exertion. 8 articles did not test a concussion population, and 76 did not meet inclusion based on study type. Included articles were organized into themes based on exercise testing. Within each theme, key points regarding autonomic testing performed, acuity of testing and findings related to linking autonomic and exercise testing were summarized. See supplmental file for bibliographies of a) included studies and b) excluded studies with reasons for exclusion.

Table 1.

Summary and analysis of the outcome measures (autonomic and exercise) performed in each of the included studies, population tested, testing timeline, methods performed in each of the included studies, and the changes found in both the autonomic measure and symptom provocation during the testing.

Author	Year	Population/Context (age years ± SD)	Testing Timeline	Cardiovascular/Autonomic Analysis Measures	Group Comparisons	Autonomic Testing Results	Symptoms Related to Exercise Testing
Squat-to-Stand
Bishop et al.	2017	12 (age 17.78 ± 2.33) male athletes with sport concussion (ice hockey)	Within 72hrs post-concussion	RR mean, SDRR, HR, SDHR, NN50, pNN50, VLF, LF, HF, %LF, %HF, LF:HF ratio, SD1, SD2, ApEn, sEn, BP, MAP	Concussion/Controls	Decreased SDHR in the concussion group during both standing and the squat-to-stand maneuver	Recorded symptoms as descriptive data
Wright et al.	2018	14 (age 19 ±1.4) male athletes with sport concussion (sport not specified)	Preseason, 72hrs, 2 weeks, 1-month	BP, sysMCAv, diasMCAv, HR, PETCO2	Concussion subjects at multiple timepoints	Decreased Phase (timing offset/response latency) during the 0.10 Hz maneuver for both systolic and diastolic measures at 72hr and 2 weeks compared to preseason. Increased diastolic Gain at 2 weeks compared to preseason.	Recorded symptoms as descriptive data
Wright et al.	2018	18 (age 18.6 ±1.5) male athletes with sport concussion (ice hockey, American football)	Baseline, 72hrs, 2 weeks, 1-month	MAP, HR, MCAv, PETCO2	Concussion subjects at multiple timepoints	Decreased Phase (timing offset/response latency) during the 0.10 Hz maneuver at both 72hr and 2 weeks compared to preseason	Recorded symptoms as descriptive data
Moir et al.	2018	19 (age 15 ±2) female (13) and male (6) athletes with sport concussion (sport not specified)	Initial visit (unknown) and subsequent visits all within 12 weeks post-concussion	MAP, MCAv, BP, HR, SV, CO	Concussion subjects at multiple timepoints/Controls	Decreased rate of regulation at first visit compared to the healthy controls. Rate of regulation at the final visit showed two distinct groups with some in the concussion group experiencing full recovery compared to healthy controls, and others continuing to demonstrate decreased rate, despite symptom resolution.	Recorded symptoms as descriptive data
Cycle Ergometer
Gall et al.	2004	14 (age 18.1 ±0.4) male athletes with sport concussion (ice hockey)	Once asymptomatic (3.6 – 6.4 days) and at 5 days post-asymptomatic.	RR mean, SDRR, LF, HF, %LF, %HF, LF: HF ratio	Concussion subjects at multiple timepoints/Controls	Decreased RR mean, LF, and HF in the concussion group at both testing timepoints	Subjects were required to be asymptomatic before starting the exercise protocol.
Gall et al.	2004	14 (age 17.8 ±0.5) athletes (gender not specified) with sport concussion (ice hockey)	72hrs after being asymptomatic and at 5 days post-asymptomatic	RHR, PEHR, MHR	Concussion subjects at multiple timepoints/Controls	Increase in all three HR measures in the concussion group at both testing timepoints	Subjects were required to be asymptomatic before starting the exercise protocol.
Memmini et al.	2021	33 (age 16 ±1) male athletes (ice hockey)	Lifetime history of concussion (6.9–47.9 months since last concussion)	mean NN interval, RMSSD, SDNN	Concussion (separated into history of 1 concussion & 2 or more concussions)/Controls	Decrease in all measures during exercise in both concussion groups and decrease in all measures during recovery period in the 2 or more concussion group	Symptoms were not reported
Harrison et al.	2021	34 (age 16.1 ±1) male athletes (ice hockey)	Lifetime history of concussion (6.4–41.8 months since last concussion)	mean NN interval, RMSSD, SDNN	Concussion/Controls	Increased SDNN and RMSSD in the concussion group at with both the cognitive and exercise task	Subjects were required to be asymptomatic before starting the exercise protocol.
Howell et al.	2021	40 (age 17.6 ± 2.2) female (22) and male (18) athletes with sport concussion (sport not specified)	Within 3 weeks of injury and still symptomatic at rest and then again at 8 weeks post injury.	HR, BP, FETCO2, MCAv	Concussion subjects at multiple timepoints/Controls	Increased FETCO2 at the same MCAv in the concussion group during exercise	Graded exercise testing based on symptom provocation
Treadmill
Kozlowski et al.	2013	34 (age 25.9 ±10.9) female (17) and male (17) individuals with sport and non-sport concussion	Average of 226 (34–949) days post-concussion	HR, BP	Concussion/Controls	Decreased HR and systolic BP in the concussion group at time of test cessation Increased diastolic BP in the concussion group at time of test cessation	Graded exercise testing based on symptom provocation
Leddy et al.	2013	8 (age range 17–52) female (4) and male (4) individuals with sport and non-sport concussion	Average of 118 (33–270) days post-concussion but still symptomatic and then again after 12 weeks	HR, BP	Concussion subjects divided into a stretching group and exercise group	Decreased HR at time of test cessation in the stretching group at 12 weeks	Graded exercise testing based on symptom provocation
Clausen et al.	2016	9 (age 23 ±6) female athletes with sport concussion (soccer, basketball, and volleyball)	Symptomatic for more than 6 weeks, but less than 12 weeks	BP, VE, MCAv, PETCO2	Concussion/Controls	Decreased VE, HR, and systolic BP in the concussion group at time of test cessation Increased MCAv and PETCO2 in the concussion group at each stage of the graded treadmill exercise test.	Graded exercise testing based on symptom provocation
Hinds et al.	2016	40 (age 15.5 ±0.9) female (17) and male (23) athletes with sport concussion (sport not specified)	Within 5 days post-concussion and still symptomatic and again once asymptomatic	HR	Concussion subjects at multiple timepoints/Controls	Decreased HR at the start of exercise in the concussion group when symptomatic compared to the concussion group when asymptomatic, and healthy controls.	Comparison between symptomatic and once asymptomatic on graded exercise testing
Leddy et al.	2018	27 (age range 14–19) female (9) and male (18) athletes with sport concussion (sport not specified)	Average of 4 (1–10) days post-concussion	HR	Concussion subjects	Lower heart rate at test cessation was strongly associated with prolonged recovery time	Graded exercise testing based on symptom provocation
Morissette et al.	2020	34 (age range 14–19) female (15) and male (19) individuals with sport and non-sport concussion	Average of 44.8 (16–144) days post-concussion but still symptomatic	HR, VO2, VCO2, VE	Concussion/Controls	No significant differences between the concussion group and the healthy controls in any of the measured variables	Graded exercise testing based on symptom provocation
Worts et al.	2022	19 (age range 13–18) female (6) and male (13) athletes who sustained a sport concussion (sport not specified)	Average of 4.5 (3–7) days post-concussion	MAP, RMSSD, LF%, LF:HF ratio	Concussion/Controls	Decreased RMSSD in the concussion group during exercise in both exercise groups Increased MAP in the concussion group during exercise in both exercise groups	Recorded symptoms as descriptive data
Cycle Ergometer and Treadmill
Haider et al.	2019	20 (age 15.9 ±1.1) female (8) and male (12) individuals with sport and non-sport concussion	Average of 5.6 (3–8) days post-concussion	HR	Concussion subjects	Decreased HR at time of test cessation in the concussion group on both the treadmill and bike exercise tests	Graded exercise testing based on symptom provocation

Abbreviations: Standard Deviation (SD), RR standard deviation (SDRR), Heart Rate (HR), standard deviation of heart rate (SDHR), the number of intervals that differ by more than 50 milliseconds (NN50), the percentage of NN50 relative to the sample (pNN50), very low frequency (VLF), low frequency (LF), high frequency (HF), normalized low frequency (%LF), normalized high frequency (%HF), the low frequency–high frequency ratio (LF:HF ratio), Short-Term Variability (SD1), Long-Term Variability (SD2), Approximate Entropy (ApEn), Sample Entropy (sEn), Blood Pressure (BP), mean arterial pressure (MAP), systolic middle cerebral artery blood velocity (sysMCAv), diastolic middle cerebral artery blood velocity (diasMCAv), partial pressure of end-tidal CO2 (PETCO2), middle cerebral artery blood velocity (MCAv), stroke volume (SV), cardiac output (CO), resting heart rate (RHR), post-exercise heart rate (PEHR), maximum heart rate (MHR), root mean square of successive differences (RMSSD), and standard deviation of NN interval (SDNN), fraction of end tidal CO2 (FETCO2), Borg Rating of Perceived Exertion (RPE), minute ventilation (VE), oxygen consumption (VO2), carbon dioxide production (VCO2).

Summary of evidence.

For purposes of literature mapping, included studies were grouped according to the type of exercise testing utilized:

Squat-to-Stand:

Bishop et al. (2017) [15]

Retrospective mixed-method study consisted of 12 (age 17.78 ±2.33) male athletes who sustained a sport concussion (ice hockey), compared to healthy teammates as control subjects. Testing was performed within 72-hours of injury and consisted of data collection at rest while standing and during a 10-second squat-to-stand maneuver. Resting and baroreflex-evoked heart rate (EKG) data were considered; and standard deviation and frequency analysis of HR and BP were analyzed in the 0–5 seconds and 6–10 seconds periods of squatting and standing. Symptoms were recorded at rest with no mention of symptom provocation during squat-to-stand maneuver. The primary cardiovascular autonomic variables collected were: standard deviation of heart rate (SDHR), low frequency (LF) [31–34], high frequency (HF) [31–34]. RR mean, RR standard deviation (SDRR), the number of intervals that differ by more than 50 milliseconds (NN50), the percentage of NN50 relative to the sample (pNN50), very low frequency (VLF) [34], normalized low frequency (%LF), normalized high frequency (%HF), the low frequency–high frequency ratio (LF:HF ratio), Short-Term Variability (SD1), Long-Term Variability (SD2), Approximate Entropy (ApEn), Sample Entropy (sEn), maximum systolic BP, minimum diastolic BP and mean arterial pressure (MAP) were also reported. Results from this study found (isolated) decreased SDHR in the concussion group compared to the control group during both standing and the squat-to-stand maneuver, without significant differences in LF, HF or other measures.

Wright et al. (2018a) [35]

Longitudinal study consisted of 136 male contact-sport (type of sport not specified) athletes of which 14 (age 19 ±1.4) sustained a concussion during their competitive season. Testing was performed during five minutes of repetitive squat–stand maneuvers at 0.05 Hz and 0.10 Hz and consisted of data collection at 72hr, 2 weeks, and 1 month post-injury. Beat-by-beat BP/HR and middle cerebral artery blood velocity were recorded. Symptoms recorded at testing timepoints, but no analysis performed. The following cardiovascular/autonomic variables of interest were: BP, systolic middle cerebral artery blood velocity (sysMCAv), diastolic middle cerebral artery blood velocity (diasMCAv), HR, and partial pressure of end-tidal CO₂ (PETCO₂); measures were used to perform transfer function analysis of (systolic BP/ diastolic BP) – (sysMCAv/diasMCAv) to estimate coherence (correlation), gain (amplitude ratio), and phase (timing offset/response latency) at follow-up compared to preseason [36]. Overall, the results showed reduced response latency (phase) at 72hrs and 2 weeks (despite clinical symptom resolution) following an acute concussion, which was relatively recovered by 1 month.

Wright et al. (2018b) [37]

Longitudinal study consisted of preseason testing of 179 male elite, junior-level (ice hockey, American football) athletes of which 18 (age 18.6 ±1.5) sustained a concussion during their competitive season. Testing was performed during five minutes of repetitive squat–stand maneuvers at 0.05 Hz and 0.10 Hz and consisted of data collection at 72hr, 2 weeks, and 1 month post injury. Symptoms recorded at testing timepoints, but no analysis performed. The following cardiovascular/autonomic variables were collected: MAP, HR, middle cerebral artery blood velocity (MCAv), which were used to determine the relationship between BP–MCAv using transfer function analysis to estimate coherence (correlation), gain (amplitude ratio), and phase (timing offset/response latency) [36]. This study found that response latency (phase) was reduced following an acute concussion, compared to preseason, at both 72hr and 2 weeks post injury, with recovery by 1 month. Athletes with a history of three or more concussions did not exhibit any greater impairment than those with zero prior concussions.

Moir et al. (2018) [38]

Longitudinal study consisted of 19 (age 15 ±2) female (13) and male (6) athletes who sustained a sport concussion (type of sport not specified) compared to healthy control athletes (9 female/9 male) who were between the ages of 12–18 years old, had not had a concussion within the last year and did not have any persistent symptoms from previous concussions. Testing was performed during two sit-to-stand repetitions and consisted of data collection acutely, following concussion (though specific window of time after injury was not specified); follow-up testing was performed within 12-weeks of the concussion injury. Symptoms were recorded and all subjects were said to be symptom free by 12 weeks post-concussion. The following cardiovascular/autonomic variables were collected: MAP, MCAv, BP, HR, stroke volume (SV), cardiac output (CO). Analysis focused on the dynamic rate of cerebral autoregulation during the sit-to-stand trial, based on the rate of change in cerebrovascular resistance relative to the change in arterial blood pressure [39]. Results from this study showed that the dynamic rate of cerebral autoregulation, was reduced in all concussed athletes compared to the healthy controls. At the concussed athletes’ final visit (9 athletes completed the final visit at 12 weeks), two themes were observed: a minority of individuals (n=2) experienced full recovery of their dynamic rate of regulation, whereas others (n=7) continued to demonstrate reduced dynamic regulation, despite symptom resolution.

Cycle Ergometer:

Gall et al. (2004a) [40]

Longitudinal study consisted of 14 (age 18.1 ±0.4) male athletes who sustained a sport concussion (ice hockey) compared against healthy teammates as a control population. Testing was performed once asymptomatic (3.6 – 6.4 days) and again 5 days post-asymptomatic and consisted of data collection during a cycle ergometry protocol of a 2-min warm-up, followed by a low-moderate intensity steady state 10-min exercise bout. Symptoms were recorded after injury to determine if athlete was asymptomatic and if not, then symptoms were recorded every 3 days until asymptomatic. A 5-min ECG sample from minutes 4 to 9 of the low-moderate intensity steady state exercise bout was used to assess heart rate variability (HRV) during exercise. The following autonomic/HRV variables were collected: RR mean, SDRR, LF, HF, %LF, %HF, and LF: HF ratio. Results from this study showed no difference at rest between the concussed athletes and their matched controls, but during both exercise tests, the concussed group demonstrated a significant decrease in the RR mean, LF, and HF compared to their matched controls.

Gall et al. (2004b) [41]

Longitudinal study consisted of 14 (age 18.1 ±0.4) athletes (gender not specified) who sustained a sport concussion (ice hockey) compared against healthy teammates as a control population. Although not clarified within this paper, there is the potential that this study population is the same as the study population in Gall et al. 2004a study [40]. Testing was performed 72 hours after being asymptomatic (4.9–8.5 days) and again 5 days post-asymptomatic and consisted of data collection during a cycle ergometry protocol of a 2-minute warm-up, followed by a low-moderate intensity steady state 10-minute exercise bout, and then a high intensity interval session consisting of repeated 40-second high intensity bouts followed by a 20-second free pedal and a subsequent 20-second rest period. The high intensity exercise bouts were repeated until the athlete could no longer maintain the workload. The addition of the high intensity intervals is an addition to the protocol used in the previous Gall et al. 2004a study [40]. Symptoms were recorded after injury to determine if athlete was asymptomatic and if not, then symptoms were recorded every 3 days until asymptomatic. The following variables were collected: resting heart rate (RHR), post-exercise heart rate (PEHR), maximum heart rate (MHR), and number of high intensity bouts tolerated. Here, while the number of high intensity bouts tolerated was similar, the average HR during steady state exercise was higher in the concussed group compared to matched controls at both timepoints.

Memmini et al. (2021) [42]

Case-control study consisted of 33 (age 16 ±1) male athletes (ice hockey) that were divided into those with or without a concussion history. Those with a concussion history were binned on total count: 1 concussion (n=11; 6.9–47.9 months since last concussion) or 2 or more concussions (n=5; 11.6–28.4 months since last concussion). Testing consisted of HR data collection during a 5min resting period, followed by 20min of cycling at a steady-state of approximately 60% to 70% of his age-predicted (220–age) maximum heart rate, and a 9-minute post-exercise cool down. Symptoms were not discussed in this study. The following HRV variables were collected: mean NN interval, root mean square of successive differences (RMSSD), and standard deviation of NN interval (SDNN). This study showed that both concussion groups demonstrated a decrease in all three measures of HRV during exercise compared to the control group, which persisted during the post-exercise cool down in the group with 2 or more concussions, while they resolved in the group with only 1 concussion.

Harrison et al. (2021) [43]

Case-control study that consisted of 34 (age 16.1 ±1) male athletes (ice hockey) that were divided into those with concussion (n=16; 6.4–41.8 months since last concussion) and without a concussion history. Although not clarified within this paper, there is the potential that this study population has significant overlap with the study population in Memmini et al. (2021) [42]. Testing consisted of data collection during a cognitive task, a 5-minute warm-up period, followed by 20 minutes of cycling at a steady-state of approximately 60% to 70% of his age-predicted (220–age) maximum heart rate, a 2-minute active cool down (unloaded cycling), a 10-minute seated resting recovery, and repeat cognitive testing. Symptoms were recorded and it was reported that all athletes in both the with and without concussion history groups were asymptomatic at the time of testing. Relative to other studies in this review, this study uniquely added assessment of autonomic function during a cognitive task designed to elicit stress. The following HRV variables were collected: mean NN interval, RMSSD, and SDNN. This study reported that during completion of the cognitive task at rest and following aerobic exercise, athletes with a history of concussion demonstrated significantly higher SDNN and RMSSD than those without a concussion.

Howell et al. (2021) [44]

Longitudinal study consisted of 40 (age 17.6 ±2.2) female (22) and male (18) athletes who sustained a sport concussion (type of sport not mentioned in study), and remained symptomatic, compared against matched (no significant difference in age, sex, height, and weight) healthy controls who had not experienced a concussion within the previous year. Testing was initially performed within 3 weeks of injury and then again at 8 weeks post injury and consisted of data collection during a modified YMCA branching bike exercise protocol [45]. Symptoms were recorded at rest and during exercise testing, and if symptoms increased by 3/10 or more on a visual analog scale, testing was stopped [46]. The following cardiovascular/autonomic variables were collected with seated rest and during exercise: HR, BP, fraction of end tidal CO₂ (FETCO₂), and MCAv. Results demonstrated a significant relationship between cerebrovascular responses to CO₂ at rest (cerebral vasoactivity) and to exercise-induced changes in FETCO₂ in concussion subjects, as well as a higher HR at which these participants at which point their FETCO₂ reached plateau compared to the healthy control group.

Treadmill:

Kozlowski et al. (2013) [47]

Cross-sectional study consisted of 34 (age 25.9 ±10.9) female (17) and male (17) individuals with concussion injuries due to sport or non-sport related incidents (e.g., car accidents, falls, non-assault related head contact) compared against healthy controls matched for age, sex and sports participation history. Testing was performed at an average of 226 (34–949 days) days post-injury and consisted of data collection during before, during and after a graded treadmill exercise test (Balke Protocol).[48] Symptoms we recorded during testing, and testing was stopped with any increase in symptom intensity, the presence of any new symptoms (Borg Rating of Perceived Exertion; RPE) [49], or when participants could no longer maintain the appropriate speed. The following cardiovascular variables were collected: HR and BP measured before, during each minute of, and immediately after the graded treadmill exercise test. Results from this study showed that mean exercise test duration was 9.4 minutes less for the concussion group than the healthy control group. At the time of test cessation, the concussion group had a lower HR, lower systolic BP and a higher diastolic BP than the healthy control group. After adjusting for differences in test duration, the RPE was the only measure to show an overall difference between groups, consistent with a symptom limited exercise test (without evidence of associated physiological limitations).

Leddy et al. (2013) [50]

Longitudinal randomized controlled study consisted of 8 (age range 17–52) female (4) and male (4) individuals with concussion injuries due to sport or non-sport related incidents that were split in an exercise treatment group or a stretching group to test the effects of aerobic training (vs stretching as a control) on post-concussive HR and symptom responses to exercise. Testing was performed at an average of 118 (33–270 days) days post-injury and then again 12 weeks after initial testing, and consisted of data collection during a graded treadmill exercise test (Balke Protocol) [48]. Symptoms were recorded during testing and testing was stopped with any increase in symptom intensity, presence of new symptoms, or if participants could no longer maintain the appropriate speed. The following cardiovascular variables were collected every 2 min during testing: HR and BP. Results from this study showed that after 12 weeks, the stretching group had a lower HR (and more symptoms) at time of test cessation than the exercise group, mirroring the pattern seen by Kozlowski et al. examining post-concussed (without treatment intervention) compared to controls [47].

Clausen et al. (2016) [51]

Case control study consisted of 9 (age 23 ±6) female athletes who sustained a sport concussion (soccer, basketball, and volleyball), and who were symptomatic for more than 6 weeks, but less than 12 weeks, compared to recreational aerobic athletes with similar age, height and weight as healthy controls. Testing was performed at an average of 9 weeks (6–12 weeks) post-injury, as well as post-aerobic treatment program in a subset of individuals (n=6), and consisted of data collection during a graded treadmill exercise test (Balke Protocol) [48]. Symptoms were recorded during testing, and testing was stopped with any increase in symptom intensity, the presence of any new symptoms, or when participants could no longer maintain the appropriate speed. The following cardiovascular variables were collected: BP, minute ventilation (V_E), MCAv, and PETCO₂. Results from this study showed that the concussion group had lower V_E, HR, and systolic BP at time of test cessation than healthy controls and altered sensitivity to CO₂ (greater MCAv and PETCO₂) compared to the healthy controls at each stage of the graded treadmill exercise test. The aerobic treatment program normalized V_E, PETCO₂, CBFV and exercise tolerance in the subset receiving therapy (n=6).

Hinds et al. (2016) [52]

Longitudinal study consisted of 40 (age 15.5 ±0.9) female (17) and male (23) athletes who sustained a sport concussion (type of sport not specified) and were still symptomatic, compared to healthy controls who were not significantly different than the concussion group in age, Body Mass Index (BMI) and mean resting HR. Testing was performed at an average of 5 days (4–6 days), and again when asymptomatic (for concussed subjects; no range of time to recovery provided); exercise testing consisted of data collection during a graded treadmill exercise test (Buffalo Concussion Treadmill Test; BCTT) [11]. Symptoms were recorded at rest and during exercise testing (along with RPE); testing was stopped if symptoms increased by 3/10 on a visual analog scale or if participants could no longer maintain the appropriate speed [46]. The following primary cardiovascular variable was collected: HR. Here, HR was lower at the onset of exercise in the symptomatic concussion group compared to their recovered (asymptomatic) concussion state, and healthy control groups; resting HR and relative HR increment with each increase in workload (during Buffalo testing) did not differ between symptomatic and recovered athletes.

Leddy et al. (2018) [46]

Prospective randomized controlled study to determine the safety of exercise tolerance evaluation using the BCTT, and whether low HR at symptom-exacerbation threshold (as reported in prior studies [47, 50, 51]) has prognostic value for predicting recovery; consisted of 27 (age range 14–19) female (9) and male (18) athletes who sustained a sport concussion (type of sport not specified). Participants were randomized to receive BCTT or none. Testing was performed at an average of 4 days (1–10 days) post-injury and again at 21 days post-injury, and consisted of data collection during a graded treadmill exercise test (Buffalo Concussion Treadmill Test).[11] Symptoms were recorded at rest and during exercise testing. Testing was stopped if symptoms increased by 3/10 on a visual analog scale, or if participants could no longer maintain the appropriate speed [46]. The following primary cardiovascular variable was collected: HR. Results from this study showed that BCTT did not appear to delay symptomatic recovery in those receiving BCTT, compared to those who did not. In those receiving BCTT, a lower HR at test cessation on the first visit was strongly associated with continued symptom presence during testing at 21 days post-injury.

Morissette et al. (2020) [53]

Case control study consisted of 34 (age range 14–19) female (15) and male (19) individuals with concussion injuries due to sport or non-sport related incidents and were still symptomatic, compared against healthy controls that were not significantly different in age or height but did have a significant difference in weight and BMI. Testing was performed at an average of 44.8 days (16–144 days) post-concussion and consisted of data collection during a graded treadmill exercise test (BCTT) [11]. Symptoms were recorded at rest and during exercise testing. If symptoms increased by 3/10 on a visual analog scale or if participants could no longer maintain the appropriate speed, testing was stopped [46]. The following cardiovascular variables were collected: HR, oxygen consumption (VO₂), carbon dioxide production (VCO₂), and V_E. Despite a significant difference in RPE during aerobic testing, this study showed no significant difference in physiological variables between the concussion group and the healthy controls.

Worts et al. (2022) [54]

Randomized controlled study to examine the effects of two different intensities of aerobic exercise vs. rest on cardiovascular autonomic [as well as oculomotor and vestibular] function and symptom burden; consisted of 19 (age range 13–18) female (6) and male (13) athletes who sustained a sport concussion (type of sport not specified) compared to age and sex matched healthy controls. Testing was performed at an average of 4.5 days (3–7 days) post-concussion and consisted of data collection during a single, 20-minute session of treadmill walking at either 40% or 60% of age-predicted max heart rate. Symptoms were recorded both pre- and post-exercise. The following cardiovascular/autonomic variables were collected: MAP, RMSSD, LF%, LF:HF ratio. This study found that RMSSD was lower and MAP higher in post-concussion athletes during exercise for both intensity groups, compared to healthy controls, but that all variables returned to pre-exercise values within 5 minutes of exercise testing (in both groups). There were no significant group effects or group by time interaction for LF% or LF:HF ratio for the low (40% age-predicted max HR) intensity exercise group, though for the moderate (60% age-predicted) group, RMSSD was lower and LF:HF, LFnu, and MAP were higher during the warm-up and exercise bout, compared to supine rest. Finally, when evaluating the effects of orthostatic position change (supine to seated), the orthostatic challenge resulted in reduced RMSSD, and increased LF:HF, LFnu and MAP across all groups.

Bike and Treadmill:

Haider et al. (2019) [55]

Case control study to test the comparability of the Buffalo Concussion Bike Test (BCBT) to the BCTT; consisted of 20 (age 15.9 ±1.1) female (8) and male (12) individuals with concussion injuries due to sport or non-sport related incidents compared to age and sex matched healthy controls. Testing was performed at an average of 5.6 days (3–8 days) post-concussion and consisted of data collection during graded treadmill (BCTT) and bike (Buffalo Concussion Bike Test; BCBT) exercise tests [11, 55]. Symptoms and RPE were recorded at rest and during each exercise test. For each exercise test, if symptoms increased by 3/10 on a visual analog scale or if participants could no longer maintain the appropriate speed, testing was stopped [46]. The following primary cardiovascular variable was collected: HR. Results from this study showed that at the time of test cessation, on both the treadmill and bike exercise tests, the concussion group had a lower HR than the healthy controls. Additionally, HR at symptom threshold did not differ between BCTT vs BCBT.

DISCUSSION

The intent of this review was to summarize the available literature in concussion populations that provide assessment of both cardiovascular autonomic function and exercise intolerance testing in order to better understand existing evidence for a relationship between cardiovascular autonomic dysfunction and exercise intolerance. Additionally, we sought to identify themes and gaps in this literature, illustrating opportunities for future study with the potential to improve approaches to therapeutic exercise in post-concussive rehabilitation.

Across the studies reviewed, altered autonomically mediated cardiovascular control – in particular, indices of heart rate variability (HRV) and HR during and following exercise – after concussion injury was reported in all but two studies [54, 55]. While a handful of studies [15, 40, 54] attempt to parse parasympathetic and sympathetic function using frequency and time domain-based analyses (e.g., LF, HF, LF:HF ratio, etc.) [31–34], none of the studies revealed by our search criteria utilized full, comprehensive autonomic reflex screening in combination with exercise testing in the same individuals. Given the breadth of measures utilized to assess cardiovascular control (e.g., heart rate or blood pressure changes, and variability of each), a common consensus on pattern or mechanism of impairment could not be extracted.

Exercise intolerance testing performed by assessing symptom provocation during exercise and/or the general presence of symptoms was discussed in all but one article [42], yet most studies did not report on the association between symptom provocation (whether it was during exercise or not) and measures of cardiovascular control. Eight studies did use symptom provocation as a termination point of exercise testing, per established protocol for formalized exercise intolerance testing [44, 46, 47, 50–53, 55]. Overall, given that the presence of symptoms is currently the hallmark of post-concussion diagnosis [56], the limited assessment of the association of activity-provoked worsening of symptoms to measures of cardiovascular control makes it difficult to find a common theme regarding the relationship of persistent post-concussive symptoms to altered cardiovascular control. Interestingly, three studies required subjects to be asymptomatic at time of testing and demonstrated differences in LF, HF, HRV, and HR during exercise between concussed and control subjects [40, 41, 43], and a single study showed impaired cardiovascular control in both symptomatic and asymptomatic concussed athletes [38] – potentially revealing persistent deficits in cardiovascular function, independent of ongoing symptoms and apparent recovery.

Gaps in Existing Literature

In general, the heterogeneity of participant selection and methodological differences in research design across the studies reviewed limits actionable conclusions from existing data. Several key themes emerged which highlight existing gaps: 1) None of the identified studies performed a comprehensive, standardized autonomic testing battery, such as the widely used Ewing’s Battery (a clinically validated assessment of sympathetic and parasympathetic function) [57–60], and many relied on indirect measures of cardiovascular and cerebrovascular function to draw conclusions about possible underlying autonomic mechanisms. 2) Despite activity related symptom reproduction being a cornerstone to diagnosis of post-concussive syndrome [56], assessment of formalized symptom driven exercise intolerance testing varied widely across studies, and was lacking in many. 3) Finally, perhaps the most problematic variability across studies was that of timeline. Studies in this review ranged from evaluation within 72 hours [15, 35, 37, 40], up to 7 months [42, 47] post-concussion, and two studies included subjects if they had “any” lifetime history of concussion (including subjects with concussion ranging from 7 months to 3.5 years [42, 43]). While several studies did utilize a longitudinal assessment approach, the variability of time since concussion and time to follow-up evaluation represents a critical gap. Of the eight studies that did utilize a longitudinal design [35, 37, 38, 40, 41, 44, 50, 52], five of these were based on number of days post-injury [35, 37, 38, 44, 50] and in the other three studies the timeline was based on resolution of symptoms [40, 41, 52], which reduced the ability for comparison across the concussion recovery timeline. From the available longitudinal studies, there is preliminary evidence that autonomic dysfunction and exercise intolerance likely differs - perhaps even markedly - at varying timepoints along the recovery process, yet clear insight into the evolution, or potential mechanistic shifts, of autonomic dysfunction is largely lacking in current literature.

Emerging Themes

1). Plausible origins of cardiovascular autonomic dysfunction following concussion.

Strain and strain rate have been shown to contribute to injury-induced neural dysfunction, possibly preferentially impacting white matter tracts [61] within the brainstem [62]. Given this susceptibility and the role of the brainstem in cardiovascular autonomic control, many have hypothesized that impaired blood flow and inability to tolerate the demands of exercise may conceivably contribute to altered autoregulation of cerebral blood flow via impaired autonomic function [13, 63–66]. Direct measurement of cardiovascular changes during an exercise task provides task-specific information about cardiovascular responses, yet without accompanying direct autonomic measures, localizing mechanistic attributions remain referential. While all studies included in this review utilized at least one measure of cardiovascular physiology (i.e. HR, BP, HRV), many served as essential surrogates of cardiovascular autonomic function. Importantly, given not a single study directly measured autonomic function using a localizing battery of standardized autonomic tests [57–60].

2). Symptom provocation

A key heterogeneity across studies was the variable approach to measurement of exertional symptoms in concussed subjects; indeed, this is also problematic clinically, as there are numerous validated tools for this task, and there is limited consensus to guide use across contexts - perhaps with the exception of sports related concussion and the symptom scale within the SCAT [67]. While eight of the studies reviewed here utilized exercise testing based on symptom provocation [44, 46, 47, 50–53, 55], five studies simply recorded symptoms as a descriptive data point [15, 35, 37, 38, 54], and none of the studies performed directed analysis of the relationship between the exercise-based or cardiovascular physiology measures and symptom scores. Therefore, our review found no data to directly support, nor refute, the hypothesis that alteration in cardiovascular autonomic function and/or associated autoregulation of cerebral blood flow contributes to symptom provocation and subsequent exercise intolerance.

Despite the lack of direct evidence, in clinical practice, symptom provocation during exercise is commonly referred to as a result of autonomic dysfunction, presumably related to altered cardiovascular autonomic responses to exercise and/or cerebral blood flow [5, 68]. Although symptom provocation can be a result of autonomic dysfunction and/or related changes in cerebral autoregulation, it can also be a result of impairment in many other physiologic systems (e.g., vestibular, oculomotor, cervicogenic system dysfunction), which were largely not assessed nor controlled for during the exercise testing procedures performed. Symptom provocation can also be influenced by psychological stress responses during exercise, which can further impact autonomic function [63, 64, 66, 69]. This further highlights the need for directed evaluation of cardiovascular autonomic function in parallel with exercise-related provocation of symptoms and, ideally, thorough clinical characterization of study cohorts; albeit, future study is needed to further parse the most informative patterns.

3). Timeline

The intent to establish a relationship between autonomic dysfunction and exercise intolerance was further limited by the range in testing timeline across the reviewed studies. Using the VA/DoD Clinical Practice Guidelines to establish a recovery timeline (Immediate 0–7 days, Acute 1–6 weeks, Post-acute 6–12 weeks, and Chronic >12 weeks) [3], three studies evaluated the immediate period [15, 46, 54], one acute [55], two post-acute [51, 53], and three chronic [42, 43, 47]; while eight studies sampled across multiple time periods [35, 37, 38, 40, 41, 44, 50, 52]. With particular reference to those studies that evaluated multiple time points, there is some indication that physiological recovery may lag behind symptomatic recovery [38], though this observation has not been consistently replicated. Thus, on the basis of this review, it remains difficult to chart the longitudinal relationship between cardiovascular autonomic dysfunction, symptomatic exercise intolerance, and physiological recovery post-concussion.

Despite the broad range of recovery periods represented in the literature, some general themes appeared. The three studies conducted in a more acute population demonstrated a greater consistency in the patterns of altered cardiovascular autonomic dysfunction – each finding reductions in HR or HRV during exercise [15, 46, 54], though this was not universally reported [42]. Studies that included more chronic populations, or populations with a larger range of time since injury, demonstrated much more variability in their results and across studies. This result is not unexpected; autonomic dysfunction and exercise tolerance likely change over the course of the concussion recovery timeline. The eight studies that did track subjects longitudinally reported improvements towards normal in measures of resting HR, HR during exercise, BP during exercise, and MCAv throughout the recovery period, though the lack of consistency in their methods limits generalizability [35, 37, 38, 40, 41, 44, 50, 52]. Finally, several studies indicated that those with a history of more than a single concussion may have more pronounced physiological abnormalities, though this relationship was not consistently demonstrated [37, 42, 43].

Limitations of the current review

The wide variability in methods and metrics utilized to measure autonomic function and the delineation of exercise measures created challenges for the evaluators in determining if the review inclusion criteria were truly met (e.g., the study included both directed exercise tolerance and cardiovascular autonomic testing). Non-cardiovascular measures of autonomic function, such as pupillary function testing and thermoregulatory sweat testing [70–72], were not specifically included due to the authors’ desire to focus solely on cardiovascular autonomic function testing, given the proposed contribution altered cerebral blood flow plays in symptom provocation during exercise [5, 68]. Exercise testing required physical activity that involved at least the movement of the lower extremities, and not a single muscle group contraction, and the measurement of HR or BP. Similarly, studies that used HR and BP as purely descriptive data, but lacked measures of cardiovascular function as outcomes, were not included. These factors, among others cited in the above sections precluded our ability to perform a systematic review and/or meta-analysis of the data related to this topic.

CONCLUSION

While symptoms of exercise intolerance are frequently accepted as clinical signs of autonomic dysfunction following a concussion, to date, few studies have directly tested this relationship [73–78]. While it is plausible that autonomic dysfunction contributes to post-concussive exercise intolerance, direct evidence in support of this remains limited. An improved understanding of the relationship between autonomic dysfunction and exercise intolerance is needed. We recommend further studies to a) appropriately utilize a localizing battery of autonomic testing performed in parallel with exercise intolerance testing and b) additional longitudinal studies to assess the evolution of autonomic dysfunction relative to exercise intolerance over the (symptomatic and physiological) concussion recovery period.