There are approximately 1.6 to 3.8 million sport-related concussions per year in the United States.1 In high school and collegiate athletics, concussions account for 8.9% and 5.8% of all athletic injuries, respectively.2 Due to the high incidence of sport-related concussion, appropriate management of concussions is critical. The current concussion consensus statement from the 5th International Conference on Concussion in Sport recommends a multifaceted approach to the evaluation of concussions, including assessments of postural stability.3 Postural stability assessments may include a baseline and post-injury evaluation, both of which should be conducted in similar conditions.3 The National Athletic Trainers' Association (NATA) recommends an objective assessment of postural control as a significant aspect of concussion management, and the National Collegiate Athletic Association (NCAA) requires baseline postural stability assessments for student-athletes participating in high-risk sports.4 Despite sports medicine associations' recommendations to include postural stability assessments in concussion management, it is historically difficult to accurately measure postural stability in a time-expedient manner.
Traditional static postural stability assessments implemented in the clinic to evaluate a population with concussion include the Balance Error Scoring System (BESS)5 and the Modified Balance Error Scoring System (mBESS), which is used in the Sports Concussion Assessment Tool 5 (SCAT5).6 Although the BESS is a quick and inexpensive assessment of static postural stability,7 it has limitations including fatigue effects,8 learning effects,9,10 and vulnerability to intra-rater and inter-rater error,11 which may lead to inaccurate scoring.12 Technology-assisted approaches, such as the Sensory Organization Test (SOT) and portable force plates, have also been used to objectively measure static postural stability.13,14 These technologies allow for improved reliability and sensitivity, but they require equipment that is expensive, time-intensive, cumbersome, and often inaccessible to the patient and health care providers.15
Overall, it is a challenge for health care providers to accurately objectify static postural stability that can be assessed clinically for baseline or post-concussion assessment.12 Wearable technology is an alternative method of quantifying static postural stability that is time-efficient, easy to use, and portable. Inertial measurement units (IMUs) are an example of clinically accessible, wearable technology that can be used to quantify static postural stability. IMUs contain accelerometers, gyroscopes, and magnetometers, which allow for the linear and rotational analysis of postural stability measures.16 Instrumenting clinical tests with inertial sensors, such as IMUs, is thought to be useful for clinicians due to the automated programmability of the device, ease of administration, and capability of detecting underlying postural stability deficits that may otherwise be overlooked in traditional evaluations.17 Previous literature has validated IMUs against the BESS in healthy adults to quantify static postural,18,19 indicating that IMUs may be an appropriate alternative method to measure static postural stability without the current limitations of visually rated clinical assessments.
Given the recent proliferation of inertial sensors and the need for valid, objective measurements of postural stability of patients with concussion, this systematic review had two purposes: (1) to compare the effectiveness of inertial sensors to measure static postural stability deficits in a population with concussion and (2) to critically assess the methodological quality of the studies that used inertial sensors to measure static postural stability deficits in a population with concussion. We hypothesized that inertial sensors would prove to be effective in identifying individuals with a concussion in a valid, reliable, and time-expedient manner.
Throughout the preparation of this article, we followed the guidelines proposed by the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement for standardized reporting of systematic reviews. An electronic search of PubMed, CINAHL, and SportsDiscus databases was performed between May and June 2017. We searched the literature with combinations of accelerometer, gyroscope, inertial sensor, smartphone, app, wearable, postural stability, sway, balance, concussion, TBI, and traumatic brain injury.
Study Inclusion Criteria
Articles were included in the systematic review if they satisfied the following criteria: (1) were written in English; (2) were available in full text; (3) were published after 2000; (4) used individuals after concussion; (5) assessed individuals who were in high school or college; and (6) used an inertial sensor to measure static postural stability. This date range was chosen due to the recent proliferation and use of IMUs since 2000 to quantify human movement.20
Two authors (MPB, CJB) independently reviewed the articles for inclusion and exclusion criteria. Each author compiled studies that adhered to the criteria for recommendation to include in the systematic review. Following the compilation, a consensus was made between the two authors to finalize the studies to include in the systematic review. Any discrepancy in scoring between the two authors was settled by the senior author.
The authors of this systematic review created and used a modification of the Physiotherapy Evidence Database (PEDro) scale.21 A comparison between the original PEDro scale and the modified PEDro scale is detailed in Table A (available in the online version of this article). Items 2 and 3 pertained to the allocation of individuals, which were not relevant to prospective studies assessing instrumented postural stability. To ensure an adequate sample size, item 2 was modified to pertain to performing a power analysis. Items 3 and 4 were modified to pertain to the validity and the reliability of the postural stability measures used in comparison to the instrumented measures. The language of item 4 was changed to modified item 5 to reflect an expected difference in baseline factors in participants with concussion versus those without. Item 6 was modified to pertain to the gathering of baseline data regarding the participant's postural stability prior to concussion. Items 7 and 8 were modified to pertain to whether or not the assessors of postural stability were blinded to concussion status and baseline measures of postural stability. Finally, original items 8, 10, and 11 became modified items 9, 10, and 11, respectively.
Original and Modified PEDro Scales
The initial search of combinations of accelerometer, gyroscope, inertial sensor, smartphone, app, and wearable yielded 45,589 studies. When those terms were combined with postural stability, sway, or balance, the results were narrowed to 1,263 studies. When the last search terms were added to the previous search, 29 studies remained. Titles and abstracts were then scanned for the inclusion criteria (Figure 1). Four studies remained following the elimination of duplicates and studies that did not meet the inclusion criteria. The entire manuscript was reviewed for inclusion criteria by each author. The authors came to a consensus that resulted in four studies for inclusion in the systematic review.
Flow chart of systematic literature search results and data abstraction.
None of the four studies included in this systematic review met all criteria within the modified PEDro Scale (modified PEDro score of 10). One study received a modified PEDro score of 5 of 10, two attained a score of 4 of 10, and one attained a score of 3 of 10 (Table 1).
Systematic Literature Review Overview
Participant Age and Time Since Concussion
Two studies included individuals who were college-aged who had suffered a concussion.17,22 One of these studies included individuals who were evaluated within 1 month after concussion,1,7 whereas the other study included individuals who were evaluated 1 to 4 days after concussion.22 One study included individuals aged 13 to 19 years who were evaluated 5 ± 3.3 months after concussion,16 whereas another study evaluated individuals who were high school-aged within 2 weeks after concussion.23 All studies included in the systematic review included healthy matched controls.16,17,22,23
Postural Stability Assessment
King et al.22 compared an inertial sensor instrumented mBESS to the visually rated mBESS, whereas King et al.16 examined these two components in the modified and full BESS format. The Balance Accelerometer Measure (BAM) was also compared to the visually rated BESS in one study.23 A final study examined the differences between groups using inertial sensors during the BESS.17
The Effectiveness of Inertial Sensors in Identifying Individuals With Concussion
Three of the research groups reported that the postural stability performance from instrumented assessments differed between individuals with a previous concussion and healthy controls.16,17,22 King et al.22 reported that mediolateral sway during double-legged stance with eyes closed was the most sensitive in identifying individuals with concussion. Similarly, Doherty et al.17 reported that bilateral stance position of the instrumented assessment was more sensitive compared to tandem stance and unilateral stance in identifying group differences. King et al.16 also reported that the instrumented BESS was more sensitive than the other instrumented and visually rated postural stability assessment used in the study. Findings contrary to the previous studies were reported by Furman et al.,23 stating that the BESS was more sensitive than an instrumented approach in identifying individuals with concussion.
The main finding of this systematic review is that inertial sensors may have an important role in objectively measuring static postural stability in individuals with concussion. Three studies concluded that using an inertial sensor during a postural assessment can help discriminate between concussion and no concussion groups.16,17,22 However, one study reported that the visually rated BESS was more effective than the BAM in identifying abnormal postural stability in individuals with concussion.23 These studies provide insight into quantifying static postural control, which is an important aspect of a concussion evaluation.
Based on our review, using inertial sensors to assess mediolateral sway during bilateral stance may have the best clinical impact for concussion assessment. Using an inertial sensor during bilateral stance was the most sensitive in discriminating between individuals with concussion and healthy controls compared to tandem stance and single-leg stance.17,22 Measuring mediolateral postural sway during bilateral stance was more sensitive over the other outcome measures in discriminating between groups.22 An important consideration in these findings is that the SOT, a commonly used instrumented postural stability assessment, only considers sway in the anteroposterior direction.24 Previous literature suggests that when the feet are close together, individuals become more sensitive to mediolateral perturbations of sensory information rather than anteroposterior perturbations of sensory information.25 Failing to include multiple measures of postural stability, such as the exclusion of mediolateral in the SOT, may not provide a comprehensive evaluation of postural control and may, in turn, overlook significant identifiers of postural deficits following concussion.
The clinical BESS was more sensitive in identifying differences between groups compared to the instrumented method using the BAM.23 The BAM was developed as a part of the National Institutes of Health Toolbox project to quantify neurobehavior measurements over a lifetime (ages 3 to 85 years), specifically for longitudinal studies and clinical trials.23 The authors concluded that the BAM was not sensitive enough to identify differences between groups. The BAM protocol consists of (1) standing on a firm surface with feet next to each other with eyes open; (2) standing on a firm surface with feet next to each other with eyes closed; (3) standing on a foam surface with feet next to each other with eyes open; (4) standing on a foam surface with feet next to each other with eyes closed; (5) standing on a firm surface in a tandem stance with eyes open; and (6) standing on a firm surface in a tandem stance with eyes closed.23 It is important to note that sway metrics during the BAM were collected independently from the visually rated BESS and only the anteroposterior path length was reported. The other studies included in this review collected sway metrics, including both anteroposterior and mediolateral sway, and visually rated data simultaneously.16,17,22 Collecting visually rated and inertial sensor instrumented measures simultaneously during a single postural stability assessment allows for direct comparison between objective and subjective measures to determine the reliability of new technology compared to traditional methods. This is a crucial practice to follow to appropriately determine if new technology (such as inertial sensors) is an effective method to clinically assess postural stability.
The measures selected to quantify postural sway differed across studies. King et al.22 used three measures: root mean square (RMS) of the acceleration time series, total power of the acceleration time series, and mean distance from the center of the acceleration. Another study only used RMS,16 whereas two additional studies used ellipsoid volume of sway.17,23 These methodological differences are important to note due to previous literature suggesting larger effect sizes between groups if RMS or total power had been used rather than ellipsoidal volume of sway,22 therefore potentially influencing the results of the study.
Although convenient and applicable, the use of technology such as IMUs may be susceptible to error. While using inertial sensors for data collection, researchers should be aware of drift error. The effect of drift on data is the accumulation of multiple sources of error over a period of time during data collection.26 One major source of error contributing to drift is bias error. Inertial sensors implement multiple integration calculations for positional information, which inherently include an error term in the mathematical operation.27 This inherent error accumulates over time during data collection and leads to bias error in the output.26 Most inertial sensors account for bias error via internal filtering, but the sensors may not mitigate all accounts of error.27 Therefore, it is warranted that researchers be aware of possible errors in the data output from inertial sensors.
One limitation of the literature included in this systematic review is that all of the studies evaluated high school- or college-aged individuals, therefore limiting the generalizability of the findings to other populations. Additionally, three studies used a fairly heterogeneous sample regarding time following a concussion.16,17,23 This heterogeneity may have resulted in the investigators measuring fundamentally different subgroups within their experimental group. Although three of the four articles supported the use of IMUs to identify individuals with concussion, these articles were of poor to fair methodological quality, as measured by our modified PEDro scale. This is an indication that further research is warranted to investigate the use IMUs in evaluating static postural stability of individuals with concussion for future consideration of implementation in clinical settings. Finally, three studies did not blind postural stability assessors to an individual's grouping (concussion or control) and therefore potentially introduced investigator bias.16,17,23
Implications for Clinical Practice
Future research should aim to study a more homogeneous group of individuals to better measure the sensitivity of using inertial sensors in acute and subacute concussive cases. Additionally, given the findings that mediolateral sway is more sensitive in identifying group differences compared to anteroposterior sway, future studies should focus on further validating which measures and conditions are most sensitive in identifying postural stability deficits following concussion. Although the validation of IMUs to assess kinetics is still being investigated, some studies have proposed the use of IMUs to predict certain components of ground reaction forces for some dynamic tasks (linear movement and sprinting)28,29 and postural control assessments as acceptable.30 Further investigation of using inertial sensors for longitudinal concussion assessments is warranted to identify postural stability progression during recovery following injury. Similar components and measurements from inertial sensors are often found in cell phone applications, which can prove to be an accessible and cost-effective method to evaluate postural stability in clinical settings. Although there is a lack of available scientific literature, the reliability of such applications should be further investigated.
This systematic review highlights the potential use of inertial sensors to identify postural deficits in a population with concussion. Provided the evidence suggests instrumented postural stability assessments are more sensitive than visually rated clinical postural stability assessments in identifying deficits after concussion, inertial sensors may provide an affordable, accurate, and time efficient method to objectify postural stability.
- Report on Trends and Participation in Youth Sports. National Council of Youth Sports. https://ncys.org/reportontrends.html. Published 2001. Accessed 2017.
- Gessel LM, Fields SK, Collins CL, Dick RW, Comstock RD. Concussions among United States high school and collegiate athletes. J Athl Train. 2007;42:495–503.
- McCrory P, Meeuwisse W, Dvorak J, et al. Consensus statement on concussion in sport-the 5th international conference on concussion in sport held in Berlin, October 2016. Br J Sports Med. 2017;51:838–847.
- Broglio SP, Cantu RC, Gioia GA, et al. National Athletic Trainers' Association position statement: management of sport concussion. J Athl Train. 2014;49:245–265. doi:10.4085/1062-6050-49.1.07 [CrossRef]
- Riemann BL, Guskiewicz KM, Shields EW. Relationship between clinical and forceplate measures of postural stability. J Sport Rehab. 1999;8:71–82. doi:10.1123/jsr.8.2.71 [CrossRef]
- Echemendia RJ, Meeuwisse W, McCrory P, et al. The Sport Concussion Assessment Tool 5th edition (SCAT5). Br J Sports Med. 2017;51:848–850.
- Starling AJ, Leong DF, Bogle JM, Vargas BB. Variability of the modified balance error scoring system at baseline using objective and subjective balance measures. Concussion. 2015;1:CNC5. doi:10.2217/cnc.15.5 [CrossRef].
- Wilkins JC, Valovich McLeod TC, Perrin DH, Gansneder BM. Performance on Balance Error Scoring System decreases after fatigue. J Athl Train. 2004;39:156–161.
- Valovich McLeod TC, Perrin DH, Guskiewicz KM, Shultz SJ, Diamond R, Gansneder BM. Serial administration of clinical concussion assessments and learning effects in healthy young athletes. Clin J Sport Med. 2004;14:287–295. doi:10.1097/00042752-200409000-00007 [CrossRef]
- Broglio SP, Zhu W, Sopiarz K, Park Y. Generalizability theory analysis of Balance Error Scoring System reliability in healthy young adults. J Athl Train. 2009;44:497–502. doi:10.4085/1062-6050-44.5.497 [CrossRef]
- Finnoff JT, Peterson VJ, Hollman JH, Smith J. Intrarater and inter-rater reliability of the Balance Error Scoring System (BESS). PM R. 2009;1:50–54. doi:10.1016/j.pmrj.2008.06.002 [CrossRef]
- Zhu M, Huang Z, Ma C, Li Y. An objective Balance Error Scoring System for sideline concussion evaluation using duplex kinect sensors. Sensors (Basel). 2017;17:E2398. doi:10.3390/s17102398 [CrossRef]
- Monsell EM, Furman JM, Herdman SJ, Konrad HR, Shepard NT. Computerized dynamic platform posturography. Otolaryngol Head Neck Surg. 1997;117:394–398. doi:10.1016/S0194-5998(97)70132-3 [CrossRef]
- Alsalaheen BA, Haines J, Yorke A, Stockdale K, Broglio S. P.Reliability and concurrent validity of instrumented Balance Error Scoring System using a portable force plate system. Phys Sportsmed. 2015;43:221–226. doi:10.1080/00913847.2015.1040717 [CrossRef]
- Guskiewicz KM. Balance assessment in the management of sport-related concussion. Clin Sports Med. 2011;30:89–102. doi:10.1016/j.csm.2010.09.004 [CrossRef]
- King LA, Horak FB, Mancini M, et al. Instrumenting the Balance Error Scoring System for use with patients reporting persistent balance problems after mild traumatic brain injury. Arch Phys Med Rehabil. 2014;95:353–359. doi:10.1016/j.apmr.2013.10.015 [CrossRef]
- Doherty C, Zhao L, Ryan J, Komaba Y, Inomata A, Caulfield B. Quantification of postural control deficits in patients with recent concussion: an inertial-sensor based approach. Clin Biomech (Bristol, Avon). 2017;42:79–84. doi:10.1016/j.clinbiomech.2017.01.007 [CrossRef]
- Alberts JL, Thota A, Hirsch J, et al. Quantification of the Balance Error Scoring System with mobile technology. Med Sci Sports Exerc. 2015;47:2233–2240. doi:10.1249/MSS.0000000000000656 [CrossRef]
- Brown HJ, Siegmund GP, Guskiewicz KM, et al. Development and validation of an objective Balance Error Scoring System. Med Sci Sports Exerc. 2014;46:1610–1616. doi:10.1249/MSS.0000000000000263 [CrossRef]
- Gouwanda D, Senanayake SMNA. Emerging trends of body-mounted sensors in sports and human gait analysis. In: Abu Osman NA, Ibrahim F, Wan Abas WAB, Abdul Rahman HS, Ting HN, eds. 4th Kuala Lumpur International Conference on Biomedical Engineering2008, 25–28 June 2008. , Kuala Lumpur, Malaysia. ( IFMBE Proceedings. ), vol 21. Berlin, Heidelberg: Springer; 2008:715–718.
- Blobaum P. Physiotherapy evidence database (PEDro). J Med Libr Assoc. 2006;94:477–478.
- King LA, Mancini M, Fino PC, et al. Sensor-based balance measures outperform modified Balance Error Scoring System in identifying acute concussion. Ann Biomed Eng. 2017;45:2135–2145. doi:10.1007/s10439-017-1856-y [CrossRef]
- Furman GR, Lin C, Bellanca JL, Marchetti GF, Collins MW, Whitney SL. Comparison of the balance accelerometer measure and Balance Error Scoring System in adolescent concussions in sports. Am J Sports Med. 2013;41:1404–1410. doi:10.1177/0363546513484446 [CrossRef]
- McCrea M, Guskiewicz K, Randolph C, et al. Incidence, clinical course, and predictors of prolonged recovery time following sport-related concussion in high school and college athletes. J Int Neuropsychol Soc. 2013;19:22–33. doi:10.1017/S1355617712000872 [CrossRef]
- O'Connor SM, Kuo AD. Direction-dependent control of balance during walking and standing. J Neurophysiol. 2009;102:1411–1419. doi:10.1152/jn.00131.2009 [CrossRef]
- Maklouf O, Adwaib A. Performance evaluation of GPS INS main integration approach. World Acad Sci Eng Technol Int J Mech Aerosp Ind Mechatron Eng. 2014;8:476–484.
- Quinchia AG, Falco G, Falletti E, Dovis F, Ferrer C. A comparison between different error modeling of MEMS applied to GPS/INS integrated systems. Sensors. 2013;13:9549–9588. doi:10.3390/s130809549 [CrossRef]
- Gurchiek RD, McGinnis RS, Needle AR, McBride JM, van Werkhoven H. The use of a single inertial sensor to estimate 3-dimensional ground reaction force during accelerative running tasks. J Biomech. 2017;61:263–268. doi:10.1016/j.jbiomech.2017.07.035 [CrossRef]
- Setuain I, Lecumberri P, Ahtiainen JP, Mero AA, Häkkinen K, Izquierdo M. Sprint mechanics evaluation using inertial sensor-based technology: a laboratory validation study. Scand J Med Sci Sports. 2018;28:463–472. doi:10.1111/sms.12946 [CrossRef]
- Lockhart TE, Soangra R, Zhang J, Wu X. Wavelet based automated postural event detection and activity classification with single IMU (TEMPO). Biomed Sci Instrum. 2013;49:224–233.
Systematic Literature Review Overview
|Author||Purpose||Participants||Assessment and Instrumentation||Results||Modified PEDro Scale|
|King et al.22||Determine if inertial sensors were more sensitive compared to visually rated clinical assessments and which stance of the mBESS differentiates between groups||College-aged student athletes who had (1 to 4 days) (n = 52) and did not have (n = 76) acute concussion||Instrumented mBESS and mBESS||Inertial sensors detected group differences. Visually rated clinical mBESS did not. Mediolateral sway using inertial sensors best classified those with acute concussion||5|
|King et al.16||Determine if modified conditions and/or using inertial sensors would improve the ability to properly determine participants who suffered mTBI||Adolescents with non-acute concussion (5 ± 3.3 months) (n = 13) and demographically matched healthy controls (n = 13)||BESS, mBESS, instrumented BESS, and instrumented modified BESS||The mBESS using inertial sensors provided the best diagnostic classification capability||4|
|Doherty et al.17||Quantify postural control using inertial sensors||College-aged individuals with acute concussion (1 month) (n = 15) and age- and sex-matched healthy controls (n = 15)||Instrumented BESS||BESS using inertial sensors differentiated between groups. Increased postural sway in bilateral stance was apparent in individuals with concussion||4|
|Furman et al.23||Evaluate the ability of the BAM and BESS to assess postural control in individuals with concussion at different time points||High school student athletes with acute concussion (8 ± 3 days) (n = 10), subacute concussion (151 ± 215 days) (n = 33), and healthy controls (n = 27)||BAM and BESS||Visually rated BESS is more effective compared to the BAM in identifying postural control deficits in individuals with concussion||3|
Original and Modified PEDro Scales
|Assessment||Revised Category||Original Category|
|External validity/applicability||1. Eligibility criteria were specified||1. Eligibility criteria were specified|
|Internal validity||2. A power analysis was performed to determine sample size
3. Reliability data were available for all applicable measures
4. Validity data were available for all applicable measures
5. Other than the measures of interest, the groups were similar at baseline
6. Baseline data were gathered for all participants
7. Assessors of postural stability were blinded to all baseline data
8. Assessors of postural stability were blinded to concussion status
9. Measures of at least 1 key outcome were obtained from more than 85% of the individuals initially recruited||2. Individuals were randomly allocated to groups
3. Allocation was concealed
4. The groups were similar at baseline regarding the most important prognostic indicators
5. There was blinding of all participants
6. There was blinding of all therapists who administered the therapy
7. There was blinding of all assessors who measured at least 1 key outcome
8. Measures of at least 1 key outcome were obtained from more than 85% of the individuals initially allocated to groups
9. All individuals for whom outcome measures were available received the treatment or control condition as allocated|
|Interpretability||10. The results of between-group statistical comparisons were reported for at least 1 key outcome
11. The study provided both point measures and measures of variability for at least 1 key outcome||10. The results of between-group statistical comparisons were reported for at least 1 key outcome
11. The study provided both point measures and measures of variability for at least 1 key outcome|
|Scoring||Assign 1 point for each item in criteria 2 to 11 that is met. The first criterion is not included in the total
All points were summed to yield a total score up to 10 points|