In People With Subacute Mild Traumatic Brain Injury, Earlier Physical Therapy Improved Symptoms at a Faster Rate Than Later Physical Therapy: Randomized Controlled Trial

Abstract

Importance

There is unclear evidence on when to initiate physical therapy after mild traumatic brain injury (mTBI) in a non-athlete, adult population.

Objective

The objective of this study was to investigate physical therapy timing after mTBI through changes in patient-reported and clinically-assessed tools and objective and mechanism measurements of sensorimotor balance control.

Design

This study was an investigator-blinded randomized control trial (NCT03479541).

Setting

The study took place at an academic research center.

Participants

Two hundred and three participants were randomized to earlier physical therapy (n = 82) or to later physical therapy (n = 121).

Intervention

After enrollment, the earlier physical therapy group started rehabilitation within 1 week and the later group started rehabilitation after a 6-week wait period. All participants received similar rehabilitation; 6-week program administered and progressed by licensed physical therapists.

Main Outcomes and Measures

The primary outcome was the Dizziness Handicap Inventory (DHI). Secondary outcomes included common patient-reported/clinical assessments of mTBI and objective/mechanism measurements of balance, including novel measures of central sensorimotor integration. Differences between and within the groups on outcomes were examined with linear mixed-effect models, t tests, and effect sizes.

Results

While both groups significantly improved and reached similar levels on patient-reported outcomes (DHI and secondary outcomes), the earlier physical therapy group had significantly larger and faster rates of improvement compared to later physical therapy. There were differential effects of physical therapy timing on the objective/mechanism-measured outcomes. Specifically, there were significant improvements in sensorimotor time delay for the earlier physical therapy group and no change in the later group. Further, the later group worsened in the motor activation components for balance control while there was no change in the early group.

Conclusion and Relevance

Earlier physical therapy after mTBI can improve symptoms at a faster rate relative to later physical therapy. Earlier physical therapy also showed improvements in sensorimotor aspects of balance control, not seen in the later group. There may be an important window to address central sensorimotor deficits after mTBI.

INTRODUCTION

Treatment approaches after mild traumatic brain injury (mTBI) are typically passive. After a brief (24 to 48 h) rest period, people are advised to progressively increase activity, while limiting symptom provocation.¹ With this passive approach, most people become asymptomatic and return to normal daily activities within 2 to 4 weeks.^2,3 However, emerging evidence suggests that early initiation of physical activity can expedite symptom resolution.⁴ Several retrospective studies in children, adolescents, and college-aged adults have reported earlier resolution of symptoms when activity is initiated within the first few weeks.^5–7 Similarly, a prospective (non-randomized) study showed that physical activity after an emergency department visit lowered the risk of developing prolonged symptoms.⁸ However, most evidence is from non-randomized trials and based primarily on children and young adult athletes.⁴ One recent, randomized controlled trial (RCT) of early activity reported that resuming physical activity 72 h post-mTBI was safe but there was no difference in symptom resolution at 2 weeks between groups.⁹

Although most people recover within 4 weeks, up to 30% of people with mTBI experience continued symptoms beyond this period and may require physical therapy.^10,11 However, few studies provide direction on when to initiate physical therapy after an injury. There is some evidence that earlier physical therapy can lead to faster symptom resolution and return to play in pediatric, adolescent, and college-aged athletes who experienced a sport-related mTBI.^12–15 However, other studies have demonstrated similar improvements in global mTBI symptoms, perceived life difficulties due to dizziness, and clinical assessments of balance regardless of the time between injury and physical therapy onset.^16,17 Evidence evaluating the effectiveness of earlier physical therapy after mTBI is limited since most studies are retrospective chart reviews and/or within specialized mTBI clinics, which may not reflect care for the general population.^12,15–17

Community-dwelling adults with mTBI typically sustain injury from other injury mechanisms (eg, motor vehicle accidents or falls) rather than sports that may change symptom presentation.^18,19 Post-mTBI care for the general, non-athlete, adult population differs from athletes and people are less likely to receive a physical therapist referral.²⁰ After an initial visit to the emergency department, people are commonly advised to follow up with their general practitioner,²¹ and physical therapy is prescribed later, only if symptoms persist beyond the typical recovery time.¹⁰ This approach results in an average of 60 days delay from the injury to the first physical therapist visit.^22,23 The delay may cause a person recovering from mTBI to miss a critical window for neuroplasticity and recovery.^24,25

Another limitation in evidence for physical therapy initiation is the frequent use of self-reported symptoms or return to activity as the primary outcome.^12,16,17 The limitations of self-report are well-known; it is documented that 48% of athletes continue to play while experiencing symptoms after mTBI.^26,27 There is also evidence of continued physiologic alterations, despite subjective recovery.²⁸ People may report no symptoms, yet their performance on objective tests may show continued deficit such as increased postural sway during quiet stance.^29–31 Persistent, subtle balance control deficits may explain the increased risk for musculoskeletal injury and repeat brain injury after an mTBI.^32,33 As a result, objective and sensitive measures of balance control may provide insights to more comprehensively define “recovery.”

Our research group characterized objective balance control impairments in people with chronic (>3 months) mTBI symptoms using a Central Sensorimotor Integration (CSMI) Test.^34–36 The CSMI test characterizes balance control as a feedback control system that involves (1) integration of weighted combinations of visual, vestibular, and proprioceptive information, (2) motor activation that generates joint torques, and (3) time delay due to central sensorimotor processing.^34–36 With the CSMI test, people with chronic mTBI had longer time delays and lower motor activation relative to healthy controls.³⁷ Interestingly, there were no deficits in vestibular weighting for balance control, but they had significantly larger visual weights, relative to healthy controls, under conditions that provided inaccurate visual stimuli.³⁷ Visuo-vestibular impairments for balance control are well documented.^37–39 However, the CSMI test provides additional metrics of motor performance that may help characterize the effects of physical therapy and shed light on when it is best to initiate physical therapy after mTBI.

Therefore, the purpose of this study was to examine both patient-reported/clinically-assessed and objective/mechanism-measured balance control after mTBI in people randomized to different initiation times of physical therapist intervention. We hypothesized that earlier initiation of physical therapy would lead to improved outcomes when compared to later physical therapy.

METHODS

Study Design

This RCT aimed to investigate the influence of timing of physical therapy initiation after mTBI and was registered at clinicaltrials.gov (NCT03479541). Participants were randomized to earlier physical therapy (within a week of baseline testing) or later physical therapy (after a 6-week waiting period). Participants were tested before and after intervention using a series of validated patient-reported/clinically-assessed and objective/mechanism-measured assessments.⁴⁰ The later physical therapy group had an additional testing session after the 6-week wait period. The protocol was approved from the Joint Institutional Review Board Committee of Oregon Health & Science University (OHSU) and Veterans Administration Portland Health Care System. All participants provided written informed consent before enrolling.

Participants

Participants were recruited from OHSU and local clinics in the Portland Metropolitan area. The inclusion/exclusion criteria have been previously outlined.⁴⁰ In summary, participants were included if they: (1) had a physician diagnosed mTBI and were within 2 to 12 weeks of injury,⁴¹ (2) were between 18 to 60 years old, (3) scored ≥1 on either balance, dizziness, nausea, headache, or vision problems on the symptom evaluation scale from the Sport Concussion Assessment Tool (SCAT 5) and had a total symptom severity score ≥15, and (4) exhibited no or minimal cognitive impairment (<9 on the Short Blessed Test).⁴² Participants were excluded if they: (1) had other musculoskeletal, neurological, or sensory deficits that could explain their dysfunction, (2) had moderate to severe substance use disorder within the past month,⁴³ (3) experienced severe pain during the evaluation (≥7/10 subjective rating), (4) were pregnant, or (5) were currently being treated by vestibular physical therapy.

Sample Size

The study sample size was determined from previous research that reported Dizziness Handicap Inventory (DHI) scores after early or late physical therapy in individuals with peripheral vestibular disorders.⁴⁴ With these values, the total sample size was calculated at 160 participants with mTBI (80 per intervention group) with 80% statistical power, and an expected 20% dropout rate.

Randomization

After enrollment and baseline data collection, participants were randomized to either earlier physical therapy (n = 82) or later physical therapy (n = 121; Figure 1). An adaptive randomization algorithm balanced the distribution of age and sex covariates with 60% of participants randomized to the later group due to an expected higher dropout rate (eg, resolution of symptoms). Allocation sequence was concealed until after people were enrolled and assigned to interventions. The physical therapists who provided the intervention were aware of participant allocation. Study outcome assessors were blinded to participant allocation.

Figure 1.

CONSORT Flow Diagram for Study.

Intervention

Physical Therapist Intervention

Participants randomized into the earlier physical therapy group started rehabilitation within 1 week of testing while people randomized into the later group were provided an educational brochure on symptoms until they started physical therapy 6 weeks later. Rehabilitation included 8 sessions over 6 weeks with a licensed physical therapist within an academic hospital setting (2 visits/week for the first 2 weeks, then weekly for 4 weeks). Each visit lasted 60 min and focused on 4 subcategories: cervical spine, cardiovascular, static balance, and dynamic balance (~15 min each).⁴⁰ Previous research has shown these subcategories to be effective mTBI rehabilitation strategies.⁴⁵ The physical therapists could progressively challenge (3 levels) each participant as tolerated. Exercises were assigned points based on difficulty/challenge to measure progression. Participants performed a daily home exercise program (HEP) with similar subcategories. The specific intervention and HEP exercises were previously described (Parrington et al, 2020 including Supplementary Material 1).⁴⁰

Primary Outcome Measure

The DHI was the primary outcome; it measures self-perceived handicap due to dizziness and is a recommended outcome measure for mTBI.^45,46 The DHI is sensitive,^44,47 has excellent test–retest reliability in vestibular populations,⁴⁶ and serves as a reliable measure to track improvement after vestibular rehabilitation post-mTBI.⁴⁸

Secondary Outcome Measures

Secondary outcomes are presented in Table 1. The primary and secondary outcomes were assessed before and after physical therapy and after the 6-week waiting period in the later physical therapy group. The Sport Concussion Assessment Tool, 5th edition (SCAT5) symptom evaluation was completed weekly throughout the study (14 weeks).

Table 1.

Secondary Outcome Descriptions^a

Test	Description	Outcomes
Patient-reported/clinically assessed
Neurobehavioral Symptom Inventory (NSI)67	Questionnaire assessing severity of 22 common mTBI symptoms in the last 2 wk, rated 0–4 (none–very severe)	Total symptom severity score out of 88
Quality of Life after Brain Injury (QOLIBRI)68	Post-mTBI quality of life questionnaire: 37 items rated from 1–6 (very unsatisfied–very satisfied)	Total score out of 100 for quality of life69
Symptom Evaluation from SCAT570	Weekly questionnaire assessing mTBI symptom severity over 14 wk with 22 items rated from 0–6 (none–severe)	Total symptom severity score out of 132
Vestibular-Ocular Motor Screening (VOMS) Tool71	Participants rate symptoms (headache, dizziness, nausea, fogginess) on a scale of 0–10 before and after 7 vestibular/ocular tasks	The total change score calculated by subtracting post task symptoms from pretest symptoms72
Mini-Balance Evaluation Systems Test (Mini-BESTest)73,74	A 14-item assessment that measures balance and gait with ratings on a scale from 0 to 2 (severe–normal)	Composite score out of 28, where higher scores indicate better balance and gait function
Objective/mechanism-measured
Standing Balance with Wearable Sensors75 (Opals, APDM, a Clario Company, Portland, OR, USA)	Two 30-s trials: standing with eyes closed, feet together on firm and foam surfaces	Sway area calculated from the wearable sensors for each EcFi and EcFo trial
Central Sensorimotor Integration (CSMI) Test37,76	Measures sway due to surface and visual perturbations, providing parameters for characterizing balance control	Sensory weights, time delay, and motor activation properties in balance control

^a The references included in the table provide additional information on test administration and outcome calculation. EcFi = eyes closed/firm surface; EcFo = eyes closed/foam surface; mTBI = mild traumatic brain injury; SCAT5 = sport concussion assessment tool, 5th edition.

Statistical Analysis

The protocol paper for this randomized controlled trial (NCT03479541) included 2 aims.⁴⁰ The first aim aligned with purpose of the current study which was to determine the effect of timing of physical therapy after mTBI. The second aim was to examine the effect of wearable sensors used during a home exercise program. Results from the second aim will be analyzed separately and published elsewhere. Therefore, analyses for the current paper focused on physical therapy timing and did not incorporate analyses for wearable sensors.

Demographic characteristics are reported and compared for each group (earlier physical therapy and later physical therapy; Table 2). A linear mixed-effects model (LME) with an intention-to-treat design was used to analyze whether the primary outcome measure differed across groups (early vs later physical therapy) over the recovery period. Specifically, the interaction effect examined the difference of change in the outcomes over time based on group. Secondary outcomes were also assessed with LME models except for SCAT5 symptoms. This statistical approach factors in the 6-week wait as well as the intervention time for the later physical therapy group and allowed us to include data from all participants even those who withdrew or were lost to follow-up.⁴⁰ Each LME model contained fixed effects for physical therapy onset group (earlier or later physical therapy), time since injury, and the group × time since injury interaction. The later physical therapy group served as the reference. Both linear and quadratic time since injury were assessed in the model, and linear time was included in the final model. The full LME models also included a priori covariates of age, gender, and initial SCAT symptom severity total score, and random intercepts to account for within-participant correlations. Outcomes were assessed for normality assumptions in the LME models and were log transformed if necessary. We accounted for participants that withdrew or were lost to follow-up using inverse probability weighting in the model.⁴⁹ Specifically, demographic information and outcomes collected at enrollment were used in a logistic model to create inverse probability weights for full participation across the study period. Inverse probability weights were calculated with the intent that observations with demographic and enrollment outcome covariate values predictive of withdrawing or loss to follow-up are upweighted in the LME models and observations with covariate values predictive of study completion are down weighted in the LME models. Sensitivity analyses were performed for inverse probability weights (see Supplementary Materials 1 and 2).⁴⁹ For all LME models, estimated change in outcome per day (with 95% CIs) in each group and p-values for interaction terms were reported. Because of the different lengths of time of the recovery period (ie, 6 weeks for earlier and 12 weeks for later groups), we present the data as rate of change (Table 3) in order to capture a more nuanced picture of the change occurring during the recovery period. However, we also report the means and standard deviations at each assessment timepoint (Table 4) as these values are more intuitive. Additionally, we used paired t tests and Hedges’ G effect sizes (ES_g) on participants that completed all study time points to understand within-group differences across the study assessment time points (Figures 2 and 3). The magnitude of the effect sizes were interpreted as absent (ES_g < 0.2), small (0.2 < ES_g < 0.5), medium (0.5 < ES_g < 0.8), and large (ES_g > 0.8).⁵⁰ Outcomes not meeting the assumptions for paired t test analysis were log-transformed. We used an α of 0.05 as the significance level for LME model and paired t test analyses. The LME model analysis was performed in SAS (version 9.4; Cary, NC, USA). Paired t tests were performed in MATLAB (Version 2022b; Natick, MA, USA) and effect sizes were calculated using the Measures of Effect Size Toolbox in MATLAB.⁵¹

Table 2.

Demographics, Injury Characteristics, and Rehabilitation Details for Both Groups^a

Descriptive Variable	Earlier Physical Therapy   (n = 81)	Later Physical Therapy   (n = 121)	Group Difference P
Demographic characteristics
Age, y	35.6 (11.6) [18.1–60.2]	36.0 (11.2) [18.6–60.0]	.802
Gender, n (%)
Man	18 (22.2%)	20 (16.5%)	.126
Woman	63 (77.8%)	96 (79.3%)
Other	0 (0.0%)	5 (4.1%)
BMI	25.1 (4.9)   [18.3–41.3]	27.3 (8.0)   [17.2–70.5]	.017
Injury characteristics
Days since mTBI at enrollment	45.9 (20.1)	45.4 (22.4)	.861
Injury mechanism, n (%)
Bike-related	3 (3.7%)	6 (5.0%)	.513
Fall	17 (21.0%)	23 (19.0%)
Motor vehicle accident	24 (29.6%)	47 (38.8%)
Other	19 (23.5%)	28 (23.1%)
Sport-related	18 (22.2%)	17 (14.0%)
mTBI history, n (%)
No	43 (53.1%)	57 (47.1%)	.372
Yes	38 (46.9%)	64 (52.9%)
Rehabilitation
Days from injury to rehabilitation	56.3 (20.7)   [19–109]	99.7 (22.2)   [65–143]	< .001
Physical therapy compliance—%	83.8 (23.7) [12.5–100]	82.5 (24.5) [12.5–100]	0.733
HEP compliance—%	52.1 (31.9) [0–100]	50.9 (32.7) [0–100]	0.826
Physical therapy exercise progression of entire program—%	34.8 (11.6) [0–56.4]	35.8 (13.3) [0–63.2]	0.623
HEP average level at end of rehabilitation (levels scored on 1–3 scale)	2.5 (0.6) [1–3]	2.5 (0.7) [1–3]	0.485

^a Values presented as means, standard deviations, and range [minimum–maximum] unless noted otherwise. Bold and italicized rows indicate a significant group difference. One participant enrolled then was terminated due to re-injury then was later re-enrolled into study with a new participant identification number. BMI = body mass index; HEP = home exercise program; mTBI = mild traumatic brain injury.

Table 3.

Estimated Outcome Change Per Day (With 95% CI) in the Earlier Physical Therapy and Later Physical Therapy Groups^a

Outcome	Change Per Day With Earlier Physical Therapy	Change Per Day With Later Physical Therapy	Group Difference in Change Per Day P
Patient-reported/clinically assessed
DHI score	−0.202b [−0.272, −0.132]	−0.105b [−0.137, −0.073]	.013
NSI score	−0.259b [−0.311, −0.207]	−0.186b [−0.210, −0.162]	.013
QOLIBRI score	0.277b [0.209, 0.345]	0.179b [0.148, 0.210]	.010
VOMS total change from pretest score	−0.131b [−0.188, −0.074]	−0.103b [−0.129, −0.077]	.387
Mini-BESTest composite score	0.019b [0.009, 0.029]	0.011b [0.006, 0.015]	.135
Objective/mechanism measured
Sway area (m2/s4)—EcFic	−0.006d [−0.010, −0.001]	−0.003d [−0.005, −0.001]	.240
Sway area (m2/s4)—EcFoc	−0.008b [−0.011, −0.005]	−0.003b [−0.004, −0.001]	.003
CSMI condition VS/EO
Visual weight	−0.0004b [−0.0006, −0.0002]	−0.0002b [−0.0003, −0.0001]	.013
Time delay (ms)	−0.217b [−0.330, −0.104]	0.018 [−0.036, 0.073]	<.001
Normalized stiffness	0.0001 [−0.0004, 0.0005]	−0.0002e [−0.0004, −0.000005]	.273
Normalized damping	0.0001 [−0.0001, 0.0003]	−0.0001 [−0.0002, 0.00005]	.247
Stimulus-evoked CoM sway (°)c	−0.004b [−0.006, −0.002]	−0.0004 [−0.001, 0.001]	.003
Internal sensory noise (°)c	−0.003e [−0.006, −0.0004]	0.0002 [−0.001, 0.002]	.031
CSMI condition SS + VS/EO
Vestibular weighting	0.0004e [0.0001, 0.0008]	0.0004b [0.0002, 0.0005]	.780
Time delay (ms)	−0.169b [−0.250, −0.089]	−0.027 [−0.069, 0.015]	.002
Normalized stiffness	0.0004 [−0.00002, 0.0008]	−0.0004b [−0.0005, −0.0002]	.002
Normalized damping	0.0002 [−0.00003, 0.0005]	−0.0002e [−0.0003, −0.00004]	.008
Stimulus-evoked CoM sway (°)c	−0.001d [−0.002, −0.0003]	0.0002 [−0.0002, 0.001]	.004
Internal sensory noise (°)c	−0.003d [−0.005, −0.001]	−0.0005 [−0.001, 0.001]	.003

^a Estimated change per day derived from linear mixed effect model with random intercepts, covariates (age, gender, and initial SCAT symptom severity total score), and inverse probability weights. Bolded  p values indicate a significant group difference in the change per day over time; P < .05. CoM = center of mass; CSMI = central sensorimotor integration; DHI = dizziness handicap inventory; EcFi = eyes closed/firm surface; EcFo = eyes closed/foam surface; mini-BESTest = mini-balance evaluation systems test; ms = milliseconds; NSI = neurobehavioral symptom inventory; QOLIBRI = quality of life after brain injury; SCAT = sports concussion assessment tool; SS + VS/EO = stance surface + visual surround/eyes open; VOMS = vestibular ocular-motor screening; VS/EO = visual surround/eyes open.

^b Denotes a significant change per day within a group P < .001.

^c Denotes variable was log transformed for analysis.

^d Denotes a significant change per day within a group P < .01.

^e Denotes a significant change per day within a group P < .05.

Table 4.

Descriptive Characteristics (Means and Standard Deviations) for Study Outcomes for Both Groups^a

Outcome	Earlier Physical Therapy		Later Physical Therapy
	Pre-Physical Therapy (n = 82)	Post-Physical Therapy (n = 63)	Baseline (n = 121)	Pre-Physical Therapy (n = 87)	Post-Physical Therapy (n = 63)
Patient-reported/clinically assessed
DHI score	29.2 (17.6)	17.8 (15.6)	34.9 (19.8)	27.6 (19.5)	20.9 (18.5)
NSI score	39.1 (13.1)	21.6 (13.2)	43.5 (13.5)	30.7 (15.3)	21.7 (13.3)
QOLIBRI score	49.7 (14.7)	69.3 (18.3)	45.7 (14.5)	56.4 (17.5)	67.4 (18.7)
VOMS total change from pretest score	14.4 (12.7)	4.9 (9.6)	14.6 (11.5)	10.3 (11.1)	5.6 (8.1)
Mini-BESTest composite score	24.8 (2.2)	25.8 (1.7)	24.4 (2.4)	25.2 (2.4)	25.6 (2.6)
Objective/mechanism measured
Sway area (m2/s4)—EcFi	0.160 (0.148)	0.106 (0.077)	0.237 (0.431)	0.155 (0.214)	0.147 (0.154)
Sway area (m2/s4)—EcFo	0.816 (1.95)	0.362 (0.209)	0.654 (0.604)	0.565 (0.483)	0.485 (0.440)
CSMI test condition VS/EO
Visual weight	0.134 (0.060)	0.111 (0.052)	0.133 (0.059)	0.122 (0.057)	0.118 (0.064)
Time delay (ms)	218 (30)	205 (28)	211 (24)	217 (29)	211 (31)
Normalized stiffness	1.219 (0.093)	1.218 (0.087)	1.228 (0.097)	1.191 (0.077)	1.221 (0.096)
Normalized damping	0.469 (0.051)	0.470 (0.046)	0.469 (0.049)	0.458 (0.046)	0.470 (0.054)
Stimulus-evoked CoM sway (°)	0.289 (0.152)	0.235 (0.121)	0.281 (0.156)	0.312 (0.175)	0.280 (0.179)
Internal sensory noise (°)	0.088 (0.041)	0.074 (0.047)	0.087 (0.047)	0.083 (0.047)	0.091 (0.062)
CSMI test condition SS + VS/EO
Vestibular weighting	0.414 (0.098)	0.441 (0.092)	0.415 (0.076)	0.445 (0.075)	0.446 (0.095)
Time delay (ms)	171 (25)	161 (20)	170 (19)	170 (20)	166 (20)
Normalized stiffness	1.381 (0.127)	1.405 (0.118)	1.382 (0.111)	1.350 (0.106)	1.365 (0.113)
Normalized damping	0.491 (0.062)	0.505 (0.064)	0.496 (0.061)	0.482 (0.059)	0.491 (0.068)
Stimulus-evoked CoM sway (°)	1.093 (0.303)	1.010 (0.238)	1.068 (0.239)	1.104 (0.246)	1.089 (0.276)
Internal sensory noise (°)	0.136 (0.053)	0.109 (0.039)	0.126 (0.053)	0.129 (0.060)	0.123 (0.058)

^a CoM = center of mass; CSMI = central sensorimotor integration; DHI = dizziness handicap inventory; EcFi = eyes closed/firm surface; EcFo = eyes closed/foam surface; Mini-BESTest = mini-balance evaluation systems test; ms = milliseconds; NSI = neurobehavioral symptom inventory; QOLIBRI = quality of life after brain injury; SS + VS/EO = stance surface + visual surround/eyes open; VOMS = vestibular ocular-motor screening; VS/EO = visual surround/eyes open.

Figure 2.

Mean (Diamond) and Standard Errors for Patient-Reported/Clinically- Assessed Outcomes Within Each Group (Closed Diamonds – Earlier Physical Therapy and Open Diamonds – Later Physical Therapy). Data presented come from participants who completed all study time points (n = 63 in both groups). T1 and T2 correspond to pre-physical therapy and post-physical therapy time points for the earlier physical therapy group. T1, T2, and T3 correspond to baseline, pre-physical therapy, and post-physical therapy time points for the later physical therapy group. The Sport Concussion Assessment Tool (SCAT) symptom evaluation total severity scores are presented at enrollment screening (Week 0) and throughout the 14-week study. Arrows indicate the direction of improvement. A significant change between timepoints within a group is indicated by *, with the number of * corresponding to the magnitude of the effect size change (*—small, **—medium, ***—large). DHI = dizziness handicap inventory; Mini-BESTest = mini-balance evaluation systems test; NSI = neurobehavioral symptom inventory; QOLIBRI = quality of life after brain injury; VOMS = vestibular ocular-motor screening.

Figure 3.

Mean (Diamond) and Standard Errors for Objective/Mechanism-Measured Outcomes Within Each Group (Closed Diamonds – Earlier Physical Therapy and Open Diamonds – Later Physical Therapy). Data presented come from participants who completed all study time points (n = 63 in both groups). T1 and T2 correspond to pre-physical therapy and post-physical therapy time points for the earlier physical therapy group. T1, T2, and T3 correspond to baseline, pre-physical therapy, and post-physical therapy time points for the later physical therapy group. Arrows indicate the direction of improvement. A significant change between timepoints within a group is indicated by *, with the number of * corresponding to the magnitude of the effect size change (*—small, **—medium, ***—large). CSMI = central sensorimotor integration; EcFi = eyes closed/firm surface; EcFo = eyes closed/foam surface; SS + VS/EO = stance surface + visual surround/eyes open; VS/EO = visual surround/eyes open.

Role of the Funding Source

This study was supported by the Assistant Secretary of Defense for Health Affairs under the Award Number W81XWH-17-1 − 0424. All interpretations and opinions are those of the authors and are not necessarily endorsed by the Department of Defense. The funders played no role in the design, conduct, or reporting of this study.

RESULTS

Between July 2018 and March 2023, a total of 203 participants were enrolled and randomized to earlier (n = 82) or later physical therapy (n = 121). Nineteen participants in the earlier group (23%) and 58 in the later group (48%) did not complete post-intervention assessments (Figure 1). Reasons for not completing the study are provided in Supplementary Material 3. There were no characteristic differences between participants that completed the study and those who withdrew or were lost to follow-up (Supplementary Material 4). Post-intervention assessments were conducted on 63 people for both groups.

Table 2 provides demographic characteristics and information on compliance and progression of rehabilitation. The groups were well matched for age, gender, injury mechanism, the number of days from injury to enrollment, and previous mTBI history. The earlier physical therapy group had a significantly lower BMI (Table 2). The groups showed no differences for intervention compliance or exercise progression. Descriptive statistics for all outcomes at each timepoint are presented in Table 4. No adverse events were reported in either group. While the physical therapists continually modified exercise intensity and rest durations per clinical judgment, there were 5 participants whose symptom provocation required the therapist to end the session short of the 60-min duration (n = 1 in earlier and n = 4 in later physical therapy groups).

Primary Outcome: DHI

Though both groups significantly improved over time, the earlier physical therapy group had a larger decrease per day in DHI score relative to the later group (P = .013; Table 3) after adjusting for covariates. Figure 2 shows the descriptive mean and standard error of the mean at each evaluation time point for participants with both pre and post measures (n = 63 for both groups). The earlier physical therapy group had a significantly decreased DHI score from pre- to post-physical therapy (P < .001; Supplementary Material 5) with a medium effect size in change (ES_g = −0.75; Figure 2A). The later physical therapy group also had a significantly decreased DHI score during the 6-week waiting period from baseline to pre-physical therapy (P = .042) as well as from pre- to post-physical therapy (P < .001). However, in contrast to the early physical therapy group, the effect sizes for the later physical therapy group were small (ES_g = −0.22 and ES_g = −0.35 between baseline to pre-physical therapy and pre- to post-physical therapy, respectively; Figure 2A).

Secondary Outcomes: Patient-Reported/Clinically-Assessed

While both groups significantly improved their symptoms (NSI) and quality of life (QOLIBRI), the earlier group had significantly larger improvements per day in both tests (Ps < .013; Table 3). Similar trends were observed in the paired t test analysis (Figure 2B and C; Supplementary Material 5). Both groups had significant improvements per day in the VOMS total change and Mini-BESTest scores. However, there were no between-group differences in the rate of change in these outcomes (Ps > .135; Table 3). The paired t test analysis of participants who completed the post-test assessments showed significant improvements in VOMS (Figure 2D) and Mini-BESTest (Figure 2E) for both groups (Supplementary Material 5). Overall, the earlier physical therapy group had larger magnitudes of change (effect sizes) when compared (without statistical analysis) to the later group (Figure 2B–D; Supplementary Material 5). Figure 2F shows data from the weekly SCAT5 symptom scale showing that while both groups reached similar symptom levels by 14 weeks, the earlier group reached this level sooner (Figure 2F).

Secondary Outcomes: Objective/Mechanistic Measures

The earlier physical therapy group had a significantly larger decrease in sway area per day relative to the later group (P = .003) for eyes-closed foam surface (EcFo) standing balance. There was no group difference in the sway area change per day for eyes-closed firm surface (P = .240; Table 3). The earlier physical therapy group significantly decreased EcFo sway area from pre- to post-physical therapy with a medium effect size in change (ES_g = −0.78) while the later physical therapy group had no significant change in EcFo sway area from baseline to pre-physical therapy and had significantly decreased EcFo sway area from pre- to post-physical therapy with a small effect size (ES_g = −0.39; Figure 3B; Supplementary Material 5).

While both groups significantly decreased (improved) their visual sensory weight, the earlier group had significantly larger decreases per day relative to the later group (P = .013). Both groups significantly increased (improved) vestibular sensory weight per day with no difference between the groups in the rate of change per day (Table 3). The paired t test analysis revealed the earlier physical therapy group significantly decreased visual sensory weight from pre- to post-physical therapy with a small effect size (ES_g = −0.40) while the later group did not change from baseline to pre-physical therapy or from pre- to post-physical therapy (Figure 3C; Supplementary Material 5).

The earlier physical therapy group had significantly larger decreases (improved) in time delay per day relative to the later group (Ps < .002). Specifically, the earlier physical therapy group significantly decreased time delay per day across both the visual stimulus (VS/EO) and the combined stance surface and visual surround stimulus (SS + VS/EO) conditions, while there was no significant change per day in time delay for the later group on either condition (Table 3). The paired t test analysis showed similar results. Within the earlier group, time delay in both conditions significantly decreased with small effect sizes from pre- to post-physical therapy (ES_g = −0.34 and −0.37, respectively; Figure 3E and F). There was no significant difference in time delay in either condition across timepoints for the later group (Figure 3E and F; Supplementary Material 5).

There were significant group differences in the change per day of motor activation parameters (stiffness and damping) in the SS + VS/EO condition (Ps < .008). Specifically, the earlier group did not change motor activation over time, while the later physical therapy group significantly decreased (worsened) stiffness and damping per day in the SS + VS/EO condition (Figure 3H and J; Table 3; Supplementary Material 5). The paired t test analysis showed a decrease in motor activation (worsening) from baseline to pre-physical therapy and no change from pre- to post-physical therapy for the later group (Figure 3H and J; Supplementary Material 5). There was no difference between groups in the rate of change for motor activation for the VS/EO condition (Table 3). However, the later physical therapy group significantly decreased (worsened) stiffness per day in the VS/EO condition (Table 3) from baseline to pre-physical therapy (Figure 3G).

Finally, the earlier physical therapy group had significantly larger decreases (improvements) per day in stimulus-evoked center of mass sway and internal sensory noise measures in the VS/EO and SS + VS/EO conditions relative to the later group (Ps < .031). The earlier group significantly decreased in both measures per day in both conditions while there was no significant change per day in the later group (Table 3). All paired t test analyses within each group across time points are presented in Supplemental Material 5. Parameter estimates for LME model fixed effects are presented in Supplementary Material 6.

DISCUSSION

For people with continued symptoms from mTBI, initiating earlier physical therapy led to faster rates of improvements in most outcomes compared to the later physical therapy group. Our hypothesis was supported with earlier initiation of physical therapy resulting in a faster rate of improvement on the DHI score—our study’s primary outcome—and a faster rate of improvement in most secondary outcomes as well.

Earlier Physical Therapy Led to Improvements in Patient-Reported Outcomes at a Faster Rate

Our results support earlier physical therapy for treating a largely non-athlete, adult population with continued symptoms from mTBI since earlier physical therapy led to faster rates of improvement in perceived handicaps due to dizziness, mTBI-related symptoms, and health-related quality of life (Figure 2A–C and F). In these figures, both groups end up with similar levels of patient-reported deficits after completing the study. However, the earlier group improved in a shorter timeframe. Symptoms related to an mTBI can have drastic impacts on daily life functioning and earlier resolution of symptoms is important.^52,53 There were no adverse events related to the earlier physical therapy group, demonstrating safety to initiate physical therapy despite a high symptom burden.

Using Visual and Vestibular Information for Balance Control

Our previous research and others have shown that visual sensory weighting for balance control is increased for people with chronic mTBI relative to healthy controls.^37,39 This over-reliance on vision for balance is problematic or symptom-provoking in situations with visual disturbances such as in moving environments.⁵⁴ Earlier physical therapy led to faster rates of recovery in visual sensory weighting for balance control. However, vestibular sensory weights recovered at similar rates within both groups. Interestingly, vestibular weighting for balance control increased (improved) without physical therapist intervention (as seen in the later physical therapy group). This suggests that central vestibular function may acutely decrease after mTBI but may recovery naturally without intervention.³⁷ Our data supports the use of physical therapy to decrease an over-reliance on visual information for balance control similar to previous work that showed improved sensory organization test scores in people with concussion following vestibular rehabilitation therapy.⁵⁵ Additionally, our data suggests that recalibration of visual-vestibular integration through physical therapy can be achieved more quickly when physical therapy is initiated early.

Potential Maladaptive Balance Control Changes With Later Physical Therapy

Our results showed that earlier physical therapy led to improvements in time delay for balance control and this improvement only occurred in the earlier group, suggesting a critical time period to initiate physical therapy.⁵⁶ The earlier physical therapy group significantly improved (shortened) their postural time delay in both CSMI conditions towards healthy control values while the later physical therapy group did not change.³⁷ With longer time delays in the later group, a decrease in motor activation (stiffness and damping) provides a compensation to maintain stable balance.^37,57 We observed no changes in motor activation in the earlier physical therapy group likely because physical therapy improved (decreased) their time delay.³⁷ In contrast, the later group had decreases in motor activation, likely in compensation for longer time delays that developed during the wait period before physical therapy.^37,57 The decrease in motor activation can be interpreted as a maladaptation with the functional consequence that patients become less able to resist external disturbances. Evidence for this reduced ability to resist disturbances is shown by the increase in stimulus-evoked sway between baseline and pre-physical therapy for the later physical therapy group (Supplementary Material 5). Further, there were still no improvements in these motor activation measures even after later physical therapy was initiated. However, it is possible that later physical therapy may have prevented a continued decrease in motor activation.³⁷

Another indicator of the benefit of earlier physical therapy is the CSMI internal sensory noise measure, which decreased significantly for the early physical therapy group but not for the later physical therapy group. This internal sensory noise measure can be understood to represent the quality or precision of central sensorimotor processing needed to transform sensory orientation information into effective motor actions for balance control.^37,57

Deficits in Objective Measures Remain Despite Symptom Improvement

Clearance for return to previous activity still relies primarily on symptoms.^58–60 However, improvement in symptoms may not correspond with recovery of sensorimotor function for balance. Decisions based solely on symptoms fail to evaluate functionality.⁶¹ A person may have enough improvements in symptoms to return to sport or duty, while still having underlying deficits in balance control, thereby exposing them to increased risk for further injury.^31–33 Notably, under reporting symptoms among high school and college athletes is well-documented.^26,27 Further, a person may subconsciously avoid activities that provoke symptoms,⁶² making it difficult to accurately determine true recovery. These well documented limitations underscore the need for more objective evaluation methods after mTBI.^30,63,64

What Happens With No Physical Therapy Intervention?

With our study design, we were able to characterize how people changed over 6 weeks without physical therapy (Table 4 and Supplementary Material 5). Participants, while waiting to initiate physical therapy, had significant improvements in symptoms, quality of life, and clinical assessments. However, with the exception of vestibular weighting for balance control, there were minimal improvements in time delay and significant degradation in motor activation for balance control in untreated patients. These observations underscore the benefit of assessing people with mTBI using a multidimensional assessment battery.^{30,37,63–66} Further degradation to time delay and motor activation may continue if patients are left untreated and progress to more chronic states of mTBI.³⁷

Practicalities on Getting People to Physical Therapy Earlier

While our study provides evidence for initiating earlier physical therapy, there are barriers that limit the practicality of this approach. Beneficial effects of earlier physical therapy have been documented in younger populations within a sport medicine clinic.^12–15 In those studies, the average days since injury to physical therapy was 8 to 27 days. In our study, the earlier physical therapy group averaged 56 days before starting physical therapy. Therefore, the timescale of initiating “earlier” physical therapy in our study was much longer, relative to research on young athletes. Instead, our study may be more representative of the general, non-athlete population.²³ We recently published data showing that an initial appointment in a specialized mTBI clinic is associated with being 75 times more likely to receive a rehabilitation referral compared to patients seeking care in the emergency department—further explaining the common delay in receiving physical therapy.²⁰

Limitations

One limitation in our paper is the higher than anticipated withdrawal and loss to follow up rate, primarily due to the COVID-19 pandemic. However, we accounted for potential attrition bias by using an inverse probability weighting approach (Supplementary Material 1).⁴⁹ We also performed a sensitivity analysis on the LME models with and without the inverse probability weights and there were no significant differences in results (Supplementary Material 2). Furthermore, there were no differences in characteristics of the people who withdrew/lost to follow-up and those who completed the study (Supplementary Material 4). While we did not reach sample size goal, we still found significant results in our primary and most secondary outcomes. Due to our rigorous statistical approach and sensitivity analysis, we have minimal concern about selection bias in our sample. Finally, our earlier physical therapy group was not “early” by most sport concussion clinic or by military standards. Therefore, it is unknown how “very” early physical therapy in the general adult population would affect outcomes. However, our timeline may be more realistic for the general public both anecdotally and reported timeframes for initiation of physical therapy.

CONCLUSIONS

People who received earlier physical therapy had faster rates of improvement in most outcomes. Earlier physical therapy also resulted in improvement in time delay and motor activation measures of balance control, while the later group did not improve—potentially developing maladaptive sensorimotor changes. Our results suggest that earlier physical therapy for people with continued symptoms from mTBI should be an important consideration for rehabilitation after mTBI.

CRediT – CONTRIBUTOR ROLES

Kody R. Campbell (Data curation [Equal], Formal analysis [Equal], Project administration [Equal], Visualization [Lead], Writing—original draft [Lead], Writing—review & editing [Lead]), Prokopios Antonellis (Data curation [Equal], Formal analysis [Equal], Project administration [Equal], Writing—original draft [Equal], Writing—review & editing [Equal]), Robert Peterka (Conceptualization [Supporting], Data curation [Supporting], Formal analysis [Supporting], Funding acquisition [Supporting], Investigation [Equal], Methodology [Equal], Writing—original draft [Equal], Writing—review & editing [Equal]), Jennifer Wilhelm (Conceptualization [Supporting], Methodology [Equal], Writing—original draft [Equal], Writing—review & editing [Equal]), Kathleen T Scanlan (Project administration [Equal], Writing—original draft [Supporting], Writing—review & editing [Supporting]), Natalie Pettigrew (Investigation [Supporting], Methodology [Supporting], Project administration [Supporting], Writing—original draft [Supporting], Writing—review & editing [Supporting]), Siting Chen (Formal analysis [Lead], Writing—review & editing [Supporting]), Lucy Parrington (Data curation [Equal], Project administration [Equal], Writing—review & editing [Supporting]), Peter C. Fino (Conceptualization [Supporting], Writing—original draft [Supporting], Writing—review & editing [Supporting]), James C. Chesnutt (Conceptualization [Supporting], Funding acquisition [Supporting], Methodology [Supporting], Writing—review & editing [Supporting]), Fay Horak (Conceptualization [Supporting], Funding acquisition [Supporting], Investigation [Supporting], Methodology [Supporting], Writing—review & editing [Supporting]), Timothy E. Hullar (Conceptualization [Supporting], Funding acquisition [Supporting], Methodology [Supporting], Writing—original draft [Supporting], Writing—review & editing [Supporting]), Laurie A. King (Conceptualization [Lead], Funding acquisition [Lead], Investigation [Lead], Methodology [Lead], Project administration [Lead], Resources [Lead], Supervision [Lead], Writing—original draft [Equal], Writing—review & editing [Equal]).

FUNDING

This study was supported by the Assistant Secretary of Defense for Health Affairs under the Award Number W81XWH-17-1-0424 (L.A.K). All interpretations and opinions are those of the authors and are not necessarily endorsed by the Department of Defense. Resources and facilities at the VA National Center for Rehabilitative Auditory Research (NCRAR) were supported by Award No. C2361/I50 RX002361 at the VA Portland Health Care System in Portland, Oregon, USA. An integrated SQL database at Oregon Health & Science University housed all the data and is supported by the Oregon Clinical and Translational Research Institute funded by a grant from the National Center for Advancing Translational Sciences, National Institutes of Health (grant award no. UL1TR002369).

ETHICS APPROVAL

The study was conducted according to the guidelines of the Declaration of Helsinki and was approved by the joint Institutional Review Board of Oregon Health & Science University (STUDY00017370) and Veterans Administration Portland Health Care System (VA#3991).

CLINICAL TRIAL REGISTRATION

This RCT aimed to investigate the influence of timing of physical therapy initiation after mTBI and was registered at clinicaltrials.gov (NCT03479541).

DISCLOSURES

The authors completed the ICMJE Form for Disclosure of Potential Conflicts of Interest and reported no conflicts of interest.

OHSU and F.H. have a significant financial interest in APDM Wearable Technologies, a Clario company, that may have a commercial interest in the results of this research and technology. This potential conflict of interest has been reviewed and managed by OHSU. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Laurie King is a member of the PTJ Editorial Board.

DATA AVAILABILITY

Data will be available to qualified investigators through the Federal Interagency Traumatic Brain Injury Research (FITBIR) Informatics System (doi; 10.23718/FITBIR/1518823) through the National Institutes of Health Center for Information Technology (https://fitbir.nih.gov/content/access-data).