Association of Multidomain Assessment Outcomes With Referral for Vestibular Therapy After Concussion

This retrospective chart review used consecutive patient data from an electronic health record via concussion specialty clinics from January 2019 to February 2020. Participants ranged from 9 to 68 years of age. To be included in the study, participants had to have a current, symptomatic concussion at the time of the first clinical evaluation for which data were extracted. Exclusion criteria were prior brain surgery, moderate to severe traumatic brain injury, neurologic disorder, treatment for substance abuse, a history of concussion within the 3 months before the first clinical visit, or a history of 3+ concussions. The study was reviewed and approved by the university institutional review board. Participants and parents (if necessary) provided written consent for involvement.

Participants self-reported demographics (age and sex), medical history (history of attention-deficit/hyperactivity disorder or learning disability, motion sickness, migraine, ocular dysfunction, depression, anxiety, and number of prior concussions [0–3]), and injury-specific information (days since injury and mechanism of injury) in a standardized clinical interview.

The Concussion Clinical Profiles Screen (CP-Screen) is a valid and reliable 29-item self-report symptom inventory. Each item is scored on a 4-point Likert scale ranging from 0 (none) to 3 (severe). Five clinical profile scores are calculated: anxiety or mood, cognitive or fatigue, migraine, vestibular, and ocular, along with 2 modifier scores (cervical and sleep).

Immediate Post-Concussion Assessment and Cognitive Testing (ImPACT) was used to evaluate neurocognitive performance. This instrument assesses symptoms using the Post-Concussion Symptom Scale (PCSS), a 23-item self-report symptom survey that is scored on a 7-point Likert scale (0 = none, 6 = severe). Neurocognitive performance is evaluated with respect to verbal and visual memory, visual motor speed, and reaction time domains. The 2 scales are reliable and valid measures of neurocognition and concussion symptom severity, respectively.

Participants were diagnosed with a current concussion by a clinical neuropsychologist per consensus guidelines including a clear mechanism of injury, initial signs and symptoms, and being currently symptomatic or impaired.³  The clinical neuropsychologist identified the patient's concussion subtype based on the results of a clinical interview and a symptom questionnaire designed to identify concussion clinical subtypes.¹⁵  Extensive information on the assignment of a concussion subtype to guide targeted intervention can be found elsewhere.^2,7,16  Consistent with the standard of care at the clinic, patients were referred for vestibular therapy by the neuropsychologist if they reported (1) critical symptoms (slow, wavy dizziness or fast, “room-spinning” dizziness; mental “fogginess”; or eye fatigue), (2) functional impairment (discomfort in busy environments, discomfort in busy visual fields such as scrolling on a cellphone, feelings of unsteadiness, avoidance behaviors, or double or blurry vision when reading), or (3) a high symptom burden after VOMS testing.¹⁷ 

The VOMS is a brief screening tool that assesses vestibular and ocular motor symptoms and impairments via smooth pursuits, horizontal or vertical saccades, NPC, HVOR or VVOR, and VMS. The NPC distance is measured in centimeters and calculated as an average of 3 trials. Participants self-report headache, dizziness, nausea, and fogginess on a 0 to 10 scale before taking the VOMS, for a total pretest range of 0 to 40. These symptoms are rescored after each test, and the pretest symptoms are subtracted from each posttest symptom score to obtain a measure of symptom provocation from the test.

For descriptive purposes, we used Mann-Whitney U tests to compare cohorts for continuous variables due to a nonnormal distribution (ie, age, time to first visit, and days to recovery). The χ² analyses were conducted with associated odds ratios (ORs) and 95% CIs for the categorical variables (ie, sex, mechanism of injury, medical history, and primary subtypes). Five logistic regressions were built to identify the association of different components of the multidomain assessment (ie, demographics or medical history from a clinical interview, CP-Screen symptom items, PCSS symptom items, computerized neurocognitive domains, and VOMS items) with referral for vestibular therapy. A sixth logistic regression model was generated using predictors retained from any of the previous 5 models to assess which variables from the multidomain assessment were most associated with vestibular referral. We adopted an all-possible regression approach for all models in this study to rule out the potential for order effects from stepwise methods. Specifically, we used the best subsets model selection method from SPSS (version 28; IBM Corp), in which the best set of predictors was selected based upon the Akaike information criterion corrected value.¹⁸  The smallest value was selected as the best performing model. Post hoc diagnostics included ensuring that the variance inflation factor of all included variables was below a mean of 4.0 and partial correlations were below r = 0.8.^19,20  For all analyses, P < .05 was considered statistically significant.

Participants (n = 331) were 16.9 ± 7.2 years old, 39.3% were female (n = 130), and they presented to the clinic 16.2 ± 46.7 days after their most recent concussion (Table 1). Regarding their medical history, 32.0% reported a prior concussion (n = 106); 21.1%, motion sickness history (n = 70); and 8.2%, ocular disorder (n = 27). Approximately 17.5% of patients demonstrated vestibular symptoms or impairments that necessitated referral for vestibular therapy.

The demographics and risk factors of patients referred to vestibular therapy versus those who were not referred are shown in Table 2. Compared with participants who were not referred for vestibular therapy, those who were referred were older, had higher odds of a concussion history (OR = 1.81), and took longer to recover. Those referred for vestibular therapy were less likely to have a sport-related mechanism of injury (OR = 0.77) and more likely (OR = 11.97) to have experienced a motor vehicle accident. Patients referred for vestibular therapy were more likely to have a vestibular ocular primary subtype (OR = 1.90–3.26) and less likely to have a cognitive or migraine primary subtype (OR = 0.15–0.36).

The results for each logistic regression model to associate individual multidomain assessment models with vestibular therapy referral are provided in Table 3. The model associating demographic or medical history variables with vestibular therapy referral (R² = 0.08; P < .001) retained 3 variables overall and 2 statistically significant variables (migraine history: OR = 1.93; motor vehicle accident: OR = 14.4).

The model associating CP-Screen variables with vestibular therapy referral (R² = 0.15; P < .001) retained 3 statistically significant variables (feeling sad: OR = 1.58; dizziness: OR = 1.58; increased headache when concentrating: OR = 1.70).

The model associating computerized neurocognitive domains with vestibular therapy referral (R² = 0.01; P = .03) retained 1 statistically significant variable (visual memory score: OR = 0.98).

The model associating PCSS items with vestibular therapy referral (R² = 0.07; P < .001) retained 2 statistically significant variables (more emotional: OR = 1.26; slowed down: OR = 1.27).

The model associating VOMS items with vestibular therapy referral (R² = 0.28; P < .001) retained 5 statistically significant variables (ORs = 1.33 for horizontal saccades [HSAC], 0.52 for NPC symptoms, 1.16 for NPC distance, 1.53 for HVOR, and 1.29 for VMS).

The model associating retained variables from each multidomain assessment with vestibular therapy referral (R² = 0.40; P < .001) can be seen in Table 4. The model retained 9 statistically significant variables (ORs = 3.25 for migraine history, 15.45 for motor vehicle accident, 1.42 for HSAC, 0.46 for NPC symptoms, 1.18 for NPC distance, 1.63 for HVOR, 1.26 for VMS, 1.37 for more emotional [PCSS], = and 1.81 for increased headache when concentrating [CP-Screen]).

Recent evidence²¹  suggested that targeted vestibular therapy for the treatment of post–concussion vestibular symptoms or impairments leads to greater clinical improvements than behavioral management alone. However, a key concern for the sports medicine professional is identifying a clinical endpoint for when and how to determine when a referral for specialized care is indicated.⁶  In the present study, we demonstrated how the administration of a multidomain assessment after concussion can be useful for identifying clinical variables associated with a vestibular therapy referral. At least 1 variable from each assessment domain was retained in the final multivariate model except for computerized neurocognition (Table 4). Consistent with our hypotheses, multiple VOMS items were associated with referral for vestibular therapy (ie, HSAC, HVOR, NPC symptoms, NPC distance, and VMS). These results can be used to inform sports medicine clinicians as to when a referral for specialized vestibular therapy may be warranted after concussion.

Common vestibular impairments after concussion are dizziness, lightheadedness, vertigo, nausea, imbalance, and disequilibrium.¹  These symptoms are typically reduced at rest but are provoked by dynamic movement or environmental stimuli such as changes in support surface or high-intensity visual demand.^17,22  As such, provocative testing of vestibular ocular and vestibular-visual integration via the VOMS is typically needed to identify vestibular dysfunction after a concussion. This notion is supported by our final regression model (Table 4), which retained 2 primarily vestibular VOMS assessments (ie, HVOR and VMS) and 2 primarily ocular VOMS assessments (ie, HSAC and NPC symptoms or distance). Additionally, none of the traditional vestibular symptoms from the CP-Screen or PCSS were retained in the final model. Clinicians should be aware of the tight integration between the visual and vestibular systems and recognize that primarily ocular symptoms or impairments may also reflect an impairment in the vestibular system.

Interestingly, patients referred for vestibular therapy in this study were associated with a history of migraine but not concussion or motion sickness (when analyzed with univariate analyses; Table 2). Medical history is an important component of the multidomain concussion assessment, as researchers have shown that certain preinjury factors can exacerbate the impairment or symptom burden during concussion assessments. For example, motion sickness predicted vestibular dysfunction after concussion and increased false-positive rates during VOMS testing.^11,13  A history of migraines can also lead to false-positive results on baseline VOMS assessments in healthy athletes.^14,23  However, our findings ultimately suggest that, although it is important for clinicians to recognize an athlete's history may predispose the individual to certain symptoms and impairments, their actual manifestation on clinical examination is most pertinent to the assignment of targeted therapies.

An important outcome was the association of multiple emotional or mood symptoms with referrals for vestibular therapy. Vestibular dysfunction can be a stressful experience, potentially resulting in activity avoidance due to an increase in vestibular symptoms. Authors²⁴  of a recent study reported that emotional symptoms on the Dizziness Handicap Inventory were related to provocation on VOMS assessment in adolescents after concussion. Further, Eagle et al²⁵  demonstrated that vestibular symptoms on the VOMS were associated with the anxiety or mood concussion subtype. Earlier researchers²⁶  also determined that patients with more sleep symptoms were more likely to have vestibular and mood concussion subtypes. Avoiding typical activities can result in internalizing affective symptoms and a potential mood disorder.⁸  Clinicians should be aware that vestibular symptoms can also increase mood symptoms and consider screening for mood disorders in patients who need a vestibular referral after a concussion.

This study had limitations worth noting. The multidomain assessment protocol described here is the standard of care for the concussion specialty clinics that were involved in this study; as such, the clinician who referred for vestibular therapy was aware of the patients' scores. Referral is a multifaceted clinical decision and should not be based upon VOMS scores alone. We evaluated a large age range of participants to increase generalizability across the population that experiences concussion. Future investigators should examine the predictive value of a multidomain assessment in high-risk concussion populations, such as adolescent and collegiate athletes and military populations. Also, we included patients across a wide range (1–384 days) of time since injury. Future authors should explore the predictive utility of VOMS item changes across recovery periods (ie, acute, subacute, or chronic). Finally, the current data were obtained from a concussion specialty clinic and may not be generalizable to other clinical settings, such as on-field assessments, emergency departments, or primary care physicians' offices.

Increasing empirical evidence shows that the application of early treatment and more rapid referrals for specialty care, such as vestibular therapies, can be beneficial after concussion.^21,22  Patients with vestibular or ocular impairments postconcussion may benefit from targeted vestibular and oculomotor therapies.^16,27  In addition to several VOMS items, in the current study, we demonstrated that patients with a concussion from a motor vehicle accident or a history of migraine and those who were feeling more emotional were associated with receiving a referral for vestibular therapy. Using validated clinical tools as part of a multidomain assessment can assist in facilitating timely referral for specialty therapies for patients with vestibular or ocular dysfunction after concussion.