Application of Commonly Used Physical Tests in a Virtual Environment in Patients With Concussion to Patients With Various Types and Severities of Acquired Brain Injury: Prospective Cohort Method Comparison Study.

Background: People who sustain a concussion and live in remote areas can experience challenges in accessing specialized assessments. In these cases, virtual approaches to assessment are of value. There is limited information on important psychometric properties of physical assessment measures used to evaluate people postconcussion virtually. Objective: The aims of this method-comparison psychometric study were to determine (1) inter- and intrarater reliability of a battery of concussion physical tests administered virtually in people with brain injury and (2) sensitivity and specificity of the virtual battery when compared to the in-person assessment. Methods: A total of 60 people living with acquired brain injuries attended an in-person and virtual assessment at the Ottawa Hospital Rehabilitation Centre. The order of the assessments, in-person and virtual, was randomized. The following physical measures were administered in-person and virtually: finger-to-nose test, vestibular ocular motor screening (VOMS), static balance testing (double leg, single leg, and tandem), saccades, cervical spine range of motion, and evaluation of effort. The virtual assessment was recorded, and a second clinician viewed and independently documented findings from the recordings twice at 1-month intervals. Results: The mean age of the participants was 45.65 (SD 16.50) years. The sensitivity metrics ranged from moderate (60%, 95% CI 30-86) to excellent (100%, 95% CI 71-100) for saccades and cervical spine right lateral flexion, respectively. Specificity ranged from 75%, 95% CI 35-95 to 100%, 95% CI 91-100 for left single leg stance eyes closed and left finger-to-nose testing, respectively. The interrater reliability ranged from poor for cervical spine extension (Cohen κ=0.20, 95% CI -0.07 to 0.47) to excellent for VOMS change in symptoms (Cohen κ=0.93, 95% CI 0.83-1). The intrarater reliability ranged from poor for cervical spine extension (Cohen κ=0.31, 95% CI 0.04-0.58) to excellent for the finger-to-nose testing on the right (Cohen κ=0.90, 95% CI 0.71-1). The wide CIs highlight variability in precision and suggest that further research with larger samples is needed before clinical use can be fully standardized. Conclusions: This study provides information on the psychometric properties associated with virtual administration of concussion measures. The VOMS change in symptoms measure appears to have the most promising properties when administered virtually when in-person visits are not possible. This is particularly relevant for patients in rural areas, for those facing access barriers, and in contexts where timely follow-up is challenging. However, caution should be maintained when administering certain concussion measures virtually. The wide CIs for some measures caution against over-reliance on single test findings, and clinicians should consider both the strengths and limitations of virtual delivery. Clinicians are encouraged to make informed decisions about which measures can be effectively used remotely, and which may still require in-person administration to maintain accuracy.