Moving from the clinic to telehealth during the COVID-19 pandemic – a pilot clinical trial comparing in-clinic rehabilitation versus telerehabilitation for persisting symptoms following a mild Traumatic brain injury

Abstract Purpose The objective of this study was to compare the effects of an in-clinic cervicovestibular rehabilitation program (education, home exercises, manual techniques, sub-symptom threshold aerobic exercise [STAE] program) to a similar program (education, home-exercises, STAE program), but without manual techniques, provided in a telerehabilitation format in adults with persisting post-concussion symptoms (PCS). Materials and methods design In this parallel-group non-randomized clinical trial, 41 adults with persisting PCS were allocated to the in-clinic (n = 30) or telerehabilitation (n = 11) program. The outcome measures, which included the Post-Concussion Symptom Scale (PCSS; primary outcome), Numerical Pain Rating Scale (NPRS) for neck pain and headache and three disability questionnaires, were collected at baseline, weeks 6, 12, and 26. Non-parametric analysis for longitudinal data (NparLD) was used. Results For the PCSS, there was a group-by-time interaction (p = 0.05) with significant between-group differences at week 6, 12, and 26 (p < 0.05) for the in-clinic group. There were also group-by-time interactions for NPRS neck pain and headache (p < 0.05) for the in-clinic group. Conclusion The study suggests that a telehealth format failed to reach the efficiency of an in-clinic rehabilitation program in terms of symptoms reduction and functional improvement. These results must be interpreted with caution given the limited number of participants. ClinicalTrials.gov Identifier: NCT03677661. IMPLICATION FOR REHABILITATION The telerehabilitation format for adults with persisting post-concussion symptoms was widely implemented in the COVID-19 pandemic without any evidence of efficacy over the more traditional in-clinic rehabilitation format. The study suggests that a telerehabilitation format failed to reach the efficiency of an in-clinic rehabilitation program in terms of symptoms reduction and functional improvement. Clinicians should try to incorporate some in-clinic appointments when a telerehabilitation format is required such as for patients in underserved area.


Introduction
traumatic brain injury is the most important cause of disability among traumatic injuries [1].Mild traumatic brain injuries (mtBi) account for 81% of all categories of traumatic brain injury [2].Most mtBi symptoms resolve within the initial weeks following the trauma [3].however, persisting post-concussion symptoms (Pcs) defined as symptoms lasting more than 4 weeks, is a well-established long-term consequence of mtBi [4].Nearly 21%-46% of adults still experience Pcs and disability 3-6 months after the injury [5][6][7].headache and dizziness are the two most frequently reported symptoms followed by neck pain and nausea [8][9][10].injury to the central nervous system may explain persisting symptoms [11,12].however, other structures near or in the head like the cervical spine and the vestibular system can also be injured by the trauma and explain some persisting symptoms [11,13].among the recommended interventions following an mtBi, sub-symptom threshold aerobic exercise (stae) programs have been shown to be effective [14,15].cervical and oculo-vestibular rehabilitation are also recommended when the symptoms or dysfunction suggest cervical spine or oculo-vestibular impairments [14,[16][17][18].
the cOViD-19 pandemic has led to major changes in health care services, including the increased use of telehealth services.telerehabilitation has shown promising results in populations with musculoskeletal and neurological diseases [19][20][21][22][23]. telerehabilitation for the treatment of patients with mtBi has been studied in pilot cohort studies and was shown effective [24][25][26].For example, a feasibility study [25] recruited three adolescents with Pcs for a 6-week telerehabilitation program consisting of stae, education, and coordination exercises.an improvement in symptoms, illness perception, and occupational performance were observed at the end of the program for all participants.another study also enrolled a small cohort of participants (n = 18) in a telerehabilitation program and found improvement over time [24].these small feasibility studies did not compare their program to a similar in-clinic rehabilitation program.another study compared a telerehabilitation program to an in-clinic program but did not include stae in the telerehabilitation program.stae programs have been shown to be a key component of evidence-based rehabilitation programs following mtBi [26].therefore, there is a need for more evidence on the effectiveness and feasibility of telerehabilitation programs for the rehabilitation of adults with persisting Pcs.
in March 2020, when the cOViD-19 pandemic lockdown was ordered in canada, we were in the recruitment phase of a Randomized clinical trial (Rct) [27,28].the recruitment of the Rct was stopped for a 5-month period.Within this 5-month period, instead of refusing participants, we offered to new incoming participants a similar rehabilitation program as the experimental program of the Rct in a telerehabilitation format.Participants of the telerehabilitation group of the current study were recruited during this 5-month period.Outcomes of the telerehabilitation group were therefore compared to the experimental group from the original Rct.
the primary objective of this pilot clinical trial was thus to compare the effects of a cervicovestibular telerehabilitation program (advice, education, home-exercises; but no manual techniques) combined with stae to a similar in-clinic cervicovestibular program (advice, education, home-exercises, manual techniques) combined with stae in adults with persisting Pcs on the severity of symptoms and disability.Our hypothesis was that a telerehabilitation program would lead to similar clinical improvement than an in-clinic program.

Materials and methods
this is a sub-study of an Rct [28] on cervicovestibular rehabilitation in adults with mtBi that resulted from the lockdown implemented in canada, in 2020 during the cOViD-19 pandemic.a modification to the ethics approval of the original Rct has been approved by the ethics committee of the centre intégré universitaire de santé et de services sociaux de la capitale-Nationale (ciUsss-cN) to complete this trial (Modification #F1-6076 of #2018-619).

Population
adults aged between 18 and 65 years with a diagnosis of mtBi/ concussion [29,30] were recruited in concussion or medical clinics in the Quebec city area as well as using mass mailing to the Université Laval community.the following inclusion criteria were applied: (1) mtBi in the past 3-12 weeks with ongoing symptoms including at least dizziness, neck pain, and/or headaches that started within 72 h of the trauma; (2) at least one of the following cognitive symptoms: feeling slowed down, feeling like in a fog, "don't feel right", difficulty concentrating, difficulty remembering and confusion that started within 72 h of the trauma; (3) at least one positive finding during the cervical physical examination (tenderness/spasm/pain on segmental testing, or reduced range of motion), vestibular evaluation (Dix-hallpike or vestibulo-ocular reflex (VOR) tests) or ocular motor evaluation (convergence, smooth pursuits, or saccades).Potential participants were excluded if they had (1) >30 min of loss of consciousness; (2) more than 24 h of post-traumatic amnesia; (3) Glasgow coma scale score lower than 13 more than 30 min after the injury; (4) radiological evidence of subdural hemorrhage, epidural hemorrhage, intraparenchymal hemorrhage, and cerebral or cerebellar contusion; (5) post-injury hospitalization for more than 48 h; (6) any fracture produced by the traumatic event or since the traumatic event; (7) neurological condition other than mtBi; (8) comorbidities of cardiovascular or respiratory systems; (9) compensation from a third party payer.

Study design
this parallel-group non-randomized clinical trial included 8 supervised treatment sessions during a 6-week rehabilitation program and four evaluation sessions over 26 weeks (baseline, week 6, 12, and 26).all participants took part in an online baseline evaluation.after giving informed consent, they completed sociodemographic questionnaires and filled self-administered questionnaires evaluating symptoms and disability.Once baseline data was collected and due to governmental-issued cOViD-19 restrictions, participants were assigned to the teleRehabilitation (tR) group if recruited between april and august 2020 or the in-clinic Rehabilitation (icR) group if recruited before april 2020 or after august 2020.all participants performed a 6-week cervicovestibular rehabilitation program combined with an individualized stae program either in-clinic or in a web-based videoconference format.Between weeks 6 and 12, participants were asked to continue their exercises and follow the recommendations given by the treating therapist.all outcomes were re-evaluated at week 6, 12, and 26.standardized interventions were given by experienced physiotherapists, neuropsychologists, and kinesiologists.

Assignation and sampling
the participants of the tR group were recruited between May 1, 2020, and august 31, 2020, in a convenience sampling method.
During this period, all potential participants who wanted to receive online treatment for their mtBi and met the admissibility criteria were recruited in the tR group.Participants of the icR group were recruited [27,28] between september 2019 and March 2021 except during the cOViD lockdown period.

Intervention table 1 describes the interventions delivered in both groups with
the between-group differences bolded.the detailed icR program is described thoroughly in another publication [27].adherence to treatment sessions were documented using an electronic appointment system.all participants were advised to avoid co-interventions (pharmacologic or non-pharmacologic) but if co-interventions were used, they were recorded.Both groups took part in two education sessions given by a neuropsychologist; these sessions were provided online using a web-based videoconference platform for the participants in the tR group and in the clinic for the participants of the icR group.One exertional tolerance evaluation session was performed by a kinesiologist at the beginning of the program (online for the tR group, in-clinic for the icR group).two physiotherapists provided 8 treatment sessions according to an individualized care program which included a cervical and/or vestibular rehabilitation program combined with an individualized stae program.the icR group received advice, education, home exercises, and manual techniques aiming to improve cervical and vestibular impairments.the tR group received advice, education, home exercises from the physiotherapists on the web-based videoconference platform and performed self-mobilization (cervical and vestibular) in front of their computer under the online supervision of a physiotherapist.Participants in the tR group did not receive any manual techniques.exercises were prescribed to both groups either on-line or in-clinic, aiming to improve cervical and vestibular function according to impairments found (on-line for the tR group or in-clinic for the icR group).
Outcome measures all questionnaires (details described in a supplementary file) were filled online on a secured platform (limesurvey).a link was sent to participants at every time-point follow-up by an evaluator blinded to group allocation.

Post-concussion symptom scale (PCSS)
the Pcss was the primary outcome measure.it is a list of 22 symptoms for which participants rate each symptom for severity on a 0 (none) to 6 (severe) numerical scale [31].the total score and total number of symptoms were recorded.Other questionnaires were the Neck Disability Index (NDI) [32], the Headache Disability Inventory (HDI) [33], the Dizziness Handicap Inventory (DHI) [34], the headache and neck pain Numerical Pain Rating Scale (NPRS).

Sample size calculation
there was no a priori sample size calculation since the formation of the tR group resulted from the government-imposed cOViD-19 lockdown in the context of the global cOViD-19 pandemic.We recruited as many participants as possible in the period for which in-clinic intervention was not possible.the sample size for the icR group was determined for the original Rct [28].

Statistical analyses
statistical analyses were performed with iBM sPss statistics (version 25.0) and R software (version 4.2.2).Baseline demographic data were compared with an independent t-test and chi-squared test.the primary and secondary outcomes were analyzed with a rank-based non-parametric repeated-measures aNOVa (Non-parametric analysis for longitudinal data [NparlD]) [35].this type of aNOVa is especially well suited for clinical data as ( 1) it has no assumption about distribution, (2) it only requires at least ordinal dependent variables, (3) the type of distribution may change across conditions or time, notably caused by the ceiling or floor effects or by the presence of non-responders, (4) it has no requirement about the additivity of effects, (5) responses to treatment may differ across participants, (6) there is no need to impute missing data, discard participants or use an intent-to-treat protocol, (7) it works with unstructured repeated measures covariance matrices, consequently, there is no requirement about the homogeneity of variance or sphericity, (8) as a rank-based routine, it is relatively unimpacted by outliers, and (9) it was especially designed to work with small samples, notwithstanding whether sample sizes are balanced or not.
NparlD calculates effect sizes and named relative treatment effects (Rte).it is a probability with 0.5 as the null hypothesis value.One Rte is associated to each condition.at baseline, as participants are expected to score high on the symptom measures, Rte measures are expected to be above 0.5 with values of 0.56, 0.64, and 0.71 as indicating respectively small, moderate, or large impairments [36].Post-treatment, Rte measures are expected to fall below 0.5 with values of 0.44, 0.36, and 0.29 as revealing respectively small, moderate, and large improvements.all analyses were bootstrapped five thousand times to get non-parametric intervals of confidence at 95% of Rte (to assess their stability) and p-values (to assess the a posteriori power).

Flow of participants
thirty participants were assigned to the icR group and 11 participants to the tR group (see Figure 1 for the consort flow diagram

and table 2 for baseline characteristics)
. the number of previous mtBi and the number of participants having sustained at least one mtBi in the past was significantly higher in the icR group.the percentage of participants who sustained a sports injury was significantly higher in the tR group.the dropout rate was 6.70% (n = 2/30 participants) in the icR group and 9.10% (n = 1/11 participants) in the tR group (reasons described in Figure 1).the number of treatment sessions was similar between groups (p = 0.16).

Between-and within-group effects each dependent variable was analyzed separately as no multivariate version exists for this rank-based repeated-measures aNOVa.
Baseline Rte varied between 0.68 and 0.84, which indicates relatively large impairments at baseline for both groups.For the Pcss total score, there was a significant group-by-time interaction (p = 0.02, a posteriori power = 72%), with significant between-group differences at week 6, 12, and 26 (p = 0.03, 0.05, and 0.01 respectively) for the icR group (Figure 2(a,B), and table 3).For the icR group, Rte felt between 0.30 and 0.37 (moderate effect) while these values remain around 0.5 (small effect) for the tR group.at the end of the study, the tR group had a mean (standard deviation) of 29.2 (18.2) points and the icR group 13.5 (12.7) points for the Pcss total score.the significant time effect (p = 0.01, a posteriori power = 74%) indicates that both groups improved during the trials.
Group-by-time interactions were obtained for NPRs neck pain (p = 0.03, a posteriori power = 69%) and headache (p = 0.04, a posteriori  power = 65%) with significant between-group differences at week 12 and 26 for neck pain (p < 0.01) and at week 26 for headache (p = 0.02) at the advantage of the icR group (Figures 3 and 4 and table 3). at the end of the study, for NPRs neck pain and headache respectively, the Rte were 0.34 and 0.26 for the icR group (moderate improvement), as well as 0.63 and 0.50 for the tR group (no improvement).a significant time effect was observed for NDi, Dhi, and hDi with post-hoc analyses showing a significant improvement at all time points compared to baseline (p < 0.01; table 3).

Discussion
in this trial, an in-clinic cervicovestibular rehabilitation program combined with stae was compared to a similar program (but without manual techniques) in a telerehabilitation format in participants with persisting Pcs.Our hypothesis was that in-clinic and telerehabilitation programs would lead to similar clinical outcomes.For symptoms severity, neck pain, and headache, our hypothesis was refuted since statistically significant between-group differences were demonstrated for the icR group.For the Pcss, the effect sizes were moderate for the icR group and small for the tR group.the difference of 15.7 points on the Pcss total score at the end of the study exceeded the minimal detectable change of the Pcss (12 points [31]) which means that the difference exceeded the measurement error.For neck pain and headache, the effect sizes were moderate for the icR group and no effect was noted for the tR group.Furthermore, the between-group difference (1.9 NPRs points) exceeded the minimal clinically important difference of the NPRs (1.5 points [37]) which suggests  a meaningful larger effect of the in-clinic intervention from the patient's perspective on pain intensity in comparison to telerehabilitation.the superiority of the icR intervention on most outcomes could be explained by several factors.the effectiveness of manual therapy hands-on techniques to improve neck pain has been demonstrated in systematic reviews [38,39].it is thus possible that the absence of hands-on techniques in the tR group could have led to the stagnation of neck pain symptoms and neck pain-related disability over time.some vestibular maneuvers were also supported manually with the hands of the therapist directing or creating movement which could have helped the icR group improve in a greater extent than the tR group.however, in the original Rct, the hands-on techniques directed to the neck and to the vestibular system were performed only in the experimental group [28].this group, however, failed to demonstrate between-group differences for the same outcomes when compared to an in-clinic stae program alone.even if the hands-on hypothesis seems plausible, other explanation needs to be explored to explain our results.Other hypotheses that could  explain the superiority of icR intervention lies in the possibility that direct in-clinic contact with the therapist could produce a stronger reassuring effect, or, the exercises were possibly better understood by the patient with an in-person demonstration and direct feedback rather than through a video conference device.Finally, knowing that time spent in front of a screen can trigger post-concussion symptoms [40], it is possible that participants of the tR group perceived an increase of some symptoms while they were having their intervention in front of their computer.
the cOViD-19 pandemic situation provided a favorable circumstance for the implementation of telerehabilitation.as a result, a few pilot studies were recently published on the effectiveness of telerehabilitation on mtBi/concussion [25,26].the most similar research paper with regards to our study compared 56 subacute mtBi participants who received 8 in-clinic physiotherapy sessions over a 6-week period to 17 subacute mtBi who received 8 telerehabilitation sessions [26]. in this trial, cervical spine, cardiovascular, and balance dysfunctions were targeted by physiotherapists.however, contrary to our protocol, stae was not included in the telerehabilitation group.as in the present study, the effect-size of the in-clinic intervention on symptom intensity was larger than the telerehabilitation intervention.these results suggest that, whenever possible, clinicians should incorporate some in-clinic sessions into telerehabilitation programs when a full in-clinic rehabilitation program is not possible.Namely, outside of considerations for the cOViD pandemic, this could apply to people living in under-served or remote areas where frequent in-clinic appointments are not possible.

Study limitations
this study has limitations.the convenience sampling procedure without any randomization for the telerehabilitation group could have led to a selection bias.a selection bias can lead to between-group differences at baseline which was the case in our study for some baseline characteristics (number of previous concussion and mechanism of injury).however, even if the number of previous concussions, a known predictor of poor recovery [41], was higher in the in-clinic rehabilitation group, this group improved significantly more compared to the telerehabilitation group.it is unknown if the between-group differences would have been larger with similar baseline characteristics.also, the global physical and psychological impact of the sanitary restrictions imposed during the cOViD-19 pandemic have been shown, in a qualitative study, to have negatively affected the well-being in persisting Pcs [42]. it may have influenced some of the outcomes measured in this study.Finally, the small number of participants in the tR group should lead to interpret the results with caution.

Conclusion
the study suggests that an in-clinic cervicovestibular rehabilitation program combined with stae was superior to a similar intervention delivered through telerehabilitation in terms of symptoms improvement in persons with persisting post-concussion symptoms.clinicians should therefore try to incorporate some in-clinic appointments when a telerehabilitation format is used by their patients.these results, however, must be interpreted with caution given the limited number of participants in the telerehabilitation group and the non-randomized group allocation.Further studies are necessary to confirm our findings.

Figure 2 .
Figure 2. Post-Concussion symptom scale total score.PCss: Post-Concussion symptom scale; iCRG: in-Clinic Rehabilitation Group; tRG: tele Rehabilitation Group; large red dots: means of the tRG; small red dots: individual score of each participants of the tRG; large blue triangles: means of the iCRG; small blue triangles: individual score of each participants of the iCRG; error bars: 95% Confidence interval.

Figure 3 .
Figure 3. numerical Pain Rating scale neck pain.nPRs: numerical Pain Rating scale; iCRG: in-Clinic Rehabilitation Group; tRG: tele Rehabilitation Group; large red dots: means of the tRG; large blue triangles: means of the iCRG; error bars: 95% Confidence interval.

Figure 4 .
Figure 4. numerical Pain Rating scale headache.nPRs: numerical Pain Rating scale; iCRG: in-Clinic Rehabilitation Group; tRG: tele Rehabilitation Group; large red dots: means of the tRG; large blue triangles: means of the iCRG; error bars: 95% Confidence interval.

Table 3 .
outcome by treatment group and time-point follow-up.