**2. Materials and Methods**

The WHO PIR adheres to methods designed from the collaborative efforts of the WHO Rehabilitation Programme and Cochrane Rehabilitation, and the directives from the WHO Guidelines Review Committee [4]. We used evidence mapping to synthesize and visualize study characteristics and evidence from CSRs on TBI. The overview was registered in Open Science Framework Registries (https://doi.org/0.17605/OSF.IO/M5XVG) and was reported following the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA 2020 statement) [18].

#### *2.1. Search Strategy*

According to the methodology developed by the Cochrane Rehabilitation [19], CSRs relevant to rehabilitation are continuously tagged to maintain an up-to-date database (https://rehabilitation.cochrane.org/evidence, accessed on 1 September 2019). We initially searched all CSRs related to TBI published between August 2009 and August 2019 and reported the results to the WHO. We subsequently searched the Cochrane Library to August 2021 to preserve the timeliness of evidence. Eligible CSRs included those assessing interventions for persons with TBI provided or prescribed by rehabilitation professionals [19].

We included only tagged CSRs that examined rehabilitation interventions on individuals with TBI, of any age and gender. CSRs focused on persons with acquired brain injury or

non-traumatic brain injury were excluded to ensure that the evidence synthesis is strictly applicable to persons who sustained a TBI.

#### *2.2. Assessment of Methodological Quality of Included Studies*

The methodological quality of each CSR was appraised by two assessors using the 16-item A Measurement Tool to Assess Systematic Reviews (AMSTAR) 2 tool. In this updated version, the 16 items are scored on a binary yes or no scale. AMSTAR-2 does not generate an 'overall score'; a high score may disguise weaknesses in 7 critical items [20]. The assessors adopted a process of 'considered judgment', which entails (1) interpreting weaknesses detected by the critical items and (2) reaching a consensus on the methodological quality of each CSR. Disagreements were resolved through discussion with a third assessor.

## *2.3. Data Extraction and Quality of Evidence Appraisal*

The authors referred to the Table of Findings presented in each of the included CSRs; these contain the following data: type of outcome, outcome measure(s), number of primary studies, sample sizes, type of population, intervention, comparator(s), and effect (i.e., no effect, in favor of intervention, or in favor of comparator). Data were collected and entered into an Excel datasheet.

In addition, the quality of evidence for each outcome was extracted using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) rating system. For CSRs that did not include GRADE ratings, two members of the Cochrane Rehabilitation team independently appraised the quality of evidence for the primary outcomes only using the GRADE approach [21]. Any disagreement was resolved through consensus decisionmaking involving a third author [22]. The GRADE appraisal approach included two steps: (1) retrieval of the original primary studies included in each CSR; and (2) tabulation of the quality of evidence provided in Summary of Findings tables using GRADEPro software.

#### *2.4. Summarizing the Data with an Evidence Map*

Quality of evidence and effect data were transferred into evidence maps developed in Excel. The evidence map integrates the outcome and rehabilitation intervention values for each comparison. The magnitude of the effect (i.e., no effect, in favor of intervention, or in favor of comparator) and the quality of evidence (i.e., very low, low, moderate, or high) were presented laterally and color-coded for each outcome in order to generate a visual aid to facilitate the understanding of the overall judgement of the evidence.

Evidence mapping was employed as a complementary method to collating and appraising evidence from the CSRs, and subsequently used to summarize the results for the overview. The instrument collated outcomes and rehabilitation interventions and resulted in a comprehensive overview of the quality of evidence and effects. Because we did not consider other outcomes and interventions in addition to the those examined in the included CSRs, evidence mapping was not used to identify evidence gaps.

#### **3. Results**

The authors identified six tagged CSRs related to TBI: one published in 2013 [23], two in 2015 [24,25], and three in 2017 [26–28] (see Figure 1).

Three CSRs included only participants who sustained a TBI and excluded people with acquired brain injury and non-traumatic injury. Two CSRs included studies with a mixed population only when disaggregated data were reported to ensure that evidence was relevant to TBI. Finally, one CSR reported including studies where the etiology of the TBI is uncertain. The characteristics of the included systematic reviews are reported in Table 1.

**Figure 1.** Flow chart displaying the tagging process of Cochrane systematic review.

Comprehensively, this mapping review encompasses 42 primary studies, 3983 participants, and 19 comparisons that examined the effectiveness and safety of either nonpharmacological or pharmacological interventions for individuals with TBI. Among nonpharmacological comparisons, four interventions (six outcomes) were categorized as very low quality of evidence, and eight interventions (16 outcomes) were deemed as low quality of evidence. Among the pharmacological comparisons, we found that four interventions (six outcomes) were rated very low and three interventions (three outcomes) were rated low in quality. The AMSTAR 2 assessment tool identified high methodological quality in the six CSRs; even when sources of funding were not reported. Results of the AMSTAR 2 assessment are displayed in Table 2.



5



6





Abbreviation: TBI = traumatic brain injury.


**Table 2.** AMSTAR 2 Quality Assessment of Cochrane Systematic Reviews.

Abbreviations: Y = Yes, N = No.

The evidence map findings were divided into two categories: (1) non-pharmacological interventions and (2) pharmacological interventions. Table 3 provides an overview of evidence map finding for non-pharmacological interventions for TBI. Table 4 provides an overview of evidence map finding for pharmacological interventions for TBI.


**Table 3.** Evidence map of non-pharmacological interventions.

High = H; M = Moderate; Low = L; VL = Very low; No effect = ⊗, Favor Intervention = ✓, Favor Comparator = ✘, Uncertain = **?**.

#### *3.1. Quality of Evidence Mapping for Non-Pharmacological Interventions*

3.1.1. Moderate Quality of Evidence

Hospital-based versus home-based cognitive rehabilitation likely has little to no effect on the return-to-work rate for moderate-to-severe TBI (1 study; *n* = 120) [26]. Similarly, cognitive didactic versus functional experiential therapy likely has little to no effect on the same outcome for moderate-to-severe TBI (1 study, *n* = 366) [26].


#### **Table 4.** Evidence map of pharmacological interventions.

Abbreviations: High = H; M = Moderate; Low = L; VL = Very low; No effect = ⊗; Favor Intervention = ✓; Favor Comparator = ✘; Uncertain = **?**; Not reported = NR.

#### 3.1.2. Low Quality of Evidence

Exercise using large muscle groups may have little to no effect on the cardiorespiratory fitness compared to usual care in severe and unspecified TBI severity levels (3 studies, *n* = 67) [27].

Cognitive rehabilitation may have little or no effect compared to no treatment on community integration in severe TBI (1 study; *n* = 12) [26], while it may have little to no effect relative to conventional therapy on return to work (1 study; *n* = 68) [26], and community integration (3 studies; *n* = 123) [26] in mild-to-severe TBI, respectively.

Electro-acupuncture as an adjunct treatment to rehabilitation training may have a positive effect on sensorimotor impairment (Fugl-Meyer Assessment) at 1 and 3 months, and on disability (Modified Barthel index) at 1 month, but not at 3 months, when the effects favored rehabilitation training alone (unspecified TBI severity; 1 study; *n* = 150) [23]. When added to conventional medical intervention, electro-acupuncture may make little to no difference to mortality rate, but it may increase the frequency of normal Glasgow Coma Score evaluations in coma patients with severe TBI (1 study, *n* = 50) [23]. Added to hyperbaric oxygen and rehabilitation training, electro-acupuncture may have an effect on the percentage of patients decreasing to moderate disability (Barthel Index > 60) but there is uncertainty on the effects on reducing its severity (Barthel Index > 40) (unspecified TBI severity; 1 study; *n* = 122) [23].

#### 3.1.3. Very Low Quality of Evidence

In mild-to-moderate TBI, the true effect of cognitive rehabilitation remains uncertain on return-to-work when compared to no treatment (1 study; *n* = 50) [26]; on activities of daily living when compared to conventional therapy (unspecified TBI severity; 2 studies, *n* = 41) [26]; on depression level versus waiting list (3 studies, *n* = 146) [24] and supportive psychotherapy (1 study; *n* = 48) [24]. There is also uncertainty on the utility on spasticity (6 h post-treatment) of repositioning splints equipped with participant-specific pseudoelastic hinges versus traditional splints with fixed angle braces for pediatric TBI (unspecified TBI severity; 1 study; *n* = 25) [28].

#### *3.2. Quality of Evidence Mapping for Pharmacological Interventions*
