Introduction: Recent randomized controlled trials (RCTs) have assessed the role of vagus nerve stimulation (VNS) when paired with standard rehabilitation in stroke patients. This review aimed to evaluate the efficacy and safety of VNS as a novel treatment option for post-stroke recovery. Methods: We searched PubMed, EMBASE, Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials (CENTRAL), and CINAHL Plus for articles published from their date of inception to June 2021. RCTs investigating the efficacy or safety of VNS on post-stroke recovery were included. The outcomes were upper limb sensorimotor function, health-related quality of life, level of independence, cardiovascular effects, and adverse events. The risk of bias was assessed using the Cochrane risk-of-bias tool, while the certainty of the evidence was assessed using the Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) criteria. Review Manager 5.4 was used to conduct the meta-analysis. Results: Seven RCTs (n = 236 subjects) met the eligibility criteria. Upper limb sensorimotor function, assessed by the Fugl-Meyer Assessment for Upper Extremity (FMA-UE), improved at day 1 (n = 4 RCTs; standardized mean difference [SMD] 1.01; 95% confidence interval [CI]: 0.35–1.66) and day 90 post-intervention (n = 3 RCTs; SMD 0.64; 95% CI: 0.31–0.98; moderate certainty of evidence) but not at day 30 follow-up (n = 2 RCTs; SMD 1.54; 95% CI: −0.39 to 3.46). Clinically significant upper limb sensorimotor function recovery, as defined by ≥6 points increase in FMA-UE, was significantly higher at day 1 (n = 2 RCTs; risk ratio [RR] 2.01; 95% CI: 1.02–3.94) and day 90 post-intervention (n = 2 RCTs; RR 2.14; 95% CI: 1.32–3.45; moderate certainty of the evidence). The between-group effect sizes for upper limb sensorimotor function recovery was medium to large (Hedges’ g 0.535–2.659). While the level of independence improved with VNS, its impact on health-related quality of life remains unclear as this was only studied in two trials with mixed results. Generally, adverse events reported were mild and self-limiting. Conclusion: VNS may be an effective and safe adjunct to standard rehabilitation for post-stroke recovery; however, its clinical significance and long-term efficacy and safety remain unclear.

© 2022 The Author(s). Published by S. Karger AG, Basel

Introduction

Stroke is often associated with a significant disease burden requiring substantial treatment, especially post-stroke care [1]. Most survivors suffer from neurological deficits and require around-the-clock care or institutionalization due to limited functional independence [2]. Rehabilitation is an important component of post-stroke care. The goal of rehabilitation is to provide extensive education and task-specific training to maximize functional abilities, thus improving the level of independence. Physical rehabilitation facilitates synaptic plasticity and cortical reorganization within the motor cortex [3].

Over the years, several large randomized controlled trials (RCTs) of increased rehabilitation regime, the use of rehabilitation devices, and brain stimulation therapies devices have not demonstrated clinically meaningful motor recovery compared to standard rehabilitation therapy only [4, 5]. Several preclinical trials have reported favourable post-stroke recovery following combined vagus nerve stimulation (VNS) and rehabilitation therapy, and several clinical trials have shown promising results [6-8]. VNS is a technique that provides stimulations to the vagus nerve via an implantable device or a non-implantable device attached to the skin overlying the vagus nerve [9, 10]. The enhanced rehabilitation model is postulated to induce a brain environment that might increase the potential for experience-dependent plasticity [11]. A recent meta-analysis of animal studies highlighted that combined VNS and rehabilitation therapy facilitated clinically significant motor function recovery following neurological injuries [10]. This study aimed to systematically review the efficacy and safety of VNS on post-stroke recovery.

Methods

The systematic review was designed and conducted in accordance with the PRISMA 2020 guidelines (online suppl. File; see www.karger.com/doi/10.1159/000526470 for all online suppl. material) [12]. This review was prospectively registered on PROSPERO (CRD42020160094). Additional outcome measures, including health-related quality of life (HRQoL), level of independence, and cardiovascular effects, were included for this analysis to provide a comprehensive assessment of the efficacy and safety of VNS in stroke survivors.

Search Strategy

Two independent review authors (R.A. and J.M.) searched for eligible publications on five electronic databases which include PubMed, EMBASE, Cochrane Library, CINAHL Plus, and Cochrane Central Register of Controlled Trials (CENTRAL). All databases were searched from the date of inception to May 2021 with no language restrictions. Our search strategy can be found in the online supplementary file. The reference lists of included studies were also hand-searched to identify potential eligible publications. Publications identified through the searches were exported to a reference management software (EndNoteX9) for screening. Duplicate records were identified and removed. Two independent review authors (R.A. and J.M.) screened the titles and abstracts of all identified publications for eligibility based on our inclusion criteria. After shortlisting, full-text articles were retrieved for further assessment by two authors independently (R.A. and J.M.) to determine if these studies should be included in our review or excluded with disagreement resolved by consensus and involvement of an arbiter (C.F.N.) if necessary.

Eligibility Criteria

Studies were considered eligible if they were RCTs, quasi-RCTs, and cluster-RCTs. We included studies in which the participants were adults (≥18 years of age) with a diagnosis of ischaemic or haemorrhagic stroke. The intervention in included studies was any form of VNS therapy with or without any forms of rehabilitation therapy, whereas the comparator was standard rehabilitation therapy with or without sham therapy or placebo.

Data Extraction and Risk of Bias Assessment

Two review authors (H.R. and R.A.) independently extracted and coded all data using a customized data extraction proforma. Missing data were obtained by contacting the authors of the published trials. The components of data extraction include authors, year of publication, the total number of participants, the details of intervention and control, duration of follow-up, outcome measures, and funding source. The risk of bias of eligible publications was assessed using Cochrane collaboration methods for Risk of Bias (RoB) assessment tool [13]. The risk of bias was evaluated independently by two review authors (R.A. and J.M.) and categorized into low, high, or unclear risk, with disagreements adjudicated by the third author (C.F.N.).

Outcome Measures

The efficacy and safety outcomes of VNS were the outcome of interest. Efficacy outcomes include upper limb sensorimotor function recovery, HRQoL, and level of independence, while safety outcomes include cardiovascular effects and adverse events. Upper limb sensorimotor function recovery was primarily assessed by the Fugl-Meyer Assessment of Upper Extremity (FMA-UE) score. FMA-UE is a core standardized instrument that explores the balancing, joint and sensorimotor function in post-stroke hemiplegic patients and has been the most widely used tool to determine upper limb sensorimotor recovery [14, 15]. Page et al. [15] developed an estimate model for the assessment of upper limb sensorimotor recovery using the global rating of change scale and reported an increase of 4.25–7.25 points as the minimum estimate. Page’s estimate model has been widely referenced and used in neurorehabilitation trials, although there is no consensus on the type of estimate model for clinically meaningful upper limb sensorimotor recovery.

Statistical Analysis

Extracted data were transferred by a review author (R.A.) into Review Manager 5.4 (RevMan5.4) for quantitative analysis and was summarized in narrative writing, forest plots, figures, and tables. Continuous outcomes were expressed as standardized mean difference (SMD), while dichotomous outcomes were expressed as the risk ratio (RR) using the Mantel-Haenszel method. Descriptive analysis was performed for outcome measures reported in only one study, and the outcomes were expressed as mean difference (MD) and standard deviation. A 95% confidence interval (CI) was calculated for all outcomes. Meta-analysis was conducted on outcome measures that were assessed in at least two or more studies. Statistical heterogeneity between studies was investigated by visually assessing the forest plot, the statistical significance of the χ2 test, and the value of I2 statistic. Data were pooled from each study using a random-effects model. Sensitivity analysis was performed to determine the robustness of the meta-analysis. The effect size between the VNS group and the control group for upper limb sensorimotor function recovery was estimated using the Hedges’ g statistic [16]. The certainty of the evidence was assessed using the Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) framework, and the summary of findings table was generated using GRADEpro software [17].

ResultsSearch Results

Our initial search yielded 755 papers, of which 286 duplicates were removed. Out of the 469 potentially relevant papers, 71 were maintained for further analysis after screening based on the title and abstract. Seven of these [18-24] met the eligibility criteria and were included in the final descriptive analysis. Figure 1 illustrates the PRISMA flow diagram. Further screening involving the reference list of included studies and published reviews yielded no additional studies. Table 1 illustrates the characteristics of included trials. The included studies were of varying methodology and were predominantly performed in a tertiary setting.

Table 1.

Characteristics of included studies

Fig. 1.Study Characteristics

Table 2 illustrates the overview of the clinical trials. The seven clinical trials [18-24] represented a total of 236 participants who had an ischaemic or haemorrhagic stroke with neurological deficits. The mean age of participants ranged from 53.7 to 72.9 years, with male predominance across all studies. The disease duration before trial recruitment ranged from 5 days to 10 years. The stroke location of patients included the supratentorial region [18, 20-22], brainstem [23], and cerebellum [23]. The mode of VNS therapy administered include both non-invasive and invasive implantation: the former involved the placement of electrodes on the skin overlying the auricular branch of the left vagus nerve, while the latter involved the surgical implantation of electrodes on the left vagus nerve. The intensity, pulse duration, and frequency of VNS therapy differed substantially across all seven studies. With only two exceptions [18, 24], the control group of the clinical trials received sham stimulation in addition to rehabilitation therapy. Only five trials [18-22] (71%) assessed the upper limb sensorimotor function recovery after stroke. All the clinical trials assessed adverse events associated with VNS therapy, except for one trial [24]. The length of follow-up varied from 1 day to 90 days post-intervention.

Table 2.

An overview of included studies

Risk of Bias

Overall, the risk of bias was mixed (online suppl. Fig. S1). There was either a low or unclear risk of selection bias where information on allocation concealment was particularly not reported in several trials [18-21, 23, 24]. Blinding of participants and assessors were achieved in about half of the included studies, but blinding of participants was not possible in three trials due to the nature of the intervention [18, 20, 22]. Attrition bias was either low or unclear, whereas most studies showed a low risk of reporting bias.

Efficacy OutcomesUpper Limb Sensorimotor Function Recovery

All clinical trials that assessed upper limb sensorimotor function recovery employed the FMA-UE [18-22]. Our meta-analysis reported a statistically significant increase in the FMA-UE score for the intervention group on day 1 (SMD 1.01, 95% CI: 0.35–1.66, p = 0.002; I2 = 61%, Fig. 2a) and on day 90 (SMD 0.64, 95% CI: 0.31–0.98, p = 0.0002; I2 = 0%, Fig. 2a). However, we found no evidence that combined VNS with rehabilitation improves FMA-UE at day 30 (SMD 1.54, 95% CI: 0.39–3.46, p = 0.12; I2 = 84%, Fig. 2a). Since the result at day 30 did not include the trial that had the largest statistical weight for the other two time-points, a pooled analysis of the two trials that had data for all three time-point was performed. The results (online suppl. Fig. S2) showed no statistically significant increase of the FMA-UE score at days 1 and 30 but an improvement at day 90 (SMD 0.99, 95% CI: 0.30–1.67, p = 0.005, I2 = 0%). Due to the presence of moderate-to-substantial heterogeneity at some of the time-points, a sensitivity analysis was conducted by removing the Wu et al. [21] pilot trial. At day 1, the heterogeneity of the FMA-UE score at day 1 reduced significantly, while the estimates of intervention effect remained statistically significant (SMD 0.67; 95% CI: 0.33–1.00; p < 0.0001; I2 = 0%; online suppl. Fig. S3). Online supplementary Table S1 illustrates the effect sizes (Hedges’ g) of between-group differences of the FMA-UE score. All trials reported a medium-to-large effect size (Hedges’ g 0.535–2.659) for upper limb sensorimotor recovery. A clinically significant response, as defined by ≥6 points increase in the FMA-UE score, was significantly higher in the intervention groups compared to control groups on day 1 (RR 2.01; 95% CI: 1.02–3.94; p = 0.04; I2 = 0%, Fig. 2b) and day 90 follow-up (RR 2.14; 95% CI: 1.32–3.45; p = 0.002; I2 = 0%, Fig. 2b). Online supplementary Table S2 illustrates the certainty of evidence using the GRADE framework for upper limb sensorimotor function recovery as assessed by the FMA-UE score. We are moderately confident that combined VNS with stroke rehabilitation will improve and produce a clinically significant upper limb sensorimotor function recovery at day 90 follow-up. The Wolf Motor Function Test (WMFT) improved following VNS therapy on day 1 (SMD 1.18; 95% CI: −0.16 to 2.52; p = 0.08; I2 = 71%, Fig. 2c) and day 90 follow-up (SMD 0.87; 95% CI: 0.50–1.54; p < 0.00001; I2 = 0%, Fig. 2c). Other secondary outcomes of upper limb sensorimotor function are summarized in online supplementary Table S3.

Fig. 2.

Forest plots for upper limb sensorimotor recovery. a FMA-UE. b Clinically meaningful recovery as assessed by ≥6 points improvements of the FMA-UE. c WMFT.

Health-Related Quality of Life

Several standardized instruments were used to assess HRQoL. While two RCTs addressed the Stroke Impact Scale (SIS) of hand, only one study addressed stroke-specific quality of life scale (SS-QOL) and EQ-5D. Paired VNS therapy with rehabilitation resulted in slight improvement in the SIS of hand on day 1 (SMD 0.09; 95% CI: −0.26 to 0.45; p = 0.60; I2 = 0%, Fig. 3) and day 90 follow-up (SMD 0.05; 95% CI: −0.30 to 0.40; p = 0.78; I2 = 0%, Fig. 3). The improvement in SS-QOL was not statistically significant on day 1 (MD 6.80; 95% CI: −1.66 to 15.26; p = 0.119) and day 90 post-intervention (MD 7.90; 95% CI: –1.86 to 17.66; p = 0.117) [22]. Although the improvement of EQ-5D was non-significant on day 1 follow-up (MD 4.40; 95% CI: −1.75 to 10.55; p = 0.163), the trial reported a statistically significant increase of EQ-5D in the intervention group on day 90 follow-up (MD 7.20; 95% CI: 0.65–13.75; p = 0.034) [22].

Fig. 3.

Forest plot for HRQoL, assessed by the SIS of hand.

Level of Independence

Two clinical trials assessed the level of independence with two different standardized instruments [22, 24]. VNS-REHAB has reported significant improvement in the SIS of activity of daily living on day 1 (MD 4.70; 95% CI: 0.43–8.97; p = 0.034) and day 90 follow-up (MD 5.10; 95% CI: 0.08–10.12; p = 0.049). Pre- and post-intervention analysis of the trial conducted by Subrahmanyamc [24] demonstrated significant improvement of modified Barthel Index (MD 3.07; standard deviation 1.71; p < 0.001) following the intervention, and the improvement was significant when compared to the control group (MD 2.07; 95% CI: −3.28 to −0.86; p = 0.002) [24].

Safety OutcomesCardiovascular Effect

Two clinical trials [19, 21] assessed the effect of VNS therapy on cardiovascular parameters including blood pressure and heart rate (HR). Wu et al. [21] reported a mild and asymptomatic reduction in systolic blood pressure with significant pre- and post-intervention group differences (−0.607 mm Hg vs. +4.273 mm Hg; p = 0.01) but negligible effect on HR and diastolic blood pressure. On the contrary, Capone et al. [19] reported no VNS effect on all cardiovascular parameters. Concerning arrhythmias, there were 2 cases of paroxysmal atrial fibrillation in patients with known prior history of this condition [18, 20].

Adverse Events

With the exception of one trial [24], all the other trials assessed adverse events. Although there were several adverse events reported, no mortality was reported. There were a total of 163 adverse events in 115 participants in the intervention group and 171 adverse events in 121 participants in the control group, with the majority (93%) reported in VNS-REHAB. Two of the six trials reported no adverse events [19, 23]. One trial reported only one adverse event [21], while the other three trials [18, 20, 22] reported a fair distribution of adverse events between the intervention and the control group, respectively. VNS-REHAB reported at least one adverse event in 81% of the intervention group and 76% of the control group (163 vs. 171 total events, respectively) [22]. The most common adverse event was post-operative pain, followed by hoarseness, coughing, headache, and wound infection. Among the participants who had a cervical implantable device [18, 20, 22], the incidence of vocal cord palsy (2.1%), left phrenic nerve palsy (0.7%), and dysphagia (0.7%) was low. These significant adverse events generally resolved within 5 weeks of conservative management [18, 20, 22], except a case of left vocal cord and phrenic nerve palsy which self-resolved after 9 months [18].

Discussion

VNS may be a potential adjunct to standard rehabilitation for stroke patients with neurological deficits. Our systematic review suggests that the enhanced rehabilitation model may have potential benefits on post-stroke recovery parameters including upper limb sensorimotor function, HRQoL, and level of independence. A small number of patients receiving VNS experienced more serious adverse events such as vocal cord palsy, phrenic nerve palsy, and dysphagia which were rare and self-limiting.

Upper limb sensorimotor function, primarily assessed by FMA-UE, may improve following combined VNS therapy with rehabilitation therapy, but more evidence is required to assess its clinical significance. Based on the page estimate model [15], our meta-analysis has suggested that a significantly higher proportion of participants achieving clinically meaningful upper limb sensorimotor recovery in the intervention group at day 1 (RR 2.01; p = 0.04) and day 90 follow-up (RR 2.14; p = 0.002). However, our meta-analysis of the SIS of hand revealed a non-statistically significant result on day 1 (SMD 0.09; p = 0.60) and day 90 (SMD 0.05; p = 0.78). Since the proposed mechanism of action is to induce a brain environment that might increase the potential for experience-dependent plasticity [11], we expect that combined VNS with rehabilitation will have an effect on impairment and possibly activity (if rehabilitation was specifically task-specific training) [25]. A further effect on participation or quality of life will likely require additional therapeutic components given the nature of rehabilitation as a complex intervention. Environment factors, including infrastructure, government policies, employment, transportation, family and social support, can influence rehabilitation outcomes [25]. The coupling of impairment-based intervention with the improvement of other therapeutic components will likely improve the HRQoL of stroke patients. We note the involvement of industry funding in most trials (n = 4, 80%) [18, 20-22]. Commercially funded clinical trials have become increasingly common and those which are well-designed with no protocol deviation are at lower risk of bias [26]. Except for the pilot study by Dawson (1a) et al. [18], the other trials were either single-blinded [21], double-blinded [20], or triple-blinded [22]. Most participants of VNS-REHAB were unclear or incorrectly inferred treatment allocation (p = 0.63) to both VNS or sham stimulation received, thus reducing the risk of expectancy bias [22]. Besides, the clinical trials did not suffer from significant protocol deviations [18, 20-22]. An independent researcher was recruited to perform data analysis and interpretation for VNS-REHAB [22].

The stimulation of the vagus nerve is postulated to promote neural plasticity in the primary motor cortex by facilitating neuromodulator production, including norepinephrine, serotonin, and acetylcholine [11]. Pairing repeated brief VNS burst with task-specific rehabilitation tripled the synaptic connection from the corticospinal motor network to the task-relevant musculature and produced long-lasting motor function recovery [27, 28]. Based on the findings of preclinical trials, animals receiving paired therapy experienced greater forelimb strength recovery compared to those receiving 2-h delayed VNS therapy following rehabilitation therapy or vice versa [8, 29]. Rehabilitation was conducted immediately following the completion of the VNS session in two trials [19, 21], while for the other three trials [18, 20, 22], patients received paired VNS and rehabilitation therapy, whereby stimulations were delivered simultaneously with each specific movement. This review suggests that the significance of the temporal relationship between VNS and task-specific training will require further attention as all clinical trials reported clinically significant motor recovery irrespective of the stimulation mode.

The majority of the trials [18-20, 22] that assessed upper limb sensorimotor recovery recruited patients in the chronic phase of stroke, with the longest disease duration of 10 years, except for the trial by Wu et al. [21] which recruited patients with subacute stroke. Our meta-analysis reported a drop in the SMD of FMA-UE score from 1.01 at day 1 follow-up to 0.64 at day 90 follow-up. Although within-group differences showed an increase in FMA-UE following VNS over the follow-up period, between-group differences reported as SMD dropped at day 90 follow-up period. We could not confidently exclude the possibility that the initial gain in motor performance was not related to the counteraction of learnt non-use phenomenon by standard rehabilitative training [30]. Constant motivation from occupational or physiotherapists during training exercises may produce positive reinforcement, subsequently leading to use-dependent cortical reorganization of the affected limb [30]. However, there is also a possibility that patients in the intervention may have achieved motor recovery plateau at day 90 follow-up [31]. To date, there is conflicting evidence surrounding the ideal time window for the initiation of stroke rehabilitation to maximum post-stroke function recovery [32, 33]. Although it is possible that the initiation of the enhanced rehabilitation during the subacute phase of stroke had resulted in better therapeutic outcomes [21], this review could not establish the association due to the limited trials available for the subgroup analysis. Our meta-analysis showed an improvement of upper limb sensorimotor function recovery at days 1 and 90 but not at day 30. This was possibly a reflection of the absence of data at day 30 from the VNS-REHAB trial which had a relatively larger sample size. Pooled analysis of the two smaller clinical trials (online suppl. Fig. S2) showed no statistically significant improvement of FMA-UE at days 1 and 30 but an improvement at day 90. It is noteworthy that except for the VNS-REHAB trial, all other trials were pilot studies. Therefore, although it is shown that the effect sizes are medium to large, more clinical trials with larger sample sizes are required to further assess the efficacy of VNS on upper limb sensorimotor function recovery over time.

Based on the findings of preclinical trials [7, 8, 34], the most optimized VNS intensity is 0.8 mA, which is employed by only three trials [18, 20, 22]. Only two participants in VNS-REHAB required lower stimulation intensity at 0.6 mA and 0.7 mA for comfort measures. The intensity could be up-titrated over a few sessions as tolerance zone boundary increases in most patients [35]. Recent studies have suggested that differential activation of noradrenergic receptors may have contributed to the inverted-U relationship between VNS intensity and cortical plasticity [34, 36]. Despite the varying stimulation intensity employed, the clinical trials reported clinically significant improvement in their respective outcome measures. Multiple factors could have contributed to the increased heterogeneity (Fig. 2a, c), including patient characteristics and treatment regime, in addition to stimulation parameters. On sensitivity analysis (online suppl. Fig. S3), the heterogeneity of FMA-UE score at day 1 follow-up reduces to zero, but the estimates of therapeutic effect remain statistically significant when the study conducted by Wu et al. [21] was excluded from the analysis. Short-term daily rehabilitation with higher stimulation intensity (mean 1.66 mA) employed by Wu et al. [21] could have contributed to the exceptionally positive result on day 1 and 30 follow-up, while the weakened result on day 90 follow-up could be due to the diminishing treatment effect. The persistent treatment effect in the other two trials [20, 22] could have been contributed by ongoing home-based rehabilitation during the follow-up period, which was not employed by Wu et al. [21]. However, due to the limited evidence, we could not confidently describe the possible benefit of higher stimulation intensity or the possible diminishing treatment effect after treatment completion. We suggest that further high-quality RCT is required to determine the optimum stimulation parameters and the treatment regime.

Given the heterogeneous treatment regime of the included trials, this review could not define the optimal number of VNS sessions and session duration for clinically meaningful post-stroke recovery. Only three clinical trials [18, 20, 22] employed a similar treatment regime: three 2-h in-clinic sessions of paired VNS with rehabilitation per week for six consecutive weeks during the treatment period. Despite the varying treatment regime, all trials reported clinically significant improvement in their respective outcome measures. Only two trials [20, 22] employed home-based rehabilitation during the follow-up period. A recent prospective cohort study has demonstrated good long-term adherence (57.4%) to home-based rehabilitation paired with VNS therapy, with an average completion of 200 ± 63 sessions lasting 2 ± 0.6 h per week in 1 year. At 1-year follow-up, the study reported significant improvement in the FMA-UE score (MD 9.2; 95% CI: 4.7–13.7) compared to baseline, and 11 out of 15 participants (73%) achieved a clinically meaningful improvement of upper limb sensorimotor function [37]. A recent RCT has demonstrated that home-based telerehabilitation comprised of arm motor therapy, and stroke education is non-inferior to in-clinic therapy in stroke survivors [38]. The feasibility of paired VNS with rehabilitation in the community setting demonstrated in the two trials [20, 22] showed a promising opportunity for the incorporation of the treatment regime into a home-based telerehabilitation programme, providing wider access to the general population.

Based on the current finding, we are unable to decipher if invasive VNS is superior to transcutaneous VNS. Only three of the seven trials [18, 20, 22] (43%) employed cervical implantable VNS. All the trials reported improved outcome measures, irrespective of the administration mode. VNS therapy is generally safe and well-tolerable with a zero mortality rate and minimal adverse events. Our systematic review reported a low rate of vocal cord palsy, phrenic nerve palsy, and dysphagia, and these cases were self-limiting. The efficacy and safety profile of VNS has been established in other medical conditions including intractable epilepsy and neuropsychiatric disorders more than two decades ago [39]. A recent retrospective study of 497 surgical procedures involving the implantation of VNS devices reported low surgical complications, including infection (2.6%), post-operative haematoma (1.9%), vocal cord palsy (1.4%), and lower facial weakness (0.2%) [40]. Moreover, histopathological analysis has revealed preserved nerve fibres and the surrounding tissues after long-term electrode implantation on the vagus nerve in animal models [41]. However, despite the promising safety profile, further prospective studies with a longer follow-up period are required to assess for possible long-term adverse outcomes.

Electrodes were placed at the left vagus nerves in all seven clinical trials. Our systematic review suggests that VNS does not induce clinically significant changes in cardiovascular parameters, including HR and blood pressure. Long-term electrode implantation does not influence vagally mediated reflexes including respiration or blood pressure [35, 41] but could promote cardioprotective effects including suppressed stellate ganglion nerve activity and reduced incidence of paroxysmal atrial tachyarrhythmias [42]. Despite the extensive safety profile over the years, there were a few case reports of transient asystole secondary to complete heart block that resolved upon VNS cessation in patients with cardiac abnormalities [43, 44]. We suggest that close monitoring is required for patients receiving the first session of VNS therapy.

We have conducted a comprehensive systematic review on the efficacy and safety of combined VNS with conventional rehabilitation on post-stroke recovery. All RCTs fulling the eligibility criteria were included without language restrictions, providing an impartial synthesis of the available evidence. Furthermore, our systematic review included participants with various stroke lesions and disease duration. However, there are some limitations of our systematic review. Only seven clinical trials with small sample sizes (n = 236 subjects) were identified from the extensive search. Besides that, several of these studies did not provide sufficient information for an adequate ROB assessment. Only two studies [20, 22] showed a low risk of bias in most domains. We also note that since three clinical trials that employed invasive VNS are funded by the same company with the same study group, it is important to see if such data can be replicated by other groups or with non-invasive VNS.

Conclusion

VNS may be an effective and safe adjunct to the standard rehabilitation of stroke survivors with neurological deficits. Combined VNS with conventional rehabilitation is shown to have the potential to improve upper limb sensorimotor function and level of independence. In terms of the safety profile, this systematic review showed that short-term VNS is a safe adjunct with no cases of mortality and comparable outcomes to the standard rehabilitation group. However, the clinical significance and long-term efficacy and safety of VNS when paired with standard rehabilitation remain inconclusive.

Statement of Ethics

An ethics approval is not applicable as this is a systematic review and meta-analysis of published data.

Conflict of Interest Statement

The authors declare no conflicts of interest.

Funding Sources

The authors declare no funding received.

Author Contributions

Roshan Ananda: conceptualization, methodology, systematic search, screening of studies, data extraction, formal analysis, software, visualization, and manuscript draft and revision. Mohd Hariz Bin Roslan: conceptualization, methodology, data extraction, visualization, and manuscript revision. Lin Ling Wong: conceptualization, methodology, visualization, and manuscript revision. Nevein Philip Botross: conceptualization, methodology, manuscript revision, and supervision. Chin Fang Ngim: conceptualization, methodology, screening of studies, manuscript revision, and supervision. Jeevitha Mariapun: conceptualization, methodology, systematic search, screening of studies, formal analysis, manuscript revision, and supervision.

Data Availability Statement

The data that support the findings of this study are openly available in the following publications [18-24].

This article is licensed under the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC). Usage and distribution for commercial purposes requires written permission. Drug Dosage: The authors and the publisher have exerted every effort to ensure that drug selection and dosage set forth in this text are in accord with current recommendations and practice at the time of publication. However, in view of ongoing research, changes in government regulations, and the constant flow of information relating to drug therapy and drug reactions, the reader is urged to check the package insert for each drug for any changes in indications and dosage and for added warnings and precautions. This is particularly important when the recommended agent is a new and/or infrequently employed drug. Disclaimer: The statements, opinions and data contained in this publication are solely those of the individual authors and contributors and not of the publishers and the editor(s). The appearance of advertisements or/and product references in the publication is not a warranty, endorsement, or approval of the products or services advertised or of their effectiveness, quality or safety. The publisher and the editor(s) disclaim responsibility for any injury to persons or property resulting from any ideas, methods, instructions or products referred to in the content or advertisements.