Botulinum Toxin Efficacy in Upper Limb Tremor: A Systematic Review and Meta‐Analysis

Abstract

Background

Essential tremor (ET) and dystonic tremor syndrome (DTS) can be treated using botulinum toxin (BoNT) injections. Previous reviews lacked an assessment of the certainty of evidence and focused solely on clinician‐reported outcomes. Additionally, studies have demonstrated interindividual variability in BoNT efficacy.

Objective

The aim of the study was to assess the efficacy and safety of BoNT injections for ET and DTS of the upper limbs, and to identify factors associated with BoNT efficacy.

Methods

We systematically searched Pubmed, Embase, Cochrane Library, and Web of Science databases for studies on BoNT injections in ET and DTS of the upper limbs. The certainty of evidence was rated using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) approach. Outcomes from randomized controlled trials (RCTs) were pooled as standardized mean differences (SMDs).

Results

We identified 5 RCTs, 6 open‐label trials, and 6 retrospective cohort studies. Meta‐analysis of post‐intervention scores showed a moderate effect on patient‐reported change (SMD: 0.58 [95% confidence interval [CI]: 0.39, 0.78]), no effect on clinically rated tremor severity (SMD: 1.69 [95% CI: −3.80, 0.42]), and no effect on grip strength (SMD: −0.63 [95% CI: −1.37, 0.10]). In contrast, meta‐analysis using change‐from‐baseline scores showed an improvement of clinically rated tremor severity (SMD: −1.12 [95% CI: −1.70, −0.54]). Certainty of evidence ranged from low to very low. No clear associations between BoNT efficacy and tremor phenotypes or injection strategies were identified.

Conclusions

Patient‐tailored BoNT injections may be effective and safe for ET and DTS. More trials are needed to confirm efficacy and safety, identify which tremor phenotypes benefit the most, and optimize injection strategies.

Essential tremor (ET) and dystonic tremor syndrome (DTS) are common movement disorders that frequently impair quality of life. ¹ , ² ET is primarily characterized by bilateral upper limb action tremor without additional neurological signs, whereas DTS combines dystonia and tremor. ³ If patients with ET have additional symptoms of “unknown certainty,” for example, mild abnormal posturing without overt dystonia, they may be classified as ET plus. ³ DTS is commonly classified into two clinical subtypes: dystonic tremor (DT) and tremor associated with dystonia (TAWD), ³ although this terminology is under scrutiny. ⁴ DT is defined as tremor present in a body part also affected by dystonia, whereas TAWD is defined as tremor occurring in a non‐dystonic body part. ³ Although ET and DTS have distinct etiologies, ³ there might be overlapping mechanisms. ⁵ , ⁶

ET and DTS can be managed through pharmacotherapy, surgical interventions, and botulinum toxin (BoNT) injections. Pharmacotherapy may include trihexyphenidyl, propranolol, or primidone. ⁷ , ⁸ Unfortunately, these drugs lack efficacy in approximately 50% of patients, offer only partial tremor reduction, and have multiple side effects. ⁷ , ⁸ For tremor not responding to pharmacotherapy, surgical interventions, such as deep brain stimulation, focused ultrasound, and radiofrequency ablation, can be considered. ⁷ , ⁸ However, these interventions are invasive, and for many patients, the risks do not outweigh the disability caused by the tremor.

BoNT injections are a promising treatment for ET and DTS. BoNT inhibits the release of acetylcholine at the neuromuscular junction by cleaving SNARE proteins, causing temporary neuromuscular paralysis. ⁹ Electrophysiological and neuroimaging studies have demonstrated additional central effects of BoNT injections. ¹⁰ These central effects are likely caused by the blockade of intrafusal muscle fibers, which may alter the output from muscle spindles and ultimately result in neuroplastic changes in the central nervous system. ¹¹

Previous systematic reviews and meta‐analyses have reported the efficacy and safety of BoNT injections in upper limb tremor ¹² , ¹³ , ¹⁴ but lacked an assessment of certainty of evidence and focused solely on clinician‐reported outcomes. Here, we present a systematic review and meta‐analysis on the effects of BoNT injections for ET and DTS of the upper limbs, evaluated by both clinician‐ and patient‐reported outcomes, and supported by the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach. ¹⁵

Moreover, studies have demonstrated interindividual variability in BoNT efficacy: 50%–75% of patients experienced an improvement in their tremor after receiving one BoNT treatment. ¹⁶ , ¹⁷ , ¹⁸ Futile treatment can be burdensome, especially if side effects such as muscle weakness occur. It remains unclear which tremor phenotypes benefit most from BoNT injections and what injection strategies are optimal. Therefore, we also summarize the evidence on the associations between these factors and BoNT efficacy. With this review, we aim to contribute to the development of more personalized treatment in ET and DTS.

Methods

This systematic review and meta‐analysis aimed to (1) assess the efficacy and safety of BoNT injections for ET and DTS of the upper limbs, using both clinician‐ and patient‐reported outcomes, and applying the GRADE approach to rate the certainty of evidence; and (2) identify factors associated with BoNT efficacy in ET and DTS of the upper limbs. The systematic review protocol was preregistered in the International Prospective Register of Systematic Reviews (PROSPERO, CRD42023489533) ¹⁹ and adhered to the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) 2020 statement. ²⁰

Database Search

We conducted a systematic search of PubMed, Embase, Cochrane Library, and Web of Science from inception to March 2025. The search strategy was developed in consultation with a librarian of the Radboud University Medical Library and is provided in Table S1. The search was extended by a forward citation search on Web of Science and a backward citation search of the included articles.

Study Selection

After removing duplicates, two reviewers (I.V. and Y.P.) screened titles and abstracts of all unique records and subsequently reviewed the selected full texts independently using Rayyan software. ²¹ In cases of disagreement, consensus was reached by discussion. We included articles that met all of the following inclusion criteria:

1. Adults (≥ 18 years) with a clinical diagnosis of ET or DTS of the upper limbs.

2. ≥ 5 participants with either ET or DTS.

3. Participants received ≥1 BoNT type A treatment of the upper limbs.

4. Randomized controlled trial (RCT), open‐label trial, or retrospective cohort study.

Quality Assessment

One reviewer (I.V.) performed a risk of bias assessment of the included studies. The risk of bias in RCTs was assessed using the Cochrane risk‐of‐bias tool for randomized trials version 2 (RoB 2). ²² RoB 2 defines the overall risk of bias as low risk, some concerns, or high risk. The Cochrane Risk of Bias in Non‐randomized Studies–of Interventions (ROBINS‐I) was used to evaluate the risk of bias in open‐label trials and retrospective cohort studies. ²³ ROBINS‐I categorizes the risk of bias as low, moderate, serious, or critical. The judgments were discussed with a second reviewer (F.N.). We used the GRADE approach to assess the certainty of evidence from the included RCTs. ¹⁵ The certainty of evidence was rated as high, moderate, low, or very low by two reviewers jointly (I.V., F.N.).

Data Extraction and Analysis

One reviewer (I.V.) extracted data on study design, study population (diagnosis, sample size, age, sex, and disease duration), intervention (number and frequency of BoNT treatments, BoNT formula, doses, and injection strategy), and outcome measures (both post‐intervention and change‐from‐baseline scores). If multiple trial arms with different doses were reported, ²⁴ we selected the outcomes from the arm with the highest efficacy. We extracted baseline‐adjusted outcomes from one RCT to account for clinically relevant baseline differences. ¹⁸ Outcomes from graphs were extracted using PlotDigitizer. ²⁵ Missing data were requested from the corresponding authors with a reminder sent after 6 weeks. Associations between tremor phenotypes and injection strategy and BoNT efficacy were extracted from texts and data when available.

We conducted a meta‐analysis of the included RCTs in R (R Foundation, version 4.4.1). R scripts are available on: https://github.com/irisvisser/Botulinum-toxin-Efficacy-in-Upper-Limb-Tremor-A-Systematic-Review-and-Meta-analysis. We calculated pooled between‐group standardized mean differences (SMD) of both post‐intervention scores and change‐from‐baseline scores using a random effects model. Our primary analysis focused on post‐intervention scores because only one RCT reported standard deviations (SD) for change‐from‐baseline scores. ¹⁷ Missing post‐intervention SDs were substituted by baseline SDs as it can be assumed that the intervention does not change the variability of the outcome measure. ²⁶

We performed a secondary analysis of change‐from‐baseline scores to account for baseline differences between treatment and placebo groups. Missing SDs for change‐from‐baseline scores were imputed using the formula: ${\mathrm{SD}}_{\text{change}}=\sqrt{{{\mathrm{SD}}_{\text{baseline}}}^2+{{\mathrm{SD}}_{\text{final}}}^2-\left(2·\text{corr}·{\mathrm{SD}}_{\text{baseline}}·{\mathrm{SD}}_{\text{final}}\right)}$, where SD_baseline and SD_final represent the SDs of baseline and final outcomes, respectively, and corr is the correlation coefficient between baseline and final outcomes. ²⁶ We assumed a correlation coefficient of 0.5 due to expected variability in treatment response, given no trial data were available for precise estimation. We also conducted sensitivity analyses with a correlation coefficient of 0.3 and 0.7 to determine the extent to which the findings depended on the assumption of the correlation coefficient.

The random effects model applied Hedges correction to correct for small sample bias ²⁷ and Knapp‐Hartung adjustment to calculate the SMD confidence intervals (CI). ²⁸ SMD values of ±0.2 were interpreted as small, ±0.5 as moderate, and >0.8 as large. ²⁹ Heterogeneity among RCTs was quantified using the I ² and τ ² statistics. τ ² was estimated by the Restricted Maximum Likelihood estimator. ³⁰ Heterogeneity was interpreted as low (I ² < 25%), moderate (I ² = 25–50%), substantial (I ² = 50–75%), or high (I ² > 75%). ³¹ We did not create funnel plots to assess publication bias, as only 5 RCTs were identified.

Results

Figure 1 provides the PRISMA flow diagram, which summarizes the search and study selection. ²⁰ Our database search identified 451 unique records. After title‐abstract screening, we assessed 29 full‐text reports for eligibility, and 20 of these met the inclusion criteria. Citation searching yielded no additional records.

FIG. 1.

PRISMA flow diagram of the search and study selection. ²⁰ DTS, dystonic tremor syndrome; ET, essential tremor; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta‐Analyses.

Characteristics of the Included Studies

Table S2 summarizes the characteristics of the 17 studies included, described in 20 reports. We identified 5 RCTs, ¹⁶ , ¹⁷ , ¹⁸ , ²⁴ , ³² 6 open‐label trials, ³³ , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ and 6 retrospective cohort studies. ¹ , ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ The majority of studies included people with ET (n = 15, 88%), ¹ , ¹⁶ , ¹⁷ , ²⁴ , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ⁴⁴ , ⁴⁵ , ⁴⁶ whereas only 5 (29%) studies included people with DTS. ¹ , ¹⁸ , ⁴³ , ⁴⁴ , ⁴⁵ Among the 619 people included, 58.0% were male (available data: n = 592), and the mean age was 65.7 years (available data: n = 599), and the mean disease duration was 24.0 years (available data: n = 426). All 5 RCTs evaluated the effect of a single BoNT injection session, ¹⁶ , ¹⁷ , ¹⁸ , ²⁴ , ³² whereas most open‐label trials and retrospective cohort studies (n = 8, 67%) studied multiple BoNT injection sessions. ¹ , ³⁵ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ The total BoNT dose in onabotulinum toxin A units was on average 92.0 U and ranged from 50 to 178.8 U (available data: n = 486). Muscle selection was based on a fixed‐dose approach, ²⁴ , ³² , ³³ a clinical assessment, ¹ , ¹⁸ , ⁴⁵ an electromyography (EMG) screening of upper extremity muscles for tremor activity, ³⁴ a kinematic method, ¹⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴⁶ or a combined approach. ¹⁶ , ³⁵ , ³⁶ , ³⁷ , ⁴² , ⁴³ , ⁴⁴ Some studies relied solely on anatomical landmarks for needle placement, ³² , ³⁴ whereas most used EMG guidance, ¹⁶ , ¹⁷ , ¹⁸ , ²⁴ , ³³ , ³⁵ , ³⁶ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ electrical nerve stimulation, ¹⁷ , ⁴³ and/or ultrasound guidance. ¹ , ¹⁷ , ³⁷ The studies used a variety of outcome measures to quantify clinically rated tremor severity, disability, clinician‐ and patient‐reported treatment effect, and quality of life. Follow‐up ranged from 4 to 16 weeks after a single injection and lasted up to 10 years following repeated injections.

Quality Assessment

We rated the overall risk of bias in the RCTs as ranging from “some concerns” (n = 4) to “high” (n = 1). The risk of bias arose from unblinding caused by side effects, the lack of prespecified analysis plans, and missing outcome data. The risk of bias in all open‐label trials and retrospective cohort studies was serious and resulted from inherent methodological limitations and missing outcome data. The detailed risk of bias assessments (Tables S3 and S4), weighted bar plots (Figs S1 and S2), and traffic light plots (Figs S3 and S4) are provided in the Data S1. We did not conduct a meta‐analysis of open‐label trials and retrospective cohort studies due to the variation in the number of treatment sessions and the serious risk of bias. We graded the certainty of evidence as very low for clinically rated tremor severity, low for patient‐reported change, and very low for grip strength (Table 1). The level of evidence was downgraded because of the risk of bias, inconsistency, imprecision, and potential publication bias.

TABLE 1.

Summary of findings for the effects of botulinum toxin injections on clinically rated tremor severity, patient‐reported change, and grip strength in ET and DTS

Population: people with ET or DTS
Settings: outpatient clinics
Intervention: botulinum toxin injections
Comparison: placebo
Outcomes	Number of participants (studies)	SMD (95% CI)	Certainty of evidence (GRADE)
Clinically rated tremor severity	140 (3 RCTs)	Post‐intervention: −1.69 [−3.80, 0.42] Change‐from‐baseline: −1.12 [−1.70, −0.54]	⊕⊝⊝⊝ Very low Due to risk of bias, a inconsistency, b imprecision, c and publication bias d
Patient‐reported change	168 (4 RCTs)	0.58 [0.39, 0.78]	⊕ ⊕ ⊝⊝ Low Due to risk of bias a and publication bias d
Grip strength	144 (3 RCTs)	Post‐intervention: −0.63 [−1.37, 0.10] Change‐from‐baseline: −0.63 [−1.37, 0.10]	⊕⊝⊝⊝ Very low Due to risk of bias, a imprecision, c and publication bias d

Abbreviations: ET, essential tremor; DTS, dystonic tremor syndrome; SMD, standardized mean difference; CI, confidence intervals; GRADE, Grading of Recommendations Assessment, Development, and Evaluation; RCTs, randomized controlled trials.

^a ‘Some concerns’ risk of bias due to possible unblinding by side effects and lack of prespecified analysis plans.

^b Inconsistency: substantial heterogeneity of I ² = 82%.

^c Imprecision: wide 95% confidence interval.

^d Possible publication bias: 75% of RCTs were funded by pharmaceutical companies.

Meta‐Analysis

Five RCTs compared the effects of BoNT injections to placebo, ¹⁶ , ¹⁷ , ¹⁸ , ²⁴ , ³² although post‐intervention outcomes were only partially available. The pooled between‐group SMD for the effect of BoNT injections on clinically rated tremor severity was statistically insignificant (−1.69 [95% confidence interval (CI): −3.80, 0.42]) (Fig. 2A). Heterogeneity was high (I ² = 82%). The pooled between‐group SMD for patient‐reported changes was 0.58 [95% CI: 0.39, 0.78], suggesting a moderate effect (Fig. 2B). I ² was 0%, likely due to low statistical power. The pooled SMD for grip strength was −0.63 [95% CI: −1.37, 0.10], crossing 0, indicating no statistically significant difference between BoNT and placebo (Fig. 2C). I² was 0%, again likely due to low statistical power.

FIG. 2.

Forest plots showing the between‐group standardized mean differences in post‐intervention clinically rated tremor severity (A), patient‐reported change (B), and post‐intervention grip strength (C). CI, confidence interval; DT, dystonic tremor; ET, essential tremor; N, number of participants; SD, standard deviation; SMD, standardized mean difference.

Our secondary analysis on change‐from‐baseline scores revealed a large effect on clinically rated tremor severity (SMD: −1.12 [95% CI: −1.70, −0.54], I ² = 14%) (Fig. 3A) and no significant effect on grip strength (SMD: −0.63 [95% CI: −1.37, 0.10], I ² = 0%) (Fig. 3B). Sensitivity analyses estimating the effect sizes using a low (0.3) and high correlation (0.7) between baseline and follow‐up outcomes confirmed a large effect on clinically rated tremor severity and no significant effect on grip strength (Figs S5 and S6).

FIG. 3.

Forest plots showing the between‐group standardized mean differences in clinically rated tremor severity changes (A) and grip strength changes (B). CI, confidence interval; DT, dystonic tremor; ET, essential tremor; N, number of participants; SD, standard deviation; SMD, standardized mean difference.

Factors Associated with Botulinum Toxin Efficacy

No conclusive evidence was found for associations between BoNT efficacy and certain tremor phenotypes, including tremor syndrome, activation condition, and body distribution, or injection strategies, including muscle selection, guidance techniques, and unilateral versus bilateral injections (Table 2).

TABLE 2.

Evidence on the associations between BoNT doses and tremor phenotypes and injection strategies

Factor	Evidence
Tremor syndrome	No differences between ET and DTS 45
Activation condition	No differences between postural and kinetic tremor 18 , 24 , 32 , 38 , 39 , 40
Body distribution	No consensus whether proximal 34 , 41 , 44 or distal tremors 1 , 35 , 36 , 38 benefit more from BoNT injections
Muscle selection technique	Trials needed to compare outcomes between clinical, EMG‐based, and kinematic methods for muscle selection
Guidance technique	Trials needed to compare outcomes between guidance techniques based on anatomy, EMG, electrical nerve stimulation, and muscle ultrasound
Unilateral vs. bilateral	Trials needed to compare unilateral and bilateral BoNT injections 41 , 53

Abbreviations: BoNT, botulinum toxin; ET, essential tremor; DTS, dystonic tremor syndrome; EMG, electromyography.

Conclusion

Our meta‐analysis on the efficacy and safety of BoNT injections in ET and DTS of the upper limbs shows a moderate improvement in patient‐reported outcomes, even after accounting for placebo effects. These patient‐reported outcomes are especially valuable in guiding treatment decisions, as they reflect the real‐life impact of treatment and are less influenced by natural tremor fluctuations than clinician‐reported outcomes. However, we did not find a consistent improvement in clinically rated tremor severity: the primary analysis using post‐intervention scores showed no statistically significant effect, whereas the secondary analysis using change‐from‐baseline scores revealed a large improvement. We hypothesize that this discrepancy may be caused by baseline differences between intervention and placebo groups, which are accounted for in change‐from‐baseline analyses. This variability likely contributed to a wide CI that crossed zero, indicating uncertainty about the true treatment effect. Still, all RCTs reported large individual improvements in clinically rated tremor severity, suggesting a possible positive effect not captured by the overall primary analysis using post‐intervention scores. Finally, grip strength was not significantly affected by BoNT injections, indicating that BoNT injections are generally safe in terms of muscle weakness.

Three RCTs also used accelerometry as an objective measure of tremor severity, ¹⁷ , ¹⁸ , ³² but variance estimates were only available for 2. ¹⁷ , ¹⁸ Therefore, accelerometry could not be included as an outcome in the meta‐analysis, especially because one RCT reported change‐from‐baseline scores ¹⁷ and another reported post‐intervention scores as medians with ranges. ¹⁸ The individual SMD of the RCT reporting change‐from‐baseline scores ¹⁷ was estimated at −1.45, indicating a large effect of BoNT injections on tremor severity. Nevertheless, additional trials are needed to confirm this finding. We strongly recommend the use of motion sensors, as it provides more precise and reliable measurements of tremor severity than clinical rating scales. Prolonged remote monitoring using wearable sensors might even offer a more comprehensive assessment, as it is less influenced by natural tremor fluctuations. ⁴⁷

The certainty of evidence for BoNT injections in ET and DTS of the upper limbs was graded low to very low because of the risk of bias, heterogeneity, imprecision, and publication bias. The risk of bias mainly arose because muscle weakness may have compromised blinding. Although grip strength was not significantly affected, individual patients may have experienced muscle weakness, or weakness may have occurred in muscles other than the finger flexors. This will cause unblinding, which is inherent to all studies on the effects of BoNT injections. Heterogeneity and imprecision might be caused by differences in injection methods between the fixed‐dose and personalized‐dose trials. Publication bias may be present, as 3 of 4 RCTs were funded by pharmaceutical companies. These findings underscore the need for more preregistered, high‐quality RCTs to establish whether BoNT injections are truly effective in treating ET and DTS.

We found no conclusive evidence that certain tremor phenotypes were associated with BoNT efficacy. First, only one large retrospective cohort study of 84 participants directly compared BoNT efficacy between ET and DTS ⁴⁵ but found no differences between these tremor syndromes. One may hypothesize that BoNT is more effective in DTS than ET, as BoNT might normalize abnormal sensory processing in dystonia via the afferent fibers of muscle spindles besides direct peripheral muscle paralysis, similar to a sensory trick. ¹¹ Possibly, larger sample sizes are necessary to detect such effects. Second, we found similar reductions in postural and kinetic tremor severity from 5 trials that reported these separately. ¹⁸ , ²⁴ , ³² , ³⁸ , ³⁹ , ⁴⁰ Third, we observed conflicting findings on the relationship between BoNT efficacy and body distribution. Some authors highlighted the necessity of injecting proximal muscles, ³⁴ , ⁴¹ , ⁴⁴ whereas others observed distal tremors to respond better to BoNT injections. ¹ , ³⁵ , ³⁶ , ³⁸ Notably, no trial reported outcomes for these distributions separately, limiting any conclusions.

Similar to tremor phenotypes, we observed no conclusive evidence that certain injection methods improve BoNT efficacy, including muscle selection methods, guidance techniques, and unilateral versus bilateral injections. Participants of the earliest trials received fixed doses to the forearm flexors and extensors. ²⁴ , ³² , ³³ These trials found reductions in tremor severity, but also a high prevalence of disabling muscle weakness. More recent studies shifted toward patient‐tailored muscle selection to minimize muscle weakness. These strategies were based on a clinical assessment, ¹ , ¹⁸ , ⁴⁵ often combined with EMG screening, ¹⁶ , ³⁵ , ³⁶ , ⁴² , ⁴³ , ⁴⁴ or a kinematic method, ¹⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴⁶ but no trial has yet directly compared these approaches. Additionally, several guidance techniques were used across studies, including anatomical landmarks, ³² , ³⁴ , ³⁵ EMG guidance, ¹⁶ , ¹⁸ , ²⁴ , ³³ , ³⁶ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ⁴⁴ , ⁴⁵ , ⁴⁶ electrical nerve stimulation, ¹⁷ , ⁴³ and/or ultrasound guidance. ¹ , ¹⁷ , ³⁷ However, no trial has compared these guidance techniques for treating upper limb tremor. RCTs comparing these approaches in focal hand dystonia and post‐stroke spasticity showed greater clinical improvement when using EMG, electrical nerve stimulation, or ultrasound compared to anatomical landmarks. ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² No differences in efficacy were observed between electrical nerve stimulation and ultrasound, although patients reported discomfort with electrical nerve stimulation. ⁴⁹ , ⁵¹ , ⁵² We hypothesize that ultrasound guidance, in particular, improves injection accuracy in upper limb tremor and thereby also minimizes the risk of injecting into neighboring muscles that may be sensitive to muscle weakness. In addition, ultrasound is more patient‐friendly and helps prevent injections into blood vessels and nerves. Bilateral injections might not further improve the quality of life compared to unilateral injections in ET, ⁴¹ , ⁵³ but this should be explored in larger cohort studies. We speculate that this phenomenon can be explained by two aspects: (1) disability may already improve by treating the dominant upper limb, and (2) tremor severity of the contralateral upper limb is, though to a lesser extent, also reduced by the central effects of BoNT. ¹¹

Strengths and Limitations

The main strengths of this review are (1) the adherence to the PRIMSA guidelines, including a certainty of evidence assessment, (2) the evaluation of both clinician‐ and patient‐reported outcomes, (3) the focus on solely ET and DTS instead of upper limb tremor in general to decrease heterogeneity, and (4) the careful distinction between dissimilar outcomes in our meta‐analysis, ensuring interpretability of our findings.

We acknowledge several limitations. First, our primary analysis used post‐intervention values as SDs for change‐from‐baseline were poorly reported. Post‐intervention values can be biased by baseline differences between treatment and placebo groups. Baseline characteristics were generally similar across groups, except for 1 RCT. ¹⁸ For this RCT, we extracted baseline‐adjusted outcomes to control for baseline differences. This approach could have biased our results. Therefore, we also performed a secondary analysis using change‐from‐baseline values with estimated SDs. This analysis should be interpreted with caution, as a change in the assumed correlation between baseline and follow‐up outcomes will affect both the SMD and its CIs. Nevertheless, we believed choosing a moderate correlation of 0.5 to be reasonable. Importantly, the sensitivity analyses indicated that adjusting the correlation to 0.3 or 0.7 did not affect the statistical significance of the effect. Second, the RCTs studied only the effects of 1 treatment session, whereas an optimal treatment effect is reached after 3 sessions. ⁵⁴ Thus, our findings are likely an underestimation of the true potential treatment effect on tremor. Third, ordinal outcomes were treated as continuous outcomes, as only summary statistics were available. Fourth, the included studies applied evolving tremor classification systems. ³ , ⁵⁵ , ⁵⁶ Consequently, patients with upper limb action tremor and additional neurological signs may have been diagnosed with ET in earlier studies, whereas they might be classified as ET plus or even DTS according to the current criteria. ³ Additionally, the terminology used for dystonia and tremor remains an ongoing debate among movement disorders specialists. ⁴ Some define DT as a tremor superimposed on dystonia, whereas others define it as dystonic jerks that mimic tremor. This limitation does not largely influence our findings, as we did not perform subgroup analyses based on tremor subtype due to the limited number of available RCTs. However, these changes in classification systems and inconsistent terminology could have contributed to increased heterogeneity across the included studies.

Future Directions

We provide several recommendations for future research. First, future trials may explore which tremor phenotypes benefit most from BoNT injections, using more fine‐grained clinical and electrophysiological subtyping. For example, tremor characteristics (eg, jerkiness), the co‐occurrence of dystonia in the tremulous limb (eg, DT vs. TAWD), the muscles that drive tremulous activity (eg, proximal vs. distal), and the effect of somatosensory input (presence or absence of a sensory trick) may all affect the efficacy of BoNT. Second, we recommend that future trials compare the effects of different muscle selection (eg, a clinical assessment, EMG screening, or kinematic method) and guidance techniques (eg, anatomical landmarks, EMG, or muscle ultrasound) when considering efficacy, safety, and healthcare costs. Third, trials may investigate the additional effects of bilateral injections on quality of life compared to unilateral injections. Fourth, we recommend that trials report the SDs of change scores to enhance meta‐analyses by accounting for interindividual differences.

Author Roles

(1) Research project: A. Conception, B. Organization, C. Execution; (2) Statistical analysis: A. Design, B. Execution, C. Review and critique; (3) Manuscript preparation: A. Writing of the first draft, B. Review and critique.

I.V.: 1A, 1B, 1C, 2A, 2B, 3A

Y.P.: 1C, 2C, 3B

B.W.: 2C, 3B

R.C.H.: 1A, 2A, 2C, 3B

A.H.S.: 1A, 2A, 2C, 3B

F.A.P.N.: 1A, 1C, 2A, 2C, 3B

Disclosures

Ethical Compliance Statement: Approval of an institutional review board was not required for this work. Informed patient consent was not required for this work. We confirm that we have read the journal's position on issues involved in ethical publication and affirm that this work is consistent with those guidelines.

Funding Sources and Conflicts of Interest: Financial support was provided by a grant from the Radboudumc‐CWZ Promotiefonds. The authors declare that there are no conflicts of interest relevant to this work.

Financial Disclosures for the previous 12 months: I.M.V, Y.P., A.H.S, and F.A.P.N. declare that there are no additional disclosures to report. B.W. received research grants from the Hersenstichting, ZonMw, Dutch Research Council, and Christina Foundation, and a speaker fee for a lecture for the Belgian Society of Neurology and a reviewer fee for Gossweiler. He received royalties from BSL/Springer Nature. He is also part of the advisory board of Vico Therapeutics and provided consultancies for Biogen, Vico Therapeutics, and Biohaven Pharmaceuticals. R.C.H received research grants from The Michael J. Fox Foundation, the Netherlands Brain Foundation, ParkinsonNL, and the Netherlands Organization for Health Research and Development, and funds for consultancy from Neurocrine Biosciences.

Supporting information

Figure S1. Weighted bar plot of the risk of bias in the randomized controlled trials. The bars display the weighted proportion of studies rated as low risk, some concern, or high risk for each domain.

Figure S2. Weighted bar plot of the risk of bias in open‐label trials and retrospective cohort studies. The bars display the weighted proportion of studies rated as low risk, some concern, or high risk for each domain.

Figure S3. Traffic light plot of the risk of bias in randomized controlled trials. Each marker represents the risk of bias judgment for a specific domain.

Figure S4. Traffic light plot of the risk of bias in open‐label trials and retrospective cohort studies. Each marker represents the risk of bias judgment for a specific domain.

Figure S5. Forest plots showing the between‐group standardized mean differences in clinically rated tremor severity changes estimated with correlations of 0.3 (A), 0.5 (B), and 0.7 (C). CI, confidence interval; DT, dystonic tremor; ET, essential tremor; N, number of participants; SD, standard deviation; SMD, standardized mean difference.

Figure S6. Forest plots showing the between‐group standardized mean differences in grip strength changes estimated with correlations of 0.3 (A), 0.5 (B), and 0.7 (C). CI, confidence interval; DT, dystonic tremor; ET, essential tremor; N, number of participants; SD, standard deviation; SMD, standardized mean difference.

Table S1. Search strategy.

Table S2. Characteristics of the included studies.

Table S3. Risk of bias assessment in randomized controlled trials.

Table S4. Risk of bias assessment in open‐label trials and retrospective cohort studies.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.