Low vs high frequency deep brain stimulation for Parkinson's disease-associated motor symptoms: A systematic review and meta-analysis of randomized control trials Authors: Bilal Salman, Lea Aridi, Vivian Faour, Wissam Abou Ali, Bilal Al Halabi, Rima Haidar, Jihane Soueid, Feten Fekih-Romdhane, Souheil Hallit, Sami Azar, Frederic Harb Published: 2026 ABSTRACT AbstractPurposeDeep brain stimulation (DBS) is an established treatment for motor symptoms in Parkinson's Disease (PD), but the optimal stimulation frequency remains uncertain. Previous studies have suggested that low and high frequency DBS may produce different clinical effects. This study aimed to compare the effects of low versus high frequency DBS on motor outcomes in PD.MethodsA systematic review and meta-analysis of randomized controlled trials was conducted. PubMed, Ovid, CINAHL, Clinicaltrials.gov and the Cochrane Library were searched from 2010 to November 26, 2025. Studies comparing low and high frequency DBS in patients with PD were included. Motor outcomes were pooled using a random-effects model, and subgroup analyses were performed based on age and disease duration.ResultsA total of 15 randomized controlled trials involving 416 participants were included. Overall, no significant difference in motor outcomes was observed between low and high frequency DBS. Subgroup analysis based on age showed no significant effect modification. However, subgroup analysis suggested greater improvement with high frequency DBS in patients with shorter disease duration, whereas no significant difference was observed in those with longer disease duration. Sensitivity analyses demonstrated that these findings were robust. Additionally, tremor improvement favored the use of HFDBS.ConclusionHigh and low frequency DBS appear to provide comparable overall motor benefit in PD. Stimulation frequency may have differential effects depending on disease duration, although these findings should be interpreted cautiously. Further prospective trials are needed to determine optimal programming strategies across disease stages.RegistrationThis study was registered on PROSPERO (CRD420251238009).Keywords: Parkinson's disease, deep brain stimulation, stimulation frequency, UPDRS-III, meta-analysis, randomized controlled trials Introduction Parkinson's Disease (PD) is a progressive neurodegenerative movement disorder characterized by the loss of dopaminergic neurons in the substantia nigra and associated disruption of basal ganglia circuitry. 1 This disease leads to a heterogenous clinical presentation characterized by motor symptoms including tremor, rigidity, bradykinesia, and postural instability, as well as non-motor features. 2 Risk factors include age, sex, environmental exposure, and genetic susceptibility. PD prevalence has been on the rise and is expected to increase by 112% between 2021 and 2050, with a global prevalence of 25.2 million people. 3 The primary treatment for PD consists of pharmacologic dopaminergic therapies, most notably levodopa, which remains the most effective agent for motor symptom control. Additional medications, including dopamine agonists, monoamine oxidase-B inhibitors, catechol-O-methyltransferase inhibitors, and amantadine, are commonly used either as monotherapy or as adjuncts. However, when motor symptoms become increasingly refractory medications, Deep Brain Stimulation (DBS) can be used. 4 DBS employs electrical stimulation delivered via electrodes embedded into exact subcortical nuclei to modify defective motor circuits in PD. DBS improves motor functions and quality of life by altering pathological firing patterns within the basal ganglia-thalamo-cortical loops. High-frequency stimulation is the most used approach and has been proved to extensively adjust rigidity, bradykinesia, and tremor. These parameters vary based on DBS location. For instance, in Subthalamic nucleus (STN) and substantia nigra pars reticulata (SNr) stimulation, low frequency is usually defined as below 60 Hz whereas high frequency is above 130 Hz.5,6 Whereas in pedunculopontine nucleus area (PPNa), low frequency is defined between 10 and 25 Hz and higher frequency is typically above 60 Hz. 7 In order to evaluate the severity and progression of PD, clinicians most frequently employ the Movement Disorder Society Unified Parkinson's Disease Rating Scale part III (MDS-UPDRS-III), which is a scoring system with a total score of 132 points spread across 18 items that measure tremor, stiffness, bradykinesia, balance, and other aspects of PD pathology. 8 Although DBS is an established treatment for motor symptoms in PD, stimulation parameters remain an area of ongoing investigation. Clinical outcomes depend heavily on stimulation parameters including frequency, amplitude, and pulse width. In particular, the impact of stimulation frequency is not fully understood, and studies comparing low and high frequency DBS have reported inconsistent findings. Determining whether specific patient characteristics influence response to different stimulation frequencies is essential for guiding programming strategies and improving clinical outcomes. The latest meta-analysis comparing LFDBS and HFDBS included data up to July 2024 and included 8 studies 9 ; We strive to perform a larger and more updated systematic review and meta-analysis. The aims of this systematic review and meta-analysis were to compare the effects of low and high frequency DBS on motor symptoms in PD and to explore potential effect modifiers, including age and disease duration. Methodology This Systematic Review and Meta-Analysis was conducted in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Eligibility criteria Inclusion criteria Studies were recruited based on the following PICOS framework” Population: Participants diagnosed with Parkinson's disease aged more than 18 years. Population: Participants diagnosed with Parkinson's disease aged more than 18 years. Interventions: Low Frequency DBS (LFDBS) * Frequency defined based on DBS target (< 60 Hz in STN DBS) Interventions: Low Frequency DBS (LFDBS) * Frequency defined based on DBS target (< 60 Hz in STN DBS) * Frequency defined based on DBS target (< 60 Hz in STN DBS) * Frequency defined based on DBS target (< 60 Hz in STN DBS) Comparators: High Frequency DBS (HFDBS) * Frequency defined based on DBS target (> 100 Hz in STN DBS) Comparators: High Frequency DBS (HFDBS) * Frequency defined based on DBS target (> 100 Hz in STN DBS) * Frequency defined based on DBS target (> 100 Hz in STN DBS) * Frequency defined based on DBS target (> 100 Hz in STN DBS) Outcomes: Primary Outcome of Interest is Motor Function assessed using UPDRS-III assed in the medication state (ON or OFF) as reported in the individual studies. Outcomes: Primary Outcome of Interest is Motor Function assessed using UPDRS-III assed in the medication state (ON or OFF) as reported in the individual studies. Study Design: Randomized Control Trials (RCTs) Study Design: Randomized Control Trials (RCTs) Exclusion criteria Studies were excluded based on the following criteria: Case reports, reviews, editorials, or animal studies. Case reports, reviews, editorials, or animal studies. Studies without extractable data or without motor outcomes. Studies without extractable data or without motor outcomes. Studies involving other movement disorders. Studies involving other movement disorders. Studies using non-surgical interventions. Studies using non-surgical interventions. Search strategy A comprehensive literature review was conducted using the following Databases: PubMed, Cochrane, CINAHL, OVID, and ClinicalTrials.gov from 1-1-2015 to 26-11-2025. The search strategy relied on the use of MeSH terms and keywords related to Parkinson's Disease, Deep Brain Stimulation, and motor symptoms. An example search string used for PubMed was: (“deep brain stimulation” OR “dbs” OR “neurostimulation” OR “brain modulation”) AND (“low frequency” OR “high frequency”) AND (“symptom relief” OR “motor function” OR “neurodegeneration”) AND (“Parkinson” OR “Parkinson's disease” OR “Parkinson disease” OR “Parkinsonism”) Screening procedure All studies were imported into Rayyan.ai and six independent reviewers screened the titles and abstracts. Potentially eligible studies were recruited and screened against the inclusion criteria. Any disagreements between two reviewers were resolved by discussion or a third reviewer. Data extraction Data were extracted using a standardized form. Extracted variables included: Study characteristics (author, year, country). Study characteristics (author, year, country). Participant characteristics (sample size, age, disease duration). Participant characteristics (sample size, age, disease duration). Intervention details (target, frequency). Intervention details (target, frequency). Outcome measures: UPDRS-III score. Outcome measures: UPDRS-III score. Data extraction was performed by two independent reviewers. Risk of bias assessment The quality of RCTs was assessed by two independent reviewers using the Cochrane Risk of Bias 2.0 tool. Each domain was rated as low, high, or some concerns. Disagreements were resolved by consensus or a third independent reviewer. Data synthesis and statistical analysis Meta-analysis was performed using Review Manager 5. We calculated mean differences (MD) with 95% confidence intervals (CI). A random-effects model was used for all meta-analyses to account for expected clinical and methodological heterogeneity across studies. Heterogeneity was further quantified using the I2 statistic, with thresholds of 25%, 50%, and 75% representing low, moderate, and high heterogeneity, respectively. Subgroup analyses were prespecified in the protocol to be conducted based on DBS target and follow-up duration. However, due to limited and inconsistent reporting of these variables across studies, subgroup analyses were performed based on age and disease duration, which were more consistently reported. Publication bias was assessed using funnel plots and Egger's test when ≥10 studies were included. Assessment of certainty in the body of evidence The certainty of evidence for each outcome was assessed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach. Evidence was evaluated across the domains of risk of bias, inconsistency, indirectness, imprecision, and publication bias. The overall certainty of evidence was categorized as high, moderate, low, or very low. Results Study selection A total of 446 articles were identified from PubMed, Cochrane, CINAHL, OVID, and ClinicalTrials.gov. 46 were flagged as duplicates by rayyan.ai, 39 of which were eliminated after manually checking the flagged articles. Of the remaining 407 articles, 384 were screened out and 23 were sought for retrieval. 4 articles were then removed for utilizing the wrong study design and for being an abstract only and an additional 3 were excluded from the meta-analysis and were only included narratively due to the raw data not being available or provided after contacting the authors. A total of 19 RCTs met the inclusion criteria. Sixteen studies provided sufficient data for quantitative synthesis and were included in the meta-analysis, while three studies were included in the qualitative synthesis only due to incomplete or inconsistently reported data (Figure 1). The screening was performed by 6 independent reviewers and was finalized with 93% alignment. The 29 conflicts were resolved by discussion with a third author. PRISMA flowchart. *19 total articles were included, 4 of which were only included qualitatively and discussed narratively, and the remaining 15 were used quantitatively. Study characteristics The included articles comprised 19 cross-over RCTs, of which 16 targeted the STN, 2 targeted the SNr, and 1 targeted the PPNa (Table 1). Of those 19, 15 crossover RCTs with a total of 210 unique participants were included in the statistical analysis with the other 4 being including narratively. Table 1.Study characteristics.Studies included in quantitative analysisStudy ID and methodsDBS targetAge (years) Mean (sd)*Disease Duration Mean (sd)LF nOutcome LF Mean (sd)HF nOutcome HF Mean (sd)Moreau, 2011 10 TEED held constant 5-year follow upSTN69*19.67 (4.44)11UPDRS-III26.67 (22.96)11UPDRS-III28 (18.52)Tsang, 2012 11 TEED held constant Short-term follow upSTN60 (6)15 (4)13UPDRS-III23.2 (2.46)13UPDRS-III13.7 (1.72)Stegemoller, 2013 12 TEED not reported 4–10 year follow upSTN61.53 (9.22)15.5 (1.8)17Motor UPDRS (items 18–31)24.5 (2)17Motor UPDRS (items 18–31)23.1 (14)Khoo, 2014 13 TEED not reported Short-term follow upSTN60.85 (9.28)16 (4.895)14UPDRS-III11.4 (7.2)14UPDRS-III16 (6.9)Nosko, 2015 7 TEED not held constant 1-year follow upPPNa64 (5.5)22.3 (6.4)9UPDRS-III23.9 (13.1)9UPDRS-III29.1 (13)Xie, 2015 14 TEED not reported Short-term follow upSTN64 (8)12.9 (4.9)6UPDRS-III15.43 (7.74)6UPDRS-III31.29 (4.79)Fagundes, 2016 15 TEED not reported Short-term follow upSTN56.65 (10.71)15.3 (4.71)20UPDRS-III34.33 (4.74)20UPDRS-III35.44 (4.3)Xie, 2018 16 TEED not held constant 1-year follow upSTN68.5 (5.9)14.2 (5.7)11UPDRS-III27.1 (9.1)11UPDRS-III32.4 (7.1)Conway, 2021 17 TEED held constant Short-term follow upSTN69.6 (7.5)12 (6.2)10MDS-UPDRS-III30.9 (9.8)14MDS-UPDRS-III32.7 (10.7)Qin, 2023 18 TEED not held constant Short-term follow upSTN62.7 (7.4)8.9 (3.2)29MDS-UPDRS-III26.48 (13.08)29MDS-UPDRS-III17.62 (9.34)Wang, 2023 19 TEED not held constant 6-month follow upSTN60 (9)10 (5)20MDS-UPDRS-III39 (15)20MDS-UPDRS-III28 (12)Artusi, 2024 20 TEED held constant Short-term follow upSNr66 (5.8)16.8 (6.3)15MDS-UPDRS-III18.6 (11.6)15MDS-UPDRS-III23.2 (13.5)Herz, 2025 21 TEED not held constant Short-term follow upSTN61.1 (4.3)10.05 (4.13)11MDS-UPDRS-III32.88 (21.35)11MDS-UPDRS-III21.77 (13.11)Imbalzano, 2025 22 TEED not reported Short-term follow upSTN56.7 (8)14.1 (3.7)12MDS-UPDRS-III23 (8.54)12MDS-UPDRS-III23.5 (12.58)Weiss, 2025 23 TEED held constant Short-term follow upSNr69.4 (6.6)20.2 (6.7)11MDS-UPDRS-III42.5 (15.86)11MDS-UPDRS-III39.9 (9.86)Studies included only qualitativelyStudy IDTargetReported OutcomeReason not pooled in the statistical analysisPhibbs, 2014 24 STNChange in UPDRS-IIIscoreUPDRS-IIIraw data not available, attempts to contact authors were unsuccessfulVarriale, 2018 25 STNUPDRS-IIIscoreUPDRS-IIIpresented in graph, raw data not available, attempts to contact authors were unsuccessfulHorn, 2020 26 STNUPDRS-IIIscoreUtilized high vs low burst DBS, all other studies employed continuous stimulationRicciardi, 2025 27 STNVerbal fluency, Composite motor score as secondary outcomeUPDRS-IIIraw data not available, attempts to contact authors were unsuccessful*Result reported only as median.DBS: Deep Brain Stimulation, LF: Low Frequency, HF: High Frequency, TEED: Total Electrical Energy Delivered, UPDRS: Unified Parkinson's Disease Rating Scale, MDS-UPDRS: Movement Disorder Society-UPDRS, STN: Subthalamic Nucleus, PPNa: Pedunculopontine Nucleus Area, SNr: Substantia Nigra. Study characteristics. *Result reported only as median. DBS: Deep Brain Stimulation, LF: Low Frequency, HF: High Frequency, TEED: Total Electrical Energy Delivered, UPDRS: Unified Parkinson's Disease Rating Scale, MDS-UPDRS: Movement Disorder Society-UPDRS, STN: Subthalamic Nucleus, PPNa: Pedunculopontine Nucleus Area, SNr: Substantia Nigra. Risk of bias The RoB assessment was performed by two independent reviewers with any discrepancy being resolved by discussion with a third reviewer. Most studies had low to moderate bias except Conway 2021, which had 3 out of 32 participants withdraw due to their inability to tolerate the treatment and Herz 2025, which had unblinded assessors (Figure 2). Risk of bias assessment summary. Motor symptoms in low vs high frequency DBS: primary meta-analysis Across 15 studies comparing LFDBS and HFDBS (Figure 3), the pooled random-effects meta-analysis demonstrated no statistically significant difference in the continuous outcome of interest between stimulation frequencies (standardized mean difference 0.13, 95% CI −0.32 to 0.58; Z = 0.56; P = 0.58), with individual study estimates being largely imprecise and crossing the line of no effect. Despite the inclusion of 210 participants in 15 crossover RCTs (206 LFDBS, 210 HFDBS), there was substantial between-study heterogeneity (Tau2 = 0.60; Chi2 = 65.65, df = 14; P < 0.00001; I2 = 79%), indicating considerable inconsistency in effect sizes that justifies the use of a random-effects model and warrants cautious interpretation of the summary estimate and exploration of potential clinical and methodological sources of heterogeneity. Sensitivity analysis excluding the studies targeting PPNa and SNr showed the same results (Supplementary Figure 1). Forest plot for motor symptoms in low vs high frequency DBS. Subgroup analyses Motor symptoms in low vs high frequency DBS based on age In the age-stratified subgroup analysis, neither patients aged ≥65 years (standardized mean difference −0.28, 95% CI −0.65 to 0.08; Z = 1.51; P = 0.13) nor those aged <65 years (standardized mean difference 0.35, 95% CI −0.29 to 0.99; Z = 1.06; P = 0.29) showed a statistically significant difference between LFDBS and HFDBS. Heterogeneity was negligible in the ≥65-year subgroup (Tau2 = 0.00; Chi2 = 1.04, df = 4; P = 0.90; I2 = 0%) but remained considerable among studies enrolling patients <65 years (Tau2 = 0.85; Chi2 = 56.99, df = 9; P < 0.00001; I2 = 84%), indicating marked inconsistency in effect estimates in the younger cohort. Similarly, the test for subgroup differences was not statistically significant (Chi2 = 2.79, df = 1; P = 0.09; I2 = 64.2%), suggesting no clear evidence of age-related interaction in the comparative effectiveness of LFDBS versus HFDBS. However, given the limited number of studies and the borderline p values, further research is warranted to clarify whether age may influence treatment response. Data is summarized in (Figure 4). Sensitivity analysis excluding the studies targeting PPNa and SNr showed the same results (Supplementary Figure 2). Forest plot for low vs high frequency DBS motor symptoms subgroup analysis based on age. Motor symptoms in low vs high frequency DBS based on disease duration In the duration-stratified analysis (Figure 5), LFDBS and HFDBS did not differ significantly among patients with disease duration >10 years (standardized mean difference −0.03, 95% CI −0.57 to 0.50; Z = 0.13; P = 0.90), whereas in patients with disease duration <10 years LFDBS was associated with a markedly better outcome (standardized mean difference 0.75, 95% CI 0.37 to 1.13; Z = 3.89; P = 0.0001), indicating a large and statistically robust effect in the shorter-duration subgroup. Between-study heterogeneity was considerable in the >10-year subgroup (Tau2 = 0.68; Chi2 = 51.78, df = 11; P < 0.00001; I2 = 79%) but negligible for duration <10 years (Tau2 = 0.00; Chi2 = 0.12, df = 2; P = 0.94; I2 = 0%), suggesting that the apparent benefit in earlier disease is consistent across studies, while results in long-standing disease remain highly variable. The highly significant test for subgroup differences (Chi2 = 5.52, df = 1; P = 0.02; I2 = 81.9%) supports a strong interaction between disease duration and stimulation frequency, although this finding should be interpreted as hypothesis-generating and validated in prospectively designed studies to mitigate the risks of spurious subgroup effects. Forest plot for low vs high frequency DBS motor symptoms subgroup analysis based on disease duration. A leave-one-out sensitivity analysis was performed and confirmed the robustness of the disease-duration subgroup analysis (Supplementary Figures 4, 5, and 6). Additionally, a sensitivity analysis excluding the studies targeting PPNa and SNr (all of which were included in disease duration > 10 years) showed the same results (Supplementary Figure 3). Four randomized studies were included in the qualitative synthesis but could not be incorporated into the quantitative meta-analysis. One study reported outcomes using non-comparable summary statistics that did not allow extraction of mean and standard deviation values. For two studies, sufficient raw data were not available and attempts to contact the authors were unsuccessful. The findings of these studies were generally consistent with the direction of effect observed in the pooled analysis; Two studies reported no significant in motor scores difference between LFDBS and HFDBS,24,27 one of which reported that LFDBS combined with HFDBS improved verbal fluency. 27 A third study reported that both LFDBS and HFDBS improved gait initiation during simple tasks but, under conditions of cognitive interference, HFDBS was associated with deterioration in gait initiation performance, an effect that was not observed with LFDBS. 25 The fourth employed Theta burst stimulation, in contrast to the 18 other articles which used continuous DBS; It found no significant difference in motor symptoms, assessed using UPDRS-III, MDS-UPDRS-III, and motor UPDRS-III (items 18–31), between HFDBS and LFDBS but found that they differentially affect the motor subscores, consistent with the findings of this analysis. Publication bias Visual inspection of the funnel plot shown in Figure 6 did not suggest the presence of significant publication bias. Funnel plot for the assessment of publication bias. Certainty in the body of evidence According to the GRADE framework, the certainty of evidence for the primary outcome (motor symptoms) was rated as moderate. The evidence was downgraded for inconsistency, due to substantial heterogeneity among studies (I2 = 79%), and for imprecision, related to relatively small sample sizes in several trials (overall n = 210). A summary of findings table is presented in Table 2. Table 2.GRADE assessments for primary outcome, UPDRS-III and secondary outcome, tremor subscores.Certainty assessment№ of patientsEffectCertaintyImportance№ of studiesStudy designRisk of biasInconsistencyIndirectnessImprecisionOther considerationsLow FrequencyHigh FrequencyRelative(95% CI)Absolute(95% CI) UPDRS-III 15randomized trialsnot seriousseriousanot seriousnot seriousnone206210-MD 0 (3.55 lower to 3.55 higher)⊕⊕⊕⊝ModerateCRITICAL Tremor Subscore (assessed with: Items 3.15–3.18 from UPDRS-III) 5randomized trialsnot seriousSeriousbnot seriousnot seriousnone6868-MD 0.84 (0.51 lower to 1.18 higher)⊕⊕⊕⊝ModerateCRITICALCI: Confidence Interval, MD: Mean Difference. a High heterogeneity (I2 = 79%). b High heterogeneity (I2 = 78%). GRADE assessments for primary outcome, UPDRS-III and secondary outcome, tremor subscores. CI: Confidence Interval, MD: Mean Difference. High heterogeneity (I2 = 79%). High heterogeneity (I2 = 78%). Secondary outcomes Tremor subscore Five studies reported tremor subscores and were included in the quantitative synthesis. Pooled analysis showed that HFDBS was better suited to reduce tremor scores than LFDBS (pooled mean difference 0.84, 95% CI 0.51 to 1.18, Z = 4.94, P < 0.00001. Heterogeneity was high (Chi2 = 18.07, df = 4; P = 0.001; I2 = 78%) justifying the use of a random-effects model. The findings are illustrated in Figure 7. Forest plot for tremor subscores in LFDBS vs HFDBS. According to the GRADE framework, the certainty of evidence for the secondary outcome (tremor subscore improvement) was rated as moderate. The evidence was downgraded for inconsistency, due to substantial heterogeneity among studies (I2 = 78%), and for imprecision, related to relatively small sample sizes in several trials (overall n = 136). A summary of findings table is presented in Table 2. Rigidity subscore Three studies reported rigidity subscores and were included in the meta-analysis. Pooled analysis showed no significant difference in rigidity scores between HFDBS and LFDBS (pooled mean difference −0.43, 95% CI −0.88 to 0.02, Z = 1.86, P = 0.06) although the point estimate favored LFDBS. Heterogeneity was negligible (Tau2 = 0.00, Chi2 = 1.4, df = 2; P = 0.5; I2 = 0%). The borderline p value and limited sample size suggests that the effect of DBS frequency on rigidity scores may warrant further research and consideration Figure 8. Forest plot for rigidity subscore in LFDBS vs HFDBS. Bradykinesia subscore Two studies reported bradykinesia subscores. The pooled analysis showed no significant difference in bradykinesia scores between HFDBS and LFDBS (pooled mean difference −1.82, 95% CI −4.63 to 0.99, Z = 1.27, P = 0.2) with high heterogeneity (Tau2 = 3.28, Chi2 = 4.53, df = 1, P = 0.03, I2 = 78%) justifying the use of a random-effects model. The findings can be found in Figure 9. Forest plot for bradykinesia subscore in LFDBS vs HFDBS. Gait subscore Two studies reported gait subscores and were included in the quantitative synthesis. Pooled analysis found no significant difference in gait scores between HFDBS and LFDBS (pooled mean difference −0.2, 95% CI −0.42 to 0.02, Z = 1.74, P = 0.08), though the point estimate narrowly favored LFDBS, which may warrant further investigation. The heterogeneity was high (Tau2 = 0.02, Chi2 = 5.29, df = 1, P = 0.02, I2 = 81%) justifying the use of a random-effects model. The results are summarized in Figure 10. Forest plot for gait subscore in LFDBS vs HFDBS. Postural stability Only one study reported postural stability subscores and was not included in a quantitative synthesis. The study found that LFDBS improved postural stability more effectively than HFDBS. LFDBS reported a lower postural stability score (mean = 1.28, sd = 0.29) than that reported for HFDBS (mean = 1.83, sd = 0.34). 15 A separate GRADE assessment was not performed for the remaining secondary outcomes due to the limited number of studies and small sample sizes. Discussion This meta-analysis found no significant difference in the improvement of motor symptoms between LFDBS vs HFDBS. Subsequent subgroup analyses also showed no significant difference between LFDBS and HFDBS based on age (<65 vs >65 years) or on disease duration greater than 10 years. However, among participants with disease duration less than 10 years, HFDBS showed a greater improvement in motor symptoms score compared to LFDBS. The absence of a significant difference between LFDBS and HFDBS may be due to several factors as DBS is a multifaceted intervention that doesn’t depend solely on frequency, but also on multiple factors such as amplitude, pulse width, and pulse duration; therefore, any change in these factors can compensate for the employment of the different frequencies, producing similar results with no significant difference. 28 Additionally, the way DBS exerts its effects is not clear-cut, as it does not just stimulate a specific nucleus, but rather entire circuits including the cortico-basal ganglia-thalamic circuit; low and high frequencies exert their effects on different parts of the circuit, which could explain the attenuation of the overall difference. 29 DBS outcomes also vary based on the target location variability. For instance, stimulation of the STN, PPNa, SNr, and various other locations can produce different therapeutic responses as they each require different parameters and modulate different circuits. Furthermore, DBS outcomes vary based on programming which varies based on numerous factors. DBS programming lacks standardized guidelines and varies between visits according to patient response, as well as across providers and clinical centers. 30 These factors introduce heterogeneity, which was found to be significant (I2 = 79%) in this study. This substantial heterogeneity may explain the absence of a significant statistical difference between LFDBS and HFDBS and hinders the ability to claim clinical equivalence. A noteworthy finding of this analysis is the greater improvement of motor symptoms scores achieved with HFDBS compared with LFDBS among participants with a shorter disease duration. This difference may reflect the progressive nature of PD. Participants with shorter disease durations are likely in earlier stages of neurodegeneration and therefore have relatively preserved neural networks and neuroplasticity. Additionally, patients in the earlier stages of PD likely still have strong compensatory changes such as an increase in functional connectivity within the default mode, sensorimotor, and executive control networks which tend to decline with disease progression.31,32 The greater improvement in HFDBS vs LFDBS may be explained by HFDBS's ability to desynchronize neural activity by producing clusters of activated and inhibited neurons, activate antidromic and orthodromic pathways which suppress beta oscillations, and modulate larger networks. In contrast, the lower frequencies used in LFDBS may fall within the beta oscillatory range, which can worsen motor symptoms. 33 Though HFDBS showed greater benefit for participants with a disease duration less than 10 years, the results should be interpreted with caution, as the subgroup had only 3 RCTs with a total of 58 participants. In contrast, the subgroup analysis based on age (more or less than 65 years) yielded no significant difference in motor symptom improvement between LFDBS and HFDBS. These findings suggest that age is a less reliable marker of PD progression or DBS outcome because it does not reflect the extent of neurodegeneration and disease progression like disease duration. Furthermore, participants enrolled in RCTs are carefully screened and selected which reduces the variability when it comes to comorbidities, potentially minimizing age-related differences in treatment responses. 34 Our analysis, in contrast, demonstrated greater improvement with HFDBS vs LFDBS in participants with shorter disease durations. The potential clinical implications of these findings should be interpreted cautiously. HFDBS has been associated in some studies with a higher incidence of stimulation-related adverse effects. 9 Therefore, in patients with longer disease durations, where the additional motor benefit of HFDBS appears limited, LFDBS may represent a reasonable strategy. Conversely, in patients with shorter disease durations, the greater improvement observed with HFDBS in this analysis suggests that the potential benefits of higher frequencies may outweigh the risks in selected individuals. Nevertheless, these considerations remain hypothesis-generating, and prospective trials specifically designed to evaluate frequency selection across disease stages are needed before firm clinical recommendations can be made. Analysis of secondary motor subscores suggested that the effects of stimulation frequency may differ across specific motor domains. Tremor outcomes, reported in five randomized trials, favored HFDBS, with moderate-certainty evidence, supporting the well-established efficacy of higher frequencies in tremor suppression. In contrast, pooled analyses of rigidity, gait, and bradykinesia, each based on two studies, did not demonstrate statistically significant differences between high- and low-frequency stimulation. These findings may reflect a true lack of differential effect, but should be interpreted cautiously given the small number of studies, limited statistical power, and borderline p values in the reported rigidity and gait subscores. Postural stability was reported in only one study and appeared to favor low-frequency stimulation; however, single-study evidence is insufficient to draw firm conclusions. Overall, these results suggest that stimulation frequency may differentially affect motor domains, with stronger evidence for tremor control and more limited or inconclusive evidence for axial and other motor features. Further randomized studies specifically designed to evaluate domain-specific motor outcomes are needed to clarify these effects. Similar to this analysis, previous meta-analyses comparing LFDBS and HFDBS in PD have reported no significant overall difference in motor outcomes. 35 Several of these analyses noted that stimulation frequency may influence specific motor domains differently, with high frequency stimulation often providing greater improvement in tremor and rigidity scores, while lower frequencies may have more favorable effects on akinesia, gait, FOG, swallowing, and verbal fluency scores.9,35 The findings of this meta-analysis may have practical implications for clinicians involved in DBS programming and follow-up of PD patients. While overall motor outcomes did not differ significantly between LFDBS and HFDBS, subgroup analyses and secondary outcomes suggest that stimulation frequency may influence specific motor domains and patient subgroups differently. HFDBS demonstrated clearer benefits for tremor control, whereas LFDBS may be considered in selected patients, particularly those with prominent axial symptoms or longer disease duration, where comparable motor outcomes and potential tolerability advantages may be relevant. These findings support an individualized approach to DBS programming, in which stimulation parameters are tailored to the patient's symptom profile, disease stage, and treatment goals. Strengths and limitations This study has several strengths. It included RCTs only, reducing the risk of selection bias and confounding and a systematic methodology following established guidelines was used for study selection, data extraction, and statistical analysis. The study also explored clinically relevant factors that may influence response to DBS, including age and disease duration, providing insights into potential effect modifiers. Several limitations should also be considered when interpreting these results. First, standardized mean difference was used to assess the change in motor symptoms between the groups as some studies employed UPDRS-III, while others employed MDS-UPDRS-III, and one used motor UPDRS which only assessed items 18–31. And although subgroup analyses were pre-specified in the protocol, the analyses could not be performed due to inconsistent reporting of relevant variables, and alternative clinically relevant subgroups were examined instead. Some subgroup analyses, particularly those based on disease duration, were derived from a small number of trials and participants, which may limit statistical power and generalizability. In addition, analyses were performed using study-level rather than individual patient-level data, preventing adjustment for patient-specific variables. Moreover, variation in DBS targets, stimulation parameters, and programming protocols across studies may have introduced clinical heterogeneity. In addition, some studies failed to adjust or report the Total Electrical Energy Delivered (TEED) after setting the frequency which introduces a major source of heterogeneity. Another important parameter-based limitation of this analysis is the cut-off of low and high frequencies, defined as <60 Hz and >100 Hz, respectively, which is not universally adopted in clinical practice, where intermediate ranges (60–80 Hz) are considered low: This variability in threshold selection may have led to misclassification and contributed to clinical heterogeneity across studies. Additionally, one study could not be included in the quantitative synthesis due to differences in outcome reporting and was incorporated qualitatively and two other studies that did not report raw UPDRS-III data were excluded from the meta-analysis after the authors failed to respond and provide the data. Furthermore, follow-up durations varied among studies, which may influence comparability of outcomes. Some eligible studies could not be included in the quantitative synthesis due to incomplete or inconsistently reported data, which may limit the precision of pooled estimates. Medication state (ON or OFF) may have differed between the individual studies and could have contributed to the heterogeneity. Conclusion and future perspectives This meta-analysis found no significant overall difference between HFDBS and LFDBS in improving motor outcomes in PD. However, subgroup analysis suggested that stimulation frequency may have differential effects depending on disease duration, with HFDBS showing greater benefit in patients with shorter disease duration. These findings raise the possibility that stimulation parameters may warrant consideration in relation to disease stage, while lower frequencies may remain a reasonable option in more advanced disease where additional benefit from higher frequencies appears limited. Additionally, secondary outcome analysis suggests that LFDBS and HFDBS affect the individual motor aspects differently, which warrants further research and DBS personalization based on the patient's phenotype. Prospective randomized trials specifically designed to compare stimulation frequencies across different stages of PD are needed to confirm these observations and guide clinical practice, as heterogeneity between studies remains high and limits the strength of these conclusions. Supplemental Material sj-docx-1-pkn-10.1177_1877718X261481663 - Supplemental material for Low vs high frequency deep brain stimulation for Parkinson's disease-associated motor symptoms: A systematic review and meta-analysis of randomized control trials Supplemental material, sj-docx-1-pkn-10.1177_1877718X261481663 for Low vs high frequency deep brain stimulation for Parkinson's disease-associated motor symptoms: A systematic review and meta-analysis of randomized control trials by Bilal Salman, Lea Aridi, Vivian Faour, Wissam Abou Ali, Bilal Al Halabi, Rima Haidar, Jihane Soueid, Feten Fekih-Romdhane, Souheil Hallit, Sami Azar and Frederic Harb in Journal of Parkinson's Disease sj-docx-2-pkn-10.1177_1877718X261481663 - Supplemental material for Low vs high frequency deep brain stimulation for Parkinson's disease-associated motor symptoms: A systematic review and meta-analysis of randomized control trials Supplemental material, sj-docx-2-pkn-10.1177_1877718X261481663 for Low vs high frequency deep brain stimulation for Parkinson's disease-associated motor symptoms: A systematic review and meta-analysis of randomized control trials by Bilal Salman, Lea Aridi, Vivian Faour, Wissam Abou Ali, Bilal Al Halabi, Rima Haidar, Jihane Soueid, Feten Fekih-Romdhane, Souheil Hallit, Sami Azar and Frederic Harb in Journal of Parkinson's Disease Acknowledgements The authors would like to acknowledge the Faculty of Medicine and Medical Sciences of the University of Balamand. Footnotes ORCID iD: Frederic Harb https://orcid.org/0000-0001-5882-1619 Author contributions: Conceptualization, B.S.; Methodology, B.S., F.H.; Literature Recruitment and Screening, B.S., L.A., V.F., W.A.A., B.A.H., R.H.; Data Analysis-Statistical Analysis, B.S.; Data Analysis-Data Interpretation, B.S., L.A.; Writing-Original Draft, B.S., L.A., V.F., W.A.A., B.A.H., R.H.; Writing-Review and Editing, B.S., L.A., V.F., W.A.A., B.A.H., R.H., J.S., F.F.R., S.H., S.A., F.H.; Supervision, F.H. All authors have read and agreed to the final version of the manuscript. Funding: The authors received no financial support for the research, authorship, and/or publication of this article. The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Data availability: All data are available in supporting information Supporting information: PRISMA checklist Leave-one-out sensitivity analysis Supplemental material: Supplemental material for this article is available online. References Associated Data Supplementary Materials sj-docx-1-pkn-10.1177_1877718X261481663 - Supplemental material for Low vs high frequency deep brain stimulation for Parkinson's disease-associated motor symptoms: A systematic review and meta-analysis of randomized control trials Supplemental material, sj-docx-1-pkn-10.1177_1877718X261481663 for Low vs high frequency deep brain stimulation for Parkinson's disease-associated motor symptoms: A systematic review and meta-analysis of randomized control trials by Bilal Salman, Lea Aridi, Vivian Faour, Wissam Abou Ali, Bilal Al Halabi, Rima Haidar, Jihane Soueid, Feten Fekih-Romdhane, Souheil Hallit, Sami Azar and Frederic Harb in Journal of Parkinson's Disease sj-docx-2-pkn-10.1177_1877718X261481663 - Supplemental material for Low vs high frequency deep brain stimulation for Parkinson's disease-associated motor symptoms: A systematic review and meta-analysis of randomized control trials Supplemental material, sj-docx-2-pkn-10.1177_1877718X261481663 for Low vs high frequency deep brain stimulation for Parkinson's disease-associated motor symptoms: A systematic review and meta-analysis of randomized control trials by Bilal Salman, Lea Aridi, Vivian Faour, Wissam Abou Ali, Bilal Al Halabi, Rima Haidar, Jihane Soueid, Feten Fekih-Romdhane, Souheil Hallit, Sami Azar and Frederic Harb in Journal of Parkinson's Disease