Effect of transcranial magnetic stimulation on perioperative neurocognitive dysfunction: study protocol for a systematic review and meta-analysis of randomised controlled trials

Abstract

Introduction

Perioperative neurocognitive disorders (PNDs), a common postoperative complication associated with anaesthesia and surgical procedures, are characterised by impairments in memory, attention, language comprehension and social functioning. Accumulating evidence from clinical studies indicates that transcranial magnetic stimulation (TMS)—a non-invasive neuromodulatory modality capable of targeted cortical stimulation—may offer therapeutic promise for PND management. To comprehensively assess the intervention efficacy and safety parameters of TMS in mitigating postoperative cognitive decline, we propose conducting a systematic review and meta-analysis of randomised controlled trials adhering to Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. Our findings aim to provide evidence-based insights into the neuroprotective potential of TMS for mitigating cognitive decline in surgical populations.

Methods and analysis

The investigation will implement a multifaceted search protocol encompassing international and Chinese scholarly resources. The search will be conducted in major databases, including Web of Science, Embase, PubMed, Cochrane Library, China Biology Medicine, China National Knowledge Infrastructure, Wan Fang Database and Chinese Scientific Journal Database from inception to 31 March 2025. To capture the latest research trends, ongoing trials will be simultaneously searched in the WHO International Clinical Trial Registry Platform, ClinicalTrials.gov and the China Clinical Trial Registry. Grey literature will be supplemented through resources such as GreyNet International, OpenGrey and Google Scholar. The inclusion criteria for this review are restricted to randomised controlled trials investigating the application of TMS as an intervention for PND. Primary endpoints comprise clinically confirmed incidence rates of postoperative delirium and delayed neurocognitive recovery. Two researchers will independently perform literature screening, data extraction and risk of bias assessment. The risk of bias in included studies will be evaluated using the Cochrane Risk of Bias Tool 2.0. Evidence certainty will be appraised through the GRADE framework with explicit justification for downgrading decisions. Meta-analysis will be conducted using STATA V.15.1 statistical software. The data synthesis process will incorporate standardised methodologies, including heterogeneity testing, sensitivity analysis and assessment of publication bias.

Ethics and dissemination

This study will not involve the collection of biometric information or medical privacy data throughout the research process, thus complying with the exemption criteria outlined in the ‘Measures for Ethical Review of Biomedical Research Involving Human Subjects.’ The findings will adhere to academic standards and be submitted for publication in reputable international medical journals following a rigorous double-blind peer-review process.

Systematic review registration

The research protocol has been prospectively registered on the PROSPERO international prospective systematic review registration platform (registration number: CRD42025636978).

Correspondence to Ms Jiarong Zeng; [email protected]

STRENGTHS AND LIMITATIONS OF THIS STUDY

• This investigation adheres to Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols and Cochrane Handbook guidelines, ensuring methodological transparency from protocol registration through literature screening to data synthesis.

• By exclusively incorporating randomised controlled trials and employing the Grading of Recommendations, Assessment, Development and Evaluation framework to critically appraise evidence quality, this analysis strengthens the clinical validity of derived conclusions.

• Combining I² and Q-test with fixed thresholds streamlines heterogeneity analysis, but arbitrary cut-offs may misrepresent variability, and random-effects models risk hiding subgroup differences.

• Predefined subgroups and sensitivity checks improve rigor, but uncorrected multiple comparisons and small subgroup sizes increase false positives and limit insights.

• Funnel plots and Egger’s test enhance bias detection, but small samples weaken power, and trim-and-fill oversimplifies complex publication bias patterns.

Introduction

Perioperative neurocognitive disorders (PNDs), encompassing postoperative delirium (POD), delayed neurocognitive recovery and postoperative cognitive impairment, represent a spectrum of transient or persistent cognitive deficits affecting memory, attention and executive function following surgical procedures.^1–3 Despite technological refinements in surgical protocols and anaesthetic modalities, the expanding geriatric surgical demographic (aged ≥65 years) has paradoxically unmasked significant clinical concerns regarding PND.^{4 5} Notably, cardiac and vascular surgeries demonstrate particularly high POD incidence rates (26–52%). ⁶PND significantly prolongs hospitalisation duration, substantially elevates healthcare expenditures and demonstrates robust associations with accelerated long-term cognitive deterioration and increased mortality risk.^7–9 Although the precise pathophysiological mechanisms remain incompletely elucidated, preclinical investigations have established the critical role of surgical trauma-induced neuroinflammatory responses, manifested through microglial hyperactivation and consequent overproduction of proinflammatory cytokines (eg, interleukin-1β), in driving hippocampal neuronal apoptosis pathways.^10–12 In addition, the resolution of inflammation and adverse anaesthetic effects may also be related to PND.¹³ Contemporary PND management prioritises multimodal non-pharmacological paradigms encompassing: (1) postoperative early mobilisation protocols, (2) preoperative cognitive prehabilitation regimens, (3) circadian entrainment optimisation through phototherapy and (4) sensory deficit mitigation strategies.¹⁴ Concomitantly enforcing stringent restrictions on perioperative administration of benzodiazepine derivatives and anticholinergic agents constitutes an essential pharmacological stewardship measure.¹⁵ These challenges underscore the urgent need for evidence-based perioperative strategies to mitigate cognitive decline, particularly given global demographic ageing trends.¹⁶

Emerging interest has focused on repetitive transcranial magnetic stimulation (rTMS), a non-invasive neuromodulation technique that modulates cortical excitability through pulsed magnetic fields.^{17 18} Preclinical evidence demonstrates that high-frequency rTMS (5–20 Hz) enhances hippocampal synaptic plasticity in rodent models, potentially mediated by the upregulation of brain-derived neurotrophic factor (BDNF) and tropomyosin receptor kinase B signalling pathways.¹⁹ Clinical studies further indicate that dorsolateral prefrontal cortex (DLPFC)-targeted rTMS improves cognitive performance in Alzheimer’s disease and major depressive disorder populations, with mechanistic associations to BDNF-dependent neuroplasticity.^{20 21} In addition, preclinical evidence has demonstrated that rTMS ameliorates spatial learning deficits and memory impairment in middle-aged murine models following caesarean section surgery, with 5 Hz high-frequency protocols showing particular efficacy.²² Clinically, perioperative rTMS application has been associated with reduced PND incidence in geriatric patients undergoing major abdominal procedures, an effect potentially mediated through upregulated BDNF signalling.²³ Nevertheless, existing research remains limited by small sample sizes, heterogeneous stimulation parameters (including frequency, intensity and duration) and insufficient mechanistic investigation. Critical knowledge gaps persist concerning optimal rTMS protocols, long-term safety profiles and pathway-specific neurobiological effects in perioperative contexts. A systematic evaluation of rTMS efficacy and underlying mechanisms in cognitive protection is therefore essential to establish standardised, evidence-based neuromodulation strategies for surgical populations.

Methods

This study will strictly follow the Cochrane Handbook for Preferred Reporting Items for Systematic Reviews and Meta-Analyses Protocol statement guidelines.^{24 25} If there is a need to modify the research design or deviate from the plan during the implementation process, we will follow the principles of transparency and standardisation to submit a modification request on the PROSPERO website.

Inclusion criteria

Types of studies

Eligible studies will include all randomised controlled trials (RCTs) using parallel-arm designs, irrespective of blinding methodology (open-label, single-blind or double-blind). Observational studies, case reports, narrative reviews and conference abstracts will be excluded.

Participant selection

Inclusion criteria: adult surgical patients (aged ≥18 years), encompassing both elective and emergency procedures.

Exclusion criteria: comorbid severe neurological disorders (dementia, Parkinson’s disease and stroke history), active psychiatric diagnoses (schizophrenia and bipolar disorder) and recent neuroactive medication use (within 4 weeks before surgery)

Types of interventions and comparisons

Intervention group: received rTMS treatment during the perioperative period (preoperative, intraoperative or postoperative), regardless of stimulation parameters (such as frequency, intensity, coil type and duration) and stimulation targets (DLPFC, posterior parietal cortex and posterior cingulate cortex).

Control group: received sham-stimulation (sham-rTMS), routine perioperative care or other non-rTMS interventions (such as drug therapy and cognitive training).

Studies that combine other neural regulation techniques (such as transcranial direct current stimulation) were excluded to avoid confounding effects.

Types of outcomes

Primary outcomes

• PND incidence: diagnosed using validated instruments such as Confusion Assessment Method (CAM), CAM-ICU, 3D-CAM, Nursing Delirium Screening Scale, DSM-5 or ICD-11 criteria.

• Cognitive performance: assessed via standardised neuropsychological batteries such as Mini-Mental State Examination, Montreal Cognitive Assessment and Wechsler Adult Intelligence Scale.

Secondary outcomes

• Serum inflammatory markers: Interleukin-6 (IL-6), (Tumor Necrosis Factor-alpha) TNF-α and C reactive protein.

• Neuroplasticity biomarkers: BDNF, S100β.

• Safety profile: incidence of device-related adverse events (seizures, headaches).

• Longitudinal cognitive outcomes: 3-month postoperative follow-up data.

Studies reporting nonspecific outcomes without validated cognitive assessments will be excluded.

Information sources

Electronic databases

This systematic review will employ a comprehensive search strategy across international medical databases (Web of Science, Embase, PubMed and Cochrane Library) and Chinese biomedical repositories (China Biology Medicine, China National Knowledge Infrastructure, Wan Fang Database and Chinese Scientific Journal Database). The search period will extend from database inception to 31 March 2025. To ensure comprehensive literature retrieval, we will perform backward citation tracking of initially identified studies. Furthermore, preprint repositories (bioRxiv, medRxiv) and non-English publications will be included to minimise publication bias and language-related limitations.

Other resources

A systematic search strategy will be implemented across international clinical trial registries, including the WHO International Clinical Trials Registry Platform, ClinicalTrials.gov and the Cochrane Central Register of Controlled Trials, as well as regional registries such as the China Clinical Trial Registry. This approach will include ongoing and completed trials across all research phases to address publication timeline bias. Specialised grey literature repositories, including GreyNet International and OpenGrey, will be used for targeted searches, supplemented by systematic searches using Google Scholar’s academic search engine to identify non-indexed scholarly outputs.

Search strategy

The search strategy is systematically formulated according to the PICOS framework, integrating both MeSH terms and free-text keywords corresponding to each conceptual domain. Taking PubMed as an example, the specific search strategy is shown in table 1.

Table 1

Search strategy for PubMed

Sequence	Items
#1	Transcranial Magnetic Stimulation (MeSH)
#2	Magnetic Stimulations, Transcranial
#3	Magnetic Stimulation, Transcranial
#4	Stimulations, Transcranial Magnetic
#5	Stimulation, Transcranial Magnetic
#6	Transcranial Magnetic Stimulations
#7	Transcranial Magnetic Stimulation, Paired Pulse
#8	Transcranial Magnetic Stimulation, Repetitive
#9	Transcranial Magnetic Stimulation, Single Pulse
#10	#1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9
#11	Perioperative Period (MeSH)
#12	Period, Perioperative
#13	Periods, Perioperative
#14	Perioperative Periods
#15	Perioperative
#16	#11 OR #12 OR #13 OR #14 OR #15
#17	Cognitive Dysfunction (MeSH)
#18	Dysfunction, Cognitive
#19	Cognitive Disorder
#20	Disorder, Cognitive
#21	Cognitive Impairment
#22	Impairment, Cognitive
#23	Cognitive Decline
#24	Decline, Cognitive
#25	Mental Deterioration
#26	Deterioration, Mental
#27	Delirium(MeSH)
#28	Delirium of Mixed Origin
#29	Subacute Delirium
#30	Delirium, Subacute
#31	Subacute Deliriums
#32	Deliriums, Subacute
#33	#17 OR #18 OR #19 OR #20 OR #21 OR #22 OR #23 OR #24 OR #25 OR #26 OR #27 OR #28 OR #29 OR #30 OR #31 OR #32
#39	#10 AND #21 AND #38

Selection of studies

A dual-independent screening process will be implemented by two trained researchers using predefined eligibility criteria. Initial title/abstract screening identified potentially eligible studies for full-text assessment. Inter-rater discrepancies will be resolved through consensus discussions or adjudication by a third researcher. The study selection protocol will follow Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) guidelines, with the complete screening workflow visually summarised in a PRISMA-compliant flow diagram (figure 1).

Figure 1. Flow diagram of the study selection process. PND, perioperative neurocognitive disorders. TMS, transcranial magnetic stimulation.

Data extraction

Standardised data extraction will be conducted using PRISMA-compliant templates developed a priori, with two independent investigators systematically recording study characteristics (author, year, country and sample size), subject characteristics (age range, surgical type and diagnostic criteria for neurocognitive disorders), intervention parameters (TMS frequency, intensity, stimulation site and treatment duration), comparator details (sham stimulation or standard care) and outcome measures (primary/secondary endpoints with effect size and evaluation time points). Missing numerical data will be requested via three email attempts to the corresponding authors at 14-day intervals, and if the author does not respond, WebLotDigitizer 4.5 (https://automeris.io/) will be used to quantify graphic data. For suspected duplicate data extracted, comparisons are made by checking the study registration number, author and institution information, study time and location, sample size and baseline characteristics and outcome indicators to avoid duplicate inclusion of the same study data. The disagreements that arose during the data extraction process will be resolved through consultation with the third reviewer.

Risk of bias assessment

The methodological quality of included RCTs will be independently assessed by two investigators using the revised Cochrane Risk of Bias Tool, which systematically evaluates five domains: randomisation integrity, intervention fidelity, completeness of outcome data, measurement validity and reporting transparency. Each domain will be classified as ‘low risk’, ‘some concerns’ or ‘high risk’. Discrepancies will undergo resolution through iterative consultation with the third reviewer.

Data synthesis

Meta-analyses will be performed using Stata V.15.0. Dichotomous outcomes (eg, POD incidence) will be expressed as pooled HRs with 95% CIs, while continuous variables (eg, cognitive assessment scores) will be harmonised through either standardised mean differences or mean differences contingent on measurement scale uniformity. A random-effects model will be employed as the primary analytical approach to accommodate expected clinical and methodological heterogeneity. Effect estimates will be visually represented through forest plots. In cases where substantial heterogeneity precludes quantitative synthesis, findings will be summarised through narrative analysis.

Assessment of heterogeneity

Heterogeneity will be quantified using the I² statistic and Cochran’s Q test. If I² <50% and p value >0.1, the heterogeneity included in the study is small, and a fixed effects model is used for meta-analysis calculation. I² >50% or p≤0.10 will indicate significant heterogeneity. In this case, sensitivity analysis and subgroup analysis will explore potential sources of variation, and if heterogeneity still cannot be explained, a random effects model will be used for meta-analysis calculations.

Subgroup analysis

Predefined subgroup analyses will examine whether intervention effects vary by surgery type (cardiac vs non-cardiac), rTMS stimulation parameters (frequency, intensity, coil type and duration), stimulation targets (DLPFC, posterior parietal cortex and posterior cingulate cortex), timing of intervention (preoperative prophylaxis vs postoperative treatment), age subgroups (≥65 years vs <65 years) and the time of evaluation of outcome. These analyses aim to identify clinically relevant modifiers of treatment efficacy and will only be conducted if at least three studies per subgroup are available to ensure statistical robustness.

Sensitivity analysis

To assess result stability, sensitivity analyses will sequentially exclude studies with high risk of bias, switch between fixed-effect and random-effects models and test alternative effect measures (eg, ORs instead of risk ratios). A ‘leave-one-out’ approach will further evaluate whether any single study disproportionately influences pooled estimates. Results from these analyses will be compared against primary findings to determine the conclusiveness of the evidence.

Assessment of publication bias

Publication bias will be evaluated through funnel plot asymmetry testing when ≥10 studies are included. Visual interpretation will be complemented by Egger’s regression test for small-study effects, with statistical significance (p<0.05) suggesting potential bias. On detection of asymmetry, the trim-and-fill method will be implemented to estimate the number of theoretically missing studies required to eliminate the observed effect. Furthermore, examination of unpublished trial registrations and grey literature will be performed to evaluate potential disparities between published and unpublished results.

Summary of evidence

The strengths and limitations of the synthesised evidence will be interpreted through the lens of the GRADE framework (Grading of Recommendations, Assessment, Development and Evaluation). We will evaluate the certainty of evidence for each critical outcome (eg, PND incidence, cognitive function changes) based on five domains: risk of bias, inconsistency, indirectness, imprecision and publication bias. Key findings will be summarised in a ‘Summary of Findings’ table, highlighting the magnitude of effect sizes, CIs and overall evidence quality (rated as high, moderate, low or very low). This will be independently completed by two researchers, and any disagreements will be resolved through discussions with the third researcher.

Ethics and dissemination

As this study synthesises data from previously published trials, ethical approval is not required. To ensure transparency and reproducibility, the full study protocol has been prospectively registered on PROSPERO. Results will be disseminated in a peer-reviewed journal.

Ethics statements

Patient consent for publication

Not applicable.

Footnote

Contributors JZ and YL designed this study. YT serves as the supervisor of this study. The research strategy for each database will be designed by all review authors. JZ and YY will independently carry out the search, selection and identification of studies and the data extraction. YL will perform the data synthesis and analysis. YT will serve as the third author for the settlement of the disagreement. YT and YL will be the advisers for methodology. All authors have approved the publication of this study protocol. JZ is the guarantor.

Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

Competing interests None declared.

Patient and public involvement Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.

Provenance and peer review Not commissioned; externally peer reviewed.

References

¹ Evered L, Silbert B, Knopman DS, et al. Recommendations for the Nomenclature of Cognitive Change Associated with Anaesthesia and Surgery-2018. Anesthesiology 2018; 129: 872–9. doi:10.1097/ALN.0000000000002334

² Hogue CW, Grafman J. Aligning nomenclature for cognitive changes associated with anaesthesia and surgery with broader diagnostic classifications of non-surgical populations: a needed first step. Br J Anaesth 2018; 121: 991–3. doi:10.1016/j.bja.2017.12.029

³ Safavynia SA, Goldstein PA, Evered LA. Mitigation of perioperative neurocognitive disorders: A holistic approach. Front Aging Neurosci 2022; 14: 949148. doi:10.3389/fnagi.2022.949148

⁴ Kirfel A, Guttenthaler V, Mayr A, et al. Postoperative delirium is an independent factor influencing the length of stay of elderly patients in the intensive care unit and in hospital. J Anesth 2022; 36: 341–8. doi:10.1007/s00540-022-03049-4

⁵ Florou C, Theofilopoulos D, Tziaferi S, et al. Post-Operative Delirium in Elderly People Diagnostic and Management Issues of Post-Operative Delirium in Elderly People. Adv Exp Med Biol 2017; 987: 301–12. doi:10.1007/978-3-319-57379-3_27

⁶ Vasilevskis EE, Han JH, Hughes CG, et al. Epidemiology and risk factors for delirium across hospital settings. Best Pract Res Clin Anaesthesiol 2012; 26: 277–87. doi:10.1016/j.bpa.2012.07.003

⁷ Steinmetz J, Christensen KB, Lund T, et al. Long-term consequences of postoperative cognitive dysfunction. Anesthesiology 2009; 110: 548–55. doi:10.1097/ALN.0b013e318195b569

⁸ Sprung J, Roberts RO, Weingarten TN, et al. Postoperative delirium in elderly patients is associated with subsequent cognitive impairment. Br J Anaesth 2017; 119: 316–23. doi:10.1093/bja/aex130

⁹ Gou RY, Hshieh TT, Marcantonio ER, et al. One-Year Medicare Costs Associated With Delirium in Older Patients Undergoing Major Elective Surgery. JAMA Surg 2021; 156: 430–42. doi:10.1001/jamasurg.2020.7260

¹⁰ Cibelli M, Fidalgo AR, Terrando N, et al. Role of interleukin-1beta in postoperative cognitive dysfunction. Ann Neurol 2010; 68: 360–8. doi:10.1002/ana.22082

¹¹ Barrientos RM, Hein AM, Frank MG, et al. Intracisternal interleukin-1 receptor antagonist prevents postoperative cognitive decline and neuroinflammatory response in aged rats. J Neurosci 2012; 32: 14641–8. doi:10.1523/JNEUROSCI.2173-12.2012

¹² Berger M, Oyeyemi D, Olurinde MO, et al. The INTUIT Study: Investigating Neuroinflammation Underlying Postoperative Cognitive Dysfunction. J Am Geriatr Soc 2019; 67: 794–8. doi:10.1111/jgs.15770

¹³ Eckenhoff RG, Maze M, Xie Z, et al. Perioperative Neurocognitive Disorder: State of the Preclinical Science. Anesthesiology 2020; 132: 55–68. doi:10.1097/ALN.0000000000002956

¹⁴ Foley KA, Djaiani G. Update of the European Society of Anaesthesiology and Intensive Care Medicine evidence-based and consensus-based guideline on postoperative delirium in adult patients. Eur J Anaesthesiol 2025; 42: 86–7. doi:10.1097/EJA.0000000000002043

¹⁵ Brown CT, Deiner S. Perioperative cognitive protection. Br J Anaesth 2016; 117: iii52–61. doi:10.1093/bja/aew361

¹⁶ Alam A, Hana Z, Jin Z, et al. Surgery, neuroinflammation and cognitive impairment. EBioMedicine 2018; 37: 547–56. doi:10.1016/j.ebiom.2018.10.021

¹⁷ Jannati A, Oberman LM, Rotenberg A, et al. Assessing the mechanisms of brain plasticity by transcranial magnetic stimulation. Neuropsychopharmacology 2023; 48: 191–208. doi:10.1038/s41386-022-01453-8

¹⁸ Ebrahimzadeh E, Sadjadi SM, Asgarinejad M, et al. Neuroenhancement by repetitive transcranial magnetic stimulation (rTMS) on DLPFC in healthy adults. Cogn Neurodyn 2025; 19: 34. doi:10.1007/s11571-024-10195-w

¹⁹ Wu Q, Xu X, Zhai C, et al. High-frequency repetitive transcranial magnetic stimulation improves spatial episodic learning and memory performance by regulating brain plasticity in healthy rats. Front Neurosci 2022; 16: 974940. doi:10.3389/fnins.2022.974940

²⁰ Lefaucheur J-P, Aleman A, Baeken C, et al. Evidence-based guidelines on the therapeutic use of repetitive transcranial magnetic stimulation (rTMS): An update (2014–2018). Clin Neurophysiol 2020; 131: 474–528. doi:10.1016/j.clinph.2019.11.002

²¹ Fu L, Ren J, Lei X, et al. Effects of repetitive transcranial magnetic stimulation (rTMS) on cognitive impairment in depression: A systematic review and meta-analysis. J Affect Disord 2025; 373: 465–77. doi:10.1016/j.jad.2025.01.014

²² Wu T, Li M, Tian L, et al. A modified mouse model of perioperative neurocognitive disorders exacerbated by sleep fragmentation. Exp Anim 2023; 72: 55–67. doi:10.1538/expanim.22-0053

²³ Zhou C, Gao Y-N, Qiao Q, et al. Efficacy of repetitive transcranial magnetic stimulation in preventing postoperative delirium in elderly patients undergoing major abdominal surgery: A randomized controlled trial. Brain Stimul 2025; 18: 52–60. doi:10.1016/j.brs.2024.12.1475

²⁴ Shamseer L, Moher D, Clarke M, et al. Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015: elaboration and explanation. BMJ 2015; 350: g7647. doi:10.1136/bmj.g7647

²⁵ Moher D, Shamseer L, Clarke M, et al. Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015 statement. Syst Rev 2015; 4: 1. doi:10.1186/2046-4053-4-1