Shirong Wei, Xin XIANG, Ming Yao, Huadong Ni
Editor, Peri-operative neurocognitive disorder (PND), encompassing postoperative delirium (POD), delayed cognitive decline and postoperative neurocognitive disorder (NCD) diagnosed up to 12 months after surgery, is a frequent complication in elderly patients following surgery. PND is associated with adverse outcomes, including prolonged hospital stay and loss of functional independence, yet effective non-pharmacological or pharmacological preventative strategies remain limited.1 Transcranial direct current stimulation (tDCS) is a non-invasive neuromodulation technique that alters cortical excitability. There are claims that it mitigates PND by promoting neuroplasticity and modulating neuro-inflammation.2,3 While a previous study suggested that tDCS may reduce POD in orthopaedic patients,4 its efficacy in preventing the broader spectrum of PND in elderly patients undergoing abdominal tumour surgery remains to be established. We aimed to investigate whether a two-session peri-operative tDCS intervention could reduce the incidence of PND in this surgical group. This randomised, sham-controlled trial received ethical approval (identifier: 2024-KY-615) and was registered with the Chinese Clinical Trial Register (ChiCTR2400094436). We enrolled patients aged >60 years undergoing abdominal tumour surgery. Key exclusion criteria were pre-existing severe cognitive impairment and long-term use of psychotropic medication. Full criteria are given in Supplement 1, https://links.lww.com/EJA/B293. Patients were randomised (1:1) to receive either active tDCS (2 mA for 20 min) or sham stimulation, administered on the afternoon before surgery and in the post-anaesthesia care unit after extubation. Guided by established efficacy and safety data,5 the anode was placed over the left dorsolateral prefrontal cortex (DLPFC) and the cathode over the right orbitofrontal area. The sham protocol involved an identical setup but with current delivered only during the 30-s ramp-up and ramp-down phases. During surgery, all patients were managed with a standardised anaesthesia protocol that included Bispectral Index monitoring (target range 40 to 60) and avoidance of benzodiazepines. The primary outcome was the incidence of PND, a composite of POD and delayed cognitive decline within 3 months. POD was assessed twice daily until discharge by blinded researchers, using a Richmond Agitation-Sedation Scale screen followed by the 3D-CAM or CAM-ICU. Delayed cognitive decline was defined as a ≥3-point decrease in the Mini-Mental State Examination (MMSE) score from baseline at either the 1-month or 3-month outpatient follow-up, a threshold supported by prior peri-operative cognitive research.6 Secondary outcomes included sleep quality, frailty, and others listed in Table S1, Supplemental Digital Content, https://links.lww.com/EJA/B294. Sample size was calculated to detect a PND reduction from 25% (based on institutional data) to 10%, requiring 107 patients per group. All analyses were performed on an intention-to-treat basis. The primary outcome was analysed using a multivariable logistic regression model, adjusting for age, baseline MMSE score, and duration of surgery, reported as an adjusted odds ratio (aOR) with 95% confidence interval (CI). Missing data for the primary outcome were handled by worst-case imputation. The robustness of the primary finding was assessed with sensitivity analyses. All other analyses were considered exploratory. All secondary analyses, including linear mixed-effects models for longitudinal data, subgroup, and mediation analyses, were considered exploratory. Full statistical details are in Supplement 1, https://links.lww.com/EJA/B294. Of the 262 patients assessed for eligibility, 214 were randomised between September 2024 and July 2025 (Fig. 1). Three patients were lost to the 3-month follow-up. Baseline characteristics were comparable (Table 1; Figure S3 to S4, Supplemental Digital Content, https://links.lww.com/EJA/B294). The primary outcome incidence was significantly lower in the tDCS group (10.3%) than the sham group (22.4%) (aOR 0.37, 95% CI, 0.15 to 0.90; P=0.029) (Table 2). The number needed to treat (NNT) to prevent one case of PND was estimated to be 9 (95% CI, 6 to 41). The incidence of POD was significantly lower in the tDCS group (9.3% vs. 19.6%; OR 0.42, P = 0.034), as was the incidence of delayed cognitive decline (5.6% vs. 14.0%; OR 0.36, P = 0.039). The Kaplan–Meier analysis further confirmed a lower cumulative incidence of PND in the tDCS group over the 3-month follow-up period (log-rank P = 0.016; Fig. 2).Fig. 1: CONSORT flow diagram of patient recruitment. Table 1 - Patient and baseline characteristics Variables tDCS (n = 107) Sham tDCS (n = 107) Age, years 70 ± 6 70 ± 7 Sex Male 60 (56.1) 51 (47.7%) Female 47 (43.9) 56 (52.3%) BMI (kg m−2) 24.0 ± 3.4 24.0 ± 3.5 Educational level (years) ≤6 74 (69.2) 74 (69.2) 6∼9 23 (21.5) 21 (19.6) ≥9 10 (9.3) 12 (11.2) Smoking 15 (14.0) 14 (13.1) Drinking alcohol 17 (15.9) 15 (14.0) American Society of Anesthesiologist classification II 60 (56.1) 48 (44.9) III 47 (43.9) 59 (55.1) History of surgical procedures 69 (64.5) 78 (72.9) Age adjusted Charlson comorbidity index 5.65 ± 1.15 5.73 ± 1.03 FRAIL Robust 64 (59.8) 60 (56.1) Prefrail 40 (37.4) 42 (39.3) Frail 3 (2.8) 5 (4.7) Hypertension 57 (53.3) 66 (61.7) Diabetes 19 (17.8) 24 (22.4) Stroke 5 (4.7) 7 (6.5) Cardiovascular disease 2 (1.9) 2 (1.9) Renal insufficiency 1 (0.9) 2 (1.9) COPD 3 (2.8) 2 (1.9) Surgical gradea II 1 (0.9) 4 (3.7) III 61 (57.0) 50 (46.7) IV 45 (42.1) 53 (49.5) Duration of anaesthesia (min) 275 ± 128 270 ± 103 Duration of surgery (min) 254 ± 125 246 ± 100 Dose of sufentanil (ug) 25 [20 to 30] 25 [20 to 30] Dose of remifentanil (mg) 2.0 [2.0 to 2.0] 2.0 [2.0 to 3.0] Blood loss (ml) 50 [50 to 100] 50 [50 to 100] Data are given as n (%), mean ± SD or median [IQR].tDCS, transcranial direct current stimulation; BMI, body mass index; SD, standard deviation; IQR, interquartile range; FRAIL, Fatigue, Resistance, Ambulation, Illness and Loss of Weight; COPD, chronic obstructive pulmonary disease.aSurgical grade was categorised based on the National Surgical Grading System of China, which classifies procedures from Grade I (minor risk) to Grade IV (major risk) based on complexity and difficulty. In our cohort, the majority of procedures were Grade III or IV. Table 2 - Primary outcome and its components No. (%) Variable tDCS (n = 107) Sham tDCS (n = 107) Odds ratio (95% CI)a NNTb P-valuea PND incidence within 3 months 11 (10.3) 24 (22.4) 0.37 (0.15 to 0.90) 9 (6 to 41) 0.029 POD 10 (9.3) 21 (19.6) 0.42 (0.19 to 0.94) NA 0.034 Delayed cognitive decline 6 (5.6) 15 (14.0) 0.36 (0.14 to 0.98) NA 0.039 Data are given as n (%) or 95% CI.CI, confidence interval; PND, peri-operative neurocognitive disorder; MMSE, Mini-Mental State Examination; NA, not applicable; NNT, number needed to treat; T0, baseline (before surgery); T1, the first day after the operation; T2, the second day after the operation; T3, the third day after the operation; T4, discharge; T5, the first month after the operation; and T6, the third month after the operation.aThe OR and P-value for the primary outcome was calculated using a multivariable logistic regression model, adjusted for age, baseline MMSE score, and duration of surgery.bNNT was calculated from the adjusted relative risk derived from a modified Poisson regression model. Fig. 2: Kaplan–Meier curve for the cumulative incidence of the composite PND endpoint (encompassing postoperative delirium during hospital stay and cognitive decline at 1 and 3 months) during the follow-up period.No serious adverse events were reported during the study. Mild, transient skin irritation at the stimulation site was more common in the tDCS group, while the incidence of other adverse events was similar between groups (Table S7, Supplemental Digital Content, https://links.lww.com/EJA/B294). The Cohen's kappa coefficient for the assessment of blinding was 0.03 (Table S8, Supplemental Digital Content, https://links.lww.com/EJA/B294). Detailed results for secondary outcomes (Table S1 to S2, Figure S2, Supplemental Digital Content, https://links.lww.com/EJA/B294), subgroup analyses (Figure S5, Supplemental Digital Content, https://links.lww.com/EJA/B294), mediation models (Figure S6, Supplemental Digital Content, https://links.lww.com/EJA/B294), multivariable risk factor analysis (Table S3), and all sensitivity analyses are provided in the Supplementary File 2, https://links.lww.com/EJA/B294. In this single-centre, randomised trial, a two-session peri-operative tDCS intervention in elderly patients undergoing major abdominal surgery, was associated with a lower incidence of the composite PND endpoint at three months. An analysis of the individual components revealed that this finding was supported by a lower incidence of both postoperative delirium and delayed cognitive decline, defined as a ≥3-point decrease in the MMSE score. Our primary composite endpoint was chosen to assess the patient's neurocognitive recovery trajectory after surgery. Our data suggest a link between these two events, as most patients who showed cognitive decline at three months had previously experienced POD. This supports the idea that by mitigating the initial delirium, tDCS may help prevent longer-term cognitive problems. While the incidence of delayed cognitive decline was significantly lower in the tDCS group, the average MMSE scores at three months were similar between the groups. This is probably due to a “ceiling effect” of the MMSE, where most patients in both groups maintained high scores. This suggests the intervention's primary effect may be the prevention of a meaningful decline in a subset of at-risk patients, rather than a general cognitive enhancement for all patients. While exploratory analyses suggested benefits in secondary outcomes like anxiety and pain, the underlying mechanisms remain unclear. The lasting effects of a brief intervention may be explained by its influence on neuroplasticity at both the synaptic and network levels, which can enhance the stability of cognitive networks against peri-operative insults.7 Our study has several limitations. First, as a single-centre trial, the generalisability of our findings requires confirmation in other settings. Second, our cognitive assessment relied on the MMSE. While our finding of a reduced incidence of a ≥3-point decline was statistically robust, the MMSE is a screening tool with known limitations, such as ceiling effects and low sensitivity to executive dysfunction. We also did not assess correlates in daily living or more detailed cognitive domains, meaning the observed statistical difference may not necessarily translate into functionally meaningful outcomes. Third, statistical limitations include a borderline events-per-variable ratio in the regression model, the wide confidence interval of the NNT estimate, and the reliance on imputation for missing primary outcome data, although we used a conservative approach. Finally, while our study followed a standardised anaesthesia protocol, residual confounding from unmeasured factors, such as postoperative sleep disruption, cannot be entirely excluded. In conclusion, in this single-centre, randomised trial, a two-session peri-operative tDCS intervention was associated with a lower incidence of PND in elderly patients undergoing major abdominal surgery. These preliminary findings are promising, but they require replication in larger, multicentre trials with more comprehensive cognitive assessments before any conclusions about clinical practice can be drawn.