We thank Lin and Gao 1 for their thoughtful comments on our study 2 and welcome the opportunity to address the mechanistic and translational questions they raise. The central question is whether the delayed behavioural benefit reflects superior analgesia or an independent neuroprotective action. Acting on this suggestion, we performed a causal mediation analysis using the mediation R package (R Foundation for Statistical Computing, Vienna, Austria) with the area under the pain score time curve as the mediator. For emergence delirium, the total effect of esketamine was an absolute risk reduction of 25.1% (95%CI 22.4–27.9%) of which only 26.0% (95%CI 21.1–31.6%) was mediated by reduced pain exposure (average causal mediation effect -6.5%, 95%CI -7.8% to -5.4%), while 74.0% reflected a direct, pain-independent effect (-18.6%, 95%CI -21.4% to -15.6%, all p < 0.001). The dissociation was more pronounced for negative behavioural changes on postoperative day 7. Only 14.7% (95%CI 11.0–19.6%) of the total effect was mediated through analgesia, with 85.3% attributable to direct effects. These findings support the hypothesis from Lin and Gao that esketamine acts predominantly through pain-independent mechanisms. This is most plausibly via N-methyl-D-aspartate-receptor modulation of neuroinflammation and glutamatergic excitotoxicity 2, and is concordant with the temporal pattern of benefit, whereby behavioural divergence emerged only after postoperative day 1 and widened through to day 30. Comparable neuroprotective benefits may therefore extend to procedures with lower baseline pain burdens. Regarding rescue medication, propofol and morphine consumption were comparable between groups, although the trial was not powered for this endpoint. To provide the translational metric requested, the absolute risk reductions of 26% for emergence delirium and 19% for negative behavioural changes on postoperative day 7 correspond to numbers needed to treat of 4 (97.5%CI 3–7) and 6 (97.5%CI 3–21), respectively. These estimates compare favourably with other pharmacological strategies in network meta-analyses of paediatric emergence delirium 3 and support institutional adoption within enhanced recovery pathways. The proposal to combine esketamine with dexmedetomidine is mechanistically attractive. Wen et al. reported that this combination reduced postoperative delirium in paediatric neurosurgery compared with either drug alone 4. However, dexmedetomidine carries well-documented risks of bradycardia, hypotension and delayed recovery 5, which warrant caution after adenotonsillectomy where airway oedema and respiratory events following tracheal extubation remain a concern. Future trials should integrate inflammatory biomarkers (e.g. IL-6, TNF-α) and neurophysiological monitoring to clarify whether transient intra-operative pharmacological modulation really reshapes neurodevelopmental trajectories or attenuates the behavioural sequelae of peri-operative distress. Our mediation analysis indicates that esketamine confers behavioural protection predominantly through pain-independent mechanisms, reframing it as a preventive neuropsychiatric intervention rather than an analgesic adjunct. Biomarker-enriched trials across diverse surgical populations are now required to define the optimal integration of esketamine into multimodal paediatric peri-operative care.
Chen et al. (Thu,) studied this question.