Equipotent Doses and ED95 Extrapolation: Methodological Concerns in Trials of Ciprofol-Esketamine/Sufentanil for Elderly ESD [Letter] Authors: Luoyun Li, Zeru Zhang, Zehua Wang, Chunyu Li Published: 2026 Dear editor We read with great interest the recent article by Qin et al, which addresses the clinically relevant challenge of optimizing sedation for elderly patients undergoing endoscopic submucosal dissection (ESD) by comparing ciprofol-based combination regimens and exploring ciprofol dose requirements.1 This work provides valuable preliminary data for this vulnerable population. However, several methodological issues warrant further discussion to support appropriate interpretation of the findings. A key concern relates to the equivalence of the co-intervention doses. The study administered esketamine 0.5 mg/kg to group E and sufentanil 0.1μg/kg to group S as background analgesics. While both doses fall within routine clinical ranges, prior work by Yang et al (cited as Ref 31) suggests that in elderly patients undergoing gastrointestinal endoscopy, esketamine 0.2–0.25 mg/kg provides analgesic efficacy roughly equivalent to sufentanil 0.1–0.15μg/kg.2 The 0.5mg/kg esketamine dose used is thus approximately twice the equipotent dose of the sufentanil regimen. Given esketamine’s intrinsic sympathomimetic effects (which counteract sedation‑related hypotension) and sufentanil’s propensity for respiratory depression and haemodynamic instability, this uncalibrated dose discrepancy may confound the observed between‑group differences in hypotension, hypoxia and induction time, rather than reflecting the true effect of the ciprofol combination. Additionally, although allocation concealment was appropriately implemented, the lack of blinding among anaesthesiologists administering interventions may introduce performance bias in operator-dependent outcomes such as induction time assessment and rescue medication timing. Another methodological issue is the specification of two primary outcomes across the study’s two phases, which deviates from CONSORT guidance for randomised controlled trials.3 A well‑powered RCT should define a single primary outcome aligned with the core research hypothesis to minimise inflation of type I error from multiple testing. In this study, Phase 1 designated the ciprofol dose–response relationship as the primary outcome (with ED95 as a model‑derived secondary metric), while Phase 2 identified intraoperative hypotension incidence as the primary outcome. The absence of a unified primary outcome obscures the core hypothesis and raises the risk of selective reporting. Furthermore, no corrections for multiple comparisons were applied to the secondary outcomes of phase 2, which limits the robustness of findings such as the difference in time to full alertness. The sample-size calculation also deserves scrutiny. The authors anchored the calculation exclusively to phase 2’s primary outcome, without addressing the precision requirements of the dose-finding component. They calculated 35 per group to achieve 80% power to detect a difference in hypotension incidence, but the 44 patients enrolled in phase 1 (24 in group E, 20 in group S) were not sized based on formal precision requirements for ED95 estimation. For probit regression-based derivation of high-percentile endpoints like ED95, larger samples are required to reduce extrapolation error.4,5 Moreover, the phase 2 sample size was not adjusted for the multiple secondary outcomes explored, meaning the trial was underpowered to detect clinically meaningful differences in outcomes such as hypoxia incidence or postoperative analgesia requirements. Equally important is the use of the Dixon up-and-down method (UDM) with probit extrapolation to estimate ED95. The UDM is inherently optimised for efficient estimation of ED50, as it concentrates allocated doses tightly around the median effective dose, with minimal data collected at the extremes of the dose–response curve. Estimating ED95 relies heavily on model extrapolation beyond the range of observed data, which introduces substantial uncertainty. As outlined by Pace and Stylianou, UDM-derived high-percentile estimates have poor precision unless the dose range is explicitly extended to cover the target percentile.2 Contemporary dose-finding studies targeting ED90/ED95 typically employ purpose‑built designs such as the biased-coin design, continual reassessment method or k-in-a-row design, which allocate more participants to doses near the target percentile to enable direct estimation rather than extrapolation. While the authors appropriately caution against overinterpreting the extrapolated ED95 values in the discussion, positioning this metric as a central outcome and featuring it prominently in the title may mislead readers regarding its reliability for clinical dosing guidance: a concern underscored by the wide 95% confidence intervals of the ED95 estimates (eg, 0.264–0.349 mg/kg for the esketamine group). Finally, the study title is not fully aligned with its core contributions. The current title frames ED95 estimation as the central output, yet the extrapolated ED95 values are secondary metrics from the exploratory phase 1, while the robust comparative effectiveness and safety data for the two ciprofol combinations represent the study’s most original and clinically useful contribution. This mismatch may lead readers to overestimate the strength of the ED95 findings. A revised title that prioritises the comparative clinical outcomes while noting the exploratory dose-finding component would more accurately reflect the study’s scope-for example, “Ciprofol Combined with Esketamine versus Sufentanil for Elderly Patients Undergoing Endoscopic Submucosal Dissection: A Randomised Controlled Trial with Exploratory Dose Finding”. In summary, while the study addresses an important clinical need, the issues of uncalibrated comparator doses, dual primary outcomes, insufficient sample sizing for ED95 estimation, reliance on extrapolated dose-finding methodology, and a misleading title collectively weaken the interpretability of the results. Future investigations targeting high-percentile dose endpoints should adopt validated dose-finding designs (eg, biased-coin design) that directly estimate ED95 rather than relying on UDM extrapolation. Pre-study validation of analgesic equivalence between comparator agents, specification of a single primary outcome aligned with the core hypothesis, and alignment of the title with the main contributions will further strengthen the rigour of trials in this field. We hope these constructive comments support the accurate interpretation of the authors’ findings and the design of future research. Funding Statement No funding was received for this communication. Disclosure The authors report no conflicts of interest in this communication. References