Interpreting Respiratory Safety in an Oliceridine Gastroscopy Trial [Letter]

Dear editor

We read with great interest the randomized trial by Hu et al, which compared three doses of oliceridine with sufentanil, each combined with propofol, for painless gastroscopy.1 The authors reported substantially less respiratory depression with oliceridine 10 or 20 μg/kg than with oliceridine 30 μg/kg or sufentanil 0.1 μg/kg and identified 20 μg/kg as providing the most favorable balance between sedation success and respiratory safety. This clinically relevant dose-finding comparison is promising. Nevertheless, three issues should be clarified before the respiratory safety conclusion is accepted.

First, the prospectively registered outcomes and design differ materially from the published report. The Chinese Clinical Trial Registry entry (ChiCTR2500097246) identifies mean arterial pressure, heart rate, and SpO₂ as primary outcomes; it does not designate the composite incidence of respiratory depression as the primary outcome. The registry also specifies 120 participants (30 per group), recruitment from March to April 2025, ASA physical status I–II, and a study end date of September 30, 2025. In contrast, the article reports a pilot-based sample size of 188, recruitment from July to October 2025, inclusion of ASA I–III, and respiratory depression as the primary outcome. The registry still indicates “not yet recruiting” and provides no measurement time points or methods for the registered outcomes. Furthermore, the registry states that individual participant data would be made publicly available with publication, whereas the article provides no repository or dataset. These changes may represent legitimate protocol amendments, but their timing and rationale are essential. CONSORT 2025 recommends identifying the trial protocol and statistical analysis plan and describing important changes made after trial commencement with reasons.2 We therefore invite the authors to provide the dated protocol, statistical analysis plan, ethics approval for any amendments, and confirmation that the revised primary outcome and sample size were established before examination of trial data.

Second, the analysis did not preserve all randomized participants. One participant in the 20 μg/kg group experienced SpO₂ <90% during gastroscopy but was excluded because a tight body-shaping garment was considered to have affected the result. Another participant in the 30 μg/kg group was excluded because the procedure exceeded a prespecified 10-minute threshold, although this criterion is not reported among the eligibility criteria. Both exclusions occurred after randomization and were related to either the primary outcome or the procedural course. Excluding an observed respiratory event on the basis of a post-randomization judgment can disrupt the protection afforded by randomization and favor the tested regimen. Inclusion of the excluded event would change the reported hypoxemia rate in the 20 μg/kg group from 0/46 to at least 1/47 and respiratory depression from 8/46 to at least 9/47. The flow diagram also states that no participant was excluded from analysis despite these omissions. An analysis of all 188 participants as randomized, accompanied by sensitivity analyses addressing the garment-related measurement concern and prolonged procedure, would provide a more credible estimate of treatment effect.2

Third, ascertainment and clinical interpretation of the respiratory composite require greater detail. Respiratory depression was defined by any of respiratory rate ≤8 breaths/min, “respiratory arrest” lasting >15 seconds, or SpO₂ <90%. Respiratory arrest was reported in 13.04%, 15.22%, 42.22%, and 51.06% of the low-, medium-, high-dose, and control groups, respectively, and therefore appears to drive much of the composite. However, the report does not state whether apnea was detected using capnography, impedance monitoring, or direct observation, nor how events were adjudicated. All participants received oxygen at 5 L/min; supplemental oxygen can delay pulse-oximetric recognition of hypoventilation,3 whereas capnography can identify apnea and altered ventilation before desaturation during endoscopic sedation.4 Because the component events can overlap, the joint distribution, number and duration of episodes, lowest SpO₂, and airway interventions should be reported by group. The term “transient apnea” may also be more clinically precise than “respiratory arrest” if no assisted ventilation or interruption of gastroscopy was required. CONSORT Harms further supports reporting event ascertainment, denominators, severity, and estimates with uncertainty rather than relying mainly on nominal P values.5 Pairwise risk differences or risk ratios with 95% confidence intervals and multiplicity-adjusted exact P values would make the dose comparisons more informative.

In summary, Hu et al provide useful preliminary evidence regarding oliceridine during propofol-based gastroscopy. However, reconciliation of the registry with the final protocol, analysis of all randomized participants, and fuller reporting of respiratory-event ascertainment are needed before concluding that oliceridine 20 μg/kg has a superior respiratory safety profile.

Disclosure

The author reports no conflicts of interest in this communication.