We thank Li and Zhao for their thoughtful comments on our study and for raising two important methodological issues regarding the analyses of postoperative nausea and vomiting (PONV) and the choice of primary endpoint.1,2

Li and Zhao correctly note that the absolute difference in PONV rates narrowed from 11.1 percentage points in the intention-to-treat (ITT) analysis to 8.8 percentage points in the per-protocol (PP) analysis, with the latter failing to reach statistical significance (p=0.092 vs p=0.040 in ITT). They question whether our explanation, that most excluded patients discontinued the intervention due to severe PONV, is mathematically coherent.

We would like to clarify this point with reference to the enrolment and exclusion data already reported in the Results section of our manuscript. A total of 142 patients were screened, of whom 16 were excluded (12 refusals and 4 with cognitive impairment), resulting in 126 patients being randomized. During the trial, 3 patients had a surgical duration outside the specified range, 1 was admitted to the ICU postoperatively, and 4 discontinued the PCA pump midway. These 4 midway discontinuations were specifically driven by PONV. Consequently, 118 patients meeting the eligibility criteria were included in the PP analysis, with 60 in the oliceridine group and 58 in the hydromorphone group (Figure 1).1

Among these 4 patients who discontinued the PCA pump due to PONV, 1 belonged to the oliceridine group and 3 belonged to the hydromorphone group. Mathematically, because more PONV events were removed from the hydromorphone arm (3 events) than from the oliceridine arm (1 event), the PONV rate in the hydromorphone group decreased more substantially, from 15.9% to 12.1% (a 3.8 percentage point reduction), compared to the oliceridine group, which decreased from 4.8% to 3.3% (a 1.4 percentage point reduction). Since the between-group difference is calculated as the hydromorphone rate minus the oliceridine rate, a larger reduction in the hydromorphone rate will naturally narrow, and not widen, the absolute difference. Therefore, the observed narrowing from 11.1% to 8.8% is entirely consistent with, and directly explained by, the asymmetric exclusion of PONV cases between the two groups.

Regarding the loss of statistical significance in the PP analysis, we attribute this primarily to the marked reduction in the total number of PONV events available for analysis, from 13 events in the ITT population to just 9 events in the PP population. With such a small number of events, statistical power is severely limited, and minor fluctuations in rates can easily shift the p-value across the 0.05 threshold. This is a well-recognized limitation of PP analyses in studies with low event rates. The ITT analysis, which preserves randomization and includes all 126 patients, remains our primary and most robust finding.

We fully agree with Li and Zhao that movement-evoked (dynamic) pain is highly clinically relevant for patients recovering from lower limb fracture surgery. Accumulating evidence indicates that movement-evoked pain correlates more strongly with functional recovery than resting pain.3,4 In a prospective study following total knee arthroplasty, reduced movement-evoked pain during performance-based tests was consistently associated with improved patient-reported function, while greater movement-evoked pain at 3 months predicted worse functional outcomes at 6 months.3 Similarly, recent research on activity-related pain burden after total knee arthroplasty found that static pain area under the curve did not differ significantly between surgical modalities, whereas activity-related pain burden did, underscoring that dynamic pain captures aspects of the postoperative experience that resting pain fails to reflect.4

We explicitly acknowledged the lack of formal dynamic pain assessment as a notable limitation in our manuscript. While resting pain is a pragmatic and widely accepted primary endpoint in non-inferiority analgesic trials, and was appropriate for the primary objective of establishing analgesic efficacy, we concur that omitting dynamic pain assessment limits the clinical applicability of our findings. A drug that provides adequate analgesia at rest but fails to control movement-evoked pain may not translate into meaningful improvements in early rehabilitation, which is of paramount importance to both patients and surgeons.

We strongly endorse Li and Zhao’s recommendation that future studies incorporate movement-evoked pain as a co-primary or key secondary endpoint, alongside functional recovery metrics such as time to first ambulation and walking distance. These outcomes would better define whether oliceridine’s analgesic profile extends to the functional domain that matters most in the postoperative period.

In summary, the ITT analysis provides the most reliable estimate of oliceridine’s PONV benefit, with the PP attenuation readily explained by asymmetric exclusion of PONV events and limited statistical power. The enrolment and exclusion details reported in our Results section document that all 4 midway PCA discontinuations were PONV-driven, with 3 occurring in the hydromorphone arm and 1 in the oliceridine arm. At the same time, we agree that the clinical relevance of our findings would be strengthened by dynamic pain and functional outcome data. We fully support the call for large-scale, multicentre trials incorporating comprehensive pain assessments, functional recovery indicators, and patient-centred outcomes to better define oliceridine’s role in perioperative care.

The authors report no conflicts of interest in this communication.

1. Jiang Q, Ouyang Y, Zheng J, et al. Comparison of oliceridine and hydromorphone for postoperative analgesia in adults undergoing lower limb fracture surgery: a randomized controlled non-inferiority trial. Drug Des Devel Ther. 2026;20:628492. doi:10.2147/DDDT.S628492

2. Li R, Zhao S. Oliceridine versus hydromorphone for postoperative analgesia: interpreting the PONV signal and the limitations of resting pain as a primary endpoint [Letter]. Drug Des Devel Ther. 2026;20:648262. doi:10.2147/DDDT.S648262

3. Wilson JM, Madden VJ, Pester BD, et al. Change in pain during physical activity following total knee arthroplasty: associations with improved physical function and decreased situational pain catastrophizing. Innov Aging. 2023;7(10):igad045. doi:10.1093/geroni/igad045

4. Xu Y, Li W, Song Y, et al. Comparison of early postoperative activity-related pain burden after conventional,navigation-assisted, and robotic-assisted total knee arthroplasty: a retrospective cohort study based on 24-72 h pain area under the curve. J Robot Surg. 2026;20(1):802. doi:10.1007/s11701-026-03768-w