Prophylactic interventions to prevent rebound pain after peripheral nerve block: a reply

We thank Mistry and Sukumaran [1] for their thoughtful comments on our network meta-analysis of prophylactic interventions for rebound pain following peripheral nerve block [2]. We agree that an important question is whether the observed reduction in rebound pain reflects true attenuation of the rebound pain phenomenon or simply a delay in pain onset beyond the observation period. To address this distinction, the incidence of rebound pain and time to onset were analysed as separate outcomes in our review. Our findings do not support displacement as the sole explanation for the observed reduction in rebound pain. The rankings for incidence and for delay were dissociated: perineural dexmedetomidine produced the longest delay in onset (mean difference 8.46 h, 95%CrI -0.46–17.40) yet ranked only fourth for reducing the incidence of rebound pain (surface under the cumulative ranking curve (SUCRA) 0.54), whereas intravenous dexamethasone ranked first for reduced incidence (SUCRA 0.91) and third for delay (mean difference 7.89 h, 95%CrI -0.92–15.11). If lower incidence were simply a function of prolonged analgesia, a closer correspondence between the two rankings would be expected. The size of the delay was also modest relative to the follow-up period of the included studies. Time to rebound pain in the control group was 12.97 h and the greatest estimated delay extended this to approximately 21 h, still within the 24-h window used by most included trials. An absolute reduction of 388 per 1000 patients is difficult to attribute to a shift of that size. It is also relevant that intravenous dexamethasone, which prolongs block duration less than the perineural route [3], was the most consistently effective intervention. This pattern fits a systemic anti-inflammatory mechanism better than displacement of pain onset. However, these arguments remain indirect and our data cannot dismiss some contribution from prolonged analgesia. Future trials that report pain trajectories relative to the time of block resolution rather than to fixed postoperative time-points, incorporate standardised definitions of rebound pain and allow adequate follow-up after complete block regression may better distinguish genuine modification of rebound pain from delayed pain onset [4]. On transitivity, we agree that it is a fundamental assumption underlying the validity of indirect comparisons and that low statistical inconsistency alone does not confirm its fulfilment. Indirectness was evaluated using the CINeMA framework across domains including study participants, interventions and reported outcome characteristics, with comparisons involving heterogeneous surgical populations downgraded accordingly [5]. This is reflected in our summary of findings table, where certainty of evidence was moderate for intravenous dexamethasone and perineural dexamethasone, but very low for several interventions; no comparison was graded as high-certainty evidence. Consequently, our conclusions are based on effect estimates, precision, certainty of evidence and SUCRA values considered collectively, rather than on rank order alone. Formal testing of block type or surgical site as effect modifiers, by sub-group analysis or network meta-regression, was not possible because most nodes were formed by only two or three trials.