Design: systematic review and network meta-analysis of randomised controlled trials
Population: 2,130 adults across 24 trials undergoing surgery with peripheral nerve blocks, eight interventions compared
Primary outcome: incidence of rebound pain after block regression
Effect: intravenous dexamethasone ranked first (surface under the cumulative ranking curve 0.91), judged definitely superior to control; pre-emptive opioids probably inferior to control

Rebound pain when a single-injection block wears off has become the main argument against regional anaesthesia in day-case orthopaedics, and practice for preventing it has been improvised. Twenty-four randomised trials covering 2,130 adults and eight interventions were assembled into a network meta-analysis with rebound pain incidence as the primary outcome.

Intravenous dexamethasone came first for reducing incidence, with a surface under the cumulative ranking curve of 0.91, and was judged 'definitely superior' to control. Perineural dexamethasone and perineural dexmedetomidine ranked highest on the different question of delaying rebound pain and prolonging time to first rescue analgesia. Pre-emptive opioids were probably inferior to control — that is, worse than doing nothing. Certainty of evidence ranged from very low to moderate, and pain severity, satisfaction and sleep disturbance were reported too inconsistently to pool.

Two things are directly usable. First, the drug most units already give for nausea prophylaxis is the one that ranks best here, and giving it intravenously rather than perineurally is the simpler route with the better evidence for the outcome that matters. Second, loading the patient with an opioid before the block wears off — a common instinct — did not help and may be harmful.

What this does not settle is severity. Incidence tells you how many patients get rebound pain, not how bad it is, and that is the question a patient asks. Certainty in the network was limited by small trials and inconsistent definitions of rebound pain itself, which remains the field's real problem.

Give intravenous dexamethasone at induction to patients receiving a single-injection peripheral nerve block, where not contraindicated
Stop pre-emptive opioid loading before block regression — it ranked below control
Consider perineural dexamethasone or dexmedetomidine where the aim is to delay onset rather than reduce incidence
Tell the patient when the block will wear off and start oral analgesia before that, not after
Check glucose in diabetic patients given dexamethasone; the perioperative rise is real even at single doses

Why it matters

The instinct to pre-load with an opioid before the block regresses ranked below doing nothing, and the drug that worked is one most theatres already draw up.

The statistics, in plain English

A surface under the cumulative ranking curve of 0.91 means dexamethasone was near the top across the simulated rankings, not that it reduced pain by 91% of anything — these are rank probabilities, not effect sizes. Network meta-analysis lets interventions never compared head to head be ranked indirectly, which is its strength and its main vulnerability. Certainty rated very low to moderate means several of these comparisons rest on small trials with differing definitions of the outcome, so the ranking is the finding and the precise ordering below first place should be held loosely.