Perioperative duloxetine has been promoted as a multimodal adjunct to reduce opioid use after arthroplasty. This meta-analysis pooled 13 randomised trials and 1,176 patients undergoing hip or knee replacement.
Duloxetine reduced pain at rest by 5.09 mm on a 100 mm scale at 24 hours and during ambulation by 4.55 mm, with effects persisting at two weeks and beyond three months. Opioid consumption fell by 22.74 morphine milligram equivalents at 48 hours and 10.58 at 72 hours. Nausea and vomiting were less frequent (risk ratio 0.73), and drowsiness more so (risk ratio 1.88).
The authors are commendably honest about the weaknesses, and they are substantial. The pain reductions are statistically significant but below the threshold usually considered clinically meaningful — roughly 10 mm on this scale. The 48-hour opioid-sparing estimate carries I-squared of 100%, meaning the trials disagreed completely, and its interval runs from 1.31 to 44.17 MME. On leave-one-out analysis, the long-term pain reduction, the nausea benefit and the fatigue benefit each lost significance when a single high-risk trial was removed. Duloxetine may have a place in a multimodal protocol, but the case for it is weaker than the headline p-values suggest, and drowsiness needs mentioning to patients.
- Pain reduced about 5 mm on a 100 mm scale — below the usual 10 mm threshold
- Opioid sparing at 48 and 72 hours, but I-squared of 100% at 48 hours
- Less nausea and vomiting; nearly twice the drowsiness
- Three findings lost significance when one high-risk trial was removed
The statistics, in plain English
An I-squared of 100% means the trials were not measuring the same thing in any meaningful sense, and pooling them produces a number with no real referent. The leave-one-out result is the other warning sign: a conclusion that depends on one trial of high risk of bias is not a conclusion. This is a good example of a meta-analysis whose abstract reads positive and whose detail reads equivocal.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for orthopaedics, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free