Background
Morphine-sparing effects are often used to evaluate non-opioid analgesic interventions. The exact effect that would warrant the implementation of these interventions in clinical practice (a minimally important difference) remains unclear. We aimed to determine this with anchor-based methods.

Methods
This was a post hoc analysis of three studies investigating pain management after hip or knee arthroplasty (PANSAID [NCT02571361], DEX-2-TKA [NCT03506789] and Pain Map [NCT02340052]). The overall population was median aged 70, median ASA 2, 54% female. We examined the correlation between 0 and 24 h postoperative iv morphine equivalent consumption and the severity of nausea, vomiting, sedation and dizziness. The anchor was different severity degrees of these opioid-related adverse events. The primary outcome was the difference in morphine consumption between patients experiencing no versus only mild events. Secondary outcomes included the difference in morphine consumption between patients with mild versus moderate and moderate versus severe events. We used Hodges–Lehmann median differences, exact Wilcoxon–Mann–Whitney tests and quantile regression.

Results
The difference in iv morphine consumption was 6 mg (95% confidence interval: 4–8) between patients with no versus only mild events, 5 mg (2–8) between patients with mild versus moderate events and 0 mg (−4 to 4) between patients with moderate versus severe events.

Conclusions
In populations comparable to this post-hoc analysis (orthopaedic surgery, median age 70 and ASA 2), we suggest a minimally important difference of 5 mg for 0–24 h postoperative iv morphine consumption.