Abstract
Abstract
Background
Postoperative pain following total hip arthroplasty (THA) and total knee arthroplasty (TKA) drives opioid dependence. Perioperative duloxetine is an option as a multimodal analgesic adjunct, yet its efficacy and safety in this population are not fully defined.
Methods
This systematic review and meta-analysis compared perioperative duloxetine to placebo in THA and TKA patients, following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and registered with the International Prospective Register of Systematic Reviews (PROSPERO) (CRD420261294277). The search strategy spanned PubMed, Web of Science, and Scopus. The Cochrane Library was also queried from inception to March 2, 2026. The primary outcomes were Visual Analog Scale (VAS) pain scores at rest and during ambulation; secondary outcomes included postoperative opioid consumption and the incidence of adverse effects. VAS and Numeric Rating Scale (NRS) pain scores were converted to a common 0–100 mm scale for pooling and expressed as mean differences (MD) with 95% confidence intervals (CI); opioid consumption was standardized to morphine milligram equivalents (MME) and expressed as MD in MME. Dichotomous data were expressed as risk ratios (RR).
Results
Thirteen randomized controlled trials (RCTs) (n = 1176) were included. Duloxetine significantly reduced pain at rest (MD − 5.09 mm, 95% CI − 8.76 to − 1.43,
P
= 0.006) and during ambulation (MD − 4.55 mm, 95% CI − 8.71 to −0.40,
P
= 0.03) at 24 h. Analgesic benefits persisted at 2 weeks (rest pain: MD − 6.48 mm, 95% CI − 11.14 to − 1.82,
P
= 0.006; ambulatory pain: MD − 8.93 mm, 95% CI − 13.16 to − 4.69,
P
P
= 0.01; ambulatory pain: MD − 3.76 mm, 95% CI − 6.62 to − 0.89,
P
= 0.01). After standardization to morphine milligram equivalents (MME), statistically significant opioid-sparing effects were observed at 48 h (MD − 22.74 MME, 95% CI − 44.17 to − 1.31,
P
= 0.04, I
2
= 100%) and 72 h (MD − 10.58 MME, 95% CI − 18.67 to − 2.49,
P
= 0.01); both estimates were subject to extreme heterogeneity, and the 48-h estimate in particular should be interpreted with substantial caution given its very wide interval. Duloxetine significantly reduced the risk of nausea/vomiting (RR 0.73, 95% CI 0.55–0.97,
P
= 0.03); a borderline reduction in fatigue was also observed (RR 0.87, 95% CI 0.76–0.996,
P
= 0.04). Duloxetine increased the risk of drowsiness/somnolence (RR 1.88,
P
= 0.001).
Conclusions
Perioperative duloxetine produces statistically significant, though sub-threshold, analgesic effects across all assessed timepoints and a selective opioid-sparing effect spanning 48 to 72 h postoperatively when incorporated into multimodal analgesia for THA and TKA. All findings are subject to substantial heterogeneity and should be interpreted with caution. Duloxetine significantly reduced the incidence of nausea/vomiting and increased the risk of drowsiness/somnolence, the latter warranting preoperative patient counseling; a borderline reduction in fatigue was also observed. However, on leave-one-out sensitivity analysis, the long-term (≥ 3-month) reduction in pain at rest, the nausea/vomiting reduction, and the fatigue reduction each lost statistical significance upon exclusion of a single high-risk trial (Inamullah 2023) and should be interpreted with corresponding caution.