Which Analgesic Should We Use to Relieve Pain After Knee or Hip Arthroplasty? A Systematic Review and Network Meta-analysis of RCTs

BACKGROUND: Effective postoperative pain management after knee and hip arthroplasties offers substantial clinical benefits; however, clinicians are faced with numerous analgesic options, and therapeutic decision-making is hindered by limited comparative evidence on the relative efficacy and safety of these treatments. Previous studies were limited to pairwise comparisons and could not integrate the full spectrum of available treatments or were focused on specific routes of administration that overlooked variability among individual analgesics within and across these routes. Therefore, an analysis comparing all analgesics that have been evaluated in randomized trials for use after knee and hip arthroplasty is warranted. QUESTIONS/PURPOSES: The purpose of this study was to perform a network meta-analysis that evaluates the relative efficacy and safety of currently available analgesics administered across different routes to determine which provide the greatest benefit in terms of (1) reducing postoperative pain, (2) improving function, and (3) minimizing adverse events. Given the distinct anatomic and biomechanical features of hips and knees and their differing postoperative pain trajectories, we conducted separate analyses for each procedure. METHODS: For this network meta-analysis, we systematically searched PubMed, Embase, the Cochrane Library, Web of Science, and reference lists from inception to July 6, 2023, with an update to July 13, 2025. We included RCTs that compared specific analgesics with placebo or other analgesics administered after knee or hip arthroplasty. Studies were eligible if they enrolled at least 10 participants per arm, administered analgesics postoperatively, and reported outcomes related to pain, function, or adverse events. We excluded secondary analyses, abstracts only, and trials focused on chronic pain starting 2 months or more postoperatively. Of the 49,400 studies retrieved, a total of 47,465 were excluded after duplication and title and abstract screening, leaving 1935 articles for full-text review. From these, 211 eligible RCTs were included, comprising 22,972 patients (median [range] age 67 years [43 to 80 years]; female proportion 24% to 98%) and encompassing 155 distinct analgesic regimens across five administration routes (intravenous, nerve-block, local infiltration analgesia, oral, and topical) and their combinations. Risk of bias was assessed using the Cochrane Risk of Bias Tool 2.0 (79% were rated as having "some concerns" and 21% as "high risk"), and the confidence of the evidence was evaluated using The Confidence in Network Meta-Analysis framework, with ratings from moderate to very low. Pain was the primary outcome, with the focus on pain at 24 hours postoperatively; pain at 48 hours was included as a secondary outcome. We prioritized pain during movement over pain at rest, as it better reflects functional recovery. When studies reported multiple pain scales, data were extracted based on a predefined hierarchy: (1) VAS, (2) verbal rating scale, and (3) numeric rating scale. We prespecified a minimum clinically important difference (MCID) of -1.8 cm on a 10-cm VAS for pain, based on a previously published study. Function and adverse events were secondary outcomes. For data synthesis, we employed a Bayesian network meta-analysis model and used a random walk model to account for the temporal structure of pain outcomes. Heterogeneity was quantified using the between-trial variance (τ 2 ) and a random-effects model was applied, given the observed small to moderate heterogeneity. RESULTS: For pain management after knee arthroplasty, compared with placebo, local levobupivacaine was the most effective analgesic (mean difference -4.9 cm [95% credible interval (CrI) -7.5 to -2.2]; surface under the cumulative ranking [SUCRA] 99%), exceeding the prespecified MCID of -1.8 cm with an 85% probability. After hip arthroplasty, compared with placebo, local ropivacaine combined with ketorolac and adrenaline was most effective (mean difference -3.5 cm [95% CrI -4.9 to -2.0]; SUCRA 73%), with a 99% probability of achieving the MCID. For postoperative functional improvement after knee arthroplasty, compared with placebo, nerve-block levobupivacaine combined with dexmedetomidine (used under monitored conditions) showed the greatest benefit (mean difference 63° [95% CrI 35° to 91°]; SUCRA 88%), with a higher degree value meaning a greater ROM. Regarding safety, compared with placebo, intravenous tramadol combined with metoclopramide reduced the risk of nausea (OR 0.1 [95% CrI 0.0 to 0.7]), whereas nerve-block bupivacaine combined with sufentanil increased the risk (OR 6 [95% CrI 1 to 31]). No other differences were observed across interventions for the remaining adverse events. CONCLUSION: For knee arthroplasty, local levobupivacaine appeared to be the most effective option. For hip arthroplasty, local ropivacaine combined with ketorolac and adrenaline appeared the best. In terms of functional improvement, nerve-block levobupivacaine combined with dexmedetomidine (used under monitored conditions) appeared the most effective after knee arthroplasty. Although most interventions had similar safety profiles, intravenous tramadol combined with metoclopramide decreased the risk of nausea. Clinicians should integrate these results with known safety profiles and patient-specific factors to guide individualized postoperative pain management after knee and hip arthroplasties. LEVEL OF EVIDENCE: Level I, therapeutic study.