Pain treatment after total hip arthroplasty: Detailed statistical analysis plan for the RECIPE randomised clinical trial. Issue 3 (12th January 2023)

• Record Type: Journal Article
• Authors: Steiness, Joakim; Hägi‐Pedersen, Daniel; Lunn, Troels H.; Nørskov, Anders K.; Lindberg‐Larsen, Martin; Graungaard, Ben K.; Lundstrøm, Lars H.; Lindholm, Peter; Brorson, Stig; Bieder, Manuel J.; Beck, Torben; Skettrup, Michael; von Cappeln, Adam G.; Thybo, Kasper H.; Varnum, Claus; Pleckaitiene, Lina; Anker Pedersen, Niels; Overgaard, Søren; Mathiesen, Ole; Jakobsen, Janus C.
• Is Part Of: Acta anaesthesiologica scandinavica. Volume 67:Issue 3(2023)
• Page Start: 372
• Page End: 380
• Publication Date: 2023-01-12
• Subjects: anaesthesiology -- dexamethasone -- glucocorticoid -- multimodal analgesia -- NSAID -- pain -- paracetamol -- total hip arthroplasty; Anesthesiology -- Periodicals; Critical care medicine -- Periodicals; 617.9605
• DOI: 10.1111/aas.14179
• Languages: English
• ISSNs: 0001-5172

Abstract

Abstract: Background: The RECIPE trial systematically investigates the effects of different combinations of paracetamol, ibuprofen and dexamethasone for pain treatment after total hip arthroplasty. To preserve transparency, minimise risk of bias and to prevent data‐driven analysis, we present this detailed statistical analysis plan. Methods: The RECIPE trial is a randomised, blinded, parallel four‐group multicenter clinical trial for patients undergoing planned primary total hip arthroplasty. Interventions are initiated preoperatively and continued for 24 h postoperatively. Primary outcome is total opioid consumption 0–24 h after end of surgery. Primary analysis will be performed in the modified intention to treat population of all patients undergoing total hip arthroplasty, and all analyses will be stratified for site. We will perform pairwise comparisons between each of the four groups. The primary outcome will be analysed using the van Elteren test and we will present Hodges–Lehmann median differences and confidence intervals. Binary outcomes will be analysed using logistic regression. To preserve a family‐wise error rate of <0.05, we will use a Bonferroni‐adjusted alfa of 0.05/6 = 0.0083 for all six pairwise comparisons between groups when analysing the primary outcome. We will systematically assess the underlying statistical assumptions for each analysis. Data will be analysed by two blinded independent statisticians, and we will write abstracts covering all possible combinations of conclusions, before breaking the blind. Discussion: The RECIPE trial will provide important information on benefit and harm of combinations of the most frequently used non‐opioid analgesics for pain after primary hip arthroplasty.