Evidence base and practice variation in acute care processes for knee and hip arthroplasty surgeries

Background Lack of evidence contributes to unnecessary variation in treatment costs and outcomes. This study aimed to identify from interventions historically used for total knee or hip arthroplasty (TKA, THA): i) if routine use is supported by high-level evidence; ii) whether surgeon use aligns with the evidence. Methods Part 1: Systematic search of electronic library databases for systematic reviews and practice guidelines concerning seven acute-care interventions. Intervention-specific recommendations concerning routine use were extracted by assessors. Part 2: Prospective medical record audit of the acute-care received by 1900 patients involving 120 orthopaedic surgeons. Surgeon use per intervention was summarized using caterpillar plots. Surgeon-specific routine and non-routine use was defined as use in ≥ 90% and ≤ 10% of patients, respectively. Primary analysis included only surgeons contributing ≥ 10 patients. Results Continuous passive motion (TKA): Routine use not recommended; 85.7% of surgeons did not use it routinely. Tranexamic Acid: Routine use recommended; 26.9% of surgeons used it routinely. Cryotherapy: Routine use not recommended; 45.7% of surgeons used it routinely for TKA; 31.8% used it routinely for THA. Intra-articular drainage: Routine use not recommended for TKA, but possible benefits for THA; 5.7% of surgeons used it routinely for TKA, 0.0% used it routinely for THA. Antibiotic loaded bone cement: Routine use for TKA not supported, recommendations for use for THA are inconsistent; 90.0% of surgeons used it routinely for TKA, 100.0% used it routinely for THA. Patella resurfacing (TKA): No recommendation could be made; 57.1% of surgeons routinely resurfaced the patella. Indwelling urinary catheterisation: Routine use recommended; 59.6% of surgeons used it routinely. Conclusion Recommendations for routine use or not exist for some of the acute-care interventions examined. Surgeon practices vary widely even in the presence of high-level recommenda