Pain score distribution: an expert elicitation study for distal humerus hemiarthroplasty and total elbow arthroplasty using a validated numerical patient rated outcome-measure in trauma

Hypothesis: With equal numbers of adults with unreconstructable distal humerus fractures undergoing total elbow arthroplasty (TEA) and distal humerus hemiarthroplasty (DHH) a randomized trial is needed to compare the interventions. Patients have identified pain as the key outcome. There are no data on pain outcomes using a validated pain instrument to plan a trial. This study aims to use expert elicitation to produce probability distributions for the Patient Rated Elbow Evaluation (PREE) pain score at 12 months in adults undergoing TEA or DHH for trauma. Methods: Using the Sheffield Expert Elicitation Framework seven experts were recruited and provided with a summary of current knowledge on pain outcomes from a systematic review. They were asked to consider their median of the PREE pain score at 12 months after a TEA for acute trauma, their difference in the median PREE pain scores between TEA and DHH, and their standard deviation (SD) of the PREE pain score after TEA and DHH. The rational independent observer concept was used to achieve consensus. Results: The elicited median value for the PREE pain at 12 months after TEA for acute trauma was 13.6/50 (interquartile range 10.6 to 16.8). The estimated median of the difference in PREE pain scores was 3.4 (interquartile range -0.3 to 7.1). The elicited standard distribution of PREE pain scores was 9 (SD 2.2), and after DHH was 12 (SD 3.0). Conclusion: Probability distributions for pain outcomes measured using the PREE pain subscale following DHH and TEA for acute trauma in adults generated by expert elicitation indicate an expectation of small differences between interventions. A noninferiority randomized trial using the PREE pain subscale as the primary outcome at 12 months could be conducted to explore whether the possible advantages of DHH, such as lower cost and reduced adverse events, are supported.