Creation of an Australian Osteotomy Knee Registry Datasheet Instrument using a Modified Delphi Method

Background: In 2015, the Australian Knee Society(AKS) agreed to initiate and fund an Australian Knee Osteotomy Registry (AKOR) in 2015 in conjunction with the Australian Orthopaedic Association National Joint Replacement Registry (AOANJRR). A concise and robust dataset was required to proceed to AKOR preliminary site trialing. Study Design: Modified Delphi consensus building. Methods: Seven participants who were orthopaedic surgeons, AKS members and AKOR working group members performed a four round modified Delphi consensus building electronic non-blinded questionnaires using Google Sheets, a cloud-based spreadsheet. Round 1 involved the attribute rating using 0-10 of surgeon assessed utility of 41 initial datasheet instrument questions with 168 stem options. Surgeons could comment on each question or stem, giving comments on their rating and or proposing new variations or additions. Round 1 questions were based on the Swedish HTO Registry dataset with additional questions from surgeon interview. Questions and or stems with a mean utility rating of less than 2 were excluded. Questions and or stems with a mean utility raking of 8 or greater were accepted. Questions and or stem with a mean utility ranking of between 8 and 2 were included in subsequent rounds. Surgeons comments and votes were non-blinded. New or alternative questions or stems could be proposed by any surgeon during the process and voted on in the next round. This process was repeated until dataset consensus was achieved. Results: A final dataset of 32 questions was created after 4 rounds. Consensus was reached on the following items; patients demographics, hospital, state, surgeon’s code, date, primary or re-operation, diagnosis at primary procedure, diagnosis at re-operation, type of primary osteotomy, type of re-operation, form of fixation, bone graft, coincidental surgery, approach to correction calculation, pre-operative status of ACL, pre-operative status of PCL, previous knee surgery, preoperative mechanical axis, planned post-operative mechanical axis, preoperative fixed flexion deformity, pre-operative Ahlbäck grading of knee OA. Questions that had a final utility rating less than 10 included; coincidental surgery, approach to correction calculation, pre-operative Ahlbäck grading of knee OA, planned post-operative mechanical axis and preoperative fixed flexion deformity. Conclusion: The final AKOR dataset requires pragmatic validation prior to nation-wide implementation. Questions with low final attribute rating may not be included in the final instrument after validatation.