Evidence-based orthopaedic surgery: what type of research will best improve clinical practice?

©2005 British Editorial Society of Bone and Joint Surgery doi:10.1302/0301-620X.87B12. 17085 $2.00 J Bone Joint Surg [Br] 2005;87-B:1593-4. Surgical research is often maligned. 1 Nevertheless, clear improvements in the surgical management of disorders of bone and joint have occurred since the advent of evidence-based medicine 30 years ago. Examples include joint replacement or arthroplasty surgery, internal fixation of fractures and arthroscopic or ‘minimally invasive’ surgery. Refinements and improvements continue to emerge further reducing complications and morbidity. Could we do better? Much of the published research associated with these changes is, according to the system suggested by the Oxford Centre for Evidence-based Medicine Framework, of poor or low quality. 2 A recent review of published evidence for shoulder surgery revealed that, during a ten-year period to 2002, only 19 (3.1%) papers in mainstream orthopaedic journals described randomised, controlled trials (RCTs) and 538 (88.2%) were single-centre case series offering low levels of evidence according to this system. 3 Most of the RCTs involved non-surgical treatments such as physiotherapy or drugs. A similar picture is probably true for most other areas of published orthopaedic research. What are the reasons for this deficiency? Some would say that orthopaedic surgeons have, at least in part, failed to embrace modern methods of research and design of trials. To a certain extent, this is true but the essence of surgical innovation and improvement has not been by clinical trials but by reliance on individual surgeons inventing new techniques or new implants and reporting their results as a case series to see if improved clinical outcomes have ensued. The fundamental issue is that the surgeon is part of the treatment and is, generally speaking, responsible for its innovation and development. The surgeon, not a third party such as a pharmaceutical company or a university researcher in a laboratory, is the inventor. His or her decision making and technical skill are intimately bound up in its success or failure. To expect an individual surgeon to be equally expert at a number of different techniques is unrealistic. With RCTs, the consent process has proved particularly troublesome and many surgeons find it difficult to participate in the randomisation process without bias. Many surgical RCTs have high rates of exclusion or withdrawal. It is perhaps not surprising that conventional RCTs of surgical treatments have proved very difficult to set up and complete. This ‘expertise bias’ is well recognised. 4