1 Deere KC, et al. BMJ Open 2019;9:e026736. doi:10.1136/bmjopen-2018-026736 Open access Assessing the non-inferiority of prosthesis constructs used in total and unicondylar knee replacements using data from the National Joint Registry of England, Wales, Northern Ireland and the Isle of Man: a benchmarking study Kevin C Deere,  1 Michael R Whitehouse,  1,2 Martyn Porter, 3 Ashley W Blom, 1,2 Adrian Sayers  1,4 To cite: Deere KC, Whitehouse MR, Porter M, et al. Assessing the non-inferiority of prosthesis constructs used in total and unicondylar knee replacements using data from the National Joint Registry of England, Wales, Northern Ireland and the Isle of Man: a benchmarking study. BMJ Open 2019;9:e026736. doi:10.1136/ bmjopen-2018-026736 ► Prepublication history and additional material for this paper are available online. To view these fles, please visit the journal online (http://dx.doi. org/10.1136/bmjopen-2018- 026736). Received 19 September 2018 Revised 25 January 2019 Accepted 25 March 2019 For numbered affliations see end of article. Correspondence to Adrian Sayers; adrian.sayers@bristol.ac.uk Research ► http://dx.doi.org/10.1136/ bmjopen-2018-026685 © Author(s) (or their employer(s)) 2019. Re-use permitted under CC BY. Published by BMJ. ABSTRACT Objectives To investigate the relative performance of knee replacement constructs compared with the best performing construct and illustrate the substantial variability in performance. Design A non-inferiority study. Setting England and Wales. Participants All primary total and unicondylar knee replacements performed and registered in the National Joint Registry between 1 April 2003 and 31 December 2016. Main outcome measures Kaplan-Meier failure function for knee replacement constructs. Failure difference between best performing construct (the benchmark) and other constructs. Methods Using a non-inferiority analysis, the performance of knee replacement constructs by brand were compared with the best performing construct. Construct failure was estimated using the 1-Kaplan Meier method, that is, an estimate of net failure. The difference in failure between the contemporary benchmark construct and all other constructs were tested. Results Of the 449 different knee replacement constructs used, only 27 had ≥500 procedures at risk at 10 years postprimary, 18 of which were classifed as inferior to the benchmark by at least 20% relative risk of failure. Two of these 18 were unicondylar constructs that were inferior by at least 100% relative risk. In men, aged 55–75 years, 12 of 27 (44%) constructs were inferior by at least 20% to the benchmark at 7 years postprimary. In women, aged 55–75 years, 8 of 32 (25%) constructs were inferior at 7 years postprimary. Very few constructs were classifed as non- inferior to the contemporary benchmark. Conclusions There are few knee replacement constructs that can be shown to be non-inferior to a contemporary benchmark. Unicondylar knee constructs have, almost universally, at least 100% worse revision outcomes compared with the best performing total knee replacement. These results will help to inform patients, clinicians and commissioners when considering knee replacement surgery. INTRODUCTION Over 90 000 knee replacements are performed annually in England and Wales, and there is a bewildering choice of total and unicon- dylar knee replacement (UKR) options avail- able from which clinicians and patients can choose. When patients are considering a knee replacement, it is understandable that many assume that the different constructs function equally. However, all constructs are not equal as evidenced by variation in revision rates between brands and knee construct types. 1 The National Joint Registry for England, Wales, Northern Ireland and the Isle of Man (NJR) is the largest arthroplasty database in the world and publishes the unadjusted cumu- lative failure rates of the most commonly used constructs in knee replacement surgery. This is a useful format for measuring absolute failure but does not allow easy head-to-head Strengths and limitations of this study ► Data presented from the largest joint registry in the world. ► A novel and systematic comparison of the perfor- mance of knee constructs to a contemporary bench- mark knee construct. ► Unambiguous presentation of data will facilitate the consenting process for patients and allows surgeons and policy makers to be more informed with respect to success and failure of different constructs options available in knee replacement. ► Residual and unmeasured confounding factors are likely to be present. ► Potential for selection bias whereby certain con- structs may be implanted for particular indications and in particular groups with different risks.