Dear Editors, We thank Dr. Jacobs for the interest in our article ‘‘Clinical and radiological outcomes of fixedversus mobile-bearing total knee replacement: a meta-analysis’’ [26] published in this journal. We thank him for the comments regarding the approach and presentation which we adopted in order to appropriately answer this controversial research question. Regarding the points that were raised: Firstly, he rightly acknowledged that the data from retrospective studies was included in this paper’s analysis. It is acknowledged that retrospective data can be subject to methodological limitations such as selection and allocation bias [5]. However, by not including all pertinent studies relating to the research question, such a meta-analysis would have exhibited selection bias [11]. By excluding studies based on quality and methodological rigour, it is true that the authors may be able to make more confident inferences on the results [5]. Nonetheless, by undertaking such a method, the conclusions drawn would not represent those findings of the evidence-base as a whole. This later factor was the overwhelming reason for undertaking a ‘best-evidence synthesis’ [24, 25]. The primary objective of this study was to compare the clinical and radiological differences of fixedand mobilebearing total knee replacement (TKR). Whilst previous studies have reported a difference in outcome between fixed-bearing TKR such as posterior cruciate sacrificing and retaining prostheses [20], an assessment of this was not the primary aim of this study. However, with this in mind, we detailed which specific implant used, using all the information provided from each paper. As Dr. Jacobs correctly points out, in a small number of papers (three out of 33), a non-posterior stabilising mobile-bearing prosthesis was compared to a posterior-stabilising fixed comparator [1, 6, 32]. Whilst this may provide some heterogeneity in the type of implant used, and therefore may underestimate a treatment effect between the two groups, the papers reviewed largely provided limited information as to the specific implants used. In response to this, it would have been inappropriate, with such insufficient information, to perform a formal subgroup analysis based on whether prostheses were cruciate retaining or sacrificing or whether the posterior cruciate-sacrificing implants were the ‘postand-cam’ variety. This, nonetheless, may be an area for further evidence synthesis to determine the potential importance of this specific type of cruciate-sacrificing implant. Dr. Jacobs rightly advocates the inclusion of unpublished or ‘grey’ literature used in our study. It is commonly considered important to include unpublished results in order to create an unbiased sampling frame for meta-analysis [22]. By not including all published and unpublished literature, the scope of a literature search is thereby reduced. Consequentially, the conclusions drawn from such reviews would be biased [11]. Dr. Jacobs should not be mislead over the simplicity of the search strategy used to assess the published or ‘white’ literature. T. O. Smith (&) School of Allied Health Professions, University of East Anglia, Norwich NR4 7TJ, UK e-mail: toby.smith@uea.ac.uk