Methods for Evaluating Medical Tests and Biomarkers: Birmingham, UK. 19–20 July 2016

Sue Mallett1, Peggy Sekula2, Douglas G. Altman3, Willi Sauerbrei2 1Institute of Applied Health Research, University of Birmingham, Birmingham, UK; 2Institute for Medical Biometry and Statistics, Faculty of Medicine and Medical Center – University of Freiburg, Freiburg im Breisgau, Germany; 3Centre for Statistics in Medicine, Nuffield Department of Orthopaedics, Rheumatology and Musculoskeletal Sciences, University of Oxford, Oxford, UK Correspondence: Sue Mallett (s.mallett@bham.ac.uk) Background and aim: Every year, thousands of articles are published on prognostic tumour markers often with contradictory results. In 2005, the REMARK guideline for reporting prognostic tumour marker studies was published. For convenience, a check list covering 20 items was provided. A review of tumor marker studies published in 2006-7 demonstrated that many lacked key information needed by readers to evaluate their reliability and clinical applicability [1]. The aim of the current study was to examine whether the quality of reporting has improved in the meantime. Methods: As closely as possible, we used the methods of the earlier review of published articles from the ‘pre-REMARK’ era. This approach includes the utilization of the same data extraction form with questions representing subitems of the original items of the REMARK check list [1]. The literature search for prognostic tumour marker studies was done in Web of Science in 2013. Altogether, we assessed adherence to REMARK for 53 publications (2007 to 12) citing REMARK (‘citing group’) and 53 publications not citing REMARK (‘not-citing group’; matched by journal and issue). Descriptive comparisons over time and between groups were done with a particular focus on 10 items of the REMARK check list. Background and reasons for the restriction to 10 out of 20 items will be provided. Results: Overall, the proportion of the 10 key items that were assessed slightly increased on average from 53% (range: 10% to 90%) in the earlier study to 58% (range: 30% to 100%) in the citing group and to 58% (range: 20% to 100%) in the not-citing group. The improvement, however, was not seen in all 10 items. While an improvement was substantial for some (e.g. item 6: ‘Study design - follow up’; past study: 40%, citing group: 60%, not-citing group: 62%), it got worse for others (e.g. item 13: distribution of demographic characteristics; past study: 58%, citing group: 42%, not-citing group: 55%). Conclusions: In principle, it should be easy to report all study details included in the REMARK checklist. However, our investigation shows that many items are still poorly reported, so there remains much room for improvement. To improve the clinical value of published prognosis research in cancer authors, editors and peer reviewers should be aware of and follow reporting recommendations. References [1] Mallett et al. Reporting of prognostic studies of tumour markers: A review of published articles in relation to REMARK guidelines. Br J Cancer 2010;102:173–80.