An extension of STARD statements for reporting diagnostic accuracy studies on liver fibrosis tests: The Liver-FibroSTARD standards

Abstract

Chronic liver diseases are highly prevalent and require an accurate evaluation of liver fibrosis to determine patient management. Over the last decade, great effort has been made to develop non-invasive liver fibrosis tests. The ensuing increase of literature is, however, impaired by extensive heterogeneity in the quality of published reports. The Standards for Reporting of Diagnostic Accuracy Studies (STARD), first published in 2003, were developed to improve the quality of research reports on diagnostic studies. We aimed to evaluate STARD statements in the setting of diagnostic studies on non-invasive liver fibrosis tests, and to propose an extended version developed specifically for those studies. Eight French experts evaluated STARD statement adequacy in 10 studies on non-invasive liver fibrosis tests and then developed an extended version with a glossary. The new checklist and glossary were independently evaluated by seven international experts. Fourteen of the 25 STARD items were considered only partially adequate for the evaluation of diagnostic studies on non-invasive liver fibrosis tests. Inter-expert agreement was at least very good for 8 STARD items (32%), moderate for 9 (36%), and poor or very poor for 8 (32%). The experts' proposals were developed into the new Liver-FibroSTARD standards including a checklist with 62 items/sub-items and a corresponding comprehensive glossary. New proposals were inserted in the 25 STARD items as a complementary module. Independent evaluation of the Liver-FibroSTARD checklist showed at least very good inter-expert agreement for 39 items/sub-items (63%), moderate agreement for 11 (18%), and poor or very poor agreement for only 12 (19%). As a supplement of the STARD statements, the Liver-FibroSTARD checklist and its glossary are new tools specifically designed for the evaluation of diagnostic studies about non-invasive liver fibrosis tests.