Céline Chapelle1,2, Edouard Ollier1,2, Philippe Girard3,4, Corinne Frere5, Patrick Mismetti1,2,4,6, Michel Cucherat7, Silvy Laporte1,2,4. 1. Unité de Recherche Clinique, Innovation, Pharmacologie, CHU Saint-Etienne, Hôpital Nord, Saint-Etienne, France. 2. SAINBIOSE U1059, Université Jean Monnet, Univ. Lyon, INSERM, Saint-Etienne, France. 3. Institut Mutualiste Montsouris, Paris, France. 4. F-CRIN INNOVTE network, Saint Etienne, France. 5. Sorbonne Université, INSERM UMRS1166, ICAN - Institute of CardioMetabolism and Nutrition, Hôpital Pitié-Salpêtrière, Paris, France. 6. Service de Médecine Vasculaire et Thérapeutique, CHU Saint-Etienne, Hôpital Nord, Saint-Etienne, France. 7. Service de Pharmacologie, HCL, UMR CNRS 5558 Evaluation et Modélisation des Effets Thérapeutiques, Université Claude Bernard Lyon 1, Lyon, France.
Abstract
BACKGROUND: Meta-analyses are widely used to strengthen available evidence and obtain more precise estimates of treatment effect than any individual trial. Paradoxically, multiplication of meta-analyses on the same topic can lead to confusion as practitioners no longer benefit from a rapid and synthetic response. This phenomenon may appear disproportionate when the number of published meta-analyses exceeds the number of original studies. OBJECTIVES: To describe an example of redundant meta-analyses published in the same area with the same randomized clinical trials (RCTs). METHODS: A systematic review was performed to identify all published meta-analyses of original RCTs that compared direct oral anticoagulants with low molecular weight heparins in cancer patients with venous thromboembolism (VTE). Forest plots were used to represent the meta-analyses results for efficacy (VTE recurrence) and safety (major bleeding) endpoints. An authors' network was constructed to explore the links between the authors of the published meta-analyses. RESULTS: In the past 3 years, four original RCTs were the subject of 20 published meta-analyses by 142 authors: five, four, and 11 meta-analyses pooled the data of two, three, and four RCTs, respectively. The results of meta-analyses were similar regarding the risks of VTE recurrence and major bleeding. The 11 meta-analyses of four RCTs were published within 6 months of the publication of the last RCT. CONCLUSIONS: The epidemic proportions of such redundant literature and authorship could be moderated by developing "living" meta-analyses and encouraging authors of new RCTs to update the corresponding meta-analysis in the same paper as their original research.
BACKGROUND: Meta-analyses are widely used to strengthen available evidence and obtain more precise estimates of treatment effect than any individual trial. Paradoxically, multiplication of meta-analyses on the same topic can lead to confusion as practitioners no longer benefit from a rapid and synthetic response. This phenomenon may appear disproportionate when the number of published meta-analyses exceeds the number of original studies. OBJECTIVES: To describe an example of redundant meta-analyses published in the same area with the same randomized clinical trials (RCTs). METHODS: A systematic review was performed to identify all published meta-analyses of original RCTs that compared direct oral anticoagulants with low molecular weight heparins in cancer patients with venous thromboembolism (VTE). Forest plots were used to represent the meta-analyses results for efficacy (VTE recurrence) and safety (major bleeding) endpoints. An authors' network was constructed to explore the links between the authors of the published meta-analyses. RESULTS: In the past 3 years, four original RCTs were the subject of 20 published meta-analyses by 142 authors: five, four, and 11 meta-analyses pooled the data of two, three, and four RCTs, respectively. The results of meta-analyses were similar regarding the risks of VTE recurrence and major bleeding. The 11 meta-analyses of four RCTs were published within 6 months of the publication of the last RCT. CONCLUSIONS: The epidemic proportions of such redundant literature and authorship could be moderated by developing "living" meta-analyses and encouraging authors of new RCTs to update the corresponding meta-analysis in the same paper as their original research.