Arne Schröder1,2, Oliver J Muensterer3,4, Christina Oetzmann von Sochaczewski5,6. 1. Klinik für Kinder- und Jugendmedizin, Elisabeth-Krankenhaus, Essen, Germany. 2. Klinik für Kinder- und Jugendmedizin, Klinikum Dortmund, Dortmund, Germany. 3. Klinik und Poliklinik für Kinderchirurgie, Universitätsmedizin der Johannes-Gutenberg-Universität, Langenbeckstraße 1, 55131, Mainz, Germany. 4. Kinderchirurgische Klinik und Poliklinik, Dr. von Haunersches Kinderspital, Ludwig-Maximilians-Universität, München, Germany. 5. Klinik und Poliklinik für Kinderchirurgie, Universitätsmedizin der Johannes-Gutenberg-Universität, Langenbeckstraße 1, 55131, Mainz, Germany. c.oetzmann@gmail.com. 6. Sektion Kinderchirurgie der Klinik für Allgemein-, Viszeral-, Thorax- und Gefäßchirurgie, Universitätsklinikum Bonn, Bonn, Germany. c.oetzmann@gmail.com.
Abstract
PURPOSE: Meta-analyses occupy the highest level of evidence and thereby guide clinical decision-making. Recently, randomised-controlled trials were evaluated for the robustness of their findings by calculating the fragility index. The fragility index is the number of events that needs to be added to one treatment arm until the statistical significance collapses. We, therefore, aimed to evaluate the robustness of paediatric surgical meta-analyses. METHODS: We searched MEDLINE for paediatric surgical meta-analyses in the last decade. All meta-analyses on a paediatric surgical condition were eligible for analysis if they based their conclusion on a statistically significant meta-analysis. RESULTS: We screened 303 records and conducted a full-text evaluation of 60 manuscripts. Of them, 39 were included in our analysis that conducted 79 individual meta-analyses with significant results. Median fragility index was 5 (Q25-Q75% 2-11). Median fragility in relation to included patients was 0.77% (Q25-Q75% 0.29-1.87%). CONCLUSION: Paediatric surgical meta-analyses are often fragile. In almost 60% of results, the statistical significance depends on less than 1% of the included population. However, as the fragility index is just a transformation of the P value, it basically conveys the same information in a different format. It therefore should be avoided.
PURPOSE: Meta-analyses occupy the highest level of evidence and thereby guide clinical decision-making. Recently, randomised-controlled trials were evaluated for the robustness of their findings by calculating the fragility index. The fragility index is the number of events that needs to be added to one treatment arm until the statistical significance collapses. We, therefore, aimed to evaluate the robustness of paediatric surgical meta-analyses. METHODS: We searched MEDLINE for paediatric surgical meta-analyses in the last decade. All meta-analyses on a paediatric surgical condition were eligible for analysis if they based their conclusion on a statistically significant meta-analysis. RESULTS: We screened 303 records and conducted a full-text evaluation of 60 manuscripts. Of them, 39 were included in our analysis that conducted 79 individual meta-analyses with significant results. Median fragility index was 5 (Q25-Q75% 2-11). Median fragility in relation to included patients was 0.77% (Q25-Q75% 0.29-1.87%). CONCLUSION: Paediatric surgical meta-analyses are often fragile. In almost 60% of results, the statistical significance depends on less than 1% of the included population. However, as the fragility index is just a transformation of the P value, it basically conveys the same information in a different format. It therefore should be avoided.
Entities:
Keywords:
Fragility index; Fragility quotient; Meta-analysis; P value; Uninformative statistic
Authors: Goutham Rao; Francisco Lopez-Jimenez; Jack Boyd; Frank D'Amico; Nefertiti H Durant; Mark A Hlatky; George Howard; Katherine Kirley; Christopher Masi; Tiffany M Powell-Wiley; Anthony E Solomonides; Colin P West; Jennifer Wessel Journal: Circulation Date: 2017-08-07 Impact factor: 29.690
Authors: Michael Walsh; Sadeesh K Srinathan; Daniel F McAuley; Marko Mrkobrada; Oren Levine; Christine Ribic; Amber O Molnar; Neil D Dattani; Andrew Burke; Gordon Guyatt; Lehana Thabane; Stephen D Walter; Janice Pogue; P J Devereaux Journal: J Clin Epidemiol Date: 2014-02-05 Impact factor: 6.437