Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
The Downside of Publication Group 2: Jing Kai, Nabilah and Soon Guan Presenter: Soon Guan Reading: Copas, J. (2005), The downside of publication. Significance, 2: 154–157. doi:10.1111/j.17409713.2005.00127. 1 Content ● Publication ● Empirical Science ● Issue with publication - Empirical Science and Selection ● Case Study: Paxil (Paroxetine) and GSK ● Conclusion 2 Publication - Publication: Bringing new ideas and knowledge out into the scientific community and the public. - Articles and studies are put through rigorous rounds of checks and review. Source:http://www.iosrphr.org/images/publication%20process.jpg 3 Publication - The Result - The system sieves through the works and studies of many researchers. - The content and credibility of the articles that are published are reassured and checked by editors and peer reviewed. - Only those that made the cut are selected. http://orig03.deviantart.net/0469/f/2010/225 /1/0/joey__s_rattata_by_spiffee.png 4 What’s the issue then? 5 Problem - Selection in Empirical Science - Empirical Sciences: Evidence based science. - Examples: Medicine, Clinical Trials and experimental studies. - Often involve hypothesis testing to test for statistical significance of the results. 6 Cause of Selection - The complication does not stem from the quality of the author’s work, but the process that it is going to be subjected to. - Not all studies that can be/ supposed to be published are published. 7 Need to publish articles that warrant sufficient interest Feedback: Unlikely to publish negative results Selectively send studies that are likely to be approved Terminate studies that show negative results Considerations: Journal’s Reputation whether it will bring merit in publishing a certain article Considerations: Unlikely to publish negative results that have little impact 8 Result of Selection - Investigators might be discouraged by negative results (e.g. failing to reject null hypothesis), terminating the studies in advance. “File Drawer Problem” - Current Incentive system: No merits for scientist to publish negative results. - Likewise they would only submit studies that are likely to be published. https://oikosjournal.files.wordpress.com/2014/05/data_ 01.jpg 9 Publication Bias: Authors submitting findings that are likely to be publish. Publishers selecting studies that are worthy to be publish. 10 Hypothetical Example - In Clinical Trial - Suppose I want to find whether Drug X works in treating depression. - I conduct the same study with the same experiment design for 100 times. - Result: 30% Negative Results, 40% No effect (or difference), 30% Positive Results Conclusion: Drug is ineffective 11 Biases - Author publishing positive results, publishers approving study worth of publication… etc. Studies Conducted Published Studies : Those we can see. 12 Result of Publication Bias Negative and ‘Neutral’ Studies published - As a result, the proportion of positive effects was inflated: from 30% to 80%. - What is presented is not representative of the actual scenario. - Because of biases, the effectiveness of a treatment can be significantly altered. Published Studies 13 Result of Publication Bias Original Distribution Observed by all 100 studies. Result: Normally Distributed success rate Drug is ineffective 14 Result of Publication Bias Original Distribution Observed by all 100 studies. Result: Drug is ineffective Published Results Strong evidence showing that the drug/treatment is effective When physician review the published results randomly, it reflects only part of the truth. 15 Case Study -Paxil 16 Case Study - Paxil (Paroxetine) - Paxil (Paroxetine) – Marketed in 1990s - Treatment of depression and anxiety problem. - Antidepressant that works by selectively inhibiting reuptake of serotonin (SSRI). - Manufactured by GlaxoSmithKline (GSK). http://www.depressionhealth.net/wpcontent/uploads/2014/06/paxil-paroxetine10mg-20mg-30mg-40mg.jpg http://emit.medschl.cam.a c.uk/wpcontent/uploads/2014/07/ GSK_LOS_RGB.jpg 17 Case Study - Paxil (Paroxetine) - The efficacy and safety of the drug was unknown in Adolescent. - The Product Leaflet states that it is not recommended in children. - Paxil is known as Seroxat in the UK 18 Case Study - Paxil (Paroxetine) - GSK conducted two trials, Study 329 and Study 377, both failed to show Paxil was effective in treating depressive disorder in children. - No amendments were made to the product summary despite these findings. - “It would be commercially unacceptable to include a statement that efficacy had not been demonstrated, as it will undermine the profile of paroxetine” GSK Internal Management. - 32,000 Paxil prescriptions were already issued to children in UK alone. 19 Case Study - Paxil (Paroxetine) - Negative effect of the trial was downplayed and not included in the conclusion when published. - Further analysis showed children taking Paxil are 1.5 to 3.2 times more likely to exhibit suicidal behaviour than those that are taking placebo. - This finding was not published as well. 20 Findings published on BMJ on Paxil "The published manuscript was biased in its conclusions, made unsubstantiated efficacy claims and downplayed the adverse-event profile of Paxil," – The British Medical Journal 21 Case Study - Paxil (Paroxetine) “ The United States alleges that, among other things, GSK participated in preparing, publishing and distributing a misleading medical journal article that misreported that a clinical trial of Paxil demonstrated efficacy in the treatment of depression in patients under age 18, when the study failed to demonstrate efficacy.” “GSK did not make available data from two other studies in which Paxil also failed to demonstrate 22 efficacy in treating depression in patients under 18.” Message - Publications are put through rigorous checks and review to ensure its quality. - The flaw of the system stems from the nature of how it works and biases are introduced which distorts the actual statistical results. - Case Study: GSK and Paxil. - Important for us to be aware of the truth. 23 http://cdn.quotationof.com/images/tip-of-the-iceberg-quotes-1.jpg 24 References Reading Article: Copas, J. (2005), The downside of publication. Significance, 2: 154–157. doi:10.1111/j.1740-9713.2005.00127.x Paxil: Article Source: The Efficacy of Paroxetine and Placebo in Treating Anxiety and Depression: A Meta-Analysis of Change on the Hamilton Rating Scales Sugarman MA, Loree AM, Baltes BB, Grekin ER, Kirsch I (2014) The Efficacy of Paroxetine and Placebo in Treating Anxiety and Depression: A Meta-Analysis of Change on the Hamilton Rating Scales. PLOS ONE 9(8): e106337. doi: 10.1371/journal.pone.0106337 Investigation of Paxil: http://webarchive.nationalarchives.gov.uk/20141205150130/http://www.mhra.gov.uk/home/groups/espolicy/documents/websiteresources/con014155.pdf Paxil Settlement: https://www.justice.gov/opa/pr/glaxosmithkline-plead-guilty-and-pay-3-billion-resolve-fraud-allegations-and-failurereport 25 The End Thank you! :) 26