The Problems of Using Unreliable Evidence in Consensus Group Decision-Making

Abstract

Landoni and colleagues report the work of an International Consensus Group and present evidence-based non-surgical recommendations to reduce perioperative mortality.1Landoni G. Rodseth R.N. Santini F. Ponschab M. Ruggeri L. Székely A. et al.Randomized evidence for reduction of perioperative mortality.J Cardiothorac Vasc Anesth. 2012; 26: 764-772Abstract Full Text Full Text PDF PubMed Scopus (84) Google Scholar The International Consensus Group has recommended that aprotinin should not be used in cardiac surgery and cites as evidence the significant increase in mortality associated with aprotinin compared to the lysine analogues that was reported in the BART study.2Fergusson D.A. Hébert P.C. Mazer D. et al.A comparison of aprotinin and lysine analogues in high-risk cardiac surgery.NEJM. 2008; 358: 2319-2331Crossref PubMed Scopus (830) Google Scholar Ultimately, the quality of any recommendation depends on the quality of the evidence upon which it is made, and evidence from the BART study may be unreliable. Health Canada reviewed and re-analyzed the data from the BART study and, subsequently, so did the European Medicines Agency (EMA). Both drug regulatory agencies found concerning limitations to the published results regarding mortality.3Health Canada. Final report – expert advisory panel on Trasylol (aprotinin) 2011. Available from http://www.hc-sc.gc.ca/dhp-mps/medeff/advise-consult/eap-gce_trasylol/final_rep-rap-eng.php (accessed 9 Oct 2012)Google Scholar, 4European Medicines Agency. European Medicines Agency recommends lifting suspension of aprotinin 2012. Available from http://www.ema.europa.eu/docs/en_GB/document_library/Press_release/2012/02/WC500122914.pdf (accessed 9 Oct 2012)Google Scholar After randomization, there had been an unexplained exclusion of 137 patients from the statistical analysis.3Health Canada. Final report – expert advisory panel on Trasylol (aprotinin) 2011. Available from http://www.hc-sc.gc.ca/dhp-mps/medeff/advise-consult/eap-gce_trasylol/final_rep-rap-eng.php (accessed 9 Oct 2012)Google Scholar Most importantly, the mortality trend of these excluded patients was opposite to that of the included patients.3Health Canada. Final report – expert advisory panel on Trasylol (aprotinin) 2011. Available from http://www.hc-sc.gc.ca/dhp-mps/medeff/advise-consult/eap-gce_trasylol/final_rep-rap-eng.php (accessed 9 Oct 2012)Google Scholar Reanalysis of the data including these originally excluded patients, reduced the mortality signal for aprotinin to statistically nonsignificant.3Health Canada. Final report – expert advisory panel on Trasylol (aprotinin) 2011. Available from http://www.hc-sc.gc.ca/dhp-mps/medeff/advise-consult/eap-gce_trasylol/final_rep-rap-eng.php (accessed 9 Oct 2012)Google Scholar Furthermore, the findings of the BART study have not been replicated; when data from other RCTs were analysed together with exclusion of data from the BART study, aprotinin was not associated with a higher risk of death compared with other antifibrinolytics.4European Medicines Agency. European Medicines Agency recommends lifting suspension of aprotinin 2012. Available from http://www.ema.europa.eu/docs/en_GB/document_library/Press_release/2012/02/WC500122914.pdf (accessed 9 Oct 2012)Google Scholar Given the major inaccuracies in the published paper that have been identified by Health Canada and the EMA, the International Consensus Group should not have considered the BART study to assess aprotinin as it has perversely influenced their decision making. Consensus groups and any other bodies making recommendations about clinical practice need to ensure that they use evidence of a reliable quality. Reply: Summarizing Randomized Evidence With Clinically Relevant Outcomes Performed in the Perioperative PeriodJournal of Cardiothoracic and Vascular AnesthesiaVol. 27Issue 3PreviewWe thank Drs Alston and McMullan1 for their comments and for the opportunity to better explain some aspects of the innovative2 paper that we recently published in JCVA.3 Full-Text PDF Randomized Evidence for Reduction of Perioperative MortalityJournal of Cardiothoracic and Vascular AnesthesiaVol. 26Issue 5PreviewWith more than 220 million major surgical procedures performed annually, perioperative interventions leading to even minor mortality reductions would save thousands of lives per year. This international consensus conference aimed to identify all nonsurgical interventions that increase or reduce perioperative mortality as suggested by randomized evidence. Full-Text PDF