Hamdi Boubaker, Bertrand Guidet, Philippe Aegerter, Yann-Erick Claessens, Alain Cariou, Service des urgences, Centre Hospitalier Princesse Grace, Hôpital Cochin [AP-HP], Assistance publique - Hôpitaux de Paris (AP-HP) (AP-HP), Laboratoire Universitaire Santé Environnement Vieillissement, Université de Versailles Saint-Quentin-en-Yvelines (UVSQ), Service de Santé Publique, Assistance publique - Hôpitaux de Paris (AP-HP) (AP-HP)-Hôpital Ambroise Paré [AP-HP], Department of Emergency Medicine, CHU Fattouma Bourguiba [Monastir] (HFB), CHU Saint-Antoine [AP-HP], Assistance publique - Hôpitaux de Paris (AP-HP) (AP-HP)-Sorbonne Université (SU), Epidémiologie des maladies infectieuses et modélisation (ESIM), Université Pierre et Marie Curie - Paris 6 (UPMC)-Institut National de la Santé et de la Recherche Médicale (INSERM), Cub-Rea Network, BMC, Ed., CHU Cochin [AP-HP], Université de Versailles Saint-Quentin-en-Yvelines ( UVSQ ), Assistance publique - Hôpitaux de Paris (AP-HP)-Hôpital Ambroise Paré, Centre Hospitalier Universitaire Fatouma Bourguiba, Service de Soins Intensifs, Assistance publique - Hôpitaux de Paris (AP-HP)-CHU Saint-Antoine [APHP], Epidémiologie des maladies infectieuses et modélisation ( ESIM ), and Université Pierre et Marie Curie - Paris 6 ( UPMC ) -Institut National de la Santé et de la Recherche Médicale ( INSERM )
Introduction Guidelines dealing with severe sepsis and septic shock mostly rely on randomized controlled trials (RCTs) to ensure the best standards of care for patients. However, patients included in high-quality studies may differ from the routine population and alter external validity of recommendations. We aimed to determine to what extent non-inclusion criteria of RCTs dealing with severe sepsis and septic shock may affect application of their conclusions in routine care. Methods In a first step, the MEDLINE database was searched for RCTs treating severe sepsis and septic shock patients between 1992 and 2008, and non-inclusion criteria for these studies were abstracted. Two reviewers independently evaluated the articles, which were checked by a third reviewer. We extracted data on the study design, main intervention, primary endpoint, criteria for inclusion, and criteria for non-inclusion. In a second step, the distribution of the non-inclusion criteria was observed in a prospective multicenter cohort of severe sepsis and septic shock patients (Cub-Rea network, 1992 to 2008). Results We identified 96 articles out of 7,012 citations that met the screening criteria. Congestive heart failure (35%) and cancer (30%) were frequent exclusion criteria in selected studies, as well as other frequent disorders such as gastrointestinal and liver diseases and all causes of immune suppression. Of the 67,717 patients with severe sepsis and septic shock in the Cub-Rea database, 40,325 (60%) experienced at least one of the main exclusion criteria, including 11% of congestive heart failure patients and 11% of cancer patients. In addition, we observed a significant trend for increasing number of patients with these criteria along time. Conclusion Current exclusion criteria for RCTs dealing with severe sepsis and septic shock excluded most patients encountered in daily practice and limit external validity of the results of high-quality studies.