• G Sarah Power, David A Harrison, Paul R Mouncey, Tiffany M Osborn, Sheila E Harvey, and Kathryn M Rowan.
• Clinical Trials Unit, Intensive Care National Audit and Research Centre, London, United Kingdom. promise@icnarc.org.
• Crit Care Resusc. 2013 Dec 1; 15 (4): 311-7.

BackgroundThe Protocolised Management in Sepsis (ProMISe) trial is an open, multicentre, randomised controlled trial (RCT) of the clinical effectiveness and cost-effectiveness of early, goal directed, protocolised resuscitation compared with usual resuscitation for patients presenting to emergency departments (EDs) in the United Kingdom with early signs of severe sepsis or septic shock. The rationale for the ProMISe trial derives from a single-centre United States RCT that reported a reduction in hospital mortality from 46.5% to 30.5%.ObjectiveTo describe the proposed statistical analyses for the evaluation of clinical effectiveness for the ProMISe trial. It is important to complete this plan before inspecting the data, and before completion of two related international studies, so that post-hoc, data-derived decisions are avoided.MethodsThe primary and secondary outcomes were defined precisely, and the approach to safety monitoring and data collection summarised, with a description of the planned statistical analyses including prespecified subgroup and secondary analyses.ResultsThe primary outcome is all-cause mortality at 90 days. The primary analysis will be reported as a relative risk and absolute risk reduction and tested with the Fisher exact test. Prespecified subgroup analyses will be based on age, baseline Medical Emergency Department Sepsis score, baseline Sequential Organ Failure Assessment score, and time from ED presentation to randomisation. Secondary analyses include adjustment for baseline covariates, estimation of learning curve effects and adjustment for noncompliance.ConclusionIn keeping with best practice, we have developed a statistical analysis plan for the ProMISe trial and place it in the public domain before inspecting data from the trial.

19 keywords...

hide…