Future research in emergency medicine: explanation or pragmatism? Large or small? Simple or complex?

BACKGROUND Recent publications have indicated that there is a crisis in clinical research. The cost of an industry-led clinical trial is now estimated at some US$100e200 million, within which an average 25% of centres will recruit no patients. The median number of enrolment (entry) criteria in a drug trial is 49, with a median of 158 different trial procedures. The number of data items per patient often runs into thousands, so it is little surprise that the average time to trial completion increased 70% between 1999 and 2006. There seems to be a self-sustaining ‘industry’ of bureaucracy around research. This system generates huge costs and a mountain of paperwork, which in turn makes the employment of trial managers essential. These costs are passed on to the industry or public funders of medical research. Regulatory rules that are designed for ‘pharma’ trials of new drugs are also applied to ‘investigator-led’ trials, even if the drug involved is already in use and has a very well-known safety profile. Against this background, as our specialty evolves, it is a good time for us to ask how emergency medicine research fits in the complex and competitive world of clinical research. Our specialty is a broad church, with emergency physicians having a legitimate research interest in almost all areas of medicine. We cannot be experts in the science of ‘everything’, but we can be experts in a particular type of methodology, which can then be applied to many different areas. The question ‘What type of primary research is best suited to emergency medicine?’ is key to the future development of the academic part of our specialty. CURRENT SITUATION Overall, the clinical trials that are performed in emergency medicine tend to be underpowered. A good example is the analysis of published trials on interventions in head injury. The authors showed that a sample of 400 patients is needed to show a 10% absolute risk reduction in the chance of death. As the mortality after severe injury is about 20%, an Absolute Risk Reduction ARR of 10% (halving the mortality) would be an impossibly fantastic advance in trauma caredyet almost all of the published trials were smaller than this, so they were not even powered to detect an incredibly large treatment effect. Only one published clinical trial in trauma care had the power to show a clinically meaningful mortality differencedthe Corticosteroid Randomization After Significant Head Injury (CRASH) trial of steroids in head injury (powered to show a 1% ARR). The 195 trials that had a sample size below 320 and used a mortality end point were in effect meaningless and illustrate the waste of resources that occurs when undertaking small complex trials. It would surely be better to cooperate and discover a definitive answer to one question rather than to spend time and resources finding inconclusive results for many questions.