The value of evidence-based risk assessments

Abstract

sumption for both of these standards that clinicians have an accurate knowledge of the relevant risks. As Rid and Wendler demonstrate, this assumption is likely false with respect to many common interventions, primarily because the data have not been collected, or exist in scattered formats that limit firm conclusions. The author’s assertion that disclosures are therefore based on intuition may be an overstatement, but individual or institutional experience is only a modest improvement. The point is that Rid and Wendler’s proposal would improve our understanding of risks associated with common clinical interventions and lead to productive management of risk reduction and improved risk communication in clinical medicine. Decision analysis suggests that people evaluate choices based on the probability of potential outcomes and the value placed on those outcomes. The approach advocated by Rid and Wendler is primarily focused on the probability component of this equation. A key question is whether changes in our understanding of the probabilities of adverse outcomes from certain research interventions will change or improve decisions made. From the investigator’s and/or an IRB reviewer’s perspectives, the question is whether additional data would alter the general ‘intuitive’ assessments that are commonly made about risk. In fact, intuitive estimates may prove to be accurate enough since we live our lives with poorly quantified risks. I believe that the risks of my morning drive to work are low and that the benefits outweigh the risks. It is unlikely that a detailed set of statistics on roadway morbidity and mortality for my route would change my conclusion. However, data might convince me to be more cautious when driving conditions are demonstrably more dangerous. Similarly, data on research procedural risks might be most useful when they reveal particularly vulnerable subpopulations of research participants. For the IRB reviewer, risks often are categorical, that is, ‘no more than minimal risk’ or ‘greater than