The effect of uncertainty on ‘risk rationalizing’ decisions

The forms and contexts in which risk analytical methods can provide useful inputs to policy decisions remains an open question. This paper assesses the role of uncertainty in cost- effectiveness estimation using the example of life-saving interventions available to regulators in the United States, as calculated by Tengs et al . ( Risk Analysis 15(3), 369--90, 1995). It identifies ‘equally plausible’ values for those interventions, based on alternative assumptions about costs and benefits. These alternative values suggest that in no case can credible point estimates indicate more than order of magnitude precision, and in the worst case, plausible point estimates range from infinite cost to net benefit for a single intervention. Some but not all of this uncertainty is irreducible. This suggests that decisions based on point-estimate cost-effectiveness calculations can give a false impression of rational, evidence-based policy. In such cases, risk assessment based decisions are ‘systematically arbitrary.’ The analysis nonetheless suggests a role for cost effectiveness analysis in rejecting interventions that under all assumptions appear extremely costly, and promoting those that in all cases appear extremely inexpensive. Finally, it affirms that risk analysis is useful and appropriate as a tool for understanding complex problems, a role that could be undermined by excessive attention to calculating point estimates.