What is the appropriate outcome for studies of treatments for out-of-hospital cardiac arrest?

Background: A good outcome following out-of-hospital medical care for cardiac arrest is survival to hospital discharge. Because a large number of patients are required to detect a minimum clinically important difference in survival, an intermediate outcome such as hospital admittance is commonly used. For an intermediate outcome to be a useful surrogate, the survival rate conditional on achieving the intermediate outcome should not depend upon the field treatment. If so, an advantage of the intermediate outcome may be a smaller sample size. However, recent trials demonstrate that survival conditional on admittance may depend upon the field treatment. Even if the resources are available to power a study for survival, is survival the right outcome? For example, no increase in survival and a large increase in admittance could be considered a bad result, as it represents a substantial waste of resources. Similarly no increase in mortality and a decrease in admittance should be considered a good result, as it represents a substantial cost savings without any sacrifice of life. Both admittance and survival are important outcomes and need to be considered jointly, that is, as a bivariate outcome. Methods: Cost-effectiveness concepts are used to distinguish between a good and bad (bivariate) outcome. Simulations are conducted to compare the impact of the univariate and the bivariate outcomes in a variety of trial scenarios. A table of sample sizes is computed for the bivariate outcome across a range of trial scenarios. Results:The bivariate outcome outperforms both univariate outcomes for most alternatives. The required sample size for the joint outcome of admittance and survival may be substantially, over 50%, Less than that for the survival outcome alone. Conclusion: Use of the bivariate outcome could provide more informed decision making about resuscitation strategies and at less cost then the current gold standard of hospital survival. (c) 2006 Elsevier Ireland Ltd. All rights reserved.