Objectives: This study reports the cost-effectiveness of interventions with nonsignificant differences in effect, and considers reporting of cost-effectiveness in situations where nonsignificant differences arise in some but not all end points.

Methods: Data on costs and effects associated with three end points (adequate assessment, risk factors, and life-years) were derived from a trial of methods to promote secondary prevention of coronary heart disease. Incremental cost per life-year gained figures were calculated, and the uncertainty around these was displayed on cost-effectiveness planes in the form of ellipses.

Results: There was a significant difference in one of the intermediate end points (adequate assessment) but nonsignificant differences in the other intermediate end point (risk factors) and the final end point (life-years). Estimation of cost per life-year figures revealed the cost-effectiveness of the interventions to be unfavorable.

Conclusions: Cost-effectiveness ratios based on final end points should be calculated even in situations where nonsignificant differences in life-years arise, to avoid publication bias and to provide decision makers with useful information. Uncertainty in the incremental cost-effectiveness ratios should be estimated and presented graphically.We thank Rebecca Turner for comments on an earlier draft; Professor Paul Fenn for permission to adapt an existing model; and Dr. Andy Briggs for advice on interpretation and presentation of results and comments on an earlier draft. In addition to the authors of the paper, the ASSIST trial collaborative group includes G. Fowler, E. Fullard, A. Fuller, M. Murphy, A. Neil, T. Schofield, R. Turner, S. Thompson, F. Wells, R. Wiles, and L. Youngman. This research was funded by a grant from the national research and development program grant of the NHS Executive.