BackgroundThe use of utility values in cost-effectiveness analysis is an accepted method for defining outcomes. Increasingly, cost-effectiveness analyses examine outcomes that involve two or more health states. There is no accepted method of mathematically combining single health state utility values into a surrogate value that represents the combined health state.ObjectiveTo test the effect of different mathematical approaches to combining single health state utility values into a surrogate value on the cost-effectiveness ratio, in a sample model.MethodsWe employed a realistic decision analysis model to test the cost-effectiveness of screening for postpartum thyroiditis. Utility values for type 1 diabetes mellitus and thyroiditis were taken from the literature and combined using different methods.ResultsThe surrogate utility values obtained using the multiplicative method were higher than those obtained with the additive method (for example, the state of both type 1 diabetes mellitus and treated thyroiditis had a value of 0.75 for the multiplicative method versus 0.73 for the additive method). The resulting cost-effectiveness ratios for the screening strategy were slightly higher, $US16 000 (1998 values) per quality-adjusted life-year (QALY), for the multiplicative method when compared to the additive method ($US14 000 per QALY). This small difference was consistently maintained during sensitivity analyses.ConclusionAll methods of combining utilities resulted in similar values. Until a consensus is reached on the method of choice, researchers should consider using both methods in sensitivity analyses and reporting both sets of results.