The underestimation of costs per year of life saved--the case of subarachnoid hemorrhage.

Abstract

calculations creates an underestimation of costs which can affect a comparison between this program and others to save lives. In many health programs the main outcome is lives saved. However, as the length of these lives saved is not always the same in all programs, the ratio-cost per year of life saved is used for comparison among programs. In the procedure of converting the costs per life saved to costs per year of life saved one has to bear in mind that individuals and society have time preferences. Usually [2-6] researchers and analysts assume that most people have a preferonce for an earlier rather than extended gain. The most widely accepted method of incorporating the time preference notion into the appraisal is the process of ‘ ‘discounting’ ‘ future effects and costs to equivalent present values. This amounts to multiplying the values of future effects (and costs) by a weighing factor, so that they can be compared as if they all occurred at the same point of time. The weights for each future year are derived with reference to the assumed rate of time preference (for more details about the techniques see [7]). By doing so we adjust the costs and outcomes and are able to compare programs in the same point of time. In calculating the ratio of costs per year of life saved, Knaus et al. simply divided the cost per life saved by the expected length of life of those whose lives were saved. By doing so they implicitly assumed that the value now of a year of life that is gained 10 years from now is equal to the value of a year of life gained now. In more technical terms it implies that they assumed no time preference, namely zero discount rate. If we accept the argument that most people and society do have positive time preference, then the worth now of the X years of life saved is less than X due to the time effect. The correct way of calculating that ratio is to calculate first the present value equivalent of this number of years and then divide the cost per life saved by that figure. This yields a higher cost ratio as the value of future gains ‘ ‘shrinks’ ‘ in present value terms. In applying this procedure one is, in effect, reducing the total number of years of life saved and that is why the cost per year of life saved rises. In doing the calculation in the way Knaus et al. did, the cost per year of life saved in the program is then underestimated. If, for example, we take 5% as the discount rate, a rate that many economists in the U.S. consider to be appropriate for these calculations, the figures in the Knaus et al. paper are changed from the range of $1 ,999-$24,71 3 per year to $3,338-$41 ,271 per year, respectively. Accordingly, their figures underestimate the cost per year of life saved by 67%.