OBJECTIVES

Many cost-effectiveness models over a patient’s lifetime incorporate background mortality. Analysts often use recently available country life tables for these estimates, for example, those from the US Centers for Disease Control and Prevention (CDC). These period life tables that provide data on mortality rates (i.e., conditional survival probabilities) from a single year. As historical data, they may not take into account future reductions in mortality rates due to technological advancement and other factors. In contrast, the Social Security Administration’s (SSA) actuarial cohort life tables attempt to account for such mortality reductions by predicting the mortality rate at each age for a birth cohort over the course of their lifetime. The purpose of this analysis was to compare life expectancy predictions based on the 2010 CDC period life table versus the 2010 SSA cohort life table and discuss the implications.

METHODS

Based on the 2010 CDC period and SSA cohort life tables. We simulated the remaining life expectancies at birth and at age 65, undiscounted and discounted at 3%.

RESULTS

The absolute difference in life expectancies (CDC versus SSA) increased with age. Undiscounted life expectancies at birth and at age 65 were 78.6 and 19.1 based on the CDC life table, lower by 5.4% (vs. 83.1) and by 14.3% (vs. 22.3) compared to SSA, respectively. Discounted life expectancies at birth and at age 65 were 30.4 and 14.1 from the CDC life table, lower by 1.6% (vs. 30.9) and 10.8% (vs. 15.8) compared on SSA-based estimates. At $150,000 per quality-adjusted life-year, CDC-based life expectancy underestimated value per individual by $75,000 at birth and $255,000 at age 65.

CONCLUSIONS

In cost-effectiveness Markov models that consider background mortality, use of period rather than cohort models will generally underestimate life years gained. These differences can substantially lower the monetary value of health gains.