Use of a mean entry age underestimates expected mean mortality rate

To explain the impact of the 10% annual increase in mortality rate in the life tables from about age 45 to 90 years on mean expected mortality rate in any follow-up (FU) group with a wide age range. Use of the mean age to enter a life table invariably underestimates the true mean expected mortality rate in small age groups with an age range of 5 to 10 years. As a result, when mean age and standard deviation (SD) are the only age characteristic given for the cohort reported in a FU study, the mean age must be adjusted upward to enter the life table to obtain a valid mean expected mortality rate of the entire cohort.

The 1989-91 Decennial US Life Table is used to illustrate the variation of expected annual rate, q', with age, x. The magnitude of the error in mean q' introduced by failure to adjust mean age to obtain mean q' is illustrated in examples of both cardiovascular and cancer FU studies. Other tabular analytical data are also presented.

From the 1989-91 US Life Tables for the white population, it is shown that the mean increase in annual mortality rate between ages 45 and 90 years has been found to be 9.36 +/- 0.79% per year for males and 9.94 +/- 1.13 for females. For age groups with a narrow range (10 years or less), a mean age can be used to estimate an accurate mean q'. But if the range exceeds about 15 years, as it does in most groups of patients in a FU study, a tabular q; obtained by entering the life table with the mean age is underestimated and is lower than the actual mean q'. The magnitude of the error increases with the magnitude of the range or SD. Examples are given of the magnitude of the error in one group as patients with coronary heart disease and in another group with cancer. Summary data on the magnitude of the error are also given for multiple groups in each category.

Recommendations are made on how to adjust the mean age, when possible, to provide a more accurate q', when data by separate age groups are not available.