Using age, period and cohort models to estimate future mortality rates

Abstract

A description of a table of age- and period-specific mortality rates may be provided by the use of age, period and cohort models. These may be extended to produce estimates of future mortality rates that allow for trends related to both birth cohort and period of death. Lung cancer mortality rates for women in England and Wales are used as an example. The period 1951-70 is used to estimate mortality rates for 1971-80 and comparisons are made with the observed values. The technique is then used to estimate mortality rates until the year 2000 based upon 1951-80. The range of conditions for which this method is appropriate is discussed.