Projection of age-specific mortality rates.

The projection of mortality rates by age has taken on added importance in recent years due to the increasing frequency of mortality changes that do not conform to generalized age patterns of mortality change implicit in model life table schema. In a number of developed countries and in some developing countries also life expectancy has risen in spite of actual mortality rises in certain age/sex groups (most notably adult males). Specific health interventions such as measles immunization prenatal and postnatal community health services or family planning lead to declines in early-age mortality greater than would be "predicted" on the basis of mortality change at older ages. The question therefore arises of how to project the age pattern of mortality for a specific country under a specific situation. The answer is clearly related to the type the extent and the quality of the data available at the moment of projection. This paper reviews a variety of methods that have been suggested by actuaries and demographers alike to project age-specific mortality rates: 1) projection by extrapolation of mortality rates (or transformations of mortality rates) at selected ages 2) projection by reference to a "law of mortality" 3) projection by reference to model life tables 4) projection by reference to another "more advanced" population 5) projection by reference to an "optimal" life table attainable under ideal conditions 6) projection by cause of death and 7) combinations of these methods. Examples of the use of these various methods are given and conclusions are drawn on their respective advantages and disadvantages. (authors)