Research on mortality differentials can serve at least three purposes. First, in research oriented towards health policy, the aim is to identify particularly disadvantaged groups in the population who would need special consideration in health and social policy. Secondly, differences in mortality can be studied from an epidemiological point of view, searching for hypotheses about the causes of specific diseases or attempting to identify the reasons underlying changes in mortality. The third use of research on mortality differentials is connected with the debate about the possibility of a further extension of population life expectancy (e.g. Olshanskyet al., 1990; Mantonet al., 1991). One way of estimating the limit to life expectancy is to examine the mortality of populations living in favourable conditions with regard to nutrition, other health habits, health care, etc., such as Mormons or Seventh-day Adventists ( Mantonet al., 1991). Socio-demographic characteristics such as gender, socio-economic position, marital status, region, and race or ethnicity may be used in an analogous way to distinguish subgroups within a national population that have particularly good health profiles. Information about the current level of mortality as well as recent mortality trends in such groups may therefore be useful for the assessment of the future development of mortality. They may also be helpful in setting targets for national mortality projections.
Although research on socio-demographic mortality differences among the elderly has not been as abundant as research on people at working ages, there is evidence that differentials according to several characteristics persist up to advanced ages. Knowledge about the trends in differences is scarcer, which is probably due largely to the fact that comparable results are seldom available for long periods of time.
To begin with, gender differences at older ages can be observed by means of basic statistics. Socio-economic mortality differentials among the aged have been reported in several countries, such as the United States ( Kitagawa and Hauser, 1973; Makucet al., 1990), England and Wales ( Goldblatt, 1990), Sweden ( Otterblad Olausson, 1991) and Finland ( Martelin, 1993). By contrast,