9. Onset of sustained mortality decline

From what has been said above it is clear that though a decline in old age mortality has taken place internationally on a very broad front, it did not begin simultaneously everywhere. We shall now try to pinpoint for each country and sex the time of onset of sustained decline. For this purpose we have calculated the 5-year means of the ageadjusted death rate at ages 80-99 in the manner of moving averages but assigning each mean to the last of the 5 years involved. These values are given in Annex Table 6.

        We may consider the year of onset of the decline to be the earliest year in which the 5-year mean fell to a level which it never exceeded again and after which the decline was never significantly broken by an increase or a period of see-saw rates.

        This rule allows some fluctuation in annual death rates but no significant break in the declining trend. Using this criterion, the years of onset of sustained mortality decline are those given in Table 10.

        Among the 25 countries in which it could be determined, the onset varied a great deal and took place generally much earlier for females than males, the average time difference being eight years. Exceptions in which the male rate began to decline first are few and not significant.

        France is unique in that mortality decline has been continuous there for the entire period starting therefore already in 1955 or earlier for both sexes. Two of its neighbours, Switzerland and Belgium, also had early starts.

        In late 1960s the decline began to reach wider areas and by 1970s it had become almost universal in low-mortality countries.

        The onset of the remarkably rapid decline in Japan was not very early and can be timed to 1966 for both sexes. In countries of traditionally low mortality (Scandinavia and the Netherlands), female death rates entered a sustained decline already in the 1950s or 1960s while the male rates showed resistance till much later, and in Denmark turned decisively down only in the 1980s.

        It can be observed that the decline began simultaneously in England and Wales and in Scotland, and that Austria followed West Germany with only a small time lag. In the former East bloc, female rates began to decline in the 1970s but male rates only around 1985, if at all. In Latvia and Poland, mortality has not yet entered a definitely declining stage and they are therefore excluded from the table, as is Singapore for which the observed period is still too short.

        Table 11 gives an overview of the speed of the mortality decline in each country during the period in which it has proceeded in a sustained fashion. The speed is indicated in the table in terms of annual percent decline, measured for each year by the mean rate for the last five years compared with the mean of the five years preceding them. It therefore gives the mean decline during the last ten years.

        We can observe several countries with relatively even rates of decline for females, in some quite rapid, as in France and Sweden, in others more moderate, as in England and Wales, Scotland, Italy and Norway, or frankly sluggish  as in the former East bloc. In several other countries the trend has clearly accelerated over time and among these are Switzerland, Japan, Belgium, West Germany and Austria while in some others, such as Denmark, the Netherlands and Finland, rapid acceleration has been followed by a slowdown. The most precipitous decline, exceeding 4 percent per year was experienced by Icelandic women in the 1970s.

        The rates of decline have been notably slower for males and no 10-year period with a sustained drop of 3 percent per year has been observed. However, also here we can frequently note an acceleration of the decline which in some cases has later slowed down but in others is continuing apace.

        This examination confirms the existence of a generalized mortality decline of recent origin but with considerable geographical variation as to its onset, speed, acceleration and occasional slowdown.


Updated by V. Castanova, 1 March 1999