ascribed to climatic factors (mild winters, cool summers). When the period of continuous mortality decline became longer, however, it became more and more unlikely that it would be due solely to milder temperatures (Garssen and van der Meulen, 2007).

As Figure 13-1 shows, from 2002 to 2008 life expectancy at birth increased by almost 2 years (from 76.0 to 78.3 years among men and from 80.7 to 82.3 years among women). A substantial part of this increase of life expectancy at birth is due to advances at higher ages, as is evident from the fact that life expectancy at age 65 has increased by more than a year (from 15.6 to 17.3 years among men and from 19.3 to 20.5 years among women). The sharpest upturn in life-expectancy trends is seen at age 85, for which 2002 marked a reversal from almost complete stagnation to a period of rapid increases.

No systematic analysis of possible determinants of this remarkable development has been made until now. This chapter aims to assess the possible causes of the reversal from stagnation to renewed decline of old-age mortality in the Netherlands. After an analysis of patterns of decline by age, gender, and cause of death, we review all main groups of determinants: biological factors, factors in the physical and social environment, lifestyle factors, and health care factors. We used readily available data to assess whether changes in these factors have occurred preceding or coinciding with the changes in mortality that could plausibly explain the change in old-age mortality trend.

Mortality data were extracted from the registry kept at Statistics Netherlands, which is derived from the municipal population registries in the Netherlands. Underlying causes of death were coded according to the International Classification of Diseases (9th and 10th revisions). No changes in coding occurred around 2002. In order to calculate the cause-specific contributions to the gain in life expectancy, we calculated, for both periods, the number of life-years that would be gained if only the observed age-specific mortality risk for a specified cause changed during the period, keeping all other age- and cause-specific mortality risks constant.

Data on determinants of mortality were extracted from various registries and surveys kept at Statistics Netherlands, which are mostly available online at http://statline.cbs.nl/statweb/ [accessed June 8, 2010]. Data on self-reported health problems, lifestyles, and medical care utilization were collected in a multipurpose survey (Permanent Onderzoek Leef Situatie) that is conducted on a continuous basis among a representative sample of the noninstitutionalized population.

Data on clinical incidence (i.e., incidence of the first clinical episode for