Deceleration in the Age Pattern of Mortality at Older Ages

From AcaWiki
Jump to: navigation, search

Citation: Horiuchi, Wilmoth, John R. (1998) Deceleration in the Age Pattern of Mortality at Older Ages.
Internet Archive Scholar (search for fulltext): Deceleration in the Age Pattern of Mortality at Older Ages
Tagged: uw-madison (RSS), wisconsin (RSS), sociology (RSS), demography (RSS), prelim (RSS), qual (RSS), WisconsinDemographyPrelimAugust2009 (RSS)

Summary

Background: The rate of mortality increase slows at older ages. This observation has led some researchers to switch from exponential (e.g., Gompertz) to logistic or power functions. There are at least two explanations for the slowing of mortality increases at older ages: the heterogeneity hypothesis and the individual-risk hypothesis. The heterogeneity hypothesis posits that "the deceleration is a statistical effect of selection through the attrition of mortality: Because the more frail tend to die at younger ages, survivors to older ages tend to have favorable health endowments and/or healthy lifestyles" (p. 391). The individual risk hypothesis argues that mortality increases may slow down at later ages because organisms often perform key processes more slowly at later ages (e.g., metabolism, energy expenditure), which may actually slow the process of aging itself and more of an organism's resources are dedicated to the repair of somatic damage (instead of reproduction) at later ages. "The heterogeneity and individual-risk hypotheses may be described as demographic versus gerontological explanations: The former emphasizes the composition of the population, whereas the latter considers senescent processes in individuals" (p. 392). This paper focuses on the heterogeneity hypothesis.

Predictions of the Heterogeneity Hypothesis: (1) Deceleration in the rate of mortality increase should occur for most major causes of death, (2) deceleration should start at younger ages for more selective causes of death (i.e., highly variable with vulnerable subgroups), and (3) deceleration should shift to older ages as the total level of adult mortality declines. Data and Measurement: National-level mortality data for Sweden (1861-1990) and Japan (1951-1990) are used to test these predictions. Mortality deceleration is measured by life-table aging rate (LAR), which is the derivative of the logarithm of the force of mortality. That is, the LAR measures the relative mortality increase with age. So, a LAR of .05 means that the death rate at exact age x is rising at an exponential rate of 5% per year. The discrete version of the LAR is K(x) = [ln M(x,5) - ln M(x-5,5)]/5, when data are given in 5 year intervals. Because cohort LARs for certain causes of death may be affected by period conditions, the effects of time trends in age-standardized death rates are removed.

Results: LARs decline with age for most causes of death, which supports the first prediction. "Overall, these patterns seem compatible with the [first two predictions]. The deceleration starts at younger ages for most degenerative diseases for which significant effects of a number of genetic and environmental risk factors have been indicated" (p. 395). Since selective causes of death appear to begin their decelerations at younger ages, it appears that selection occurs relatively early. This is consistent with the second prediction. However, evidence is mixed for the third prediction. To address this issue, the effects of senescent mortality (i.e., "natural" mortality) from background mortality. Using a Gamma-Makeham model, Horiuchi and Wilmoth conclude that "the data from both Sweden and Japan lend support to the third prediction of the heterogeneity hypothesis, when modified as follows: Mortality deceleration tends to shift to older ages as the level of senescent mortality declines" (p. 404). That is, as intrinsic mortality declines, selection occurs at later ages which delays deceleration.

Conclusions: The findings in this paper are broadly consistent with the heterogeneity explanation. A problem with the individual-risk explanation is that it is "quite flexible and could be compatible with almost any observed pattern . . . To test the individual-risk hypothesis in a comparable fashion, we would need to specify the changes in senescent processes that account for mortality deceleration and to state falsifiable hypotheses that can be tested with available data" (p. 404).