Approximating the Life-Table With a Single Characteristic Number. K. P. Brand, University of Ottawa, Canada
Life-table analyses are playing an increasingly prominent role in weighing chemical safety. The analyses can be used to compute several indices of life-time detriment including (but not limited to) life-time risk, population attributable risk, and measures of impact upon life-expectancy. Such measures lend themselves to the more meaningful comparison of risk factors (interventions), and can extend to the incorporation of quality/disability adjustments (e.g., QALYs or DALYs). In this paper we present novel approximations of the cause-modification (a generalization of the cause-deletion) life-table calculation. We restrict our focus to multiplicative models of excess-risk, seeking simplifying relationships between age-specific impacts and the indices of life-time impact. Our approximations, based on Taylor series approximations, perform remarkably well across a wide set of conditions. Although the approximations could in theory reduce computational burden, that is not our main purpose. Rather, the approximations provide valuable insights, which (i) reveal previously unrecognized relationships among the various indices (allowing a more meaningful interpretation of each index), (ii) identify heuristics for extrapolating estimates from one population to another, (iii) suggest a way to classify causes of death (e.g., lung cancer versus cardiovascular disease) based on properties relevant to cause-modification life-tables, and (vi) permit the consideration of uncertainty importance calculations as part of uncertainty analysis. Results are demonstrated using 1994 Canadian Mortality rates for several causes.
The McLaughlin Centre for Population Health Risk Assessment.
Go to . . .
2002 SRA Annual Meeting Table of Contents
2002 SRA Annual Meeting Author Index
Main Abstracts Menu Page
RiskWorld Home Page