Inferring frail life expectancies in Chicago from daily fluctuations in elderly mortality

Susceptible sub-populations with existing disease have exhibited stronger relationships between air quality and mortality in time-series studies, but their associated life expectancies have largely been overlooked. Murray and Nelson developed a new time-series model that estimated a small unobserved (frail) sub-population and their resulting life expectancies in Philadelphia, including environment relationships. As a further example in a different geographic area, we used this model with 1987-2000 daily mortality data in Chicago and found a stable frail population at risk of ∼900 persons with a mean life expectancy of ∼11 days; fewer than two daily deaths were associated with air pollution. We considered daily concentrations of CO, NO₂, O₃, PM₁₀ and SO₂, and found PM₁₀ and O₃ to have stronger associations with frail mortality. Our estimates of life expectancy and air pollution and temperature relationships are similar to those found in other studies that used different methods. Temperature was more important than air pollution during the 1995 heat wave, when mortality risks increased dramatically after 2 d exposure and life expectancies decreased to 3-5 d. Modeling this event separately had substantial effects on lagged mortality--air pollution relationships and the population at risk. The premises of the Murray-Nelson model were supported by simultaneously considering an additional subgroup of non-frail individuals; they contributed only ∼1% of total elderly deaths. We conclude that frail life expectancies estimated by the Murray-Nelson model are robust, and that under these conditions non-frail persons have little risk of acute mortality, with or without contributions from air pollution.