Assessing Differential Area Mortality Trends via Bayesian Random Effects

Abstract

Changes in area mortality are important for assessing spatial health inequality. They are likely to be differentiated by age as well as spatially and may vary by demographic strata (e.g., gender, ethnic group). A simple approach assumes linear improvement in log mortality risks, with noninteracting area and age coefficients. By contrast, this article considers parsimonious models for mortality change allowing nonlinear trends and interactions between ages and areas in mortality levels and trends. A case study considers trends in mortality in 32 London boroughs over the 8-year period 1999–2006 for deaths data disaggregated by age, sex, and area.

Keywords:

• Age Group
• Bayesian
• Mortality
• Spatial
• Trend

Mathematics Subject Classification:

• Primary 62F15
• Secondary 62P10

Notes

*Observations where Pr(y _rep < y | y) +0.5 Pr(y _rep = y | y) exceeds 0.95 or is lower than 0.05.