Page 118 - Contributed Paper Session (CPS) - Volume 3

CPS1956 Areti B. et al.

Figure 1: Local authorities in England that are detected as unusal

Figures 2 and 3 show the time trends of mortality rate and relative risk for two
of the detected areas, Southwark (prob < 0.05) and Leeds (prob > 0.95),
plotted against the national time trends. We can see that the life expectancy
trend for the local authority of Southwark differs in year 2017 to the national
trend, showing a lower rate than expected, while the opposite is observed for
the local authority of Leeds in years 2013, 2015, 2016 and 2017.

4. Discussion and Conclusion
In this paper we analysed mortality data on females to estimate life
expectancy trends in England from 2001 to 2017. We were interested in
detecting areas whose time trends deviate from the overall national trend,
suggesting unusual behaviour. For the analysis we used a Bayesian hierarchical
model, allowing for spatial, temporal, and age effects, and we based our
detection rule on exceedance probabilities of the space-time interaction. We
found 21 areas to be unusual with a decreased slope compared to the national
one, and 32 with an increased slope respectively. These findings call for further
investigation as to the reasons behind these patterns. Potential extension of
this work is to allow for deprivation, the major determinant of life expectancy
(Woods et al. 2005; Smith, Olatunde, and White 2010), in order to
assess potential associations with the unusual time trends.

107 | I S I W S C 2 0 1 9