Spatiotemporal modelling of all-cause and cause-specific mortality in England

Abstract

High-resolution data for changes in mortality and longevity are scarce. Estimating mortality for specific combinations of spatial units, time periods, age groups, and causes of death poses statistical and computational challenges which typically result in compromises in the granularity of the results. I applied Bayesian hierarchical models based on patterns of mortality over age, space, and time, to obtain robust yearly estimates of life expectancy and cause-specific mortality for small areas. Using vital registration data held by the UK Small Area Health Statistics Unit, I investigated trends in mortality for subnational units in England from 2002 to 2019.

I examined trends in life expectancy in England’s 6791 middle-layer super output areas (MSOAs). In the years 2014-19, 1270 (18.7%) MSOAs for women and 784 (11.5%) MSOAs for men saw declines in life expectancy. The same analysis was performed for the 4835 lower-layer super output areas which comprise London. At this smaller level, issues with the population data in the older age groups affected the reliability of the life expectancy estimates.

I modelled cause-specific mortality for 314 districts in England. The inequality in life expectancy increase since 2010 was driven largely by that of deaths from dementias and the residual group of non-communicable diseases, as well as ischaemic heart disease in men. The analysis was extended to look specifically at the top ten leading cancer causes of death. Preventable cancers showed the greatest spatial inequality in 2019. Unlike areas in the rest of the country, mortality in London from several cancers did not increase in poorer districts, suggesting that some features of London weaken the relationship between poverty and mortality.

England has seen increasing inequalities in all-cause mortality over the past two decades, driven by rises in mortality for a large number of communities and from several preventable causes of death.

1 Rationale, aims and objectives

1.1 Rationale

Although the UK has, by global standards, a relatively high life expectancy, the country still suffers vast inequalities in mortality. While there have been subnational studies on trends in life expectancy in England, these have been limited in resolution to the district level (Bennett et al., 2015). There is vast heterogeneity in health outcomes and their determinants within districts. As a notable example of mortality from recent years, the district of Kensington and Chelsea, which contains both the most affluent areas in the country in its south and some of the poorest in its north, saw a tragic fire in a high­rise residential building in a council housing complex, Grenfell Tower, in 2017. There is a clear need to uncover further variation in mortality below the district level.

Since around 2010, although national life expectancy has continued to increase, there is evidence that the rise in female life expectancy has reversed in the most deprived deciles of the population (Bennett et al., 2018; Marmot et al., 2020). There is a need to identify the specific places where these declines are happening. There are likely to be important changes in mortality from specific causes of death in these areas that have driven this reversal.

1.2 Objectives

The aim of my thesis is to estimate trends in mortality for small areas over the past 20 years in England. There are two specific objectives which will help achieve this aim:

1. To estimate trends in life expectancy for very small areas in England from 2002 to 2019.
2. To investigate trends in cause-specific mortality for districts in England from 2002 to 2019.

1.3 Structure of the thesis

Chapter 2 reviews the literature on spatial methods for mapping disease and mortality for small subnational regions, followed by the literature of separating total mortality into different causes of death. I will then explore inequalities in the UK over the past few decades through to the present. Chapter 3 presents the data sources, and Chapter 4 the statistical modelling choices common to all objectives of this thesis. Chapter 5 concerns the first objective of the thesis - estimating trends in life expectancy for very small areas in England. Chapter 6 extends the first objective by focussing on London at a finer scale than the previous chapter in an attempt to gauge whether higher-resolution analyses are possible. Chapter 7 addresses objective two of this thesis, modelling specific causes of deaths at a coarser scale, and looking at potential drivers of the observed trends in life expectancy. Chapter 8 follows the methods of Chapter 7, but focussing solely on deaths from cancers. Chapter 9 concludes with a discussion on the public health implications of the findings and areas for future research building on the work in this thesis.

Bennett JE, Li G, Foreman K, Best N, Kontis V, Pearson C, Hambly P, Ezzati M. 2015. The future of life expectancy and life expectancy inequalities in England and Wales: Bayesian spatiotemporal forecasting. The Lancet 386:163–170. doi:10.1016/S0140-6736(15)60296-3

Bennett JE, Pearson-Stuttard J, Kontis V, Capewell S, Wolfe I, Ezzati M. 2018. Contributions of diseases and injuries to widening life expectancy inequalities in England from 2001 to 2016: A population-based analysis of vital registration data. The Lancet Public Health 3:e586–e597. doi:10.1016/S2468-2667(18)30214-7

Marmot MG, Allen J, Boyce T, Goldblatt P, Morrison J. 2020. Marmot Review: 10 years on. Institute of Health Equity.