|
Background
An individual's current health reflects what has happened to them
over their lifetime. A pensioner celebrating their 80th birthday
in 2000 was born in 1920, attended school during the Depression,
experienced wartime rationing as a young adult, raised a family
during the long boom of the 1950s and 1960s, and perhaps retired
early due to the rising unemployment of the late 1970s and early
1980s. How have the differing experiences of these children of the
Depression affected their health? How have unemployment, slum clearance,
de-industrialisation and so on affected whole communities and the
health of those who lived then?
Analysing such
long-term trends, and disentangling the local impacts of economic
circumstances on individuals' lives has been greatly complicated
by large variations in the geographical units for which the relevant
statistics have been gathered. However, a new information system,
the British Isles Historical Geographical Information System is
enabling researchers to examine health data over time for consistent
sub-county geographical areas. The historical GIS holds Census and
vital registration data from 1920 onwards for circa 1,500 local
government districts.
Aims and
Objectives
The British Isles Historical Geographical Information System
will, for the first time, make data on mortality and socio-economic
conditions over the twentieth century available for consistent sub-county
geographical areas. The aim of the project is to explore the historical
antecedents of geographical health inequalities and their evolution
since the 1920s.
The objectives of the project are to:
- collate and
map data on health inequalities and their relation to socio-economic
conditions, at the small area level and over time for the whole
country;
- collaborate
with the Centre for Longitudinal Studies to link longitudinal
data on individuals with data from the historical GIS concerning
their places of origin in early life;
- carry out
analyses to demonstrate historical geographic dimensions of health
inequality over the period from 1920 to the present;
- make the
data available to other researchers.
Study Design
There are three stages to the project. We will begin by:
researching
the immediate health impact of the 1930s recession, by analysing
time series data at local government district level (circa 2,000
areas) for infant and adult mortality and relating these to statistics
on housing (over-crowding), urbanisation (population density) and
unemployment. We will employ multi-level modelling using both county-level
unemployment time series data and 1931 census unemployment data
for individual local authorities.
We will then
move to analysing the longer-term impact of the inter-war recession
on today's population. For this we will use information based on
place of residence in 1939, derived via National Insurance numbers,
to link information from the ONS Longitudinal Study on survival
and health status of individuals in 1991 to inter-war data on unemployment,
overcrowding and infant mortality in the areas in which these individuals
lived.
As the historical
GIS becomes fully operational, we will extend our time series post-1945,
adjusting for the effects of boundary changes to create essentially
new series for constantly-defined geographical units. Post-1974,
we will aggregate very detailed recent data into the same unchanging
units. While all areas will show overall improvement, we will identify
those areas which experienced the most rapid and the slowest improvement
in health, and relate these patterns to long-run socio-economic
data.
Policy Implications
Current policies aim to reduce health inequalities through improving
employment opportunities and housing conditions in deprived areas.
Evaluation is complex because the effects of such policies are necessarily
long-term. By studying such long-term relationships over most of
the century, the project will create a better understanding of how
targeted area regeneration can influence health.
Project Summary
In 2001, the government set national health inequalities targets
to underwrite its commitment to reducing health inequalities. One
of the two national targets is to reduce inequalities between health
authorities, by narrowing the gap between those areas with the poorest
health and the population as a whole.
The project
provides an important historical perspective on this goal. Its aims
include enhancing the historical datasets necessary to map spatial
inequalities in health over the last 100 years. The major problem
with comparing information across long time periods is that the
spatial units for which the data are published changed significantly:
for around 600 registration districts in England and Wales until
1911; then for local government districts, but dropping from over
1,800 in 1911 to under 1,400 in 1974; and thereafter much more detailed
digital data. New methodologies are required to compare information
from these different sources. One approach uses the Great Britain
Historical Geographical Information System (GIS) developed by Ian
Gregory and Humphrey Southall to interpolate the data onto a single
standardised set of spatial units; this system maps changing administrative
units down to parish-level from the mid 19th onwards. Studies of
shorter periods use geography conversion tables constructed directly
from boundary change lists.
Work for the
Programme involved extending the linked statistical database (i)
by computerising mortality data for every local government district
for every year from 1921 to 1973 and (ii) adding census data on
social structure and housing conditions, and enhancing occupational
data. A subset of these data were redistricted to 1939 units and
then linked to the Office for National Statistics Longitudinal Study
(ONS-LS), which contains information on place of origin in early
life. Building on this work, the project has gone on to explore
geographical trends in health improvement and health equity over
the last 100 years. It has examined whether the rapid decline in
mortality across the century has been accompanied by a narrowing
of health inequalities. Infant mortality, a major proxy for health
in historical studies, has been the main focus for this analysis
of trends.
Key findings
- In the 1890s,
infant mortality rates were high. There were clear core-periphery
and rural-urban differences in rates. The south and east of England
and Wales had lower rates than the regions surrounding this prosperous
core. Rates were also higher in the main urban and industrial
areas (inner London, south Wales, Birmingham, Tyneside and the
industrial parts of Lancashire and West Yorkshire) than in rural
areas. Over time, the strong rural-urban contrast appears to weaken;
the core-periphery divide, however, remains pronounced.
- Infant mortality
rates in England and Wales fell from 153.3 per 1000 live births
in the1890s to 65.1 in 1928 and 22.6 in 1958 (by the early 1990s,
the rate was 6.3). Between the 1890s and 1958, the reduction of
mortality was comparatively rapid in the east and south Midland
regions and these regions improved their ranking over time. The
northern and northwestern regions remained comparatively disadvantaged
with relatively high rates throughout the period. The position
of Wales worsened from the 1890s, when infant mortality rates
were middle ranking, to 1958, when it had the highest infant mortality
rates.
- Population
density (urbanity) showed a stronger relationship with geographical
inequalities in infant mortality in the early part of the 20th
century; social factors (measured by such factors as overcrowding)
became more strongly associated with infant mortality rates between
1928 and 1958. This pattern is consistent with the idea that the
greatest initial gains in infant survival during the 20th century
were achieved by reductions in the very high levels prevailing
in some urban and industrial areas. Towards the middle of the
century, the major cities began to stand out less strongly. However,
after 1958, the major cities did less well compared to rural areas.
The largest relative increases were in cities like Bradford, Sheffield,
and Newcastle-upon-Tyne and in inner London districts like Hackney.
In Bradford, for example, the infant mortality rate was 15% above
the national average in 1958 (at 25.5 per 1000); by the 1990s,
it was 84% above the average (at 11 per 1000).
- The rapid
decline was accompanied by changes in the pattern of geographical
inequality in infant mortality. Between the 1890s and 1928, the
decline was accompanied by a slight reduction in some indicators
of inequality. Between 1928 and 1958, the absolute reduction in
the infant mortality rate was less, but the reduction in inequality
was much more marked and was significant for all measures of inequality.
The extent of change varies between the inequality indicators
used, but all measures of inequality point to an unambiguous decline
in geographical inequality between 1928 and 1958.
- Analyses
of the effect of the economic recession in the 1930s on infant
mortality found that districts with high unemployment (over 20%)
had infant mortality rates that were over two-thirds higher than
in districts with unemployment rates below 6%. But there was little
evidence that unemployment contributed to a widening of spatial
inequalities in infant mortality. The continuing improvement in
average infant mortality was not accompanied by slower falls than
average in high mortality areas: rather, the high mortality areas
had sharper falls in mortality risk.
- At some point
in the period since 1958, there was a re-emergence in inequality.
All measures show a return to the levels of inequality apparent
in the 1890s and in 1928, a result which is not an effect of the
generally declining number of infant births. The analysis therefore
supports those who argue that inequality in infant mortality has
not improved during the 20th century.
- Further analyses,
based on annual data for 414 districts in the north of England
for the period 1921 to 1973 and a wider range of explanatory factors,
confirm that the 1950s were a low point for spatial inequality
in infant mortality. There is also evidence that the area increases
in unemployment associated with the depression years of the late
1920s and early 1930s did not contribute to a widening of area
differentials in this key measure of health and health inequality.
- Linkage to
the ONS-LS shows that, even after allowing for a wide range of
factors concerned with more recent living conditions, people brought
up in areas of high unemployment in the 1930s were significantly
less likely to have survived from 1981 to 1991 than the children
of more prosperous districts. Being brought up in areas of over-crowded
housing, areas with high proportions in social classes IV and
V, and areas categorised as Depressed Areas all made people who
survived to 1991 significantly more likely to be affected by limiting
long-term illness.
- In addition
to these findings, the methodological aims of the project have
been achieved. Linkage between the Historical GIS and the ONS-LS
is complete, the project having identified 66,397 individuals
within the ONS-LS who were aged 16 and under in 1939, who survived
to the 1981 census and who have a valid 3-digit NHS code, allowing
linkage to 1931/1939 location data.
|
