Have school league tables led to more socially segregated neighbourhoods?
By Chris A Garrington, on 24 January 2023
In the early 1990s, parents in England were given access to league tables based on school performance which gave them more information when deciding where their kids would go to school. In this blog Dan McArthur and Aaron Reeves share findings of research which used the ONS Longitudinal Study and which found quantifying school quality had the unintended consequence of increasing the geographical concentration of advantage, potentially entrenching inequalities.
Where we grow up matters: a childhood in a neighbourhood where the majority of people are materially deprived can lead to lower earnings, a less prestigious occupation and even a lower life expectancy. Neighbourhoods matter because they contain important local services such as hospitals that can have a profound impact on what happens in our lives. Parents sense that neighbourhoods (and the local services available within them) matter and so areas perceived to have high quality services typically have higher house prices, and this can to segregation because less affluent families find it harder to stay in or move to neighbourhoods where the quality of these local services is perceived to be good.
Education is particularly important because school quality can have a major influence on adult life chances, and British parents pay a substantial house price premium to gain access to good schools. This can be self-perpetuating. School league tables at least in part reflect the social composition of the pupils: middle class children tend to perform well, and this pushes their schools up the rankings. Better-off parents move to be near successful schools, and in doing so help those schools to become even more successful. Conversely, the process leads to less successful schools facing a downward spiral as parents who are able to move away from them choose to do so.
The very act of measuring school quality, then, has the potential to deepen divisions between advantaged and disadvantaged residential areas. To explore this possibility, we carried out two studies to explore what happened after the introduction of GCSE league tables in England, which provide a clear and simple (albeit controversial) measure of school quality.
Our first study compared census data from 1981 to 2011 on the social class composition of neighbourhoods with school performance data, and showed that areas with better-performing schools saw greater increases in the proportion of professional and managerial people among their populations.
Our second study used the ONS Longitudinal Study, a one per cent sample of English and Welsh census records and life-events data, to study individual-level patterns of residential mobility. We found people with professional or managerial occupations became more likely to move to areas with better schools after the introduction of school league tables. However, this only occurred if those people had school-age children: this provides us with strong evidence that those who were most able and most incentivised to benefit from the introduction of league tables did indeed alter their patterns of residential mobility in response to their introduction.
We could not directly ask parents why they moved to a particular area, and we could not see which schools their children subsequently attended, but our study does provide substantial indirect evidence that league tables drove a change in class-specific patterns of residential mobility. We do, however, consider alternative explanations for the changes we see over time. For example, it might be the case that middle class parents became more concerned about their children’s educational prospects during this period; but we find no evidence of a change in this direction. By ruling out these alternative explanations, we can be more confident that the change was driven by the change of policy rather than by other factors.
Would these findings be replicated elsewhere? We believe they would: England’s schools are funded via a centralised system which also redistributes funding to areas where there is educational disadvantage, and in places where this is not the case, such as the United States, the effect may in fact be more pronounced.
We believe the alluring simplicity of league tables based on exam performance may reduce the importance of more informal sources of information for parents. Before the league tables were published, families gleaned background on schools through local networks, as well as ad hoc publicity around the particular successes of individual schools. League tables changed their perspective – what may previously have seemed a good local school may have suffered in comparison with a more successful school further away, and in fact the average distance children travel to school has almost doubled since the 1980s.
We know from other research that there are class differences in how families respond to league tables. Those in less advantaged homes may put more weight on proximity, for instance, and may give their children more say. Those with professional and managerial jobs are more likely to take the lead in the decision, and also to be focused on entry to an elite university. But there are also financial reasons why a middle-class family is more likely to move into the catchment of a good school: those on lower incomes are often unable to afford to do so.
Our research showed that the share of professional and managerial residents increased fastest in local authorities with high-performing schools. It also found those in advantaged social classes became more likely to move to areas with high-performing schools after the introduction of league tables, but only if they had school-aged children.
This does not necessarily mean that league tables will always lead to a greater geographical concentration of advantage – other background factors such as school funding policies and the level of house price inequality will also play a part. But we can say that quantifying quality in this way reveals inequalities in performance which were previously opaque. It also increases the likelihood that good schools will do even better as successful families move into new catchment areas.
This matters because the measurement of school quality is a political choice, albeit one which – as we have shown – has unintended consequences. League tables enable parents to make informed choices, and may even improve teaching quality, according to some research. But in doing so they deepen the geographical concentration of disadvantage, and they potentially affect the life chances of children whose parents are unable to move into the catchment of successful schools.
‘The Unintended Consequences of Quantifying Quality: Does Ranking School Performance Shape the Geographical Concentration of Advantage?’ by Daniel McArthur and Aaron Reeves, was published in The American journal of sociology: https://www.journals.uchicago.edu/doi/10.1086/722470
Health and place: How levelling up health can keep older workers working
By Chris A Garrington, on 21 November 2022
As part of its levelling up agenda, the UK Government has set itself an ambitious target to add five additional healthy years to the average UK lifespan by 2035. In this blog Dr Emily Murray highlights lessons from the Health in Older People in Places project (HOPE), which she leads. HOPE uses data from the ONS Longitudinal Study to showing the link between levels of employment and health in a place.
We know place matters when working to extend healthy life expectancy (HLE) – there are large inequalities in older people’s health, depending on where they live. The Government recognises this and has set target of narrowing the gap between those living in the ‘healthiest’ and ‘unhealthiest’ local authority areas by 2030.
There are strong links, too, between the health of the population in a local area and levels of employment. So if we want people to be able to stay healthy and to work for longer, narrowing these gaps can make a real difference.
Staying in work
If the UK had achieved the current levelling up agenda goal of reducing the HLE gap by five years between 2001 and 2011, older people’s participation in the labour market would have increased by 3.7 per cent between 2001 and 2011. That would have meant 250,000 additional older people in paid employment. The HOPE project used Disability-Free Life Expectancy (DFLE) g as a proxy for HLE, as HLE data for local authorities was not available in 2001.
While disability-free life expectancy (DFLE) improved overall in the UK from 1991 to 2011, there was still a significant gap between the local authority areas considered the ‘healthiest’ and the ‘unhealthiest’. In 2011, DFLE at age 50 varied from 13.8 to 25.0 years – that’s a gap of 11.3 years between the healthiest and unhealthiest areas, which widened during the study period.
Unfortunately, over a decade later, the conversation hasn’t moved on much further. Health Equity in England: The Marmot Review 10 Years On, the 2020 follow-up to Sir Michael Marmot’s landmark study, found that the health gap between wealthy and deprived areas had continued to grow.
The HOPE project has built on this research by using Census data for England and Wales to show the link between levels of employment and health in a place.
- The higher the proportion of older people with poor health in a place, the less likely it is that any adults in that place will be in paid work. For example:
- Older workers from the unhealthiest areas were 60 per cent more likely to be out of work than those who live in the ‘healthiest’ areas
- Women aged 50-74 living in the ‘healthiest’ areas re 5.6 per cent more likely to be in paid work than those living in the ‘unhealthiest’ areas.
- Men aged 50-74 living in the ‘healthiest’ areas were 7.1 per cent more likely to be in paid work than those living in the ‘unhealthiest’.
- How we measure health in a place matters: links between health in a place and employment are stronger for self-rated health measures, compared with life expectancy figures or mortality indicators.
- Historically disadvantaged areas continue to struggle: areas where people left paid work at a younger age due to poor health in 1991 were much more likely to experience this trend in 2011 as well.
- This disproportionately affects people in manual occupations: they’re much more likely to experience ill health, and they can expect four fewer years of healthy life beyond age 50, compared with workers in administrative or professional roles.
- There’s a correlation between health in a place and younger people being in paid employment: for example, the probability of a woman aged 16 to 49 not being in paid work was 33.7 per cent in the ‘unhealthiest’ areas compared with 26.3 per cent in the ‘healthiest’ areas.
- Those working in professional occupations were more likely to be in work 10 years later than those working in elementary occupations or doing repetitive manual labour: this gap in employment outcomes was most marked for people living in ‘unhealthy’ areas.
The fallout from the COVID-19 pandemic and the current cost of living crisis are likely to widen existing inequalities. So it’s unclear how the Government intends to achieve its ambitious goals to increase healthy life expectancy and to narrow the gap between those in the ‘healthiest’ and ‘unhealthiest’ areas, especially given its recent decision to abandon the promised white paper on health disparities.
We recommend that The Government should:
• Increase spending on preventative health programmes to at least 6 per cent of the national health budget. This is in line with Canada, who currently invest the most in prevention across the G20 and continue to raise this proportion in accordance with the rise in preventable diseases.
• Earmark part of the £4.8 billion levelling up infrastructure fund for projects that will create jobs suitable for older workers in the ‘unhealthiest’ local authority areas, especially in those where a high proportion of employment is in manual work.
• Collect, monitor and publish data every year on health in a place, in particular self-rated health measures and labour market participation for people over the age of 50.
• Confirm that there will be another census in 2031 and add detailed questions about health and labour market participation for people aged over 50.
• Improve access to medical services to allow older people in poor health to remain in work. This includes reducing wait times to see a GP and for referrals, treatments and A&E.
• Provide support, including career training and advice, to help older workers transition to less physically demanding roles, especially those in manual roles.
Local authorities should:
• Develop a five-year strategy to increase employment rates for people aged over 50 in the ‘unhealthiest’ communities, in partnership with business. This strategy should recognise that older women often face additional barriers to employment apart from health barriers.
• Include local targets to improve population health in line with the national average for people aged 50 to 74 as part of their annual planning exercise.
• Increase support for older workers in manual occupations to stay in employment. For example, training and financial support, either through the benefits system or apprenticeship schemes, can help older workers transition to less physically demanding jobs as they age.
• Strengthen local tailoring of prevention programmes to ensure that services fully cater to local population health requirements.
• Address ageism at a local level, by educating and informing people on how to receive the best care to prevent or manage health conditions, regardless of age. The aim is to challenge the perception that long-term conditions are an inevitable consequence of old age when many are preventable. Local authorities should also work with businesses to challenge employer perceptions that older people’s health is a barrier to their participation in the labour market.
Although the prevalent narrative is often that individual health is an individual problem rather than a societal one, the whole community is affected by poor health. It’s not just about helping people live longer, healthier lives but supporting local economies and economic growth.
The levelling up agenda is more important now than ever, and it’s vital it is not sidelined.
The Health of Older People in Places (HOPE) project is a multidisciplinary research project funded by the Health Foundation under the Social and Economic Value of Health in a Place (SEVHP) programme. The research team includes scientists from the Department of Epidemiology and Public Health at University College London (UCL) and the School of Geography at the University of Leeds. The full report, Health and place: How levelling up health can keep older workers working,
is available here. The report was written and published by the International Longevity Centre, UK.
The work was launched on October 19, 2022, at an event whose keynote speakers included Lord James Bethell, Parliamentary Under Secretary of State at the Department of Health and Social Care. Slides from the event are available here: https://ilcuk.org.uk/hope-project-report/
Dr Murray discuss the work further along with Dr Brian Beach in this Linking our Lives podcast.
Mental health service use and local crime – how are they associated?
By Chris A Garrington, on 26 September 2022
Living in neighbourhoods with higher crime rates is linked with a higher prevalence of mental health problems – but what is the relationship with mental health service use, especially with psychotropic medications? Also, are there any groups of people more vulnerable to the impact of crime, and how do changing crime levels help to understand this association? In this blog, Gergő Baranyi discusses new insights using data from the Scottish Longitudinal Study.
In the United Kingdom one adult in six is affected by common mental health disorders at any given time, and the cost associated with mental illnesses adds up to four per cent of the national gross domestic product. We know the physical and social environment plays a part: Residential areas with high levels of deprivation and social disorganisation tend to have more crime and violence, and that might impact the mental health of residents. Personal experience of being a victim or witness of crime and violence is more common in high-crime neighbourhoods. However, people in these areas might experience more stress, avoid public areas and reduce social interaction with others, even if they are not affected directly by crime – and that can influence their mental health.
We were able to use the Scottish Longitudinal Study (SLS), a five per cent sample of the Scottish population which enabled us to link participants’ census responses with levels of crime in their neighbourhood and with psychotropic medications prescribed through their GP. Our study captured people who responded to both the 2001 and 2011 censuses and who were aged 16 and over in 2001. Based on addresses available in the census and in other administrative records we linked information on neighbourhood crime to these participants. This included police-reported crimes and offences such as assaults, crimes of violence, domestic break-ins, drug offences and vandalism, which were aggregated across 6,500 small areas in Scotland with a population of 500-1000 individuals. Although the 2011 census in Scotland included a question on self-reported mental illness, the SLS enabled us to link NHS Scotland records on prescribed antidepressants and antipsychotics to our participants.
The findings showed that living in a higher crime area was linked to reporting mental illness in the census and receiving antidepressants or antipsychotic medication, and also extended our understanding of crime and mental health.
Psychotropic medication as a proxy for mental health problems
After excluding from the sample those with pre-existing mental health conditions – identified as receiving any psychotropic medications in the first six months of the study period – we had a sample of almost 130,000 adults with an average age of 51 in 2009.
During the follow-up period between 2009 and 2014, 22 per cent of our sample received at least one new prescription for antidepressant and two per cent at least one new prescription for antipsychotic medications.
After taking into account key personal characteristics, we found these proportions differed significantly according to crime levels in residential neighbourhoods. Those living in areas with high crime were at a significantly higher risk of having a new prescription for antidepressants. The odds were higher for people in young and middle adulthood, especially women. These associations were present even after we controlled for area disadvantage.
When we looked at antipsychotic medications we found a similar association, though the risk was higher among men and in middle adulthood. Area deprivation was not associated with antipsychotic prescription.
While we used antidepressants and antipsychotics as a proxy for mental health problems, they are often prescribed for other health problems and not all individuals with mental illnesses receive pharmacological treatment. Still, findings with self-reported mental illness led to similar conclusions.
Utilising changing crime levels to understand underlying mechanisms
Scotland experienced a significant drop in crime during the study period, but not all neighbourhoods benefitted equally from this drop and there were even areas where crime increased. In our second study, we used information on changing crime levels, adding neighbourhood crime rates between 2004 and 2013 to our dataset. Participants’ addresses during this time were available from GP registration records.
Our findings revealed that young adults who stayed in the same neighbourhood while crime levels were increasing were more likely to report mental illness in the 2011 census and to receive antidepressant prescriptions from their GPs. This provides stronger evidence for the impact of neighbourhood crime on individual mental health.
We also found that middle-aged adults who moved into higher-crime areas were more likely to report mental illnesses and have antipsychotics prescribed during the study period. Although it is difficult to tease out exact pathways, this can be due to people with more severe mental health conditions moving to more affordable but often disadvantaged and higher crime areas.
Our studies confirm that local crime is an important predictor of mental health service use, independent of other individual- or area-level risk factors, but the associations differ across type of medication, and between sex and age groups.
Crime and violence reduction programmes, targeting crime hotspots and rehabilitating deprived areas, might be beneficial for population mental health. Mental health promotion in local schools, prevention initiatives for high-risk individuals and enhanced mental health services in high-crime areas might provide opportunities for those most in need.
The project team included Gergő Baranyi, Mark Cherrie, Sarah Curtis, Chris Dibben and Jamie Pearce from the Centre for Research on Environment, Society and Health, University of Edinburgh. Gergő Baranyi presented this research at the UKCenLS conference in Cardiff on September 20. This blog post is based on two papers:
Baranyi G, Cherrie M, Curtis S, Dibben C, Pearce JR. Neighborhood Crime and Psychotropic Medications: A Longitudinal Data Linkage Study of 130,000 Scottish Adults. Am J Prev Med. 2020;58(5):638-647 https://doi.org/10.1016/j.amepre.2019.12.022;
Baranyi G, Cherrie M, Curtis S, Dibben C, Pearce JR. Changing levels of local crime and mental health: a natural experiment using self-reported and service use data in Scotland
J Epidemiol Community Health 2020;74:806-814. https://jech.bmj.com/content/74/10/806;
Language in Northern Ireland: Who has lost, gained or retained knowledge of Irish?
By Chris A Garrington, on 1 September 2022
In 2020, the New Decade New Approach (NDNA) deal for Northern Ireland outlined a strategy for the Irish language. Then in May 2022 the Identity and Language Bill was introduced in Westminster, providing for the strategy to be granted official status. But who knows Irish, and what changes have occurred? Dr Ian Shuttleworth discusses findings from a research project using Census data to look at changes between 2001 and 2011.
There is considerable political, media and policy interest in the use of both the Irish and Ulster-Scots languages in Northern Ireland. And because Census data from the Northern Ireland Longitudinal Study (NILS) tracks a large sample of the population – 28 per cent – over time, it can provide us with valuable insights above and beyond what official statistics offer.
We set out to answer a series of questions, aimed at adding new evidence on key socio-demographic, household, and health factors:
- Is Irish language knowledge associated with socio-economic status, type of household or health?
- How did self-reported Irish language knowledge change between the 2001 and 2011 Censuses?
- What changes could be observed among young people over the 10-year period in the knowledge of Irish language?
We found the main factors linked to having Irish language knowledge were being aged 11-15 years, being born in the Republic of Ireland, being Catholic, having no religion of upbringing (compared to Protestants), having Irish national identity, having degree-level education and living with others who had Irish language knowledge. However, about 8% of those who knew Irish in either 2001 or 2011 were Protestant.
We also found people living in the 20 per cent most deprived areas and those living in the West and South of Northern Ireland were more likely to have Irish language knowledge.
When we looked at change between 2001 and 2011, we found the highest proportion of people learning Irish were aged 3-19 years (13.6 per cent) in 2001, while the highest proportion of those losing Irish were those aged 11-15 years in 2001 (13.3 per cent).
Change was strongly connected with changes in religious affiliation: 45.7 per cent of those who said they were Catholic in 2001 but not in 2011 lost Irish over the same period, while 43.5 per cent of those who were not Catholic in 2001 but were Catholic in 2011 gained the language.
Over a 10-year period, around a third of Irish-speakers retained their knowledge, around a third lost it and a further similar number gained knowledge despite not having had it at the start.
For both Censuses – and for the previous one in 1991 – the peak age for Irish language knowledge was 13. In each Census year, between 20 and 30 per cent of 13-year-olds had that knowledge. But in each year, the proportion of those in their mid-20s who had it was much lower, at between 10 and 13 per cent.
Why would this be? The Census can’t tell us, but we can offer some insight – those with Irish language, like those with Ulster Scots, tended to have a higher level of education than the general population. Those who spoke Irish at age 13 or less would have been likely, over the next 10 years, to have moved on to university and possibly to be living in student accommodation or not in Northern Ireland at all. So the Census will have been unable to pick up some of them. We might also speculate, of course, that their parents may have marked them down on the Census as Irish speakers, whereas they may not have felt strongly as adults that they had that ability.
Given the policy context, our work has offered a useful picture of the socio-demographic, household and health associations of those who have knowledge of the Irish language, and we hope it will inform further Government initiatives in the future.
The research on profiling the Irish language in Northern Ireland was led by Dr Ian Shuttleworth from the School of Natural and Built Environment at Queen’s University, Belfast, supported by researchers in the Northern Ireland Statistics and Research Agency (NISRA) and endorsed by the Department for Communities. Dr Shuttleworth will be presenting the work at the British Society for Population Studies conference in Winchester on September 7.
Person or place? Finding out more about what drives health inequalities
By Chris A Garrington, on 25 July 2022
It is known that life expectancy is higher in some areas of the UK than in others. These inequalities in health are linked to the socio-demographics of the area: poorer health and shorter life expectancy tends to be a feature of less affluent areas of the country. In this blog, the third in a series on cancer and social inequality, Fiona Ingleby discusses research which uses data from cancer patients included in the ONS Longitudinal Study to assess the evidence on health inequalities and cancer outcomes.
The NHS has set out a plan for healthcare over the next decade that specifically aims to reduce inequalities. The research that this plan is based on uses area-level statistics. In other words, the UK is split into small areas about the size of a single postcode, and each area is given a score according to a variety of measures, such as how many people in the area are unemployed or on income benefits. These scores are used to identify the more deprived areas of the country that might be in greater need of healthcare resources.
This system is convenient and widely used, but by design it is a simplification of reality. Actually, communities tend to be a mix of people from all types of occupations, educational levels and income groups. So, what does the NHS’ plan mean for people who, for example, are unemployed or on a low income, but live in an area that is scored as being quite affluent overall? Will these people be overlooked by healthcare policy that is focussed only on less affluent areas? Or do these people experience health benefits from the overall affluence of the area they live in?
We set out to explore these questions in the context of cancer. Cancer is a major health issue in the UK, with a large proportion of NHS resources dedicated towards it. Around 350 people die of cancer every day in the UK. Strikingly, it has been estimated that as many as 1 in 10 cancer deaths in the first 5 years following diagnosis are due to health inequalities.
Our study used data from cancer patients included in the ONS Longitudinal Study to assess whether there is evidence for health inequalities in terms of cancer outcomes. We used census data to group patients according to their individual circumstances, including their type of occupation, the qualifications they held, and their estimated income. We used statistical models to determine if there were differences in cancer outcomes across these occupation, education, and income groups. Our analysis therefore considered health inequalities in terms of differences among individual circumstances, as well as using the type of area they live in.
The results revealed that there are inequalities in cancer survival across different individual circumstances, in addition to the well-known inequalities across different areas. For example, women on low incomes tended to have lower survival from breast cancer than women on high incomes. In addition, for both men and women with colorectal cancer, people without many qualifications tended to experience poorer outcomes than those with a higher level of qualifications.
We expected to find that inequalities in cancer outcomes would be larger when estimated at an individual level than when estimated based on area-level scores. This is because area-level scores are a sort of average across all the people within each area. Because of this, we thought that area-level scores might only show a diluted measure of the real, underlying inequality. But we didn’t find this. In fact, inequalities were of a similar size when estimated at an individual-level and at an area-level. This suggests that a person’s individual circumstances are just as important as the overall affluence of the area they live in. It is likely that both approaches, individual- and area-level, are useful and might highlight different types of inequality, rather than one approach being more accurate than the other.
We also used our statistical models to address the question of whether the health inequalities we found across individual-level circumstances were the same across more and less affluent areas of the country. Is the experience of a cancer patient on a low income living in an affluent area the same as a cancer patient on the same income but living in a relatively poor area?
We found that the answer to this question depended on the type of cancer. For colorectal cancer, for instance, the answer was simply that differences across individual circumstances were the same, no matter where someone lived.
On the other hand, for breast and prostate cancers, inequalities between people with different types of occupation were much bigger in the poorest areas, and much smaller in the most affluent areas: a kind of amplification of inequalities in poorer areas. This means there is a sub-group of individuals who are both individually deprived and live in a poor area who experience the poorest cancer outcomes by quite some margin. These people may benefit somewhat from existing efforts to allocate NHS resources to areas of the country in greatest need. However, it’s likely that even with such policies in place, these cancer patients will still be at a disadvantage compared to more affluent individuals within their community. Additional healthcare policies that focus on individual-level inequalities could be used to target all possible sources of health inequalities.
This kind of innovative approach to healthcare policy could help to reduce inequalities more effectively. Of course, individualised health policies are often more complicated and expensive to carry out than broader, area-level approaches. Our research helps to pinpoint specific types of cancer and types of inequality that could be targeted in order to make policies more cost effective. Future research of this kind, using individual-level resources like the ONS-LS, will help to identify specific policies that could help to reduce inequalities in health overall.
The project team at the Inequalities in Cancer Outcomes Network are Fiona Ingleby, Aurelien Belot, and Laura Woods from the London School of Hygiene and Tropical Medicine, Iain Atherton from Edinburgh Napier University, Lucy Elliss-Brookes from Public Health England and Matthew Baker from NCRI Consumer Forum. The research is described in two papers:
- An investigation of cancer survival inequalities associated with individual-level socio-economic status, area-level deprivation, and contextual effects, in a cancer patient cohort in England and Wales BMC Public Health
- Are deprivation-specific cancer survival patterns similar according to individual based and area-based measures? A cohort study of patients diagnosed with five malignancies in England and Wales, 2008–2016 BMJ Open
- Project website
Does social position affect our chances of contracting bowel cancer?
By Chris A Garrington, on 13 July 2022
We know cancer incidence is linked to socio-economic status, but that this differs according to types of cancer. In the second of three blogs on research using the ONS-LS to explore cancer and social status, Charlotte Sturley has examined diagnoses of bowel cancer, and found some clear evidence of a social effect.
Bowel cancer – also known as colorectal cancer – is the fourth most common cancer in the UK. Over 42,000 people are diagnosed with it in the UK each year so it is a major public health problem.
Cancer incidence varies between different groups of people, and differences have been found based on gender, age, ethnicity and where people live.
In England, for most cancer types, incidence is higher in the most deprived areas compared with the least deprived. The deprivation gap is largest for lung cancer, reflecting the fact that more deprived groups are more likely to smoke. Conversely some cancer types, such as breast cancer in females and prostate cancer in males, are more common in the least deprived areas.
The association between colorectal cancer and deprivation is less clear, and findings from previous studies have been inconsistent. In the 1980s, affluence was associated with an increased risk of colon and rectal cancer in Europe. But more recently, evidence has emerged of links between this type of cancer and living in a deprived area.
Understanding the causes
Given this apparent shift in the relationship between socio-economic deprivation and colorectal cancer, it is important for researchers to monitor recent data to see if the patterns are changing. We also need to understand the extent to which inequalities are associated with both individual and area-level factors to better target their underlying causes.
Most research on inequalities in cancer incidence has focussed on indicators of deprivation at area level, largely because cancer registries do not collect data on indicators of socio-economic position, such as the patient’s level of education or occupation.
Using Census Data to dig deeper
The Office for National Statistics Longitudinal Study (ONS LS) offers the opportunity to investigate variations in cancer incidence using information gathered in the census on individuals’ socioeconomic positions. My study used measures of educational attainment, occupational social class and housing tenure, along with an area-based measure of deprivation called the Townsend deprivation score.
My sample were LS members who were present at the 2001 Census and were aged 50 years or over, as incidence of colorectal cancer is very low among people aged under 50.
Among the study sample of 178,116 individuals present at the 2001 census, there were 4,418 cases of colorectal cancer recorded by the end of 2015. Because the ONS LS links census responses to cancer diagnoses, we could measure the average length of time between the 2001 census and the diagnosis.
The study found evidence of socio-economic inequalities in colorectal cancer incidence and that these differences varied by indicator of socio-economic position. LS members with a degree were less likely to have a colorectal cancer diagnoses compared to those without a degree, after accounting for differences by age, sex, ethnicity and area deprivation. A statistically significant association was also observed between housing tenure and colorectal cancer incidence, but only for those in social rented housing, who were at an increased risk of colorectal cancer compared to owner-occupiers.
Those employed in manual occupations were more likely to have a colorectal cancer diagnosis, compared to those in non-manual occupations – however this association was not statistically significant when adjusted for other variables. There was no statistically significant difference in colorectal cancer risk among study members in private rented accommodation compared to those in owner-occupied housing. No significant variation in colorectal cancer incidence was found by the level of area deprivation.
So, we can say individual measures of socio-economic position based on educational attainment and housing tenure are associated with colorectal cancer. My finding that there is not a link with area-level deprivation differs from other recent research which reported an emerging association between this type of cancer and deprivation, measured at the area-level. But these other studies used a different measure of deprivation which means comparisons are more difficult. The longitudinal nature of the LS data and the long-follow up period enabled time-to-event analysis to be employed in my study, whereas previous studies have tended to be more of a snapshot.
Not all individuals living in deprived areas will experience the same level of deprivation, and that could explain why I did not find area effects even though I did find individual ones.
One explanation for an association between colorectal incidence and socio-economic position could be different levels of exposure to risk factors such as poor diet or smoking. There is strong evidence to link socio-economic disadvantage with such behaviours.
My study highlights the complexity of the relationship between socio-economic circumstances and health outcomes and the need to investigate socio-economic inequalities by a range of different indicators in order to implement targeted policy interventions to reduce cancer incidence.
An interesting next step using the LS would be to investigate if and how change in individual socio-economic position and area deprivation over a person’s lifetime might influence their risk of having a colorectal cancer diagnosis. Linking the LS to data from the bowel cancer screening programme to investigate the impact of screening on colorectal cancer incidence and socioeconomic inequalities would also provide valuable insight.
Charlotte Sturley, who carried out this study as part of her PhD research, presented the work at the 19th International Medical Geography Symposium 2022, which is being held at the Royal College of Surgeons of Edinburgh from 19th-24th June
Her presentation is available here: (PDF) Contrasting socio-economic influences on colorectal cancer incidence and survival (researchgate.net)
Cancer risk and social status: what are the links?
By Chris A Garrington, on 23 June 2022
How does our social environment influence our chances of getting cancer? New research using Census data by Professor Robert Hiatt and colleagues shows there is a link between socio-economic status and cancer incidence, but also throws up some unexpected findings. In the first of a series of three blogs on socio-economic links to cancer, he discusses his work.
It’s well known that specific health outcomes are affected by socio-economic status – for instance the Whitehall IIlongitudinal study of civil servants clearly demonstrated this relationship as did subsequent Marmot reviews including the 2010 Marmot review of health inequalities. Recent research using the ONS Longitudinal Study has shown us in more detail the connections between socio-economic status and cancer survival, but we also wanted to look at how our background might be connected with the onset or incidence of different cancers as well as with their outcomes.
We now know that people from more deprived backgrounds have lower survival rates when they have cancer, although the causes of this are multi-faceted. We know that those with higher social status have access to critical knowledge, money, power, prestige, and social connections, which plays out to their benefit for cancer as well as any number of other health outcomes. There is a complex interplay between these fundamental aspects of their lives and their incomes, occupations, education, income, education, culture migration status and sexual orientation. And that in turn affects the material resources they have, their food security, their internet access, the type of healthcare they can access, their exposure to discrimination and stigma, and the support networks on which they can rely.
Those with lower socio-economic status on the other hand are missing out on many of the fundamental underpinnings leading to good health and in addition are more likely to be exposed to things like environmental toxins and climate change. In some cases, they may also be more likely to adopt unhealthy behaviours, such as smoking or a lack of physical activity. So, when it comes to the social determinants of cancer, the picture is multi-level, complex and a challenge to understand.
The ONS-LS can’t tell us everything about the possible links between social determinants and cancer, but it does allow us to look at different types of cancer diagnoses by occupational status, level of education and household characteristics, as well as by the type of area in which a person lives.
We were able to take data on almost 140,000 individuals who were alive in 1971, based on a one per cent random sample from the census, and look at the types of cancer diagnoses among them taking into account some of these social factors.
Two major questions
We asked two major questions: How does socio-economic position, measured both by area of residence and by individual characteristics, relate to cancer incidence and mortality? And how do social factors such as education, social class, occupation, home location and prior health status contribute to the relationship between socio-economic position and cancer outcomes?
We looked at all the major types of cancer – the biggest are lung cancer, breast cancer, colorectal cancer and prostate cancer. Overall incidence (onset of new cancers) tended to be higher among those with lower socio-economic status. But of course, behavioural factors such as smoking could be instrumental in that relationship and there is no data on smoking status in the ONS LS. So we looked again after eliminating those cancers with links to tobacco – and still found the same effect.
When we looked at individual cancers rather than overall rates, a rather different picture emerged. For those from lower social groups, there was a higher risk of being diagnosed with lung cancer, cervical cancer and stomach cancer. But those groups actually had a lower likelihood of contracting breast cancer, prostate cancer or melanoma. The relationship went in opposite directions depending on the cancer site. There were some also major cancers including colorectal cancer for which there wasn’t a clear social pattern.
We examined this relationship both through the lens of respondents’ income and education, and by the area in which they lived, and we came up with similar findings.
So what are the implications? At this stage, the full picture isn’t clear. We can say that overall, those from poorer backgrounds are more at risk of contracting cancer. And we know the reasons for that are complex. Lung cancer is certainly linked to smoking, so that must be a factor – and these social groups may also have higher exposure to the human papillomavirus (HPV), the major cause of cervical cancer, or to H. pylori bacteria, which live in the digestive tract and are linked with gastric cancers.
We can also say that those from wealthier backgrounds are more at risk from certain types of cancer. We may speculate that in some cases there will be behavioural and social factors at work – for instance, women who have children later in life or remain childless are at greater risk of breast cancer. These may tend to be the more highly-educated women who want to pursue their careers. But at present we don’t have any good theories on why prostate cancer might be more common among men from higher socio-economic groups. So our research raises questions as well as answers.
Our measures of individual socio-economic position are based on crude categories of income, education and occupational status, but may not give us a full picture. And we haven’t yet interpreted our findings in the context of the societal changes which has taken place since 1971.
There is much more work to be done in this area. We plan to publish three papers; one on socioeconomic status and cancer incidence, a second on socioeconomic status and mortality and a third comparing incidence with mortality. There will be detailed studies of individual cancers, too – so the ONS-LS will continue to prove a rich resource for cancer researchers in this and other areas.
Robert Hiatt’s presentation on Social Gradients in Cancer Incidence (and Mortality): the ONS Longitudinal Study was given on May 23, 2022 at University College London, and was based on forthcoming research with Nicola Shelton, Wei Xun and Eduardo Santiago-Rodriguez.
ONS research plays key role in children’s social care review
By Chris A Garrington, on 25 May 2022
by Fran Abrams
Three pieces of research using the ONS Longitudinal Study were cited in the recent report of the Independent Review of Children’s Social Care. The findings from the Looked After Children Grown Up project played a key part in providing evidence to the review – and were reflected in its report and the Government’s response.
The government’s 2019 Manifesto included a commitment to a review of the care system, and in March 2021 the independent review of children’s social care, led by Josh MacAlister, was given a year to produce a report.
The Looked After Children Grown Up project, which was funded by the Nuffield Foundation and led by Professor Amanda Sacker with colleagues at the UCL Institute of Epidemiology and Health Care and Kings College London, had already begun in February 2018 and was able to provide important evidence to the review.
The adverse consequences of being looked after as a child were already well recognised, but this research project was set up to address a lack of evidence on what happened to looked-after children later in life. Studies tended to follow them into early adulthood and no further, but using census data from the ONS Longitudinal Study the researchers were able to explore outcomes for those who experienced care from the 1970s onwards, up to the age of 50.
The study allowed researchers to track those who were children at the time of each census, and to identify whether they were living in residential care, as an unrelated member of an individual household, as a biological or adopted child in a parental household or as a child in a relative’s household.
By tracking care-experienced children into mid-life, the project was able to look at their later outcomes from a variety of different angles: it was able, for instance, to look at their likelihood of long-term illness, their employment, their education, their housing tenure, the type of family relationships they had and even whether they were at greater risk of dying early.
A fuller picture
The research, which was shared with Josh MacAlister at an event last July, was also able to drill deeper into the later experiences of children who experienced different types of care: it compared the outcomes of those who experienced residential and foster care with those who remained living with relatives, both parental and other.
Findings from the project were shared in an earlier Linking our Lives blog, and were welcomed by Josh MacAlister at the event last Summer.
In his report he focused on many aspects of the research, both citing it explicitly and reflecting its findings in his narrative.
In particular, he focused on a 2021 report from the study which showed lower rates of long term illness and higher rates of employment for adults with a history of kinship care compared to those that grew up in foster or residential care.
He also highlighted a second report from the team which showed that care leavers who were in residential care had the highest prevalence of limiting long term illnesses (around 32 per cent on average), followed by adults who lived in foster care (around 16 per cent on average) and adults who lived in kinship care (12 per cent on average). This was significantly higher than the average prevalence of limiting long term illnesses amongst individuals who had not been in care (7 per cent) , he said.
The review report also focused on mortality rates among care leavers from different types of care, and cited a 2020 report which used the ONS Longitudinal Study to link childhood out-of-home care status with all-cause mortality up to 42-years later.
It highlighted findings which showed adults who spent time in care between 1971-2001 were 70 per cent more likely to die prematurely than those who did not, and were also more likely to experience an unnatural death through self-harm, accident, mental or behavioural causes.
The review report made a number of recommendations which chimed with the research findings, including:
- Support for families to cut down referrals and help to keep children in their family homes or with relatives – £2 billion over five years.
- Unlocking wider family support networks including payments for relatives to act as foster carers.
- Support for a ‘new deal’ with foster carers to help larger numbers of children to be cared for in families rather than in residential care – 9000 new carers over three years.
In his response for the Government, the Education Secretary Nadim Zahawi promised more support for family hubs which offer early help and intervention. This would add seven new areas to an existing network of centres in 75 areas that receive a £302 million pot of funding for family hubs. A further 5 areas would receive part of a £12 million investment to deliver on a manifesto commitment to a network of family hubs around the country, he said.
Addressing concerns about the educational outcomes of children who had been in care – the research found those in parental care had a 28 per cent chance of achieving an NVQ level 3 qualification compared with just 11 per cent for those in residential care – Mr Zahawi promised funding for local authorities to help them keep vulnerable children in education.
Funding would be provided to local authorities for continued delivery of the Social Workers in Schools and designated safeguarding lead supervision programmes, which launched in September 2020, he added.
An evidence-based approach
In both the review report and in the Government’s response, there was a strong focus on the need for reforms to be underpinned by evidence.
The review suggested the Office for National Statistics should collect and report data on the mortality rate of care leavers and care leaver health outcomes, and that the Government should also launch a new cohort study which tracks the health outcomes of care experienced people and helps to gather other missing data on housing, education and employment outcomes.
In his response, Mr Zahawi promised support to help the most at-risk families to stay safely together, and a focus on early help, preventing them from reaching crisis point.
As part of this, he said, the government would set up a new National Implementation Board of sector experts and people with experience of leading transformational change and the care system. This would boost efforts to recruit more foster carers, increase support for social workers including on leadership, recruitment and retention, improve data sharing, and implement a new evidence-based framework for all the professionals working in children’s social care.
“Everything we do to raise the outcomes for children and families must be backed by evidence,” he said. “This report will be central in taking forward our ambition to ensure every child has a loving and stable home and we will continue working with experts and people who have experienced care to deliver change on the ground.”
The Looked After Children project involved Professor Amanda Sacker, Dr Emily Murray, Professor Barbara Maughan and Dr Rebeccca Lacey.
LS shows richer moving picture than Levelling Up White Paper
By Chris A Garrington, on 28 March 2022
by Ian Shuttleworth, Queens University Belfast
The Government’s new Levelling Up White Paper focuses attention on trends in the movement of people within the UK. Ian Shuttleworth was cited by its authors – and he says longitudinal Census data can show us a richer picture than was revealed by the document.
When civil servants from the Department of Levelling Up, Housing and Communities were putting together their new White Paper, I was one of the researchers they consulted.
Had migration rates within the UK been falling in the long term as they had been in the US, they wanted to know? The issue is a matter of concern because it is considered a key factor in levelling up the labour market between regions.
If people are unwilling to move, that can lead to them working at a level that doesn’t fulfil their potential. And it has an effect on places, too. Over the years, patterns of mobility have left the UK with ‘steaming-ahead places’ in the South East and ‘left behind’ places elsewhere, the argument goes.
In particular, graduates are more likely to move to find suitable work – that means some areas tend to be abandoned by the more highly-skilled, leaving them with depleted levels of human capital and lower productivity. This has negative effects for richer areas, too, because it puts pressure on housing costs and living standards, which can also lower growth and productivity.
NHS Register data
The White Paper quoted a research paper by myself and my colleague Tony Champion from 2016, which used data from the National Health Service Register on people moving between areas within the UK. However, the companion paper – which was also seen by the White Paper’s authors and which used Census data from the ONS Longitudinal Study, wasn’t directly quoted.
It’s worth looking at that second paper, though, because it gave a more nuanced picture.
The White Paper asks whether a long-term decline in inter-county mobility in the United States was replicated in the United Kingdom. It reports, correctly, that the overall level of internal mobility has fallen in the UK since 2001. It points to rising housing costs in London as a major factor, saying they lessen the wage premium and therefore dampen the incentive to move but that is not the full picture.
In fact, our LS research shows migration has dropped not just for graduates but for all groups of people. But we find that when it comes to moving, there’s still the same difference between graduates and non-graduates that there always was.
When we looked further back at the health service data to see if the USA experience of decline in geographic mobility since the 1970s had been mirrored in the United Kingdom, we were surprised by the result: In England and Wales there had been no substantial long-term decline in the overall intensity of between-area migration over that period, unlike for inter-county moves in the USA.
This was the case for both between-region migration and also the rate of migration within regions. There was a drop in migration for those aged 65 and over, but the rates for the other four age groups we looked at had been essentially stable or increased.
That surprising result was the reason we decided also to look at Census data from the ONS-LS: this confirmed the overall result but showed a more nuanced picture.
The Census data in the ONS LS showed us that as in the USA, there was a marked reduction in the level of shorter-distance moves – less than 10 kilometres – for almost all types of people. But in contrast to the US experience, the proportion of people in England and Wales making longer-distance address changes had declined much less.
So if we want to know what’s driving the trends, we could look more closely at the causes of the sharp reduction in shorter-distance moving in Britain as well as all moves in the USA. That, too, could help inform policy on ‘levelling up’ in the future.
Are People Changing Address Less? An Analysis of Migration within England and Wales, 1971–2011, by Distance of Move, by Tony Champion and Ian Shuttleworth, was published in 2017: . Popul. Space Place, 23: e2026. doi: 10.1002/psp.2026.
Can housing policies affect assimilation of the children of migrants?
By Chris A Garrington, on 17 February 2022
by Fran Abrams
Immigrant families often choose to live in neighbourhoods where there are others from similar backgrounds. But does this affect their children’s prospects? New research using the ONS- LS suggests policies aimed at desegregating neighbourhoods could make a difference.
Many immigrant children grow up in segregated ethnic enclaves, which raises a question: Does this have an impact on their cultural assimilation, and if so, how?
There is a difficulty in answering this question. New migrant parents who choose to live outside of such enclaves may be different from those who choose to live in ethnic areas. They may be better educated and may have a wider variety of different contacts. Maybe their language skills are better and that might open up a range of opportunities for them. But we don’t know which of those who do live in ethnic enclaves also have these skills.
Hard to unpick
So it’s hard to unpick what comes from children’s environment, rather than directly from their parents. It’s a complex picture.
But using data from the Census for England and Wales alongside information from the Fourth National Survey of Ethnic Minorities, which tells us about first-generation immigrants’ cultural preferences and their neighbourhood choices, new research has come up with some answers.
The researcher, Yujung Hwang, used information on families with South Asian origins living in England and Wales, and asked if the area where they grew up made a difference to their children’s prospects later on. She was able to produce a picture of the outcomes of offspring from Indian, Pakistani, and Bangladeshi families in Census 2011.
She asked if living in an ethnic enclave influenced whether these second generation migrants remained in a similar area, whether they kept the South Asian religion of their parents and whether they married someone from the same religion or ethnicity. For women, she looked at whether neighbourhood was linked to success in education and in work.
While some of those factors did not seem to be affected, she found, some of were significantly altered by neighbourhood effects.
Overall, the outcomes which showed the strongest effect were the likelihood of continuing to live in an ethnic enclave: those who grew up in such an area were 44 percentage points more likely to continue living in one as adults. Women who grew up in those areas were also less likely to graduate from college and less likely to go to work.
There were differences between the two age cohorts, though. Those born in the 1970s were 22 percentage points more likely to live in an ethnic enclave if they grew up in one. For women in that age group growing up in an enclave meant a 25 percentage point lower employment rate and a 29 percentage point lower rate of college graduation.
For those born in the 1980s, the only significant neighbourhood effect was on residency – the likelihood of staying in such an area was 55 percentage points higher for this group, though that could be at least partly because they were still unmarried and living with parents in 2011.
Once parental characteristics such as education, ethnicity, religion, employment status during childhood, and years since arrival were taken into account, there was no significant neighbourhood effect on the intergenerational transmission of religion or marriage preference.
This, the author suggested, could be explained by the fact that second-generation migrants were highly likely to remain and to marry within their parents’ religion and ethnic group, regardless of where they lived: there was no neighbourhood effect because both groups tended to conform to the ethnic and religious norm.
The results of the study point to possible policy implications: measures which support the ethnic desegregation of neighbourhoods could lead to the wider cultural assimilation of immigrant groups.
For instance, building social housing in diverse neighbourhoods could support the children of migrants in their educational and employment journeys.
Neighbourhood Effects on Intergenerational Cultural Transmission is a working paper by Yujung Hwang.