 Before we start, I'll give a couple of introductions. I'm Vanessa Higgins, and I work for the UK Data Service. And my co-presenter and our guest speaker today is Dr. Alan Marshall from the Frail Group at the University of Manchester. Frail stands for Frailty, Resilience, and Inequality in Later Life, and the group do lots of interesting and exciting research work on aging. So we're delighted that Alan can join us today. So this is what we'll cover today. I'll spend around 15 minutes introducing the UK Data Service and the data that we hold. I'll go through some useful data on aging, some useful resources, and also fairly quickly how you can access the data. And then Alan will cover the research potential of the data, giving some really good examples of his research from Frail. Now, we cannot hear you, so you need to type questions into the text box. We will have some time to questions at the end. He can type in questions at any time, but as I said, we'll take the questions at the end of the presentation. So first of all, what is the UK Data Service? Well, we're a comprehensive resource funded by the Economic and Social Research Council. We started about a year and a half ago, and we made up of the former services, the ESDS, the sensors.ac.uk, the question bank, and the secure data service. So these former services are now under one heading as the UK Data Service. So it's really good because now we provide a single point of access to this wide range of social science data. And the data is free at the point of use, apart from if you're a commercial user using it for commercial purposes. And the data comes with a really comprehensive documentation, which is a good starting point for those new to the data. In addition, we also provide support, training, and guidance in the form of online materials, so those are written on video, and training events. And we have a knowledgeable help desk as well. So if you're stuck with accessing the data or using the data, you can contact them. Well, who's it for? Well, it's basically for everyone to use. I've listed lots of different types of users here, but this is for everyone. So they're not just restricted to academics, which is often a misconception, although there are one or two exceptions to this rule where some of the data are only available to higher education or further education workers. That is the exception to the rule. So we have a wide range of different data types, including UK government surveys. So these are the large-scale, mainly cross-sectional surveys that are nationally representative. They're often funded by government departments, and some examples are the crime survey for England and Wales, the health survey for England, the labour force survey, and others along those lines. We also have longitudinal data sets. The longitudinal studies, as you probably know, follow the same subjects over time, and they allow researchers to analyse change at the individual level. We have several key longitudinal studies in our collection, including the birth cohort studies, the British household panel studies, and the understanding society. So these are large samples, and they're also nationally representative. We also hold data from the census. Now, this comes in a variety of different formats. It's the population census that's held every 10 years in the UK, and we have a separate part of the website dedicated to just the census data, because there are quite a number of formats that you can get the census data in. So if you're interested in the census data, I suggest you go to the URL, which is highlighted on that screen there. Cross-national surveys we cover as well. These are surveys across different countries where the same survey instrument and methods are used. So, for example, the Eurobarometer surveys, which monitor social and political attitudes and have been running across the EU since the early 70s. We also cover qualitative data, so this is non-numeric data. The most common kind of qualitative data is interview transcripts, but there are many other kinds, such as diaries, and they're also mixed-method studies, so qualitative and quantitative. We also have country-level macro data. So this is international data that's been aggregated to a country or regional level. It's time series data, and depending on the database, it will be available annually, at quarterly, or monthly. It's usually produced by international governmental organizations, such as the World Bank or the International Monetary Fund. And the databases cover a range of socioeconomic themes, such as industry, trade, employment, demography, and lots of other things. And then finally, we also cover business data. This is business micro data provided by the Office for National Statistics. Now, these data are relatively identifiable, so they have to be accessed through our Secure Lab, which I'll talk about later. Okay, so I wanted to give you some examples of data on aging. So this isn't a comprehensive list at all. There is a huge amount of data on aging available from the UK Data Service, but this is a list of some of the key and most relevant studies. I do suggest that you go and have a look for yourself using our search engine, but this will give you an idea and a taste of what's out there. So probably the key one is the English Longitudinal Study of Aging, which is the acronym that helps us. And this is a longitudinal study, and it's dedicated just to the subject of aging. Very wisely used. It's a really good place to start. It collects data from a representative sample of the English population age 50 and older when they start in a sample. And it started in 2002, though in 1998 through to 2001, it was collected on the Health Survey for England. It's now in way six, so it's followed participants over this whole period. So there's some really juicy information on aging in it collects objective and subjective data relating to health and disability, biological markers of disease, and other things such as socioeconomic circumstances, well-being, social networks. And Alan will talk a little bit more about that in his presentation because he uses ELSA in his research. So some other examples of data that you might want to use for aging are some of the longitudinal or cohort studies. So I think the longitudinal studies kind of lend themselves towards research on aging because they are following people all the time. So a good one that you could use that's kind of up and coming is the National Child Development Study, which is a longitudinal study that follows the lives of people living in Great Britain who were born in a particular week in 1958. So the reason I say it's up and coming is because the participants are now in their fifties, so it's quite a good one for aging research. And ELSA and the National Child Development Study are now becoming more comparable in terms of the questions and the instruments they use, so they can be used to validate one another's results. So that's a good one to use. I just wanted to highlight the work by CLOSA and the URL is there. So CLOSA is the Cohort and Longitudinal Studies Enhancement Resources. And it's funded by the ESRC and the MRC, which is the Medical Research Council. And it covers the nine birth cohorts that are funded by these councils. So as well as some of the data held but the cohort data held by the UK Day Service, it also covers other cohort studies, such as the 1946 birth cohort study, which is funded by the Medical Research Council and the people who were originally in that cohort are coming up to 70 years old now. So it's just to be aware that even though that's not a UK Day Service study, if you know what I mean, it's actually got some UK Day Service staff working on it and we do work closely with them. Okay, the British Household Panel Study, that started in 1991 and later became part of the Understanding Society study. And the aim is to understand social and economic change at the individual and the household level in Britain. And it follows people all the time, so it tracks people as they age. So that's another good longitudinal study. But as I say, there are quite a few that we hold and which might be useful. So go and have a look for yourself and I'll show you just over in a moment. Okay, so the cross-sectional surveys, which I've mentioned already. One of the good ones to use is the Health Survey for England, which is designed to monitor trends in the nation's health. And you can look at the health of particular age groups and particular conditions across different years. Also, every so often it has a boost and sometimes it has a boost of elderly people. So in 2000, it's included a sample of care home residents. And in 2005, it included a boost sample of people aged 65 and older who are resident in private households. So even though it's a little bit old that data now, you know, it's still there and there's a chance that there may be another boost in the future. Also, some other surveys that you think may not contain data because the titles don't sound like they do. So you might find that they actually come up trumps and have data on aging. So things like the opinions and lifestyle survey. In 2008, there was a quality of life module aimed at elderly people. And in 2011, there was a module on attitudes towards different age groups and societies. So it is worth having a good look around to see what's there. The census data can be useful for aging research because it covers the whole population. So for example, with the aggregate census data, so these are tables from the census. You can look at age distributions across the country or in small geographical areas. The things like limiting long-term illness, provision of care, things like employment and occupation. And also, the sample of anonymized records is microdata with a very large sample, so this can be useful for the elderly population as well. There's international aggregate data, such as the World Bank data on life expectancy. So this is available for over 200 countries, I think, and it's available for quite a number of years. So you can select the countries and the years that you want to make your own with these folk tables. And we do have a video tutorial on our website, which I'll show you the link to shortly, especially dedicated to this data. We also have some qualitative data. So for example, a really good and very well-known one that we hold is The Last Refuge by Peter Townsend and in the 1950s, he did a national study about the provision of long-stay institutional care for old people in England and Wales. So we do have some really good stuff in there. And then there's other more specific studies, so I've just got the example there, which I found very interesting was migration, nutrition and aging across the life course in Bangladeshi families. So I'm gonna whizz through now because I think I'm running over slightly. So to find data, if you go to our website, that's our main homepage and you'll see I've circled the Discover catalog. If you press Go here, you get taken through. And I've typed in here into the text box aging and I've pressed Go and it's searched. And you can see that it produces 4,505 results on aging. So there's an awful lot of stuff in there. But don't be overwhelmed with the number of results that you might get. Obviously here, we've got the first few that have come up are the really relevant, the else is to do wave one, note five and wave one, and then another aging project here. But what you can do is you can do an advanced search to reduce your options. You can narrow things down by refining things in the types and subjects and date boxes. You can also sort by other things other than relevance as well. So that's just to say, go and have a look and a play and don't feel overwhelmed with the number of results you get. Okay, so useful resources. We have some theme pages on aging. So again, back to our main page. And if you click on Get Data and go to Data by Theme, then you can see that we've got a number of themes. We're building these up gradually. And there's a few tabs here. I'm not going to go into the detail of each of these now because we don't have much time. But this is a really useful place to start looking for data on aging. If you don't know where to start in our website, then that's a quite good place to go to. Other useful resources. The links are in here. So the video tutorials. We have about eight up on the side for the moment, but we are expanding these rapidly. So these are kind of two or three minute video tutorials on really simple things like how to download data and how to access census aggregate data and things like that. We also have some written guides which are available through Discover. So we have things like a guide to waiting data, a guide to complex survey design, guides on specific data sets. So those can be useful. We have case studies of research on aging. Again, if you go through Discover and search on case studies, you can look at what other people are doing with the data. And these case studies are written in conjunction with the principal investigator or the author of the study as well as us. So yeah, so those can be really helpful to see what else people are doing, what work people are doing. We have some teaching resources. So if you want to teach with the data, there's a suite of web pages there to help you. And we have lots of things for new users. So we have some web pages with advice for new users. We also have some help pages and frequently asked questions. And if all else fails, we have a help desk, which if you go through to our help pages, you'll find information there about how to contact the help desk. And so if you're stuck with accessing or finding something or stuck with using the data, you can email us and we can help. So lack of advice from me, how to access the data. This is a real quick summary. Accessing Discover doesn't require registration, but to download any data, you do have to register. Registration and authentication is required for most data, but there are some data sets that are available under the Open Government license. So you don't need to register or authenticate for those. Most data is available directly from the website unless you want to apply for more detailed, more disclosed data, which are available from either the Secure Lab or the Special License. Data-free apart from commercial usages and there's some important web pages for access. These slides will be up afterwards so you can always access the URL first. And, sorry, the access page is on the website. If you go to get data and then how to access then that will tell you everything. Sorry about that. Okay, I shall just pass you over to Alan. Okay, well, first of all, thanks everyone for joining this webinar on aging. I'm Alan Marshall. So I'm one of the research and authority resilience and inequality in a later life project. And what I want to do today is talk to you about how data from the UK data service can be used to explore health inequalities in later life. So I'm going to be drawing here mostly on research from the frail project and I've put a link there if you want to find out more about that. But I should also acknowledge my collaborators who worked with me on this, who are listed at the bottom of the slide there. So the aim is to show the research potential of data on health and circumstances of older people. And I'm going to focus on three research themes to do this. So the first is to look at trends in the health of older people. And here I'm going to focus on frailty as a health outcome. And I'm interested in looking to see if there are differences in frailty across cohorts. So in more simple terms, if we compare frailty of a 60-year-old in 2002, is it the same or different to the level of frailty of a 60-year-old in 2010? Second theme is retirement and health. So here I'm interested in whether we see improvements or deterioration in health after retirement compared to before. And I'll be looking at a self-reported measure of illness in that part of the presentation. And finally, I want to look at national context and health care. And in this part of the presentation, I'm going to focus on hypertension or high blood pressure and the management of that condition in the US and England, two countries with very different health care systems. So the main data source I'm going to use is the English Longitudinal Study of Aging, which I might refer to as ELSA throughout the presentation. I'll also use the Census and the Health and Retirement Study, which is a United States harmonized version of the English Longitudinal Study of Aging. So first of all, why should we be interested in health inequalities in later life? Well, one reason is that the burden of poor health is greatest at the older ages. So if we're interested in tackling health inequalities and understanding of health inequalities in later life is really crucial. Recent research has suggested that inequalities in health continue to grow with age into later life. An absolute socioeconomic inequality and mortality lies with age. And one theory that's been proposed for this is a life course theory, which suggests disadvantage or advantage is accumulated over the life course from childhood to the older ages leading to stark inequalities in later life. So if we look at two examples of these, if we compare life expectancy at 65, we see that this is 21 years in Harrow compared with 14 years in Glasgow. And as I'll show you later in this presentation, we see a 10-year gap in levels of frailty between the richest and poorest older people in England. So Vanessa's already given some information on the English Longitudinal Study of Aging, so I won't spend too long on this slide, but it's a panel study. And in the first wave of the study, we had approximately 11,000 people who were aged over 50. The survey involves face-to-face interview, but there is a biomedical assessment carried out by NERS every four years. The survey, as you'll see later, contains a really rich set of information from demographics to social relationships to cultural participation. And there are sister studies in different countries, for example, the Health and Retirement Study, which contain comparable information for cross-national work. So I'm going to move to my first theme now, which is trends in frailty in older people. So we've seen a steady increase in life expectancy over the past century, and commentators have noted that this will bring a set of associated challenges, for example, linked costs of care provision. But the extent of these future care challenges really depends on how the health of the older population changes. So if you look at projections of care costs, these are really sensitive to the assumptions on the health of the older population in the future. But the evidence on these trends is really mixed. It depends on the methods that are used, the country, the particular health measure, and vary according to social class. So in this first part of the presentation, I want to look at whether there are differences in the levels of frailty across cohorts. So does a 70-year-old in 2002 have the same or different levels of frailty to a 70-year-old in 2010? And do we see differences in these frailty cohort effects according to wealth? I'd like to, if I do that, include a poll in this. So what do you think? Do you think that an average eight-year-old in 2002 is more, less, or equally frail compared to an average eight-year-old in 2010? So I'll give you a few seconds to vote on that. Some people still voting here. Gradually going up, we've got 77% voted. Okay, I'll give a few more seconds and then... Yeah. I'll close the poll now. So the results of that show me that most people think that an eight-year-old in 2002 will be more frail than an eight-year-old in 2010, which is very impressive because that is actually the correct answer. That's what our research also tells us. But what our research using also tells us is that this is particularly the case for the poorest individual. So I'd like to show you the research which illustrates this. But first of all, I have to spend a few minutes talking about frailty itself. So specific definitions and models of frailty are contested. But there is broad agreement that frailty is a non-specific state, that it reflects age-related declines across a range of systems and that these lead to adverse outcomes, for example, mortality and hospitalization. There are many ways that frailty can be measured. I'm using here a measure as a frailty index developed by Ken Rockwood and Coise in Canada. Now, this is based on what's an accumulation of deficits. So these deficits cover a range of domains, such as activities of daily living, cognitive function, falls, depression, joint replacements. And all of these deficits are measured on a 0-1 scale. So if they're equal to one, that indicates the person has that particular deficit. And the frailty index is then just the proportion of deficits held. So the closer this proportion is to one, the more frail that individual is. So I'm interested in whether frailty changes across cohort. And I want to show a hypothetical example of how I model this. So here I have a graph which shows age along the horizontal axis and the level of frailty on the vertical axis with more frail being higher up at axis. And I've got two lines here. And these lines show the average or mean level of frailty for two cohorts as they pass through the alpha survey between 2002 and 2010. So the red cohort are a more recent cohort. So at the start of the survey, they were aged 62. And you can see that by the slope of that red line, the level of frailty for that cohort increases over the course of the survey up to the point where they reach age 70. So the blue cohort shows a later or older cohort. They were 70 in 2002. And you can see that their level of frailty also increases through alpha, so that it's a faster rate, which makes sense in that they're an older group. But we can compare the level of frailty in each cohort here at age 70. So this is the same age, but different time points. And what we see is that this shows quite an optimistic scenario. So 70-year-olds in 2010 in the red line are less frail than 70-year-olds in 2002. So this is optimistic in the sense that lower care needs in the future, it would imply lower care needs in the future as older people are becoming less frail over time. But clearly we might see a situation where those lines overlap, which would indicate no change or where the 70-year-olds in 2010 were more frail than in 2002. So let's go on to some results. This slide shows frailty trajectories by cohorts using LC data for all people. And again, I have age on the horizontal axis and frailty on the vertical axis. The higher that axis, the more frail the person is. So if we focus first on between the ages of 50 and 70, we can see that the change in frailty for the cohorts overlap, which indicates there's no change in frailty over our period 2002 to 2010 for those age groups. But if we actually look at the older age groups, 70 to 90, we see higher frailty in more recent cohorts. So a pessimistic result. It's interesting to look at how that varies by wealth. So here I show two sets of results for poor groups, which the poorest quintile of the health sample and the richest quintile. The poorest quintile have higher levels of frailty in general compared to the rich. And what we can see from this is that the trajectories or changes in frailty across cohorts for the richest group overlap across all ages. So for the richest, there's no evidence of change in frailty across cohorts at any age. But for the poorest individuals, we see higher levels of frailty in more recent cohorts compared to later cohorts. So it's the poorest individuals that are driving this cohort difference that we are observing. I've highlighted two lines here in red, and this is to show you this 10-year gap in frailty between the poorest and richest groups. So we can see that the evolution of frailty for a poor individual who's aged just under 70 is very similar to that for a rich person who's aged just under 80. So it shows a stark extent of inequality in frailty in England. So in summary, we see comparable levels of frailty across cohorts at the younger, well, at the ages of 50 to 70. And then higher levels of frailty in more recent cohorts compared to later cohorts over the ages of 70. And this is driven mostly by the poorest groups. And it's a pessimistic outcome in the context of rising life expectancy. But it's not a finding that hasn't been shown elsewhere. So a similar finding has shown in the U.S. And if we think about why this is happening, one reason could be that improvements in medical and care services across the life course are improving the survival probabilities for frail individuals over time. Or alternatively, we could be seeing the effect of unhealthy lifestyle choices in the more prevalent in young cohorts. But what this results, this theme does seem to be showing is that social conditions are influencing the rate of frailty or deficit accumulation in older populations. And the stronger cohort differences for the poorest may reflect deterioration in their relative socioeconomic circumstances. So I'm gonna move to my second research theme now, which is looking at retirement and health. I've got a set of research questions here. Does retirement have an effect on subsequent health? Does any retirement affect vary according to the type of work? My proposal is to increase retirement age, exacerbate health inequalities at older ages. Elsa is really well suited to answer such questions. We have information on retirement, why people retired, and detailed information on the work characteristics and health and circumstances of older people as they approach and pass through retirement. And we can also use other sources. So the census enables us to look at subnational variation in patterns of self-reported illness at retirement. And I want to start off by looking at some census data. But before I do that, I'm going to give another poll. I think this is a slightly easier question to answer. So after retirement, do you think that an individual self-reported health will increase, decrease, or stay the same? So the poll is open. I'll give you a bit of time just to answer that. Okay, we've got 63% voted. Do we count down maybe five, four, three, two, one? We'll close the poll. Okay, so the results of that, most people are saying that self-reported health will deteriorate after retirement. And again, everyone's exactly right with that. That was the result that we found in our research. But interestingly, we found that the deterioration in self-reported health actually increased at a slower rate after retirement compared to before. And if we looked at people in manual occupations or routine occupations, we actually found evidence of a leveling off in self-reported health after retirement. So I'll go over the research that shows this. So here I've got some census data for three districts in England and one, two districts in England and one in Wales. And along the horizontal axis, I have age and along the vertical axis proportion with a limiting long-term illness. And you can see here that we have, we have very different levels of limiting long-term illness in each district. The similar patterns with age, we have an increase with age, but we see that levels of illness are much higher in Merthyr Tidville compared to Berry, which is higher in turn compared to South Buckinghamshire. But what I think is particularly interesting is the differences in the shape of the limiting long-term illness curves at retirement. So if we focus on South Buckinghamshire first, which is the bottom thin black line, you see that there's a smooth increase in limiting long-term illness throughout retirement. In Berry, the dotted black line, we see there's a slowing down of the increase in illness rates after retirement. Whereas if we look at Merthyr Tidville, we actually see a drop in the levels of illness after retirement, which might imply an improvement in the health of people after retirement compared to before. So I mapped the different patterns of limiting long-term illness after retirement and dividing districts into three groups. So the white groups here show districts like South Buckinghamshire, which have no post-retirement health improvement. The dark gray parts of the map show the districts that had a large improvement in limiting long-term illness rates after retirement. And we see that those areas that have the improvement in health after retirement coincide with more deprived and often formerly industrial areas, whereas the districts that had no improvement in health after retirement are clustered around the more affluent southeast around London. One of the problems with this census analysis and the conclusions that I've tentatively suggested is that we have different populations on either side of retirement in this aggregate data source. So the census is just a snapshot. So we can't be certain that a process such as differential mortality in districts such as Merthyr Tidville or differences in migration at either side of retirement are causing the differences we observe. So to answer this kind of question, we need data on individuals. We need to track individuals through retirement and see how levels of limited long-term illness or their risk of having a limited long-term illness changes as they move through retirement. And this is what we've done using the English Long and Tunal Study of Aging. So here in this graph, which is based on ELSA data, along the horizontal axis, I have time to retirement, going from 10 years before to 10 years after, and I've marked the point of retirement in the middle. On the vertical axis, I have the probability of having a limiting long-term illness. And you can see two sets of lines here. The jagged line show the observed probabilities of a limited long-term illness on the English Long and Tunal Study of Aging as a person moves through retirement. And the solid line shows the model probabilities. And what we did was to fit a slope representing the increase in illness rates leading up to retirement, and then a slope representing the change in illness rates after retirement. And we can see here that for the general population, there appears to be some evidence of still increases in risk of limiting long-term illness after retirement, but that increase is at a slower rate compared to before retirement. We can look at that according to different types of occupations. So here I present the same kind of information for people working in managerial and professional occupations and those working in routine occupations. So in the routine occupations, we see a steeper increase in probabilities of illness in the lead-up to retirement, and then a leveling off in the rates of illness after retirement. Whereas for management professionals, we just see a smooth increase in illness rates throughout retirement. So this in some ways corroborates the spatial research using census data because we know that those in routine occupations are more clustered in more deprived parts of the country. So just to summarize that theme, we saw strong spatial distributions in patterns of illness rates at retirement. And for individuals working in routine occupations, there were faster increases in the probabilities of having an illness in the final years of employment and then a leveling off in probabilities of illness after retirement. And this is in line with other research such as that by Hugo West, London colleagues in France. And what it suggests from a policy perspective is that increasing retirement age may well exacerbate inequalities we see in self-reported health at the older ages because this improvement we see for those in routine occupations, the improvement in illness rates after retirement will be delayed till older ages. So the final theme that I want to look at is aging in a national context. And here I want to focus on hypertension or high blood pressure healthcare. And I'm going to compare the US and England and ask the question of whether different healthcare systems lead to different care outcomes for high blood pressure. We know that the US system is dominated by private healthcare provision, especially in the age 65, whereas England has a universal healthcare cover through the National Health Service. And in order to answer this question, I'm going to combine two different data sources, the English Longitudinal Study of Aging and the Health and Retirement Study which do contain harmonized information which facilitates this kind of comparison. So a bit of background on hypertension. One billion people worldwide have a condition which is usually asymptomatic. And the rise in the hypertension population has been attributed to increasing longevity and population growth. But also unhealthy lifestyle factors have been linked to this. Hypertension is a key risk factor for cardiovascular disease and subsequent mortality from these diseases. But hypertension is controllable and it's cheaper than interventions to deal with subsequent health problems. So it's a good help for them to look at to compare different health systems in that it's asymptomatic and it's also controllable. So here's a question just to start off this section. What do you think? Do you think that levels of uncontrolled hypertension are higher or lower in England compared to the US? So by uncontrolled hypertension, I mean a situation where someone has been told that they have hypertension in a previous, but by a doctor, that they have measured high blood pressure at a later date. So hypertension is high blood pressure? Yeah. 53% voters going up quickly now. We do a countdown. Five, four, three, two, one. We'll close the poll. Thanks, everyone. Okay, so this one, the correct answer here, is actually the same, which is the least popular answer here. So let me go over now the research to show how we looked at this. So the data that we used to look at this, as I've said, is the Health and Retirement Survey and the English Long Tunel Study of Aging. The good thing about both these surveys is they include nurse visit where blood pressure is measured. That allows us to get a measured blood pressure through which we can diagnose high blood pressure, which we use a cutoff of 140 systolic blood pressure or 90 diastolic blood pressure. We also have information on diagnosis of hypertension. So we know if people have been told in the past that they have high blood pressure by their doctor. So in this research, we're working with a total hypertensive population. This is anyone diagnosed with hypertension or measured with high blood pressure during the survey. And what we do is we try and predict in each country three hypertension care outcomes. Hypertensive control. So this is people who have normal measured blood pressure in the survey, but they've been diagnosed with high blood pressure in the past. That's a good care outcome. Hypertensive uncontrolled. This is where someone was measured with high blood pressure in the survey that they've been diagnosed with high blood pressure in the past. So that's a bad outcome. And the final bad outcome is hypertensive undiagnosed. This is where a person was measured with high blood pressure in the survey, but they'd never been diagnosed with high blood pressure in the past. So here are some model probabilities of each of those categories controlled, uncontrolled and undiagnosed hypertension. I've slipped the results here into the under-65 group and the over-65 group. So focusing on the under-65 group first, if we look at controlled hypertension, which is the blue bar, you can see that there are higher levels of controlled hypertension in the US compared to England. But this difference isn't driven by uncontrolled hypertension. The levels of uncontrolled hypertension are similar in each country. What does drive the difference is undiagnosed hypertension. And we see that that is higher in England compared to the US. And a similar set of results apply for the over-65 population. We can look at how this differs across US insurance categories, private insurance, government insurance, which is basically Medicare and Medicaid, and those who hold no insurance. So if we compare private insurance and government insurance, if we look at controlled hypertension, the blue bars, although there looks to be a difference there, we didn't actually find anything statistically significant. So comparable levels of controlled hypertension. And similarly, we found comparable levels of uncontrolled hypertension across US private and government insurance. The key difference that we did note is that undiagnosed hypertension is much lower in the US government insurance group compared to those private insurance. If we look at the over-65 group, we can see that comparing the US private insurance and the US government insurance group, the hypertension health care outcomes are very comparable. So if you are aged over 65, there's no clear advantage here in terms of hypertension care of buying private insurance. So the final slide I want to show in these results is just looking at model probabilities of these care outcomes across wealth quintiles. So in this graph, I have wealth quintiles going from one, which is the least affluent group, to five, which is the most affluent group in England and the US. The black lines show the prevalence of controlled hypertension, the gray uncontrolled hypertension, and the dotted black lines is undiagnosed hypertension. And the key point I want to focus on here is for the US. So if you look at the solid gray line here, which is uncontrolled hypertension, you can see that the probability of uncontrolled hypertension declines with increasing wealth. And that isn't the case in England. So we see here that the English health system appears to be more equitable compared to the US system in terms of uncontrolled hypertension. So a few conclusions. There are potentially lower risks of undiagnosed hypertension in the US compared to England. And we think that that stands for some differences in guidelines around diagnosis and treatment. And we would note that there isn't really clear evidence that treating people who are borderline hypertensive has any advantages, particularly at the very oldest ages. There's no clear advantage to private health care systems and hypertension care appears to be more equitable under government-funded systems such as the NHS. So here I've talked about a national context. You can also use data sources such as English, French, international studies for aging to look at local contexts. So how aspects of neighborhoods such as neighborhood deprivation influence health inequalities. So here we see two images in the top left topstiff, which has a life expectancy of 70, and in the bottom right of Kensington and Chelsea, which has a life expectancy of 85. And you can see the very different environments. And a number of researchers have suggested that aspects of neighborhoods have an effect on health over and above individual characteristics. And in the bottom of the slide, I just give you an example of some research on the failed project, which we've done to look at this. So in conclusion, I'd argue that a complex set of factors contribute to the health inequalities we observe in later life. So social-recognized circumstances, events such as retirement or the death of a spouse, local and national contexts, and also earlier life-course circumstances. And these are mediated by genetic and metabolic factors. And on the failed project, we are trying to link all of these factors together. Longitudinal data sources such as English, Longitudinal study for aging, which now has five waves, can allow us to model changes in health over time and test causal hypotheses. And we can combine this data with other sources from other countries from the census to develop deep understandings. So with all this data, it's really quite an exciting time for research on aging. And I'll finish there. That's brilliant. Thank you, Alan. Thanks so much for coming, telling everybody about your research and giving up your time to do so. You just bear with us if there's a few noises in the background while we just shift around, because we're going to take the questions now. So I think we have a few questions coming in just before we start looking at them. They'll just say that the slides will be up on the website with the recording of this webinar as well. So you can go back and look at, for example, the references that Alan put up, and this should be up within the next few days. Okay, so we had a question about whether or not there was a longitudinal study of aging for whales. Not that I know of. I've never heard of one. I think so. No. I'm sure it would have come up strongly in the search if there was one. We've not heard of one. There is the Health Survey for whales, the Welsh Health Survey, it's called. So you might want to go and look at that and see if that has any topics on elderly people or aging. What else have we got? Is it possible to get soft copies of the slides after the session? Yes, covered that one. Yeah, there's a question for Alan in here. Oh, has it gone? Which potential confounding variables are included? Right, do you want to? Yes, I've had a question here which asks about potential confounding variables, and I think this concerns the research on cohort effects in frailty. So that particular search doesn't control four aspects such as education, lifestyle, or smoking. So the only really social economic variable that is in there is wealth. And part of the reason for that is an issue of sample size. So the more variables that are included, the sample size drops and reduces the power to kind of draw conclusions. So it'd be really interesting to try and look at include variables such as smoking and drinking or education to try and understand the differences further, but it's quite problematic to do that. Okay, thanks Alan. Just looking through these. Somebody's actually made a useful comment here. There's a Scottish Longitudinal Aging Study in Scotland which is due to pilot this year. But some useful information. Have a look. A lot of people are asking for the slides. Do you know of any good data sources for health and safety at work? I think the labour force survey covers that. I'd have to go away and have a good look at the data sources, but I can get back to you on that one. Is that one for you Alan? So just bear with us while we're reading them. So there's a question on an expected release date for Elsa Wave 6, which I'm afraid I can't answer exactly, but I think it will be very soon. Okay, great. There's another question on which data are open data. The data that are open are census aggregate data. So the census tables, we've got some of the World Bank data. The international aggregate data is open and we have some teaching data sets which use survey data and also some qualitative teaching data sets that are open data as well. So you can get those without registering. Okay, there's another one for Alan here. Have you considered variations in inequality between urban-rural areas? Yeah, so we haven't considered urban-rural differences. We have looked at differences according to deprivation and we focused in that particular piece of work on looking at depression and we tried to see if there was a neighborhood health effect on depression in terms of area deprivation. It will be possible to look at urban-rural inequalities. To do that, you would need to apply to have a version of Elsa that can be linked to some measure of the level of rurality or urban-ness in the particular areas that you're interested in. So that will be an interesting and possible research question to look at. Thanks, Alan. Can the HRF be accessed through the data service? That's the U.S. study, isn't it? No, it can't be. It's accessed through the equivalent of the data service in America which is at the University of Michigan. Did you analyze black and white people separately for the U.S., Alan? Yeah, so when we looked at the results that I presented for hypertension did control for ethnicity in the U.S. and England. It was very broad. We just divided the groups into white and non-white in each country. Okay, great. There's another one about ethnicity. Oops, keep flicking back. Is there evidence of the effect of ethnicity on inequalities in health outcomes? Yeah, so one of the problems with using Elsa to look at ethnic inequalities is that there's a very small sample of ethnic minorities. So a better source to do that is going to use one of the Health Survey for England data sets where they have a boost, a boosted sample of ethnic minorities. Okay, I think there are a couple more questions, but what we'll do is try and answer them separately on email because we have your email addresses from registration. So I think because we're getting close to three o'clock now. So I'd like to say thank you to Alan for coming and giving up his time to do this. It's been really good of you to show everybody how these lovely data sets can be used. And thanks to everybody for attending. And yeah, the slides will be up on the website. Come back and attend more webinars. Okay, thank you very much. Bye.