Health Studies User Conference 2026: Lightning talks

Okay everybody it's 11:35 so we will make start with our next session. Um my name is Mars Smith. I'm the director of health and biomedical surveys at the national center for social research. Um and I'm chairing the session of lightning talks um looking at data linkage uh on different longitudinal health related studies. Um so in this session we will have four short talks um and we will take questions at the end. So as we go as you have your questions put them in a Q&A box and we'll take them all at the end. So first up we have uh Richard Silverwood. Um welcome Richard. Uh Richard will tell us about uh health data linkage uh in at CLS's national longitudinal cohood studies and Richard is professor of statistics and chief statistician at USL center for longitudinal studies. He has applied interest across health and social research and his methodological work covers handling missing data the analysis of linked administrative data and various aspects of making causal inferences from observational data. So I'm handing over to you, Richard. >> Great. Thank you very much. Just put a timer on myself. Um, okay. So, thank you. Yeah, the the title here does major on health data linkages, but as I was putting this together, I realize there's a a lot to cover in terms of recent survey data as well. So, there'll be a bit of everything. I'll have to go through quite quickly because 10 minutes isn't very long. [snorts] Um, okay. So, what do we do at CLS? We run uh a number of different um uh nationally [clears throat] representative birth cohorts. So these are uh studies following individuals born in a specific period of time. So maybe in one week or in one year across the rest of their lives. So I'm going to be talking mainly today about these four core birth cohorts that you see here. So the first the National Child Development Study were all born in one week in 1958 uh across all of Great Britain. The British cohort study were born in one week in 1970. Again, across Great Britain. I'll skip to the Millennium Cohort Study who uh a sample of those born uh right across the UK between 2000 2002. Uh and then next steps were not recruited until they were age 13 or 14 in 2004, which means they were born in uh 1989 to 90, but we don't see them until they're about age 14. And all of these are around 16 to 20,000 individuals at initiation. So they're reasonable sample sizes. This figure shows um the ages or the points in time at which we've collected data on the individuals in these four studies. Um so you can see that for NCDS, BCS70, and MCS, they're we're collecting data at pretty uh regular points in time right across the life course. Uh but you'll see that for next steps, we don't collect any data up until their recruitment at age 14 as I just explained. And you can also see from this figure uh the current age that these cohorts are at. So NCDS are in their 60s and retiring or approaching retirement. BCS7 are in their 50s and next steps in their 30s and MCS in their 20s. So the data that we collect in these cohorts are, you know, very uh multiddisciplinary. Um but today we're going to be focusing on the the health data. We collect fairly um similar information across these different cohorts at similar points in time. Um but the the information that we collect obviously differs a bit with time. So we're interested in different things in childhood versus adolescence versus adulthood for example. Uh but some um some domains that we're or some constructs that we're collecting data on those ones down the bottom are things that we're collecting information on consistently throughout the life course. So you'll see that there's some or many different health and well-being measures that are a relevance today. Um so this figure shows the recent and near future data collections across these uh across these four studies. So the NCDS recent data collection uh was interrupted by COVID and eventually completed in 2024. BCS70 was out in the field recently until 2023. Uh next steps similarly 2022 to 2023. And MCS uh completed the most recent data collection in 2024. And all of [clears throat] these the data from all of these data collections are now available uh largely from UKDS. I'm going to talk about these in a bit more detail over the next few slides. Um, looking forward a little bit, we can see that later on this year, we have a planned joint data collection, a web survey between NCDS and BCS70. [snorts] And further ahead, still in 2028, we have a planned a27 survey in MCS. So the next few slides I'm going to very briefly give an overview of some of the health data in each of these most most recent sweeps of data collection. Uh so next steps age 32 these data have been available I think since 2024 now. Um there's lots of the the typical things that we ask about in terms of uh general and long-standing health problems some um anthropometry and lots about um sort of health behaviors. And importantly here we collected um saliva which means that we now have DNA and genetic data on next steps cohort members for the first time and I'll come back to this later on. um BCS most recent BCS70 sweep. These data have been available since last year. Again, a similar sort of um profile of information in terms of the health data across general and specific health conditions, health behaviors and some more kind of um age age specific measures. Thinking a bit about uh about menopause and fertility here. The NCDS data again they've been available since last year or the survey data at least. So this um slide is looking at the survey data. In the next slide we've got some separate biomedical data that were collected at a health visit. Again similar sort of profile of of variables here with some age specific um differences. Uh we had here a specific biomedical data collection from a a health visit. So we can see there's um a variety of different uh variables on the left hand side. information collected from a nur nurse visit. Again, there were bloods taken here and analyzed. So, we've got a number of biomarkers that you can see there listed on the right hand side. Um, there have been similar um biomedical sweeps, one earlier in NCDS when they were in their mid-40s and one at a similar age in BCS70, which means that we can do some nice kind of longitudinal biomarker analyses in NCDS and some crosscohort work too. um the biomarkers collected in in those different sweeps are um fairly similar but here the um metabolomics is a novel addition at this sweep. MCSH23 data only become available early this year. Again very rich data lots of different um health uh variables that you might be interested in uh with a few more age specific uh details there. So now we got on to the actual linkages. So um this is broken down by country in the first column because the the data are often available only on that basis. So in England we have hospital episode statistics available for all four of the studies that I'm talking about here. So that includes um admissions into NHS hospitals, outpatient visits and A&E attendances. Um so at the moment you can access all those data at UKDS. um that that will be transferring to UK LLC, the UK longitudinal linkage collaboration from that point onwards. And there's already some other health data, administrative health data that are available at UK LLC. So, I'm not sure if anyone's mentioned this yet, but UK LLC is a recent um collaboration and initiative um that's been used by UK longitudinal studies uh to help enact administrative data linkages. Until recently, it's only been available for COVID research, but this is now being opened out. So, it's become going to become a very rich data resource for the analysis of linked uh administrative data. Uh in Scotland, we have linked health data, linked health administrative data for MCS, NCDS, and BCS70. You'll note there's no next steps here because it's England only. Uh there's a variety of different health records that are linked which depends a bit uh on which study we're talking about and these are all available from UKDS. In Wales we only have linked health admin data for MCS. Um but the data that we do have are quite rich across a number of different admin sources and these are available either at the sale data bank which is based in Wales or UK LLC or uh UKDS as well. um genetic data. So all of the um four core CLS studies that I've been talking about now have genetic data available. As I say, that's only recently been the case for next steps. These [snorts] are free for researchers to access. If you want to get hold of the raw genetic data, big data files, then you can do so by an application to our data access committee. But [snorts] the major advent recently in terms of the genetic genetic data is the derivation and deposit of uh polygenic scores um across all of these four core studies for a number of different health and social phenotypes and these are all available from UKDS and this gives you more you know genetic data in a more manageable format for the I guess the the non-expert potential user of genetic data. Uh we've recently published a genomic data cohort profile in the international journal of epidemiology and there's a link on the left hand side there for our website which covers all of this in lots of detail. Um and as you can imagine there's some selection into who provides genetic data in our cohorts due to attrition and people not providing blood samples and we're in the process of uh deriving some weights which can be used to uh rewe the sample with genetic data back to the original cohort. Um and finally, just moving on from those four core um cohorts that I've been talking about so far, um very exciting recent news at CLS is that we've been funded um to set up and run for the first um two waves of data collection, the first UKwide birth cohort study in 25 years, which will be called Generation New Era. We're aiming to follow the lives of around 30,000 children who will be born throughout this year and their families. and the scientific focus is really on thinking about uh inequalities in in development and how you know subsequently that plays out throughout life. Um it be a nationally representative sample across the UK, boosts for ethnic minorities in England and Scotland, boost for low-income areas across all four nations and larger samples in the devolved nations relative to England. data collection starting at the end of this or towards the end of this year, but we're in a pilot phase of data collection at the moment. So, we're already very busy with this. Thank you very much. >> Excellent. Very much. Any >> any questions you have for him? Daniel, please put them in the Q&A. Richard, sorry, can you maybe mute yourself because I think it's creating an echo as well. Thank you. Excellent. Okay, so we are moving now on to understanding society and we have Jack Nishaw from University of Essex telling us about recent developments. So Jack is the executive director for understanding society and leads on his data linkage activity. He has over 20 years of experience working on the managing a wide range of national and international social science data infrastructure projects, services and tools. So handing over to you Jack. >> Thank you uh Mari. Uh so I have a broadly similar set of slides to Richard and and thankfully he's introduced some of the topics and and concepts I was going to talk about. Um understanding society uh sponsored by the uh economic and social research council undertaken at the run by the University of of Essics uh where I'm based. Oops. Ah is my is it moving forwards at all? No. Oh okay. Um I have three or four slides where I just want to talk through what the study is and then I'll talk about three or four slides where I talk about the recent data linkages and the uh and the things that can be done with that. Uh UK household longitudinal study um is a a household panel survey. Um we follow uh households annually. We have a um a questionnaire that's that's delivered annually. We also collect um we ask our respondents for consent to to make linkages and I'll talk about those later. And we also collect um from time to time bio samples and collect many of the kind of same analytes and genetic and epigenetic data that uh that Richard was talking about. Um we have various samples. The main study has been running since 2009. Um but we include if those of you might remember the BHPS. The BHPS is a sample that continues through the study. We have data for both participants right back to 1991. Um, and we have boosts for the uh constituent nations. So there's a Scottish boost, a Welsh boost and a Northern Irish boost which allows you to um anal analyze those countries um on their own with a sufficient number. And we also have an ethnic minority boost and a further immigrant and ethnic minority boost that allows you to um conduct analysis on um ethnic groups which are powered to a uh because we have a a larger end for those. We're a multi-topic multi-purpose study. So not not a health survey as such. We collect information on a wide range of areas. So uh listed there, education, employment, family and household and income, wealth and expenditure as well as health and well-being. We collect information on participants statuses, their behaviors, their attitudes, belief and expectations. And the idea is that this provides a kind of rich um contextual information that you can use if you're if you want to study the um the health data. you you can understand how that fits into the household, the family behavior, people's um labor market statuses um and partnerships and and histories like that. In terms of health, um we have a a health and well-being module that's covered each year. We have a a range of of kind of standard modules. So there's the we have the GHQ12, the Warrick and Edinburgh mental and health well-being uh scale, the SF12, uh the CAGE alcohol consumption um SDQ for measuring children's mental health and as I mentioned we um from time to time have a a biocolction wave. We've got one happening right now in wave 16 though will make the data available for that early next year. Um but we have a wave two uh and three biocolction which is now about 10 15 years old where we've taken a uh measured a range of blood analytes similar to the ones that Richard had listed um and also taken genetic and epigenetic data from DNA from uh those studies. Uh as with the cohorts, we also had a COVID uh s uh survey data that ran um study that ran between uh April 2020 and September 2021 um which you can you can use to kind of cover people's um health and well-being in uh in the COVID uh pandemic times. But to focus on the on the linked NHS records um so Richard explained what the LLC was. Um it's a place where now many longitudinal studies are putting their data to allow them to be linked to NHS records. Um I don't need to say that it was born in the COVID times because Richard mentioned that um but initially um and right now the people who we've got there who you can link are those who consented in the uh COVID survey about 8 and a half thousand participants. Um but we're excited that from this summer we're able to link uh a huge uh number more. So we have people who consented at wave 1, four and 16 of our uh uh study. Um they're now able to be linked to the NHS records which will take the end up to about 38K. So there's uh substantially more uh numbers that you can uh analyze and we're also adding the nurse visit data. So all those um blood analytes and biomeasures from wave two and three will also be added to the resource LLC. Um this is just a slide that gives you um a list of the linked data. Many of you who uh analyze NHS records will be aware of uh many of these data sets. All of these are currently linked for uh England only. Um, but the Welsh and Scottish uh linkages I think are working their way through the system and you should be able to um analyze Welsh and Scottish participants alongside English ones with their uh health records later in 2026. Um so just briefly what this means so that hard outcomes that you get from NHS records, the hospital visits, the cancer registrations, death, all that coded information um can now be linked with the wider survey responses. So that give you the the context of their individual family, labor market, um education, all that kind of rich information that um is not contained in the in kind of the administrative records is there to be analyzed alongside the um the harder the coded stuff. Um, the other thing that I think is worth saying that's quite exciting is the first time at the LLC, you'll be able to link not just um, NHS records with the survey responses, but there linking the DWP and HMRC data and I think later on the national pupil database data. So you ought to be able to very soon um not just have the survey responses and the bio measures but also the NHS records also track pe the same people through their benefits history um HMRC income data and education data. So it's we're kind of a a cusp of a really exciting time. Um so you know hopefully people will be able to get their hands on these all these data in the second half of 2026. Um this is my final slide. I'm think I'm just about on time. Um, didn't want to just talk about the LLC. I'm at a UK data service conference. So, these linkages are ones from Understanding Society that are available by the UK data service secure lab. Um, we've linked um consenters to their credit histories. So, we've got data from credit reference agencies um linking their personal finance, their mortgage um payment histories, their credit card debts, whether they've taken out emergency payday loans. Um we have a month monthly credit history data um and people are already using that to kind of link um debt to mental health. Um we've got a project currently done by someone looking at the personal finance of of people who are carers. Um so there's a rich sort of seam of of data there. We've also linked to Nest autoenrollment pension histories and again I think that's a a monthly data set. Um and so you can look again there people are looking at people's pension wealth um and those who don't have pensions um and comparing it and looking at you know their health trajectories through um we have property level data at the moment it's on a on a smaller study that we call the innovation panel um but we're linking energy performance certificates we've got smart um energy data we've got some data on air quality sensors and again all this kind of information can be used to to answer um questions in health research. Um the first two are already available by the UK data service secure lab. Um the third one is on its way to the UK data service um in the next couple of months. Um and that's my final slide. I've not been timing, but I'm hoping that's kept fairly close. >> That's excellent. You actually made up the time that you started late for. Thank you very much. really exciting ter developments for understanding society and data linkage. So I think we have some questions related to this as well later. Okay. But we are moving on. So um we now will hear from Steven Yash from University College London telling us about UK census longitudinal studies. Uh so Steven is professor of uh quantitative social science at UCR Institute of Epidemiology and Healthcare and is director of economic and social research council funded center for longitudinal study information and user support. So I'm going to hand over to you Stephen. There you go. Brilliant. Can you see that Mari? >> All good. >> Fantastic. Yeah, I managed to share the slides but then forgot to unmute myself and which I actually found quite difficult to do after I done the share. Anyhow, here now. Um, thank you Mari for the introduction. Um so as Mar says my name is Steven Chibraj and I um lead a ESRC funded investment um which is sort of broadly known as Celsius and as Mari said we um essentially provide um user support and access to something called the ONS longitudinal study um which is part of a consortium of three census based longitudinal studies. So they include the Scottish longitudinal study, the Northern Ireland longitudinal study and uh the study that we support, the ONS longitudinal study. And so um we come together under the banner of this um title, UK census longitudinal studies. Um if everything I've already said to you already um is completely old hat, then you can probably spend the next sort of 5 10 minutes checking your email. But if it's not and you're not aware of these things called censusbased longitudinal studies, I'd really appreciate your attention cuz I think these these data could be potentially really useful if you're someone who does research or is involved in commissioning research on the sort of broad area of um population health. So uh first question I guess you have if you haven't heard of these is what are they? What are these census based longitudinal studies? Well um clues in the title they are all census based. So they are linked individual records from the census in the case of the England and Wales data going back to 1971 in case of the northern Irish data going back to 81 and the case of the Scottish data going back to 1991. And so these records from the census are linked over time giving you a longitudinal history of a sample of people. Now what the real great thing about these data are which um are very much in contrast to the um co-op studies and the um uh understanding society data that we've just looked at is the beauty of these data is that they are incredibly large in size. So in the case of the ONS longitudinal study for example, it's a 1% sample of people who submitted a census return in any given census. So we're talking here of more than half a million people in any given census. And as I said, those records are linked over time. In the case of the Northern Irish and the Scottish data, they also have incredibly large samples. In fact, proportionally greater than the England and Wales data in that in Northern Ireland, it's about a 28% sample. And in Scotland, it's just over a um 5% sample. >> [snorts] >> Now, we don't have anything like the depth of information that's in the cohort studies or in understanding society. In fact, um if you're aware of census measures on health, um they are pretty limited to things about um long-term limiting illness and self-rated health. But we do have linked data which I'll describe in in the next slide in terms of looking at measures of health. But it is fair to say these data are are nowhere near um provide us in-depth information. But as I said, what they do provide is that really large sample size. So if you're interested in looking at underrepresented groups in society, so let's say you're interested in particular ethnic groups, maybe you're interested in people who provide unpaid caring, who are very underrepresented in society, then these census based longitudinal studies are really useful because we have sufficient sample sizes with which to look at those groups and look at what's happened to those people um over time. should the census measures have been collected um for uh for long enough. And also another aspect of these studies is we don't suffer from the same type of attrition as other studies in that the way people are um um put into these censusbased studies is by taking uh random birth dates across the year which are not known to um to myself and in fact are only known to a small number of people in each of the constituent statistical agencies in uh different parts of the UK. Um and I'll talk about why that matters in terms of data access um in a couple of slides time. But um the way in which people are included is if they have a birth date on one of these random uh dates, their census record is taken. And so in effect, what happens here is nobody knows other than those very small number of people who know what the birth dates are. Very very few of the people that are in the study actually know they're part of it. Um and no one knows if they're not part of it unless they know what the birth dates are. Um in terms of data linkage there is a whole host of data linkage across the three studies. Uh the core information that's available universally across all three. Uh the primary thing is vital registrations and vital registrations are used to provide data to researchers but actually also are really important in terms of maintaining the cohort. So the critical thing here for researchers and cohort maintenance is whether a member who is part of the longitudinal studies if they are known to have died then we don't need to retrieve their census records in in later time periods. So that helps with the cohort maintenance but also uh mortality data is provided in this data set for researchers to analyze both or cause mortality and cause specific mortality and historically those data have been available annually since co there's been a bit of a hiatus in terms of the availability of those but generally they are certainly up to date as of the latest census if not updated more regularly than that. Also um each of the studies has various health service information. So um across all three studies there are there's information on cancer registration um but in the Northern Irish and the Scottish data there's a whole host of other um health service information things like hospital episode statistics prescribing data um but those data aren't available in the England and Wales data at the moment. also weather information which perhaps may be not interested to health researchers as outcomes but potentially as exposures of predictors of health. So education data is available both in the Northern Irish and the Scottish data and there is also weather and pollution data uh which has led to some really innovative research both in the Northern Irish and the Scottish data and specifically in relation to the northern and Irish data there's also land and property service information. Just one thing I forgot to say in relation to the vital registrations. Um the uh members that are part of the um longitudinal studies if they are known to have a birth that information is also recorded in the uh in the data. Now this is the hardest part of the cell of this data which is data access which is much more incredibly difficult compared with um your sort of standard enduser license um data set that might be available on the UK data service which you'd only need to um have a UKDS account create a project and then probably sort of within the hour have your data downloaded the access to these longitudinal studies um are are much more um um involved um in that they require an application to use the data and they also require users to obtain ONS accreditated researcher status um which is it like I say it it it it it does require something a bit more but then um as I hope I'm trying to convince you it provides you access to a data set which has its um its own um unique selling points um but for each of the studies there is um quite substantial investment from the SRC which we're very appreciative of which enables us to help researchers throughout the whole project life cycle. So we recognize that accessing these data are difficult. So what we recommend to potential users is that they contact the support users in the first instance when they're interested in using these data. And once you make that contact, if you do want to submit an application and ultimately also obtain accreditated researcher status, each of the support units will guide you through that process and help you. And ultimately, when it comes to your project being approved, what will then happen is each of the support units for the constituent data that you've applied to will create a bespoke data set for you. So again, unlike data that you might download for the 1958 birth cohort from UKDS, which will be the same data set that everyone downloads for the sort of standard data. In the case of the longitudinal studies, every data set that a user um has is bespoke in that we only provide access to the variables that people um request rather than access to the uh the full data set. I should say that for anyone who's interested in doing research on individual census based records but doesn't require any longitudinal linkage over time or access to the data linkages that I've described on the previous slide. There is data available on the UK data service um um repository. Um in fact some of the most um least confidential data is actually available open access and the more confidential data with uh more detailed um variables and also larger samples is also available through the various um access um requirements on on UKDS. So if you are interested in using individual census based records then don't think that the um longitudinal studies are the only place in which to um to obtain those. Um final thing is just to uh flag up um where to look out for support for these data. So if you're interested in using one of the uh constituent data sets then uh these are the websites of the three support units. So, as I said, if you're interested in using these data, I'd strongly encourage you to reach out to one of the support units, again, if perhaps you're not a researcher yourself, but you're interested in commissioning data or commissioning research, sorry, um, but and you think these data might be useful for that that commission, then again, feel free either to encourage your researchers to to reach out to us or again, contact us directly. But if you go to any of those links, you'll find lots of information about the data and how to contact us and ultimately submit a project application. Thank you. >> Thank you very much, Stephen. And very very powerful interesting data source there. So I hope people to uh to take time to to get access to it for their research and dies very nicely as well to our next presentation and final one in this session uh which is by Esther Larry from Queens University of Belfast and uh speaking about nils that was just mentioned um and uh the linkage with Northern Ireland maternity services system. So Estelle is a director of the Northern Ireland Longitudinal Study and a senior research fellow in the school of natural and built environment at Queens University of Belfast. She's a health geographer specializing in health inequalities and life course modeling. All right, so I'm going to hand over to Esta. >> Yeah, thank you. Um, so this actually leads on then quite nicely from uh Steven's talk. Um, because I'm the I work on the Northern Ireland longitudinal study. So this is almost then like a case study of use and of the kind of things that you can do with um our data and the kind of things you can explore when you add on different linkages. Um yeah, so I'm going to talk a little bit about maternity research and how we can use the Northern Ireland longitudinal study to do kind of like to look at maternal um and child health. Um so I'll talk a little bit about the kind of objectives of this study and how it came about. Um I'll touch on the data although I think Steven has covered uh most of it. Um but I'll just talk about the kind of the added value that that linkage to the maternity services system has brought us and then I'll give a few preliminary results um although I'll not focus too heavily on these as um we're here more to think about what the data can do or what you can find out from the data. Um so just to give a little bit of a background then on this study. Um despite a decrease in the birth in the number of births in Northern Ireland, we found that the sociodemographic profile of women giving birth was actually changing and that was bringing um more complications uh and the actual needs of women giving birth were changing. So it's really important to consider this in the level of demand on the maternity and the health services that it's not just about the number but the the types of births too. Um the latest kind of figures in Northern Ireland at that time showed us that over half of women entering pregnancy were now overweight or obese. So that was from 202122 when we started the process to this new linkage. Um so that showed that almost 60% of women um were entering pregnancy with a higher BMI and we know that increased BMI during pregnancy can be linked to a number of short and long-term complications to both the mother and the offspring. So um it was really important to kind of address these uh in terms of complications it can be it can increase the risk of still birth um it can increase delivery risks um and it can also increase the risk of obesity to the child in the longer term as well as metabolic um abnormalities and and many other uh complications. And as well as that then it also places a significant burden on what is an already stretched health system. So it was um an important topic to kind of look at. Um our objectives then were were sort of twofold. The first and kind of primary objective of this was to really um test the linkage. Um so it was piloting this new pathway of linking the Northern Ireland longitudinal study and the Northern Ireland maternity system. Um so this really provides like an ideal and unique combination of socioeconomic and clinical data to allow um us to explore the most socioeconomic determinance of maternal obesity because we all know that administrative data brings us such a wealth of information and of knowledge. Um and as Steven said our studies are are really large sample size. So it does let you see um kind of under under represented grips. Um but sometimes the actual real value uh comes when we link data sets and we can um maybe see some hidden insights when we link things in different ways. Um in terms of the actual research objectives then this pilot study was aiming to to firstly just explore the demographic, social and economic determinance of maternal obesity or maternal overweight. Secondly, to really kind of um give something novel and something which isn't looked at a lot is we could identify a subop of women with more than one pregnancy. And we could actually look at this um interim period termed the inter pregnancy period and look and see whether women are losing weight, gaining weight or remaining the same between their pregnancies. um and then to look at some determinants of those kind of weight changes and see if they were different from those who maintained a normal weight. I'll probably just take you through the very um high level results uh in a moment, but if you are interested in any of those objectives, please do reach out. Um so, as Steven said, he talked a lot about the kind of the core data that we have available within the UK census longitudinal studies. So I'll maybe just touch on what um Northern Ireland has that's maybe a little bit different in our census in Northern Ireland. I guess the thing to highlight is the health questions. Um so we actually have uh a higher number of health related questions. Um I think there's 11 of them at 2021. We had a bit less at the 2011. Um but they will ask about very specific conditions as well. So we have um you know problems with vision, problems with hearing, breathing difficulties um and we have a a whole range of kind of questions about specific conditions. We also have the Northern Ireland mortality study which is like a sister study to the nils and that includes 100% of um those enumerated at a specific census time and any subsequent mortality uh linked on to that. So it um just gives a better way of looking at if your study is focused on mortality or cause of mortality. Um and then again we wanted to link this with the sort of the clinical information all the things we can get from the maternity notes. So this was before uh they moved to encompass. So it was all from um I'm not sure if in other parts of the UK you have the green folder but this was like the green maternity notes in Northern Ireland. Um and the way we designed it, so you see we're not using the latest census. Uh but we do have that available. It was just this study came before that that was available. So we took kind of demographic, socioeconomic, household and area level information um because it's also geographical studies or spatial analysis is um really good to do with these kind of UK census longitudinal studies. And then from the maternity notes, we took things like the number and the timing of the birth, any pregnancy complications, gestational age, birth weight, and maternal BMI. Um, so this first slide of results is just really showing like what could we actually track from using this data set. Um, so we took I forgot to say, but we took the 2011 census information and we linked that right up to I think it went to midway through 2019 at the time of the maternity notes. Um, so we were able to get 40 just over 44,000 pregnancies and that was to um just slightly over 31,000 mothers spanning that kind of 8-year period. And you see I've got in orange um is the whole Northern Ireland trends and uh you can see they're very slightly decreasing year on year and that is Mered in the the Niels members or the Niels pregnancies only a very slight decrease over the years. So just um kind of a way of checking that we are catching things which are representative of the total trends and also then looking at the parity of the mother when she first appears in the data set. So, we had um the majority were um the mother was appearing for the first time or for the first baby. We had about 8,000 appearing for the first time with her second baby. Um and then slightly less as it went on. And we've just rounded up. The yellow bar shows fourth and subsequent babies. Um and again just looking at the risk profile of pregnancy. So some highle results then. um things which probably will not come as a shock but is nice to see that these trends are kind of duplicated in our data as well. So there is a shift in the age profile of mothers. Um so we use the first year available which was the blue line 2011. The last year is the gray line of 2019 and the orange we took a mid which is uh 2015 and we can see that there is this shift um and the mean age changed from 28.7 to 30.1 um in just those 8 years and the clustered bar chart here is showing our obesity and overweight trends. So the blue um is the normal weight and you can see we're seeing a a decrease here in those who are normal weight going into pregnancy by 2019 and the black is the obese um box and again you can see this upward trend. Um also then I wanted to show this um I know it is a lot of numbers but I'll just talk through it just because this is uh one of the more novel things that you can actually get with this specific data linkage. Um so it's these inter pregnancy periods so between pregnancy 1 and two between two and three etc. And um although we have many smaller numbers coming up to that final interpreg period, the numbers for the rest are pretty robust and will allow some good analysis. Um so we have like almost 8,000 and then almost 4,000 for the first two interpreg periods. um you can see the kind of the mean duration um is slightly decreasing over time and the the inter pregnancy period one so between pregnancies one and two um we're seeing the highest weight gain. We also linked this in with deprivation measures and it was interesting to see kind of uh how people in certain areas or um the the trends of weight gain in those areas. So the next steps then are really to identify whether the social and demographic determinance of interpreg weight change are different from those who have maintained normal weight. So going back to the actual research and also then to use this data linkage pathway now that we have it more streamline for further research which might inform the new um Northern Ireland maternity strategy. And just to acknowledge then the um staff in the research support unit for really helping out with this linkage and pushing it forward. um and to my co-authors um Neil Roland as well. So, thank you