 Hello and welcome to today's mental health data webinar. Now, our kind of plan for today, we promise will be finished by 3 o'clock. I'm going to talk first, I'm Sally McManus, I'm based at the National Center for Social Research and I'm going to talk about a range of different cross-sectional surveys but focusing on specialist mental health surveys. And I'll also say a bit about the Longitudinal Understanding Society Survey. Then Owen McElroy from Leicester University is going to talk about the birth cohort studies for about 20 minutes and then we should have about 15 minutes which is going to be facilitated by Sarah King Heal who is a UK data survey who's organised this whole webinar. But if you have questions as you go through, you'll see the panel on the right-hand side. There should be a box where you can key in questions. So as we're going through, while we can't take the questions as we go through, we will have that time at the end to go through them. And if there are any questions that we don't get to, we can see who's asked the questions and so we can email you later and do our best to answer that way as well. Okay, so today is part of a series of webinars. The other webinars in the series are focusing on things like political behaviour, religion and languages being spoken and those webinars are going to be taking place in the new year. So keep a lookout for that. But today we're going to be focusing on primarily different types of survey data which we can use to understand mental health. Now good place to start when you're looking for new data is of course the UK data service's own website and the website address is there. And we're incredibly fortunate in the UK to have the UK data service which has been running for many decades and brings together survey data from across a range of different sources and a range of different topics. Now whilst I'm going to be focusing for the first 20 minutes now around specialist mental health surveys, in particular the adult psychiatric morbidity survey series, there are a series of other types of surveys that you can access through the UK data service which may well be more relevant to the research questions that you have. Now the general health surveys are a key resource and they're often run every year for many years. The health survey for England for example includes the GHQ12 so it's a very useful resource if you want to look at trends in mental health. The attitudinal surveys can also be a really useful resource. Both the British social attitudes survey and the Scottish social attitudes survey have included modules in particular years that are focused on things like attitudes towards mental illness and attitudes towards things like talking therapies and treatment. Finally, while our focus today is not really on subjective wellbeing measures it is an available data source to you that things like the annual population survey and the crime survey for England which have phenomenally large sample sizes have over the last decade included things like the Warwick-Headingsworth measure or the ONS4 so they'll often include information on whether or not people have experienced anxiety yesterday how happy they were yesterday. Those sorts of effective subjective measures included on that. But mainly today I wanted to talk about the Adult Psychiatric Mobility Survey which is a terribly named survey and I do apologise for its name. The Turquoise squares that you see on this timeline here, APMS that stands for the Adult Psychiatric Mobility Survey they've been carried out every seven years since 1993. So you've got four major surveys there that are available for secondary analysis. Now the yellow squares on the screen there refer to the mental health of children and young people survey which I'm not going to talk about so much today but is also an available resource which follows a very similar approach to that taken on the adult surveys but using different screening tools. Now the survey series is funded by the Department of Health and the two most recent surveys were commissioned by NHS Digital. As I said it's a series of repeated cross-sectional surveys carried out in 1993, 2000, 2007, 2014 which actually makes it the longest running mental health survey series of its kind in the world to use consistent measures over time. These are the probability sample so it's a high quality general population sample drawn to the household population usually about seven to eight thousand adults in each wave interviews take place in people's own homes and quite substantial lasting about an hour and a half. Now what makes the survey series distinct and distinctive is that the assessment of mental disorder is done approximating the diagnostic criteria that are used in the standard manuals. So what this means is that rather than just having a simple screening tool like the GHG12 is great but it just is a measure of generalized psychiatric distress whereas on the mental health surveys we use the clinical entity schedule revised which is a much more rigorous approach. It has about 130 questions covering 14 different types of neurotic symptoms and for each of these symptoms it asks about its presence in the last seven days how many times, the severity, the length, the duration of symptoms. So Bay detailed amount of information is collected. We don't apply algorithms to that data and we can produce six different types of common mental disorder and we can also produce which will be very useful in a lot of the secondary analyses that you might be planning a dimensional severity score which is called the CISR score which could be very useful if you need continuous outcome measure. Now, Owen shortly will be talking about the beautiful cohort surveys and those cohort studies are wonderful for understanding change across the life course and are ideal for starting to disentangle some of the cause and effect. However repeated cross-sectional surveys are better for some kinds of research questions. So for the sorts of research questions which you might want to use APMS data after looking at population prevalence. So because these haven't suffered from attrition in the same way that say a cohort study might, it's much better for looking at prevalence in the population and it also covers the whole household population. So it can tell us about how much of something there is and because we have repeated cross-sectional surveys that use broadly the same methods it can tell us about temporal changes over time. Now all of the surveys that we're talking about today will be really powerful resources for looking at treatment gaps, where we have information collected about what treatment people are accessing and what symptoms are present. Also the surveys collect really rich information that enables us to identify different subgroups and enables us to look at inequalities and circumstances. So what I'm going to do over the next sort of 5-10 minutes is talk about examples, worked examples of where we've used the mental health data to answer some of these sorts of research questions. So firstly around population prevalence. This is a short paper that we put together recently where we wanted really to disentangle what do we mean when we talk about population prevalence in mental disorder. And we wanted to show that this really depends on the types of disorders that you're measuring and the severity level and timeframe in which they're measured. So if we take for example a very rare condition, a very severe condition such as the experience of psychosis or psychotic episodes in the past year, then we're going to achieve prevalence of around one in 100. But if we add into that group, those who've experienced severe anxiety and depression at a level at which we would certainly consider treatment to be warranted, then we're looking more like a presence of around one in 10. And if we add in those who are experiencing any kind of anxiety and depression at a level likely to be clinically relevant if not necessarily warranting intervention, we're getting to more like one in five. And of course substance dependence is another area of mental disorder that can be considered which includes things like alcohol dependence and drug dependence. And then if we add into our part those experiencing some sorts of neurodevelopmental disorders like ADHD, then we can certainly identify more than one in four in the population likely to be affected. So alongside examining different disorders to diagnostic criteria, the beauty of the specialist mental health surveys is that they cover a range of different types of mental health indicators. So you might want to base an entire study on any one of these, or you might want to look at the relationships between different types of disorders. So the data's ideal for exploring comorbidity in this way. Now as I said, we have a series of repeated cross-sectional surveys and we've tried very hard to keep the approach taken in each survey comparable. So this is an example of where we've used the survey series to look at temporal trends over time in the proportion of the population reporting having self-harm. Now the things to bear in mind when you're using the mental health survey series to look at temporal trends is that you do need to check that the question or the measure being used is consistent. And there's always a trade-off within the survey team that we want to keep the measure consistent, but we also want to use the measure that's meaningful in a contemporary setting. So you need to keep an eye on that. The other thing is to check that you're looking at a consistently defined sample. So when we were looking at temporal trends and self-harming, we had to exclude the data from the 1993 survey because there self-harming was only asked of people who'd expressed depressive ideas. You'll also need to check about consistency in mode. So with this paper here looking at trends and self-harm, we've always asked about self-harm face-to-face, but in the more recent surveys we've also asked about self-harming in the self-completion part of the interview. But because reporting is higher in the self-completion, we had to not use that information when constructing temporal trends. The other things to bear in mind are ensuring that we have a consistently defined sample in terms of geographical cover. So the earlier surveys in the series covered all of Britain, so Scotland and Wales as well, whereas the two most recent surveys covered England only. The age group has also changed. In 93 went up to age 64. In 2000 up to 74. And in the more recent surveys, we have no upper age limit. So this is some of the results that we published in that paper. It shows a steep increase in the proportion of the population reported that they were using self-harming as a coping strategy. The chart on the left shows the men and on the right for women. We can see that the increase has been steeper in more recent years amongst women. So the other sorts of things that we can use survey data for in this way are to look at the treatment gap. So that is amongst people who have symptoms of a disorder, what proportion of receiving treatment and are there inequalities in that. What this chart here shows is the proportion of people reporting that they're self-harmed saying that they have received medical or psychological support as a result of that self-harming. We can see that at least half of people don't go on to receive medical or psychological support. We can also do regression analyses to look at the characteristics that predict not getting support. And we found that if you're relying on treatment data, you might want to be aware that younger people, men, and those in debt who had self-harmed were less likely to receive treatment as a result. Okay, and then the final sorts of questions that you might want to use the mental health survey series to address are around exploring differences between groups. So on the surveys, a wide range of different characteristics and contexts of respondents are asked about. These are some of the different things that we ask people about. So you could base the paper on any one of these topics. In this paper, what we did was we looked at the data we had that enabled us to identify whether or not people were students or not students. And we focused in on 16 to 24-year-olds. And this chart shows amongst men the proportion of students and not students who reported or who met the criteria for common mental disorder, such as anxiety or depression disorder. And we're not seeing a key problem amongst students in this data. However, when we looked at female students compared with female non-students, there was indications in the data that female students might be experiencing an increase in anxiety and depression. Other sorts of outputs that you can produce, this is just a very simple descriptive report that we did focusing in on the information we have on intellectual impairments in the sample. And we used that data to show, for example, that amongst people with lower levels of predicted verbal IQ are much more likely to experience anxiety and depression than amongst people with higher levels of verbal IQ. And third sector organisations and lobbying groups like Money and Mental Health have also made use of data from the series. For example, in this policy briefing that they recently produced which highlighted that problem debt is much more common amongst people with more severe common mental disorders. Okay, so before you start work, I recommend that you have a look at this website which is the mentalhealthsurvey.org site and it's basically nothing fancy, it's just a repository of where we put all the papers that we're aware of that have used the data. So avoid you replicating work that's gone on elsewhere. So in order to get hold of the data, the data up to APMS 2007 is quite straightforward. You just go directly to the UK Data Service website. You'll need to register or log in if you're based in an institution. You'll often have institutional login information. Now there's a range of different types of access conditions but open access you don't even need to register. End user license is the approach for accessing the mental health surveys up to the 2007. You'll have an interface like this where all of the documentation and the reports, information about the derived variables all available and if you click on the right-hand side there's information about how you can just immediately download the data. Now for the more recent surveys, those GDPR and other changes, you have to first get approval from the Data Access Request Service based at NHS Digital. So what you need to do is complete an application with then the permission to use the data and once you've got that permission then you can go to the UK Data Service and download the data in the usual way. Now that's about the specialist mental health surveys. I'm just going to say a few slides before we go on to about understanding society which of course is a longitudinal survey. Absolutely huge longitudinal survey initially started with something like 40,000 households. Of course it also draws on sample from the Bridgehouse Health Panel Survey so there are individuals within the sample who have been followed up for over 25 years. Sample members are interviewed every year and there's a course set of questions that are asked so you can look at change over time. And due to say wide geographical coverage, large sample size and ethnic and immigrant boosts, lots of different subgroups can be studied and compared. Some of the key mental health and wellbeing indicators in the data set include the GHQ, General Health Questionnaire, which is a really useful summary measure and the SF12 is also a really useful measure which includes both physical and a mental health component. In terms of children, the strengths and difficulties is one of the key data sets for looking at mental disorders in children. Now there's information in these slides so you can use those links later on for finding the different variables. And on the understanding society website there's also lots of information about training and webinars and you can submit questions as well to the research team there. There's even a YouTube channel, I should check that out, I haven't seen that. And again so that you don't jukescape work that's already underway elsewhere, all the latest research and publications are also listed on that website. And you can sign up for information and updates about understanding society using these links here. So that's me done and now I'm going to hand over to Owen who's going to talk about the amazing first cohort studies. Hello. Yes. Hello. Thank you, Sally. Really, really interesting presentation, really so much information there. So what I'll be talking about, so Sally covered mainly the kind of cross-sectional and specialist mental health surveys and obviously understanding society there towards the end. I'm going to be talking more about the wealth of longitudinal data and longitudinal studies that have been conducted in the UK with a particular focus on the British birth cohorts, a really useful data resource. So quickly just run through what I will be talking about here. So first of all I'm going to talk about the actual studies that have been conducted, the measures available, how to find information about those, followed by a brief overview of the British birth cohort studies and the kind of strengths of these fantastic studies and the types of research questions that they are particularly useful for answering. I'll then talk a little bit about a project I've been working on with colleagues at the Centre for Longitudinal Studies in UCL about maximising the comparability of mental health measures in the British birth cohort. So how can we be sure that we're measuring the same thing across time and across different generations in these studies? And I'll wrap up just with a little bit more about how to access these studies. So until very recently, searching for information on all of the available mental health data resources or longitudinal data resource in the UK was a bit of a chore. There's an awful lot of studies out there, so researchers often found themselves searching in isolation, having to go through individual studies to find out the data resources that can answer their questions. However, if you click on the below link, I should say, or scan the QR code provided here, you'll be taken to a fantastic new data resource that has been launched just a few weeks back. This is a comprehensive catalogue of longitudinal and cohort studies in the UK with a particular focus on the mental health measures that are available to researchers. So this project was led by Louise Arsenault at King's College London, Louise and Bridget Bryan, both at King's College. This was a closer funded project and what they developed is essentially a comprehensive catalogue containing information from a wide variety of longitudinal studies. So I think there's about 47 studies in there at the moment moving up to I think there's potential for nine or more studies to be added. So it includes studies that look at things like birth cohort studies and other longitudinal studies like household panel studies like understanding society and stuff like that that Sally already mentioned. So if you follow this link, you will find it provides a search engine for finding information on mental health and well-being measures collected in the existing UK cohort and longitudinal studies. It provides information about particular studies themselves such as, for example, here the English longitudinal study of aging, ELSA. So it will tell you things like when the study commenced, the geographic coverage, the types of measures that are collected, the sample details and the overall aims of the studies, lots of information about the actual studies themselves. And very usefully there is a tutor for searching detailed information or finding detailed information about mental health and well-being measures including individual items, response scales, so sort of general screener questionnaires, more diagnostic focused interviews and all those different types of, those different methods of assessing mental health so you can get lots of information there. It's a very easy to use website, very user friendly. So to discuss all of the studies that are included in the mental health catalogue I just talked about would take far longer than 20 minutes. So I'm going to focus my discussion on the British birth cohort studies. So these are studies that have been essentially following the same individuals or groups of individuals from birth throughout their lives. I'll be focusing specifically on those studies that are nationally representative. We have five nationally representative cohort studies in the UK. I don't believe sort of this type of data resource exists anywhere else in the world. And I think they can be very, very powerful data resources in answering questions not only about mental health but about other relevant health phenomena. So the first study I'm going to briefly cover is the NSHD, the National Survey of Health and Development which is the oldest of the British birth cohort studies. It is unique in having data from birth on the health and social circumstances of a nationally representative or a representative sample of over 5,000 individuals, men and women born in England, Scotland or Wales in March 1946. And they've had 25 data collection points over the participants' lives so far. So they're due to be surveyed again next year when there will be roughly aged around to about 74 years of age. The next cohort study I'll talk about is the 1958 National Child Development Study. This, again another nationally representative study, follows the lives of over 7,000 individuals who were born in England, Scotland and Wales in a single week in 1958. So this represented 98% of the total births across England, Scotland and Wales in that week. And since then the cohort has been followed up 10 times. There's been 10 assessment waves for this data set. Roughly 12 years later, so early on in the process of the British cohort, a new cohort was launched roughly every decade or so. So in 1970 the British cohort study or BCS 70 was launched. It follows again around about 17,000 people who were born in England, Scotland and Wales in a single week in 1970. And to date they've been followed up 8 times. And the most recent wave has just been released at age 46. Next steps, breaking from the traditional birth cohort mould here. It's formerly known as the Longitudinal Study of Young People in England. It's not strictly a birth cohort. However it follows around about 16,000 participants from early adolescence in England. So it began in 2004 when the cohort members were aged roughly 14. And sort of efforts are now being made to retrospectively collect data from earlier waves. So what's unique about this cohort is it is linked to the National Pupil Database. So it has individual cohort members scores on the Key Stage 2, 3 and 4 administrative... Key Stage 2, 3 and 4 tests. And for individuals who are maybe interested in looking at mental health and education, it's a very useful cohort. More recently, or the most recent birth cohort study, the Millennium Cohort Study, is roughly 19,500 children born over a roughly two-year period, September 2000 to January 2002. They've been followed up five times with the next survey coming in 2020, when they'll be aged 17 years of age. So I've talked a lot about what surveys are available and what's out there, what are the measures, what sort of information is collected in these surveys. Well, again, I mentioned briefly these are not mental health specific surveys. These are broader data resources for both the medical and social sciences. But they collect a range of useful information on a range of useful variables if you're interested in mental health. So across the cohorts, similar information has been collected from birth, things like household composition, information about the parent's background, socioeconomic status, physical characteristics. And also over time, different measures have been collected as individuals age, including things like mental health and well-being, also things like cognition and other physical measurements. So what are the types of research questions that are particularly suited to this type of data? Well, Sally covered the types of research questions we can ask with cross-sectional data and with repeated cross-sectional data. I guess the main advantage of longitudinal data in general is that by having repeated measures we can explore patterns of change within individuals over time. So for example, our mental health, are there certain trends or developmental trajectories we can find? This might help us identify key periods over the life course for the onset and development of these difficulties. Also given the wide range of information collected on various other aspects of life, we can explore the role of different antecedents or risk factors. What are the things that maybe influence the development of mental health or set people in motion on particular developmental trajectories across their life? We can look at risk factors, not only fixed risk factors, things like gender and things like that, that don't change over time, but also dynamic risk factors. So we can ask questions, well, are there particular changes in circumstances over time that can impact the mental health of individuals? And finally, given that these data are following the same people for a very, very long period of time, we can get information on the consequences of mental ill health. So not only in terms of things like the continuity of mental health, are people continuing to suffer mental health problems for a long time, but how does this impact other aspects of people's lives as they age? If you're interested in those types of things, then longitudinal studies, particularly the birth cohorts, are very useful in that regard. What are other methodological strengths of this type of data? Again, I mentioned there's not specialist mental health service. There's a lot of rich information on a variety of factors that we can control for. We can more accurately or more easily get at temporal ordering, because we can have a temporal ordering built into the nature of the design. The fact that we're assessing things prospectively reduces the risk of recall bias. We're not asking, for example, people to recall things from their own childhoods. We're using things like, for example, for mental health, we're using maternal and teacher proxy reports of mental health and capture that when people are children. And also the particular cohorts that I mentioned today are also nationally representative, so the findings can be generalized to the broader population. So one of the unique things about the longitudinal studies in the UK is that I'm not really aware of any other countries in the world where we have such rich birth cohort studies staggered across different periods of time. So this affords us a kind of unique opportunity to answer some very interesting research questions. Okay, as I mentioned, the kind of main benefit of longitudinal studies is we can study longitudinal development within birth cohorts over time. But the fact that we have multiple birth cohorts and they're all nationally representative allows us to look at cohort differences over time. So this allows us the potential to potentially tease apart age effects. For example, in mental health are there key periods of development that are kind of universal. It doesn't matter when you were born or are there cohort effects. Are there being born in particular points in time or in history, does this impact on mental health? And that's something that the cohorts are really well positioned to do. However, it's not necessarily straightforward, although very potentially powerful and impactful. Measures from different cohorts, they're not always directly comparable. Sally touched on this with the cross-sectional data. So it's important to pay careful attention to the equivalence of measures. And I'm going to talk very briefly, I think I only have a couple of minutes left here, about our attempts to maximize, I guess, the comparability of measures of mental health that are available in the British birth cohorts. So this is a project I was working on with colleagues at the Centre for Longitudinal Studies. In UCL, led by Professor George Plubidis. And this project was funded by Closer. And what we aim to do is retrospectively harmonize the available measures of mental health in the British cohorts. In other words, we've been measuring mental health using questionnaires in the cohorts for a very, very long time. Why can't we simply take, say, for example, a mean level of depression in adolescence in the 1946 cohort and compare it to a similar cohort, for example, the Millennium Cohort Study? Will this project aim to maximize our ability to make valid comparisons across and within these cohorts? So for example here are some examples of just the different ways in which we've assessed mental health problems over the years. We can see there's different barriers to comparing different measures. So for example, different content of questionnaires, different symptoms were asked. These might have been reported on different scales across and within cohorts. And also different reporters. So for example, in childhood, we typically use parent or teacher proxies, whereas in adulthood, we are more likely to self-report. So what we did, first of all, to deal with the issue of content, was even though lots of different questionnaires have been administered over the years, they might ask about different types of symptoms and stuff like that. A number of these questionnaires tap similar or common symptoms. So what we did is we had difference to raters independently kind of screen all of the available questionnaires and the cohorts and identify common symptoms that are assessed within these cohorts. Basically assigning a code of what's being assessed in each question. So here's an example from the GHQ 28 item version. We found that independent of one another, the two raters rated the first question felt constantly under strain as kind of reflecting tension or stress, something like that. And this allowed us to develop a kind of schematic of all the common symptoms that are assessed across the different cohorts. And this will be made available for researchers to use. You might find depending on when exactly you're looking at, whether you're looking across cohorts or within cohorts or at certain developmental periods, you will have different permutations of common items. We've created a searchable tool that can help you identify these. We also looked to empirically test the measurement equivalence of these measures. And we have some guidance on how to do this. I'm running short of time so I'm not bore anybody with the technicalities of this. But essentially what we tried to do is test, are we measuring say across cohorts or within cohorts over time, are we measuring the same constructs, psychological stress for instance, and are we measuring this to, are people interpreting the questions that we're using to measure this in similar ways. So all of this information will be available in the new year. We're just in the final stages of publishing a resource report on this that will provide guidance for people and this will be available on the closer website in the new year. And just very, very briefly, some examples of kind of capitalising on this measurement work to make reliable comparisons to some of the work we're currently doing, looking at kind of trends in adolescent mental health in adolescents, also trends in the development of mental health in adulthood. And by kind of maximising the comparability of these measures, we're minimising the likelihood that any observed differences are solely due to measurement error. In other words, we're more confident that we're seeing true differences within and across different cohorts over time. So very quickly, before we move on to questions, I will talk about access. Again, most of this was covered by Sally. There will be some variation here depending on the particular study you're interested in. Most of the birth cohorts I've talked about today, particularly the 1958 cohort, the 1970 cohort, next steps and the millennium cohorts are available, freely available to researchers on the UK Data Service website. I believe all of the mental health measures are covered under a standard end user licence. Other cohorts, some of the more maybe region specific cohorts or, for instance, the 1946 cohort, the MRC National Survey of Health and Development. This is housed at the Unit for Lifelong Health and Aging at UCL. This is available, again, free of charge to bona fide researchers who have academic credentials. This can be accessed directly through the link I've included here today. Thank you all for listening to me. I believe now I will pass you over to Sarah, who has organised this webinar. She will take some questions or put some questions to both myself and Sally. Thank you. That's great. Thank you very much then. Those were really good presentations. We have, thus far, about six questions. We should have until three o'clock to answer questions. If anybody has a question they want to ask, do type them in now. If you can't see where to type them in, I think it's normally the top right of your screen. You should see a red box with a white arrow. If you click on that you should be able to see where you can type questions in. Okay, so questions. First one we have was from MA, or Amy. Can you talk a little more at the end about the Data Access Request Service and when we need to ask permission versus when we can use the data straight from the UK Data Service website? I wonder if that's one for me to try and answer. Is that okay, Sally and Ewan? Sure. Yeah, that's fine. I'll just make myself a presenter then. So what I've done is I've gone to the UK Data Service website and I've typed in APMS and this has brought up the results. And what you'll see is that the top result there is the 2014. It says it's the special license access version. And you'll see there's a 2007 version that doesn't say anything about what its license is. And if I wanted to see all of them I would just click on the bit that says series here and it brings up all the psychiatric morbidity surveys. If I click on that and then click on access data and then on the title you can see all of the different psychiatric morbidity surveys that are in this series. Okay, so to answer the question, this one, for example, the one before 2014 to 2007 one, sorry, that's available under the end user license, which means that you can just register with the UK Data Service and anybody can register. And then you would just go to the catalog page, click on access data and add it to your account once you've registered. And then you can just download it straight off the website. So basically that goes for all of the ones before 2014 and I think that also applies to the ones that you were talking about, Owen. However, if I go back to the psychiatric morbidity surveys and I go to one that says special license access, these are the ones where you're going to need to get permission. So again, I will click on access data. And although it says you can add it to your account, in reality, you're not going to be able to download this particular data set until you've got permission. And that means you should read the information that's here basically that explains what you need to do in order to get it. And I think that's what Sally was referring to. Did you have anything to add to that, Sally? No, I thought that's a great summary. Okay, yeah, and if you want to find, for example, the surveys that Owen was talking about, you would go to the data service website, go to get data, key data. You can also search using the search box at the front if you know the name of the particular survey that you're interested in. And they should take you to all the different kinds of surveys that we've got. These are just the key data, as I say. Some of them you can, if you can't find the key data, then you'll want to actually search the data in the data catalog. Click on longitudinal studies, and that's where you'll find links to most of the studies that Owen was talking about. And again, I think with most of these, we're just talking about end-user license, which means that you can just add it to your, that one. So, for example, so you've got all the information, all the documentation, and then click on access data, and you just add it to your account, okay? The important thing to remember is if it doesn't say what kind of license it is, then it means that you can just download it from the website after you've registered. Okay, then, so another question from Sam. What is the most appropriate way to describe the one in four prevalence, please? This is preferring, I think, to your slides, Sally. Have you reported on different ways in the past, each year in any given year, over a lifetime? Yes, so it's really interesting, and this is something we kind of wanted to deconstruct a little bit, because the one in four is such a ubiquitous figure, and we wanted to kind of understand, well, what is it that we mean, because it's cited in all the government documents, it's used by mind, it's used by many of the charities, the other organizations. So we just wanted to kind of tease it apart a bit using the latest mental health survey data. And basically what we said is that, well, we found that the one in four is a useful summary measure, so long as we understand that it represents broadly a range of different types of disorders, and it is quite inclusive in terms of its severity, and that we would describe it as referring to people currently affected by mental health symptoms, because whilst we did include, because of the measures that are used, for the very rare conditions, prevalence over the past year, what really drives the prevalence is anxiety and depression, and that was measured in terms of their current symptoms. So it's fine to say that about one in four of the population at any one point in time is likely to be experiencing some kinds of mental health symptoms, but so long as we understand that to mean a broad range of different types of disorders and quite an inclusive measure in terms of its severity. Okay, excellent. I see we've got quite a lot of questions now, so we'll go through those. So the catalogue mentalhealth.ac.uk site looks great, thank you. This might seem like a silly question, but is this where we can get the actual questions on scales? For example, if I wanted to do a study that measured depression amongst university students who experienced risk factor, is this where I can, I think, find a validated measure of scale of depression, and can I use those questions in my study? That's from MA. I think that one's for Owen. Yeah, so I believe the catalogue will at least have information as to the source of the questionnaire. So if a researcher was interested in a particular measure and thought it was appropriate for a new study they were conducting, it would certainly point them in the right direction as to where to find more information on that particular measure. But it wouldn't, I don't believe it has a direct download or anything like that of the actual questionnaire itself. Can you think of any data sets that might include variables around child abuse, corporal punishment, and, nevertheless, integration, cross-sectional or longitudinal? Well, I can just come in on the cross-sectional. So the adult psychiatric morbidity survey does ask about, retrospectively, about experience of both sexual and physical childhood abuse. And it also asks questions to enable diagnosis of conduct disorder. So this relates to getting into fights and bullying and things like that as a child. Now, of course, these are retrospective reports. So these are questions that are asked of adults about experiences that they had in childhood. Okay, did you have anything to add, Erin? Yeah, so let me think. With more severe sort of variable things like child abuse, off the top of my head, I believe there is some questions, but they're retrospective. I believe, I think possibly in the 1950, I would need to look into, if you give me the person who asked this question, if you give me their email address, maybe I could look into that in a little bit more detail. As for sort of adolescent aggression, well, the SDQ, the strengths and difficulties questionnaire, very common measure of kind of difficulties in children includes a kind of behavioral problem subscale, so things like fighting, bullying, aggression, and kind of a hyperactivity scale. So that might be, that's available in a number of the longitudinal cohorts, particularly ALSPAC and the Millennium Cohort Study. They have multiple ways of SDQ data, so that might be something to look at. Okay, Doki, Musa asks, will we get emailed these slides? And I'll answer that. They're going to be here in news and events. If you go in there, you'll go to events, then past events, and the slides will be available probably later in the week. Okay, so Anne asks, in the data sets that we're referred to within the webinar, are there questions related to drug use? Is it difficult to compare data from mental health data sets and drug use data sets to look at questions around dual diagnosis? So from the perspective of the mental health surveys, yes, we both assess, we've got the audit, we've got the severity of alcohol dependence measure, and we also assess for use of a range of different types of illicit drugs and also dependence. So we've got five different types of signs of dependence on drugs. So you can look at those things by presence of anxiety and depression and other types of mental disorder as well. I know there's information in the Millennium Cohort Study. There's a lot of information around those sorts of behaviors. But I believe that the Mental Health Catalog would have, again, a very good overview of what's available in the longitudinal data sets. Okay. Parisa asks, what are the nationally representative state-of-the-art and publicly available longitudinal studies of youth adolescents on youth development, mental health behaviors? Thanks. Well, I think any of the studies that I covered there, obviously with the more recent studies such as the Millennium Cohort Study, we have a lot of information and some of the measures might be slightly more relevant. Again, as Sally mentioned, this is something that kind of changes with the time. We don't... Our strategies for measuring things changes over time. But again, the harmonization project that we're working on is should provide guidance on kind of how to maximize comparability across cohorts if researchers are interested in looking at these things in nationally represented cohorts from different decades. Great. Okay. Joe asks, thanks to all. A question for Sally. I'd like to merge APMS Survey results to create a larger sample size to analyze. Do you think there are any major reasons not to do this? A larger sample size will allow for interesting effect modifications that is around gender, the local environment, and CMDS. I think it's a brilliant idea. We've got reduced weight specifically for merging 2007 and 2014 data. In the survey reports, we use this approach for looking at low-prevents disorders like autism and for psychosis because otherwise we've got so few positive cases in the sample. So it is a bit of a management task. But once you've got your head around that, yes, it's a really good idea. Great. Bo asks, is there any longitudinal dataset that has free text of any kind by the participants? Off the top of my head, I know there is some free text data in the 1958 cohort. I know Professor Elissa Goodman at the Centre for Longitude and most of this has been working on some research, looking at the association, I believe, between letters to do with children's aspirations and their actual socioeconomic outcomes in adulthood, I believe. So that's off the top of my head. That's all I'm aware of. But I believe there are other examples of that right there. Great. So there's still a fair number of questions left. So are more training sessions on how to use the data, for example, APMS, Planned in Future of Victoria? So I think people are always welcome to email me. If you've got a particular question about how to use the data, we do provide quite detailed documentation, which describes what variables used for waiting and for controlling the complex survey design, things like that. But yes, nothing specifically planned. I don't know if the UK data service has got anything coming up. Not that I'm aware of about this. No. Hazel asks, are you aware of any data sets that are supposed about mental health measures and measures of sexual risk taking? I believe that the Millennium Cohort Study has some information on sexual behaviors. So that might be a cohort to look at in the most recent sweep, certainly. Also, the 2014 APMS, we included some questions on sexual risk taking. So in particular, numbers of partners without condom. And I think that's the main in the last five years so there's an opportunity to use APMS 2014 to look at this. But it wasn't asked in the earlier APMS surveys. John asks, I believe the next MCS wave will include a self-completed version of the SDQ. What potential issues may arise from using both parent-completed and self-completed versions in a longitudinal study? Well, that's something we could test empirically. So at the moment where we're doing something similar, we're working on a project where we're looking at the kind of equivalence across teacher reports and parent reports. I believe there's a lot of literature out there about sort of the invariance across parent and self-reports. But yeah, that's something we could sort of address empirically and try and get an idea of how comparable those measures are. And Emma asked, do we need permission to use the cohort data if it includes questions around mental health? No. I'll follow up. Most of the cohort data, the mental health data is the majority of it is standard end user license. Again, it's only if you're looking to sort of look at its association with more sensitive data. For example, genetic data, educational data, geographic data, that sort of thing. But the actual mental health data itself is very easily accessed. Okay. We have just three more questions. So I think we'll just transit those in. So this is quite a long one from Lawrence. I'm an enthusiast for secondary data sets and have used and including mental health. For sophisticated researchers who use multivariate analysis or decision trees, it's possible to disentangle the apparent effects on some outcome variable, e.g. physical health conditions of mental health and other variables which are associated with it, such as age, smoking, et cetera. However, there is a danger that crude users will nearly do a mental health versus others analysis, which is actually misleading. Can you recommend any text which would help people to avoid the crude approach? I would make one comment, which is that this is a very pertinent observation when you want to look at causal relationships. But sometimes for government resourcing, for understanding, we simply need to know about prevalence in order to target groups. So it's not necessarily to say that a particular characteristic is what drives a higher level of mental disorder, but simply that people with a particular characteristic may well have a higher level of need. So it's really just to say this is, I absolutely agree, and it's a very good point. But sometimes there is also value in that very simple descriptive approach. There's a question. It's not really directly related. I don't think it's how to prevent the situation that somebody who has used the same data as I download publishes similar results before my publication. If you're planning on working with APMS, I'm aware of a lot of the work for people who are doing that are working with APMS data. So we try to sort of keep a bit of a log. It's very informal. So, I mean, people can publish anything they like. There's no obligation to alert other people. But if you want to just email me, I can tell you if I'm aware of someone who's currently working on it. And the value of that is if there is someone else who's currently looking at the same thing, A, you may want to collaborate. So you might want to publish something together. Or B, you might just want to focus your analysis slightly differently so that you can avoid duplication in that way. But at the moment, that is a risk. The cohort studies, again, it depends on the type of or who houses the data, essentially, for some of the more specialist cohort data sets. For example, the ALSPAC, the Avon cohort, they keep a very strict record of who's working on what. As do the 1946 cohort, other cohorts, I believe that that could be an issue. There's nothing to my knowledge that that's incredibly systematic. Okay, great. I think we've got through everything. One final comment, which was just a compliment to say that this was a good webinar that was fuller and more informative than many other similar types of webinars. So there you go.