 This is the afternoon session of the Health Studies user conference. We've got a thematic link to the talks over the next hour. These were all focused on different aspects of mental health. First up, we have Natasha Chilman. Natasha is a second year PhD student at Kings College London. She's funded by the SRC. Her PhD is a mixed methods project focusing on inequalities in mental and physical multimorbidity for people who've experienced homelessness. And this analysis of nationally represented data from the Adult Psychiatric Mobility Survey forms the quantitative part of her PhD. So, Natasha. Thank you so much, Sally. And thank you everyone for coming and for having me. I'm excited to present some of the initial results of my ongoing PhD research today. So firstly, a bit of background about this area. We know that health follows a social gradient. And what I mean by this is that there is a gradual slope of association between deprivation and worse health outcomes. However, health inequalities experienced by people who are homeless have been described as a cliff rather than a slope of inequality, as illustrated by the graph on the right hand side. So this is the case across both physical health conditions and also mental health conditions. There was a recent systematic review and meta analysis by Gatwinski and colleagues, which found a 76% prevalence of any current mental health condition among people experiencing homelessness. But something else that the systematic review also found that in the included studies, there was a lack of research looking at comorbidities of mental health or multimorbidity. So the presence of multiple mental and physical health conditions. Where studies have looked at multimorbidity, they have usually recruited samples of people using specialist homelessness services, such as hostels, night shelters, or specialist primary care services. And this raises the question, what about people outside of these services? And what about people who have previously been homeless, but are now living in private households? Do we still see these cliffs of inequality in mental and physical health and multimorbidity? So these are some of the gaps which my research aims to address using data from the adult psychiatric morbidity surveys. So this is the APMS. And the APMS is a nationally representative survey that was conducted in England in private households. So it's just worth highlighting that when I say nationally representative, this is a people in private households, not people outside of private households, which I'll talk about later on. But they use stratified probability sampling to recruit participants and then apply survey weights to account for selection probabilities and non-response. I'm using data from both the 2007 and 2014 surveys, which is a total sample of almost 15,000 participants. And the aim of the current study is to describe the prevalence, presence and nature of the mental and physical health multimorbidities in individuals who have previously experienced homelessness, who are now living in these private households. And these are the measures which they use in the APMS, which I am now looking at within the data set. So homelessness was self reported by participants. They were shown a list of a number of different life experiences, for example, being expelled from school or experiencing violence within the home. And one of these experiences was quote being homeless. So it's worth noting that the definition of homelessness was left up to the participant. And if they indicated that they had experienced homelessness, the interviewer asked when they last experienced homelessness. To assess for common mental disorders, the APMS uses a structured validated scale for the CISR, which is widely used within the search. And this asks after symptoms of common mental disorders, such as anxiety, depression and phobias in the last month and week. And they also use the audit scale to assess for alcohol use and alcohol use problems. physical health conditions were self reported by participants from a list of 21 physical health conditions. And it's worth noting that in my analysis, I'm only in this presentation, I'm only looking at conditions which participants said they had experienced within the last 12 months. And so while these measures were collected face to face by the interviewer, this last measure on substance dependency was administered using a computer assisted tool. So these are questions based on the DSM for substance dependency. So firstly, looking at the prevalence of homelessness within private households. So out of the total sample of almost 15,000 people, 599 people reported a previous experience of homelessness. When we apply the survey weights, this leads to a prevalence estimate of almost 4%. And when they asked when homelessness last happened, and if it happened multiple times, when was the most recent time, and the majority of people last experienced homelessness over six months ago, but after the age of 16. So in terms of the demographics for age and sex, and these are presented here on the left hand side, we have age and in the lighter bars are the formerly homeless group and the darker bars of the group of people who never experienced homelessness. And we can see that the formerly homeless group is slightly younger than the rest of the sample. And when we look at sex distribution, so for men and women, in both the formerly homeless group and the never homeless group, men and women were pretty evenly distributed between both groups. So all of the prevalence estimates I'm going to now present are adjusted for age and sex. So firstly, looking at the prevalence of common mental disorders, including anxiety, depression, phobias, and other common mental disorders. So this is using a score threshold of 12 and above on the CISR scale. And we can see that the prevalence is much higher within the formerly homeless group, quite a drastic difference compared to the never homeless group, which is about 15% compared to 44% prevalence in the formerly homeless group. But so a strength of using the CISR is that we can actually look at the severity of common mental disorder symptoms. As a higher score indicates a higher severity. And this is what we're looking at within this slide here. So on the right hand side, we can see that the groups are for people who scored between 12 and 17. So they met the criteria essentially for a common mental disorder symptoms, and 18 and above, which indicates symptoms of a level very likely to warrant intervention. So we can see out of this 44% of the formerly homeless group who met the criteria for a common mental disorder, a large proportion were actually experiencing the most severe symptoms of a common mental disorder. So these are some of the physical health conditions. So these are five of the 21 physical health conditions that they asked about, stratified by age for the formerly homeless group and the never homeless group. I won't go into too much detail about each of these conditions, but I just wanted to highlight a comparison with previous work. So this is the study referenced in the background section. So they recruited people from hostels and night shelters who are currently homeless. And we can see sort of similarities between these two studies. For example, there appears to be a high prevalence of asthma across the age groups for the formerly homeless group. And I'm looking at this sort of intersection between mental and physical health and mental and physical multi morbidity. So this is defined as the presence of a common mental disorder and at least one physical health condition. Again, we can see that the prevalence is much higher within the formerly homeless group. But we can go into more sort of detail with this by looking at the multi morbidity condition counts. So I'm looking at within this slide here. So on the x axis here on the bottom axis, we have the number of health conditions that people have. So these are people with one health condition to health conditions, etc. And in the blue bars and never homeless group and orange bars as a formerly homeless group. And we can see that having one or two health conditions is fairly prevalent within the sort of general population compared to the formerly homeless group. But actually, we start to see inequalities for the formerly homeless group when we reach three or more conditions, particularly when we get to this more severe end of the scale with five or six or more conditions. And I think this is interesting when we think about our definition of multi morbidity, which is often just looking at a binary cut off of two or more conditions. And lastly, looking at substance use problems. And so on the left, we can see the prevalence for dual diagnosis, which is defined as a common mental disorder, and a substance use problem, which includes either an alcohol problem, and or substance dependency. And on the right, we can see the prevalence of tri morbidity. So experiencing all three of these types of health conditions. And again, we can see the prevalence is higher in the formerly homeless group. Although it's worth noting this isn't as prevalent as the sort of mental physical multi morbidity. So strength of this work is that we are using data from a nationally representative household survey. So a real strength is that we can look at comparisons between the formerly homeless and the never homeless groups within the same data set. And also by using a structure validated scale to assess for common mental disorders, this can identify people who would meet the criteria for a common mental disorder in the community, but who may not be diagnosed. And this is important as we know that people who experience homelessness can also experience barriers to accessing care. In terms of limitations, this is a cross sectional study. So we can't make inferences about causality between homelessness and mental or physical health and multi morbidity. And as this is using data from private households, this does not include people who are long term homeless, living in institutions, or currently homeless. The key takeaway from this work so far is that a previous experience of homelessness is associated with the severe health inequalities and multi morbidity and mental and physical health. And this highlights the importance of integrated care for this group. During the experience of homelessness, but also beyond when they are living in private households. And I wanted to thank everyone who's continuing to support me in this work, my supervisors, Jay and Peter, Sally, he's sharing the session provide advice on the project and the experts by experience at Breathing Mental Illness and Pathway, who I've been consulting with, and will continue to speak with throughout the course of the project. And thank you everyone for this name. Thank you, Natasha. This absolutely fascinating. And we're now going to turn straight to Ruth Plackett from UCL. Dr. Ruth Plackett is a research fellow at UCL in the Department of Primary Care and Population Health. She is currently undertaking NIHR Three Schools Mental Health Programme Fellowship at UCL, where she's exploring the relationship between social media use and the mental health of young people. Ruth is a background in psychology and mixed methods research. Thank you. You've got 10 minutes. Thanks, Sally. And hopefully you let me know if you can't see my slides, okay. But yeah, thank you for the introduction. And yes, I'm going to talk about a bit of research I did in my previous role with the Air, the School of Public Health Research on the impact of social media use on young people's mental health using understanding society data set. So I'll give you a bit of context around why this topic was chosen and why it's important. So the mental health of adolescents has been identified as a real concern, as I'm sure you're all aware, both before the pandemic, but perhaps even more so after the pandemic, it's become a real concern. And we know that young people's mental health problems have been increasing over time, and there's been more and more referrals to mental health services with time for referrals increasing as well. And at the same time, there is a lot of concern, as you can see in the media and elsewhere about how social media use is affecting young people's mental health. And this is both coming from policymakers as well, clinicians and young people. And we know that social media use is becoming ubiquitous among young people, a lot of people are using it every day now. And just to mention that by social media in this context, I'm talking about kind of in a broad sense, so any online platform that you use to share information, whether that's kind of words, pictures, videos and includes all the popular things that young people are engaging with right now, like TikTok, WhatsApp, Snapchat and Instagram. So the games of this also, the background for this study was that we know from a lot of cross sectional studies that social media use has been linked with common mental health disorders like depression, anxiety with young people. We also know that there has been some positive associations found as well. So social media has been very helpful to people to connect with others and reduce loneliness too. So there's a bit of disparity in the literature about the benefits and harms. There is very limited longitudinal evidence to kind of disentangle and see whether there's any kind of causal relationship between social media and young people's mental health. Nevertheless, people are very concerned about the role that social media plays in mental health and making mental health problems worse in particular. So there's definitely a need to do some more longitudinal research in this area to help us understand the complex relationship and also inform public health interventions in this area. So the aims of the study were to basically understand the longitudinal relationship between sexual media use and mental health adolescents, but also trying to explore some of the mediating factors that might be involved. So it's a way to kind of disentangle this complex relationship and understand how it might be working. So the two mediators that I was particularly interested in are self-esteem and social connectedness. So self-esteem is, as we most know, is very related to poor mental health and has also been linked with problematic social medias. So things like comparing yourself to others, for example, can be common. And social connectedness, which is essentially that subjective feeling of kind of belonging to others and being part of a group, has also been kind of very linked with mental health and social media use as although social media use can kind of help us with those social connections and help us to increase the amount of connections we have. Not all of the connections that we make are then positive and there's lots of things like cyberbullying and stuff that can happen and can lead to poor mental health. So just to summarise really the research questions for the topic were to look at the relationship between social media use at age 12 to 13 and mental health two years later in the understanding society data set. And also to look at whether this is mediated by feelings of social connectedness and self esteem. And just to note that the reason why this kind of age group was chosen is a couple of reasons, partly because at sort of age 13 years around that age where it's kind of the minimum age that social media platforms like people to take part in but we do know that the younger age groups at 12 year olds do also commonly use social media. And it's also obviously a really key developmental stage where young people's identity is starting to you know form a bit more and peer relationships become a lot more significant so social media is likely to play quite a key role for this age group as well. So the sample from the understanding society data set was taken from wave 1 to 10 and there was just over 3000 people in it and they had data basically when they were on social media use when they were 12 or 13 and some mental health data two years later. And then the specific measures that were used for mental health so the sdq total difficulty school was used which I think was mentioned in the plenary earlier and is a very valid reliable measure of mental health for children and young people. And the social media use was measured on one question which asked how many and the number of hours they spend on social media on a weekday and was on a scale of kind of none to seven or more hours and self esteem was measured on a variety of aspects of self esteem over eight questions and social connectedness was measured on kind of two questions looking at number of friends and also how people feel about their friends how happy they were with their friends essentially on a scale from one to seven and we looked at lots of covariates as well and so we included things like age, sex, ethnicity of the young person and we also included at the year they took part in the survey because we were looking at this a trend in time of this increased mental health reporting over the years and we also accounted for some household variables and some such as household income and the number of employed people in the household as well as some mothers, marital status and with this highest educational high educational attainment as well and all of these factors in sort of the previous cross sectional analysis are found to be related so that's why they were chosen. So we did a multi-level linear regression in the end to whether social media use at 12-13 predicted mental health at 8-14-15 and a path analysis with structural equation modelling to investigate the mediation. So very briefly the results found that those who spent quite a lot of time on social media so seven or more hours did tend to have slightly poorer mental health but this relationship was quite weak and was really attenuated after we included the covariates. It does seem that strongest predictors are things like poorer mental health at baseline and the year they took part in the study so this may be unsurprising given that we kind of we know that poor mental health at baseline is likely to predict mental health outcomes later and that we are seeing mental health problems increase over time so that could explain those findings. So in the mediation analysis we looked at in just a data we saw that more social media use was associated with having lower self-esteem which in terms was associated with poorer mental health and so around 68% of the effects of social media use on mental health to use data was mediated by self-esteem but again this relationship was attenuated after including covariates and I did some sensitivity analysis while I multiply the data set of this and again found no significant relationship and just to note as well that social connectedness was not found to be a significant mediator. So I just wanted to touch on basically the limitations of the study and I think one of the biggest limitations to kind of highlight here is that the variables and measures of particular social media use may not be particularly valid or helpful in this sense. So although many cross-sectional studies have found that there was a relationship it doesn't seem to be stacking up in a few cross longitudinal studies that are coming out now and I guess it's where the time spent on social media so number of hours is really a good way of measuring kind of how we use social media. So there's been a lot of new work that's looking at how we use social media more specifically so things like whether it's passive use or active use or something like that which might help us to kind of understand a bit more about this relationship and we may see more things happening there. The sample size is also relatively small and some of the based on generalizability issues it does really depend on how you cut the data with what you find with some of these things so if you look at a different age or time differences perhaps and that could be you see different results. So to conclude this study suggests that there is not much evidence of a causal relationship between UK adolescent social media use and mental health problems two years later. It has implications for practice and policy in that we think that maybe reducing time specifically spent on social media alone may not help to improve mental health outcomes for adolescents and we need to think probably in these public health interventions more about how we use it whether it's active or passive use and perhaps there's really important related factors like self esteem and we need to take this into consideration if we're creating guidance for policy makers and clinicians around how they can support people who who are impacted by social media and their mental health and thank you for listening that's all. Next going to swiftly move on to an intersectional analysis of inequalities in young people's mental health within specifically within the higher education context. So here it's Dr. Kiran Baloo who's a senior lecturer at the University of Southern Queensland and a visiting lecturer at the University of Surrey. He's co-investigator and project coordinator of the student wellbeing and life outcomes project on which this presentation is based. Thank you. Thanks Sally. OK, yeah, so I'm just going to go through one of the studies from this project which has been funded by the ESRC for the last couple of years. My colleague Anisa is also here today, so she can probably help with some of the questions and in this particular study we conducted an intersectional analysis looking at mental health outcomes at age 25 from the longitudinal study of young people in England and we particularly looked at the higher education context. So just to give you a bit of the background to this, so we were particularly focused on social determinants of mental health inequalities. So taking the position that mental health outcomes are not the same for all individuals and the particular background variables such as sex, socioeconomic status, sexual identity and ethnicity can predict differential outcomes for in terms of mental health and this position that we take here is the idea that people's social identities and positions, so their socioeconomic status, their sexual identity, they actually are kind of effectively acting as proxies for systemic marginalisation. So the idea being that it's not a intrinsic factor that's causing people to have negative outcomes but societal and structural aspects which are kind of imposed on individuals because of these backgrounds, so for instance, policies that disproportionately disadvantage particular people from low socioeconomic groups, they are likely to have a negative impact on mental health outcomes. But one of the things we were quite interested in is taking a position of an intersectional approach which argues that these particular social determinants do not act in isolation individually and independently, but they actually combine to have a kind of compounded effect in terms of predicting health outcomes. So this position is known as intersectionality and just very briefly, Kim Lee Crenshaw who is a scholar who developed the theory of intersectionality proposed that when you combine multiple social categories together, that you actually get a compounded effect that is separate from if you just looked at social determinants independently. So ethnicity and gender, for example, as two different variables combined, is likely to have an impact on outcomes that's different just for gender, just looking at gender or just looking at ethnicity. And this is particularly important when thinking about health outcomes and health inequalities. Now intersectionality traditionally has been looked at qualitatively, but more recently, particularly when looking at the population health context with large data sets, there have been quantitative approaches to intersectionality that have been developed. And a quantitative intersection approach takes the notion that these social identities or positions, so sex, social economic status, can be used as kind of analytical categories and we kind of provisionally adopt these categories in order to combine variables together and look at things in a quantitative way. And there's two particular models that you can focus on when looking at it quantitatively. So the first is an additive model, which is the idea that things that these variables, these categories act independently. So they're kind of cumulative. So there could be a negative impact for sex separately from sexual identity, separately from ethnicity. And that's an additive approach. So they might add up to have a negative impact on health outcomes. And then the other alternative approach is that there could be a multiplicative model. And this is the idea that there are interactions between these variables that occur so that these identities and positions combine and the characteristics multiply and amplify each other. So one of the approaches you can use with quantitative approach to this is used as a new approach called intersectional multi-level analysis of individual heteronarity and discriminatory accuracy or measure. And this uses a multi-level modeling approach. Now, in traditional multi-level modeling, the idea is that we look at individuals as being clustered within particular larger structures. So often this is used with, say, schools. And we say that individuals within one school, there may be common things about that context compared to individuals within another school. And we use multi-level modeling to take into account the fact that there is that common factor about them. With the major approach, what it does is it actually positions the combined categories at that higher level. So the social identities and positions, it positions them as structural issues, not just intrinsically in the person and positions at this high level, which they call social strata. And the idea is using this multi-level approach, we can look at things intersectionally to understand whether there's multiplicative or effects or just answer it. We were also interested in looking at the university as a social context. So wondering whether there are social determinants of mental health inequalities are additive and multiplicative, but also whether these effects differ depending on whether individuals go to university or not. And the main reason for that being that there is evidence that shows that students may have differential mental health issues that are different to those who don't go to university. So the university may shape mental health inequalities. As I said before, we used the long-student study of young people in England, the LSYPE. And this is so far, there's eight waves of data. And as you can see, we collected in this survey, data was collected for variables at different stages. And the things I've put in blue are the different social determinants that we focused on. And then we have mental health outcomes at different points. So some of these GHQ12 general health questionnaire waves two and four we use as a measure of adolescent mental distress. And then we had some outcomes, long-term mental health outcomes at age 25 in terms of mental distress, whether they declare the mental illness and whether they're self-harmed. So firstly, we took each of these social identities and positions and looking at multiplicatively, so we looked at whether each of these combinations of categories had an effect on mental health outcomes at age 25. So firstly, we didn't find any evidence for multiplicative effects, so combined. So that obviously doesn't mean that there might not be, there might also still be additive effects. So individually when we, or independently when we look at these social determinants, we were still interested in whether they could predict mental health outcomes. So firstly, so what we did is we looked additively at these effects and we looked at them split by whether they attended university or didn't attend university. So on the left, we've got those who didn't attend university. And some of the main differences I've just kind of highlighted and involved. So I'll just point out now. So for those who had experienced mental health, mental distress as an adolescent, they were slightly more, they were three to three and a half times more likely than those who hadn't experienced mental distress to have mental health problems at age 25. And it's not too surprising that that history predicts longer term mental health issues. But what we were surprised by is that actually those who went to university, although they were still more likely to experience mental stress later on than those who hadn't had it as an adolescent, it was slightly smaller in size, so 2.3 to 2.9 times. Female's about the same regardless of where they went to university. But the main difference was those who had come from a more socially deprived background, if they hadn't been to university, they were one and a half times more likely to experience mental distress and we didn't find any effect for those who had been to university. For self harm, they were 7.2 times more likely to experience mental distress as an adult, but still more likely for those who have been to university, but it's a slightly smaller in size and black and Asian individuals were less likely to declare a mental illness if they hadn't whether they've been to university or not. And some of these things that sort of explanations for this is that there could be potentially some positive outcomes of having attended universities. So slightly smaller effects in terms of long term mental health outcomes for those who've been to university and some of these could be structural things. So the opportunities that afforded say people who've from more socially deprived backgrounds, having been to university potentially that could be more positive for section minority individuals. It might be that there's some evidence that shows that universities can have a can be a more kind of embracing environment where people can feel more able to be themselves. So there's some potentially findings potentially pointed towards university having some kind of potentially protective impact for some individuals based on their background. Just in terms of the conclusions, we do need to be careful or cautious about interpreting the impact of combined multiple identities because we didn't find any evidence of that. And the main reason we need to be cautious about that is because it can be stigmatizing for certain groups if we just assume that particular, you know, combined people who have come from multiple marginalized identities are like are definitely going to all more likely to have negative outcomes. And it might just be independent effects that are going on. So the university environment could be having a positive effect, but obviously we don't know if this is causal. And it might be that there's some interventions that happen at university that are particularly beneficial, particularly particularly groups. And we should probably look at whether these maybe can be replicated in the general population. We have been lucky enough to just publish this study in SSM population health is open access. And yeah, the references and that's it. Thank you. Thank you now for our last session, our last presentation of this this hour long session. This presentation is focusing on the association of common mental disorders and oral health in a representative older population, age 50 and over. I just had the terrible realization just before this session that actually I'm just just eligible in a couple of months, I'll be eligible for the ELSA ELSA groups. I'm awaiting my invitation to take part. Afshan Merza from UCL is a part-time general dental practitioner and a part-time research assistant in dental public health at UCL. So we're combining both practitioner expertise and research expertise at the UCL Institute of Epidemiology and Healthcare. Thank you. Thank you, Sally. That introduction. I'm just checking. Everyone can hear me. I can see my slides. Perfect. So I know times of the essence, so I'll go straight into the presentation. So my data, my study was using data from the English longitudinal study of aging. And I was looking at the association of common mental disorders and oral health outcomes in a older adult population. So I will just give you some background to my study. So older, you know, older health, you know, older age is something that we're all going to experience. And oral health is good oral health is a really important aspect of healthy aging. You know, in England, in the UK by 2036, the number of older adults in the UK is expected to rise nearly 25% of the population. So good oral health is a really fundamental aspect of healthy aging. We need from a functional perspective, we need teeth to chew with, swallow, speech. We also have the psychosocial component of oral health as well, linked to confidence, self-esteem, smiling. So generally, older adults are at higher risk of poor health, just really due to the retention of teeth for longer. You have physiological changes associated with reduced library flow, which is linked to dental decay, pre-adorned disease. We have co-morbidities, multi-morbidity at the age. So we're looking at poly pharmacy, increased medication use and sugar is a prime ideological factor for tooth decay. We know that mental health is a public health issue, a global public health concern. And amongst those age 60 plus, depression, anxiety are the most common mental disorders, affecting 7% and 3.8% of older adults respectively. I have data here from the UK National Wellbeing Survey. And we can see that the older adults, the percentage of older adults being anxious or depressed range about 14% to about 22%, which is relatively high. So I did some background research on this topic. And there are a number of gaps that were identified. So just generally, there's a very limited study on the oral health of older adults. And we need to fill that gap. Consequently, there are limited studies on the oral health of older adults with mental ill health. There really is not much out there at all. The majority of studies that are present, they focus on either depression or anxiety, as separate disorders. There's not really a composite measure on common mental disorders. And the studies that are available, there are some limitations in study design, such as in a small sample size, convenient sampling, unadjusted analysis. So based on this information, I went ahead and formed my research question. So is this in the relationship between common mental disorders and oral health outcomes in the LSAT study? So in this study, I defined common mental disorders as a composite term. So that includes anxiety, depression, low self-esteem, low confidence. And there are a number of objectives with the study to describe the prevalence of common mental disorders amongst the study sample. To assess a relationship between common mental disorder and through oral health outcomes, oral health related to quality of life, self-rated oral health and complete tooth loss. And I'll talk about those measures in a bit more detail for you. So just a little bit of background on ELSA. So it's an actually represented cohort study of older adults, age 50 class. And that's one reason I chose this study because the older adult population is very defined. So it began in 2002. Participants are followed up every two years. And I chose ELSA Wave 3 because it has a common mental disorder variable. And it has the three oral health variables also. And that's not always common to the Wave. So just a brief participant flow diagram. So from the initial sample, I excluded partners from the study. I excluded those people living in residential facilities. And the analysis was run as a complete case analysis. So I ended up with quite a high number of 6,555 observations there. So I just wanted to talk about the oral health outcomes. So one of the negative points about ELSA, it doesn't have any clinical oral health measures. So we had to use subjective oral health measures, but nevertheless still really important. So the oral impact on daily performance. So there's a question. In the last six months, have you had any problems with your mouth, teeth and dentures? And of course the following. So this is really important because it tells us the impact of poor oral health on daily things, such as speaking, smiling, laughing, showing teeth, emotional stability, things that we pretty much take for granted when we have good oral health. The second variable we had was also a subjective health measure, self-weighted oral health. And that was also dichotomized as a binary variable. And the importance of this question is those that often have right there, oral health is poor, often have high clinical need. So it gives us some insight into that aspect. And the third outcome was teeth presence. So it was either no teeth present or one or more teeth present. And the hypothesis was that if you have a common mental disorder, then your oral health is going to be worse. So my exposure variable was the common mental disorder, sort of briefly discussed in the other presentation. So there's a 12 item general health questionnaire. So it's not a definitive diagnosis, but it takes into account patients could be depressed, anxious, unhappy, poor self-esteem and how this was coded. So negative answers were given the code zero, positive answers, very positive answers were given code zero, negative answers, a code one. So there's no cutoff score for the general health questionnaire, sort of a definitive one. So basically those that had a GHQ12 score of four or more consistent with other studies, they were defined as having a common mental disorder. So just briefly some of the results. So around about 40% of the old adults had a common mental disorder, which is quite high. Unsurprisingly, those with a common mental disorder were more likely to be women, non-white, low social health physicians, smokers have a longstanding illness. Nearly 10% of the sample reported at least one or impact on daily performance. And when I looked closer into these numbers, around 10% of the sample reported problems with eating around 8% of the sample reported problems with smiling. And that was that. And nearly 20% of the population rated their health as fair or poor, which is quite a large percentage. 16 and a half percent of the sample had a total two for us. So after adjusting for demographics, social and health factors, older adults with a common mental disorder had a 1.86 higher odds of reporting at least one oral impact, you know, which is relatively high. And the results were significant. 1.45 higher odds of reporting fair for self rated poor or health. So nearly 50% shots of that. But there was no association between having a common mental disorder and complete tooth loss. And, you know, one of the reasons is probably because tooth loss is a very gradual chronic condition. And here we're dealing with cross sectional data. So you really get information at one point in time. So as a door study, there's strengths and limitations. And one of the main limitations with the study is that we understand there's a bi-directional relationship to mental health and oral health. It's very possible that poor oral health can lead to poor mental health. That can't be ruled out. There were no measures on dental anxiety, dental attendance in ELSA. And, you know, these may be on the causal pathway and we couldn't explore those. Explore that. All the measures in my study, they were self-reported. So we can't rule out bias. They're reporting bias. There's no clinical data that we're relying on self-reported behaviors. So I think the take home message there was that older adults with a common mental disorder are at high risk of poor oral health. And we need maybe a collective viewpoint with dental professionals, mental health professionals, general practitioners, you know, to be aware of that so that we can all help older adults with a common mental disorder know to maintain oral health. I've said it's a really important aspect of a healthy aging. Thank you very much. Thank you. That was absolutely fascinating. OK, so welcome, everyone. Just about time now for our first presentation, which is on housing, financial conditions and mental health during a pandemic, which is by Marco Flichi, who is a PhD candidate at the University of Cambridge, Department of Land Economy and Darwin College. His research spans household finance, housing, subjective well-being and mental health. So thank you very much. Thank you, Seldi. Thank you, everyone, for joining in. So, yeah, I'll discuss my research on housing, financial conditions and mental health during the COVID-19 pandemic. I'll start with some context. So the COVID-19 pandemic has been recognized to have major impacts on mental health across the world and the policy response to the pandemic. So lockdowns and social distancing implied a lot more time spent at home as compared to before. So while the importance for mental health of the living environment was significant in pre-pandemic times, this has likely to have grown during the pandemic because of the nature of the pandemic and the policy responses. In addition to this, the increased financial stress because of negative shocks to employment and income could compound the total mental health. And in this respect, too, housing is prominent since housing payments being a rent or mortgage payments are a sizeable part of household expenses and something that is difficult to adjust as compared to the issue for instance. So my contribution is to test whether the gradient in mental health in terms of housing tenure that existed before the pandemic and this was recognized empirically in many countries. So I'll focus on the UK, but in many places we see that outright homeowners fare the best in terms of mental health then come mortgages and then renters. So I want to test for the UK if these stay the same during the pandemic and connected to this if two possible channels of these are affected specifically the ability to keep up with housing payment and access to other space. The data use is understanding society. So eight waves except the last one in September during the pandemic and the 10 waves between 2009 and 2020 of the normal understanding society's module. This panel data depending on the model that I use I have, I use about between 128,000 and 338,000 observations for about 32,500 individuals. Yeah, the data provides insight on many things but what I'm mainly interested in is mental health, housing tenure the ability to keep up with housing payments and access to other space. So this is descriptive it's pretty packed with information it also contains all the main insights I'd say from this research. So on the Y axis we have an indicator of mental distress based on one question from the general health questionnaire is the question on unhappiness and depression. But in fact, repeating the analysis with the 12 items general health questionnaire as a continuous measure yields a qualitatively similar results and irrelative terms also quantitatively rather similar results. On the X axis, we have a timeline a benchmark pre COVID period with data up to March 2020 between 2019 and 2020 and then all the pandemic waves. So monthly up until July 2020 and then by monthly until March 2021. So these are averages by housing tenure. So if you see we have four groups outright homeowners those that own with the mortgage those that rent from local authority or housing associations and those that rent privately. What can we observe? The first thing is that there's a clear relative ranking before the pandemic and this relative ranking states the same all throughout in particular outright homeowners fare the best then come mortgages then come private renters and finally social renters. The second thing we can observe is that also the relative distance between the categories is not greatly affected during the pandemic we see some dynamics but not that much. And because of this also there is a synchronization in how this indicator moves across the different groups. So it seems to increase and then decrease following the severity of the pandemic waves. So it increases going into the first wave then improves going into the summer and then again it worsens as we go into the second wave. This is the same thing basically by dividing between people that could be up to date with housing payments and those that couldn't and also separating those that had access to outer space and those that couldn't. I'll jump straight into the results discuss this a bit. These results are really to test a bit more formally what we saw in the descriptives. What I do is event study types analysis with so-called difference in differences. So there's an event that is the onset of the pandemic and then I systematically divide in two groups for each analysis. As an example, the first one here separates our tri-tome owners and the rest of the housing tenures. Zero is the this is a relative time. So it's periods to the pandemic to the onset of the pandemic. So April 2020 in this data. On the left, so the red dots are the difference between our tri-tome owners and the rest before the pandemic started. And on the right, we have the difference after the onset of the pandemic. In fact, zero here we can think about it as the difference is constant while going away from zero is a variation. So for our tri-tome owners, what we see is that there seem to be a relative decline in mental distress but although it's clearer once the pandemic has started the trend seems to have started before the onset of the pandemic. Similarly for mortgages but the opposite, the mirror-like image that seem to have been a relative worsening of mental distress but also with a non-set that predates the pandemic. And anyway, converges back to around zero in both cases. We then have public renters. We see some dynamics but not an increase that is not very well identified. Finally, private renters for which we see instead a pretty constant difference before the pandemic and then a clear jump down. So a relative decrease in mental distress that converges back. These four comparing those that were up to date with housing payments to those that were not, we see an increase but again, not very well identified. And comparing those with and without access to outer space, again, some dynamics but not very well identified. To sum up a bit the results, we see some sizeable short-term variations. For instance, the gap between private renters and the rest but these changes we absorb pretty quick. And looking back at the descriptors, we see a largely synchronized trends across tenures. So worsening and improve them but then moves together across tenures. And these similar to other what's been seen for the trend over all trends in other studies. So to conclude the pre-pandemic gradient in mental health across tenures stays largely the same during the pandemic. And the channels of financial distress and access to outer space don't seem to be affected much. So the pre-existing gradients seem to be persistent and resistant to this kind of multifaceted shock that the covenant in pandemic was. And therefore mental health seems to be inequality seems to be rather entrenched and structured. Thank you. I will see that I look forward to the discussion. Thank you, Marco. Absolutely fascinating. I think tenures is absolutely one of those socioeconomic indicators that really ought to be used so much more in our analysis because it's incredibly powerful predictors as you showed there and persistingly so. That now and I can introduce our next speaker who is Chiara Costi, a third year PhD candidate in economics at Lancaster University. Her research is focused on health economics with a particular interest on topics related to the economics of aging, long-term care and mental health. And her presentation is caring for others good for our mental health evidence from the COVID-19 pandemic in the UK. Thank you. Thank you a lot Sally for the introduction. I hope you can only hear me okay and you see the slides. No, just tell me, perfect. Thank you. Okay. So today, as Sally said, I'm going to present this work. It is a joint work with Bruce Fiennes in Central Virginia. And it works, okay. So the aim of this work is to investigate informal caregivers' mental health during COVID-19 and we focus particularly on caregivers with different caregiving experiences. So with different durations of caregiving. Why is this topic important? We know that informal care is crucial for the sustainability of most health care systems worldwide. It is a low-cost alternative to formal care. And due to COVID-19, there has been a sudden disruption of formal care services. Nowadays it is estimated that in the UK alone around 26% of the UK population now provides some form of informal care. And out of those people, around 4.5 million started to provide care after the COVID outbreak. So just to give you a brief preview of the results, we've seen that mental health fluctuated according to social restrictions. Existing caregivers, those who were already informal caregivers before the start of the pandemic seem to cooperate very well. Whereas the new caregivers, those who started after the COVID outbreak seem to have a deterioration, a significant deterioration in their mental health, especially during lockdown periods. So this might suggest that there is a need to psychologically support informal caregivers, especially at the start of their care provision. So the contribution of this paper would be to examine the role of a caregiving experience. So looking at these groups of existing and new caregivers versus the control group of a never-carried. We employ a mixture of propensity score matching to make the groups of informal caregivers never-carries more comparable based on observable characteristics related to the decision of providing care. And after this propensity score matching, we employ a different approach to obtain causal estimates. So first of all, we employ the traditional to a way fixed effect, considering two groups at a time. And then we also employ the new advancements as proposed by the framework of Caliou and Santana, where we can see different treatment groups at different periods in time. In terms of the literature review, we know that providing informal care is associated with our mental and physical health deterioration of informal caregivers. However, in all this literature, there is always a problem of self-selection into caregiving. Propensity score matching has been used recently to try to account for this self-selection issue. However, most of these studies are only use propensity score matching. And there is always the argument that if you only use matching techniques, then of course, you observe the bias is reduced, but maybe the results at the end are still a bit biased due to unobservable characteristics which are not controlled for. In terms of literature within COVID time, now there has been mental health research with the informal carers. These studies are mainly focused on a convenient samples and therefore they have limited external validity. Some are based on the concessional desires. So far, or at least as far as I know, there are only two published papers that try to account for self-selection into caregiving while looking at longitudinal data sets. Mak et al is one where they compare the mental health of informal caregivers and non-caregivers. However, their sample is not representative and they do not have a pre-COVID data and they only use a propensity score matching. And the other study is Bergman and Wagner which use two waves of share, one before and one after COVID-19. And they use a mixture of matching analysis and the linear regressions. However, they only analyze the two waves and the matching variables that they use are not related to the provision of care. Here I also want to mention that nowadays and very recently there has been even more researchers looking at the mental health and informal caregivers particularly. Also at WEB which is in our session now has done a really good work in terms of mental health and home care. So his work and his work is kind of complimentary. So it's really nice to see that mental health of informal caregivers is a topic that is becoming widely popular among researchers. In terms of the data set to use in this study, I've used understanding society, in particular the three regular main stage questionnaire collected before COVID-19, so wave eight, nine and 10 and the eight COVID-19 survey questionnaire waves from April 2020 and March 2021. So the final sample analyzed is composed of around 4,700 respondents. They are interviewed in all the 11 waves and this final sample has been obtained after dropping those answer to mental health which is our outcome of interest and to the informal care status. Also I dropped the dose we're already informal caregivers in wave eight which is our first wave analyzed. So in wave eight, nobody is providing informal care and I can do the capacity score matching technique. Also I dropped them carers because this paper only focus on external caregivers so those who provide care to somebody living outside their houses. And I also make sure that if somebody starts to provide informal care, they carry on later on. So I dropped those people without a continuous pattern. For the variables used, mental health is measured with the well-known GHQ cases. So the main analysis is done with this variable. I've also conducted another analysis with a binary indicator of a score of four or more. The covariates included in the analysis are the usual demographic and socioeconomic characteristics but also some variables specific to COVID to capture the COVID situation at each time point. The pre-treatment variables used as you can see from this table, I hope they are divided into three main areas which according to past literature, these main areas are related to the decision of providing informal care which is the need to, the willingness to and the ability to provide the care. In terms of the analysis, we know according to the recent literature about different frameworks that to a way fix effect is not robust to treatment effect at the regenerative. So as I already mentioned, our approach is to pre-process the data using propensity score matching. So as you can see from this table after the matching, we do not find the significant differences between groups. So after this matching, I conducted the separate 2FA regressions considering two groups at a time. And then as I mentioned already, I've also used the Kalo-Encentana framework which basically allows us to simultaneously estimate the average treatment effect on different treatment groups based on different time periods. So in this case, based on the exact time period in which these people start to provide care. Before doing this different approaches, we checked the critical assumptions of different designs which is basically the parallel trend. So we checked those with the visual inspection. As you can see that from before, the treatment happened that the trends in between the two groups are really similar. And we also find the insignificant pre-treatment coefficients in the table of the results as I will show you later on. The analysis is conducted using one-to-one PSM was the kernel of PSM is conducted and we found the similar results. So this slide I just showed you the classic approaches the two-way fixed effect and difference in different multiple time periods. So I'm not gonna go through this in detail but I'm just gonna show you the results. So this table here hopefully is not too small for you. Shows you the results of the two-way fixed regression models. So the first three columns here shows you the regression for existing carers as compared to another carers. So we can see that the estimates are positive and when we see positive estimates, it means that there is a deterioration in mental health. However, for existing carers that are all insignificant apart from COVID wave seven, which is collected in January, 2021 when the UK was in its third national lockdown. So it seems that for existing carers they were coping relatively well during the first year of the UK pandemic. On the opposite side, when you look at new carers in the last three columns of the table we see that they had a significant mental health deterioration especially during lockdowns which are captured by these shaded areas here in the table. So we see that it is also a kind of big magnitude which is comparable according to past literature as major life events such as unemployment for instance. So here really suggests that there is a need to psychologically support the caregivers especially at the start of their care provision and also when they do not have support because for sure when lockdowns were in place also there was a lack of this support from somebody outside. These figures here shows you the results for the different different with multiple time periods which is the Kallou and Santana approach. The first few figures three and four shows you the existing caregivers results when they start in wave nine and wind 10 respectively and we see that the estimates are all insignificant which is a line with what we found in the previous slide and the final picture figure five shows you the results for new carers and we see that again this is a line what we found previously there are significant mental health deterioration in the first period after the COVID-19 outbreak but in this analysis we see that at the end this the estimates turns out to be insignificant. So in the last minute I believe we I'm just gonna show you some limitations of this study. As I said at the beginning we only look at external informal caregivers so it might be that the estimates found are in the lower bound. I only look at the GHQ-12 because of data. We know that it is a self-assessed indicator so maybe if we had the chance to have a look at other mental health outcome and maybe more objective it would be interesting to see if we find similar results. The intensity of care is not examined so we do not look at the type of care provided but we only look at the period, the duration of care. It is a representative sample of the UK population and it might be interesting to see what happens in other countries too. And just to conclude as I said already we see that the psychological well-being fluctuated according to social restrictions but the new carers were the most affected especially during lockdowns and there is really no need to psychologically support these informal caregivers especially the staff. So thank you a lot for being here. Thank you Chiara. I love the way you got the embedded links to your visualizations that kind of pop up and that was as beautiful. The graphics were amazing. So our next presentation is from Gemma Shields from University of Manchester. She is presenting on key worker health status pre-anduring the COVID-19 pandemic and explorative analysis using the EQ5D5L. Gemma Shields is a lecturer at the University of Manchester focusing on the design and implementation of economic evaluations in mental health and chronic disabilities. Brilliant, so hi everyone. As introduced I'll be talking about key worker health status pre-anduring COVID pandemic. Just a quick thank you to the abstract co-authors and also the wider project team. So this is a very, very tiny piece of work conducted as part of the resilience hubs project funded by the NIHR. So just a quick disclaimer for everyone to read. So some background. It was noted during the COVID pandemic that health and social care workers would be at high risk of acute and long-lasting mental health conditions due to things like experiencing trauma at work, working long hours, having to isolate from friends and family, risking giving your friends and family COVID and so on. So the NHS clinical leaders network issued an urgent call for action. And as part of those, there were a number of research priorities including monitoring mental health in the key worker population, improving fine posting to mental health services and in general trying to ensure that key workers put access evidence-based interventions to support their wellbeing and mental health. So some background on the resilience hubs project. It's a mixed method study evaluating resilience hubs set up to support key workers during the pandemic. Resilient hubs aim to increase screenings for mental health conditions and also facilitate access to evidence-based interventions for mental health. They do target different population groups but we're focusing specifically on key workers looking for support for their wellbeing and mental health during the COVID pandemic. It started in October 2020 and we submitted the final report last June but there's still more exploratory analysis going on as I'll talk about. This is a small portion of the health economics that was conducted and we're focusing on health data but you can see there's a lot more work done and I've included the Twitter for people who are interested. But in terms of our aim, we're looking at summarizing the health status of key workers who access to support. So that's using our own data from the resilience hubs project and we want to compare them to a pre-pandemic sample of health and social care key workers and we did this using 2018 data from Health Survey for England. So our methods in brief for the resilience hub's clients we've sent out a service use questionnaire. It's titled that because we also included fields on access to NHS and social care use, mental health and wider physical health services too. We emailed hub clients who had consented to be contacted for further research about six months after they were screened by resilience hubs. So at that point some of them will have had interventions for their mental health others will currently be receiving it some will still be on rating list. And we recruited 299 clients from four resilience hubs who were mainly NHS staff at screening. When they screened with the resilience hubs many were showing several and substantial mental health difficulties including anxiety, depression, PTSD and so on. In the service use questionnaire we included the EQ5D which is a generic measure of health status that includes five domains. We've got mobility, self-care, usual activity, pain and discomfort, anxiety and depression. And we use the five level version. So there's five levels within each of those domains ranging from no problems to extreme or unable depending on the domain. We converted responses to utility values for the non-health economists a utility value is a numeric value we attach to a health state. It ranges between zero which is equivalent to dead up to one which reflects full health or perfect health so it allows us to look at health states by assigning numeric values. And we had 270 complete cases. This wasn't the sample size we wanted and so we summarized using simple descriptive stats. Now one of the main limitations of the resilience hubs data was that we didn't have the control group. We didn't have data on key workers not accessing mental health services during the pandemic or accessing alternative services. We also didn't have our resilience hubs client data at screening so we didn't have their health status when they entered resilience hub support nor did we have it for pre-COVID and pre-the pandemic. So we started looking at identifying relevant groups of comparison. As is fairly standard we started with looking at general population norms. So we have typically accepted general norms for the UK in terms of utility. We also looked at comparing to different mental health conditions so published evidence from groups of anxiety and depression, people accessing IAPT and so on. But what we really wanted to do was look at a group of key workers. Ideally majority health and social care professionals. So we did some targeted literature reviewing and we couldn't identify any published studies but that's where HSE data came in. In 2018 the EQ5D5L was reported and occupation groups were also reported which allowed us to restrict to your health and social care professional sample. Much smaller numbers than you'll have seen in the other studies presented here today. We only had 348 to compare to. And it's still very useful for us given the complete lack of data published elsewhere. And so again, stuck to simple descriptive stats and we applied the interview way to adjust for potential buyer. So looking at the results. So in the purple box, you'll see our resilience hub client sample mean EQ5D. So the mean utility is 0.755. So remembering that this is about six months after they start accessing mental health services that's still quite a long way off the one which would signify full health or perfect health. And compared to the general population means for similar age groups, you can see it's lower in comparison and quite substantially too. We'd expect it to be 0.85 upwards. Compared to people with mental health conditions, those they published anxiety and depression data. It was typically higher. So the resilience hub client sample has generally better health status. As you might expect because they started receiving intervention for their mental health. Now most interestingly compared to the HFE sample, resilience hub's client's health status is lower than might be expected. So I'm looking first at reporting across domain. So the proportion reporting at each domain and level, we can see that her clients consistently reported having slight or more problems across each domain. It's particularly noticeable for usual activities and anxiety and depression because we're focusing on mental health. We just got a graph of the anxiety and depression domain. We can see for our hub client sample over two thirds are reporting having some problems versus only about third for the HFE sample. So despite this intervention, they are still having quite significant issues with that domain. Now looking across the age group groups across utility values, we can see consistently that by age group resilience clients report lower, well have lower EQ5D values or lower utility. Although as you can see here, it's one of the issues we hit with sample size. We're relying on very small samples of data for these comparisons, which is why this analysis and perhaps isn't as complex or interesting as we had originally planned for. So on to the discussion. This is the first known EQ5D data related to a sample of key workers. And so one of the things we want to focus on is encouraging more future research, linking mental health and health status in key workers specifically, perhaps comparing to other groups and so on. It does suggest that the hub client sample EQ5D values were lower than would be expected, but having reflected on this as a research team, we're wondering whether it is actually lower than would be expected. So looking at the wider literature, we're seeing an ongoing effect of the pandemic on key workers, so impacting their mental and physical health. We still also have a proportion of our population who are receiving mental health treatments or are on waiting lists for things like IAPT and so on. We also have the ongoing impact of COVID-19, which we know from the general population if having an impact on health status. And there's also interestingly some publications on delayed dysfunction, which suggests that following disasters, people's health will reduce more over time and it will take some time for the most severe symptoms to impact. Now, as you can see, this is very exploratory and simple in terms of the analysis. We have loads of challenges making comparisons in particular sample size. It's still interesting because it is the first EQ5D data we could find for key workers, but it does really limit the conclusions. I think in particular, what I would like to see is kind of more historical data on the mental health of key workers so we could compare more between the different data sets and inevitably because we are using different data sources with loads of issues around having things in different formats, having asked questions very slightly differently and so on. So yeah, recommending more research in particular, no relevant to HSE data. We're thinking about if there are future data cuts that could be used, maybe we could look just focusing on HSE data, what the kind of pre and post pandemic health status of key workers is and also maybe comparing key workers to other occupations as well. I've noted some people have been focusing on occupations during this conference today. So it'll be interesting to see what else is available. Thank you. Any questions? That was absolutely fascinating. Thank you, Gemma. And now we move to our final presentation of the hour, which is COVID-19 lockdown, unemployment and mental health evidence from the UK. Sinjoy, sorry. Yeah, Sinjoy. Sinjoy, perfect. Sinjoy Sen is a PhD student from the Department of Economics and University of Warwick. His research interests lies in assessing the impact of policies in the UK. He uses modern econometric methods, notably causal inference to address his research questions. One strand of his research focuses on quantifying the impact of certain UK austerity measures, while another strand focuses on mental health concerns arising from the UK government policies during COVID. He uses UK administrative data or large scale, longitudinal surveys for his research. Hello, everybody. So I'm Sinjoy. I'll be presenting my paper on COVID-19 lockdown, unemployment and mental health evidence from the UK. So before we move, let me give a brief background. So as we all know that COVID hit us, hit UK pretty hard in the face of rising cases. The then UK Prime Minister Boris Johnson had announced nationwide lockdown on 22nd of March, 2020. As we can see from the Office of National Statistics, the unemployment rate went up during the lockdown period, which is from April to June of 2020. And then it kept increasing, and it was the highest since 2015 as per awareness. So even though this table depicts a macro-level picture of the post-lockdown overall unemployment situation in the UK, there are similar micro-level findings that are also revealed from the COVID-19 surveys, especially the understanding society survey. Before we move on, let me give a brief background. So this is an image that I got from Ethnogen Spandix paper in 2020 that pandemic has not only affected the physical health of people, but also has its effects on mental health and well-being. As we can see, this is the mean well-being score. It's the average well-being score over the periods 2010 to 2020. So we can see there's a sharp fall in mental well-being, especially it's more prominent for female as compared to males. It's a sharp fall in 2020, and this could be attributed to the COVID-19 lockdown. So the research question that I'm looking at currently is basically to find a causal link between reduction in work hours due to the COVID-19 induced lockdown and mental well-being. So there are a lot of papers in the literature who've looked at a link between COVID-19 and mental health. They've looked at the effect of childcare responsibilities, household work, the differential effect of mental health on people from different ethnic groups on gender. But what we try to establish is that COVID-19 affects mental health through the channel of unemployment. This is what we are trying to identify. And our empirical identification strategy allows us to find the impact of foreign employment which can entirely be attributed to reasons led to COVID-19 in post lockdown on mental health. So unemployment that arises because of the lockdown and not because people have decided to quit or people decided to just work less. So as far as my knowledge goes, there's no studies that has looked at this particular impact especially the channel of unemployment on mental health due to COVID-19 lockdown and has established a causal mechanism. So the data that we use in this paper is the UK House of Long Literal Study. So I've looked at the yearly waves that happen pre-COVID and as also the monthly waves that happen in April, May, June, 20, which are the post lockdown months. So the UK HLS released a COVID-19 study which was basically an online self reported survey of individuals. And they wanted to know about the state, social and economic state in the COVID-19 phase. So what we zero in are 10,000 individuals who have, who have been tracked from wave seven, eight, nine. So wave seven, eight, nine goes back to 2015, 2016, 2017, 2018, 2019. So pre-pandemic era to the April, May, June, 2020 of the COVID-19 service, which is the post pandemic or the post lockdown period. So for, we will look at graphs in the latest, we look at graphs and for those graphs, we refer to April, May, June as waves 10, 11, 12, even though it's technically not correct because we have 10, 11, 12 waves in the UK HLS survey. But for simplicity, we will refer to the April, May, June as the waves 10, 11, 12. Okay, so the main outcome variable that we focus on is mental health. And mental health we can gauge from the generalized health questionnaire. So generalized health questionnaire has been covered in all waves of the UK HLS studies, both pre-COVID and post-COVID. So as we can see, this is a snapshot of the generalized health questionnaire. So it is basically a combination of 12 different questions that basically look to gauge a person's mental health, like concentration, lack of sleep, capable of making decisions, whether they are enjoying data activities, whether they are unhappy or depressed, losing confidence. And for all these questions, they have four options to choose where, like for example, like on your concentration, have you recently been able to constrain on whatever you're doing? So they have four options, like not at all, no more than usual, rather than usual, or much more than usual. So to get more numeric measures of mental health, we look at this Likert score. So these three measures that we have Likert score, Likert score, standardized Likert score and the mental health problems indicator, these are basically drawn from the psychology literature, which is a way of gauging a person's well-being. So for example, Likert score yields a total score between zero and 36, zero being least distressed and 36 makes most distressed. So then standardized Likert score, we basically standardize it. So we subtract the mean and divide it by the standard. So that the differences in the GHQ 12 can be measured in standard deviations. And the mental health problems indicator is equal to value one, if the individual is at the risk of mental health problems as we could see from the Dali et al. for each one of the paper. So it's a simple indicator variable that takes the value one if the individual is the risk of mental health and zero, otherwise. So the empirical strategy that we use is a difference in different strategy, which the previous presentations alluded to. So here in difference in difference, we want to look at treatment and control. What I take as a treatment is the individual who realized a fall in work hours in the post-COVID lockdown survey in comparison to January February 20. So January February 20 was right before March. So that was the period right before the lockdown got initiated. So we can filter how these respondents, because the respondents in the survey, they have also mentioned why they experienced reduction in work hours. So some are due to non-COVID reasons, which could be that they are retiring or the employer or they just took a sabbatical or the seasonal variation of the job basis that they worked less in April, May, June, 2020. And then there are reasons, COVID-19 related reasons where it is outside their control whether the employer cut their tasks, they were being redundant, they were being laid off or they were being furloughed. So we focus on the treated individuals, those who experienced a reduction in work hours in at least one of the post-COVID months, which is April, May, June, or the post-lockdown months, April, May, June. So we use pool DID model, which tries to average in the effect of the people, average in the mental health effect on people who've experienced reduction in work hours in at least one of the post-COVID months. And then even study design is a visual method of showing the definitive strategy which shows the evolution of mental health over time. So this is the regression equation. The good thing about the definitive model is that it controls for unobservables, which are time invariant, which are corrected by the fixed effects, the individual fixed effects and the time fixed effects. Also we use some time variant controls as well, which where we interacted time effects with education, gender, even location as well, so that if there's any unobservable shocks that affect the location, and by location I mean the 12 regions in the UK, so the West Midlands, East Midlands, North West, those regions. If there's any sort of employment shocks there that is being controlled by using the XIT variable. So we have tables, but I think it's best to look at a visual representation of the even study design. So as I mentioned, as you can see, so higher coefficients mean worsening mental health. So if we index wave nine, so remember wave nine is 2019 wave, so wave nine, if we index it as zero in wave nine, so we can see that in 10, 11, 12, which is 10 is April, 11 is May and 12 is June, we see there's a definite significant increase in the likelihood score, which means that individuals who experience reduction work hours are experiencing worsening mental health. So assumption that needs to be, so to identify this effect, one assumption for different different model is that there should be the trends, the pre-trends should not diverge, and which we can see the pre-trends hold in this case. So this is our dependent variable or the outcome variable is likelihood score. For the standardized likelihood score, we can see that on average in the post-lockdown period, mental health worsened by around 0.05 standard deviations. Again, pre-trends hold here, so the different identification strategy holds in our case, and this is even more prominent. So we can see that on average people, so this is, so we look at the mental health problem indicator, which means whether we want to see whether a person, how likely the person is to suffer from mental health problems. So we can see that on average in the April, May, June post-lockdown period, individuals who experienced reduction in work hours, and at least one of the post-lockdown months are three percentage points more likely to suffer from mental health problems. So just to conclude, we have done more robustness checks, and our results have been statistically significant and have been robust. And we try to disentangle two different, but relating back. So we look at the impact of the fall in employment, which can entirely be attributed to reasons related to COVID in the post-lockdown period on mental health. The results are that people who experienced fall in work hours due to reasons outside their control, like being made redundant, being furloughed, were had significant, were three percentage points more likely to be at the risk of mental health, or their likers were increased by 0.05 standard deviations. And as we expected that deterioration of mental health is much stronger for individuals who experienced a fall in work hours due to COVID-19 related reasons as compared to those who experienced fall in work hours due to non-COVID related reasons. And by non-COVID related reasons, I mean, whether they were being, whether they're seasonal variation, they decided to take a sabbatical, or they decided to retire, or they decided to just work less because of their own accord. So yes, so that's about it. So that concludes my presentation. Thank you. That was beautifully clear and brilliantly presented.