 My name is Jennifer Boyko. I'm the Manager of Scientific Operations with the Canadian Longitudinal Study on Aging, or CLSA for short, as most of us, I think most of you know. Thank you for attending our May webinar on this beautiful hot May day entitled, Comparing Measures of Obesity in Relation to Health Care Use in Adults from the Canadian Longitudinal Study on Aging. As I mentioned last month, if you were here, the CLSA launched a COVID-19 study in April to date. More than 27,000 CLSA participants have actually taken part in either our web or telephone-based questionnaire. I encourage you to visit our website at www.clsaelcv.ca backslash coronavirus. If you want more information, we're trying to keep that regularly updated. I also want to acknowledge that the CLSA National Coordinating Center, where we normally are, is located on the traditional territories of the Mississauga and Haudenosaunee nations, and within the lands protected by the dish with one spoon wampun agreement. As attendees of this webinar, I encourage you to learn more and to acknowledge the original inhabitants of the land where we currently and you currently have the privilege to do research, live, and work, and wherever that may be. So now to today's webinar, Comparing Measures of Obesity in Relation to Health Care Use in Adults from the Canadian Longitudinal Study on Aging. Let me now introduce our speaker, Alessandra, sorry, Alessandra Andriaki. Hopefully, I did justice to that, to your name. She is a Masters of Public Health graduate from McMaster University. She completed her master's thesis with the CLSA on this project under the supervision of Dr. Laura Anderson. Her research interests include aging, obesity, and body measurement, as well as chronic disease prevention. So I will now give it over to you, Alessandra. Perfect. Okay. Thank you, Jennifer, for the land acknowledgement and introduction. And thank you, Shirley, for arranging this and for making sure that all of this is set up properly. Hopefully, we don't run into any technical difficulties, but as Jennifer introduced me, my name is Alessandra Andriaki, and I am a recent graduate from the McMaster Master of Public Health Program. And today I'm going to be sharing with you some of the work that I completed for my master's thesis. So, before we begin, and I get into any of the research, I just want to quickly set the stage for this presentation and touch on how we need to reframe the discussion around obesity. So, obesity is known to be a disease according to the Centers for Disease Control. So it's important that we use person-first language to eliminate weight bias and discrimination by not labeling a person by their condition. And this goes hand in hand with using non-stigmatizing images and messaging. So, throughout this presentation, I'm going to try my best to be using person-first language and the images will hopefully be non-stigmatizing ones. For any more information, if you'd like to learn some more, you can visit the links that I've posted on this slide at the Obesity Canada website and World Obesity. So, as a very brief overview, I'm going to begin by giving a little bit of a background. I will then introduce my research objectives. I'll describe the methods that I use to investigate them. I'll present some key findings, and then I'll finish with some implications and conclusions of this work. So, as a brief background about obesity, we know that obesity affects 27% of Canadian adults, and this is based off of body mass index. Obesity is associated with many different chronic conditions ranging from cardiovascular disease type 2 diabetes and even some forms of cancer. And obesity is also associated with increased healthcare use and costs. So, I think it's important to really understand what does it mean to have obesity? So how do we actually measure obesity? And the first way we can measure obesity is using an anthropometric measure called body mass index or BMI for short. And BMI is the ratio of one's weight to height squared, and it's the most commonly used measure in epidemiological studies and in clinical settings. But there are some implications with using body mass index such that it doesn't consider one's body composition. So, BMI doesn't differentiate between the type of fat, muscle, or skeletal weight. And additionally, there are some implications with using BMI in older adults as adults experience age-related height decline that can inaccurately register an increase in BMI when an individual remains at the same weight. And also, adults go through changes in body composition with age that aren't accounted for by BMI. So, the most commonly affected cut points for obesity is greater than or equal to 30 kilograms per meter squared. And those are cut points that are recommended by Health Canada and the World Health Organization or the WHO for short. Next, we have two measures of abdominal obesity that we could use to assess one's obesity status. So, the first is waist circumference. And one's waist circumference is a good measure of abdominal obesity. And we know what's important to look at abdominal obesity because it's been demonstrated to be associated with many different cardiometabolic diseases than just fat in general. The most commonly accepted cut points for obesity are greater than or equal to 88 centimeters in females and 102 centimeters in males. And those cut points are recommended by Health Canada and the WHO. Third, we have waist to hip ratio, which is the ratio of one's waist circumference to one's hip circumference. And although we have this measure, it has been recommended by many guidelines and reports to use waist circumference over waist to hip ratio for many different reasons. One of the reasons being is that waist to hip ratio is a relative measure, not an absolute measure. So, two people with very similar waist to hip ratios may have very different body mass indexes and waist circumferences. Secondly, waist to hip ratio may not identify if someone were to lose or gain weight proportionally from their waist and their hips. We may not, we might not necessarily see a change in one's waist to hip ratio, but we would see a change in waist circumference. And lastly, a reason why we would use waist circumference over waist to hip ratio is because waist to hip ratio involves two measures. So, it increases the chance of measurement error. And some commonly accepted cut points for waist to hip ratio are greater than or equal to .85 in females and greater than or equal to .9 in males. And those were recommendations made by the WHO. And lastly, a fourth measure that we can use to assess obesity is percent body fat. And many consider this to be the gold standard of assessing adiposity using dual energy X-ray absorbed geometry. Many call this DEXA. And we may not see this measure being used as often as we'd like in epidemiological studies because it is very costly and time consuming to obtain percent body fat on large populations. The most commonly accepted cut points are greater than 35% in females and 25% in males. And these are cut points that are based off of use in previous literature. So next, I conducted a small literature review of studies that looked at the relationship between obesity and health care use. And I identified 19 studies. Most of these studies reported increased health care use amongst individuals with obesity, but I found the results very difficult to compare and summarize for a few different reasons. The first reason being is that most of these studies only used body mass index to assess obesity. And there were only two of these 19 studies that used waist circumference in addition to BMI to assess obesity status. And of these studies, most of them used self reported body measures to assess obesity. And we know that there's some issues with self reported measures like height and weight because height tends to be overestimated, weight tends to be underestimated, leading to a general underestimation of one's BMI. And this is especially true for females. Another reason why the results of these studies were very difficult to compare is because they all affect health care use outcomes in very different ways. Some of these studies looked at any health care use in a given period of time. Some looked at the number of visits in a given period of time. And this period of time that they were asking their participants to report their health care use also varied. Some asked for health care use in the previous six months and some asked for health care use in the previous 12 months. These studies all used very different statistical analyses and adjusted for very different covariates in which most of these studies conducted legit logistic regression analysis and reported odds ratios. And lastly, many of these studies lack the inclusion of older adults in their analyses. And we know that it is important to capture this relationship in an older adult population since the proportion of older adults is expected to rise in the future in Canada. So this leads me into my research objectives. And the primary objective of this work was to evaluate the association between anthropometric measures, including body mass index, waist circumference, waist to hip ratio and percent body fat, with health care use in the past 12 months amongst community living Canadians 45 to 85 years of age. So I looked at how all four of those anthropometric measures were associated with four different indicators of health care use. And those health care use indicators were any visits with a general practitioner in the previous 12 months, any visits with a medical specialist, any visits to an emergency department and having been a patient in a hospital overnight. As secondary objectives, I wanted to investigate if the associations between anthropometric measures and health care use differed by sex and by age. And another secondary objective was to evaluate the associations between anthropometric measures, including body mass index, waist circumference, waist to hip ratio and percent body fat. So in terms of methods, I completed a secondary data analysis using data from the Canadian Longitudinal Study on Aging, as you are all aware, or CLSA for short. And very briefly, the CLSA is a national longitudinal study of Canadian women and men aged 45 to 85 years at enrollment. And there are just over 50,000 participants that either fall into the tracking or the comprehensive cohort. And exclusions to Canadians being able to be a part of the CLSA is anyone living in the Canadian Territories or on Federal First Nations Reserves, members of a full-time members of the Canadian Armed Forces, people residing in institutions, and anyone that was unable to respond in English, French, or provide informed consent. So specifically, I used the baseline data from the comprehensive cohort, which is just over 30,000 participants that have measured anthropometrics by trained staff. And I excluded five participants based off of implausible waist circumference or waist to hip ratio measures, which left me with a final sample size of 30,092 participants. So a little bit about the data design. The data is prospective in nature. And if we look at the timeline on the left here, at baseline or month zero is when the anthropometric measures were recorded. And about 16 to 18 months later, the CLSA had administered a maintaining contact follow-up in which participants self-reported their healthcare use in the previous 12 months. And that little gray shaded area on the timeline represents the 12-month healthcare use recall period, although this time period varies for participants depending on when their maintaining contact questionnaire was administered. And if we look at the histogram on the right, this shows the distribution of the days that participants had followed up and reported their healthcare use. And for the majority, about 69% of participants had their healthcare use reported at least 12 months after baseline, meaning that for those participants that healthcare use recall period won't overlap with the baseline and their data is truly prospective. So now I'm getting into the methods I use to analyze my data in terms of all my analyses were conducted using SAS and R. And in terms of descriptive, I computed weighted descriptive statistics. So means and standard deviations were computed for continuous variables, frequencies and percentages for categorical variables. After I computed my descriptors, I then wanted to go in and compare these anthropometric measures. So I compared BMI, Waste Reconference and Waste to Hip Ratio all to percent body fat, which was my reference standard for this study. I computed scatter plots, piercing correlation coefficients and sensitivity and specificity analyses. And then next, I got into a regression analysis. And this was done to estimate the associations between obesity and healthcare use. I first computed relative risks. And then I also computed risk differences as an absolute measure of association to compute my risk ratios. I computed generalized linear models with a log link distribution. And then for risk differences, I used an identity link distribution. And both of these analyses were weighted using analytical weights that were provided by the CLSA. I have two models here where I adjusted for different covariates. And all of these covariates were selected a priori based off of rationale and previous literature. And I also use the Anderson's model of healthcare use as a guiding principle. In my first model, I adjusted for age, sex, education, household income, urban and rural living, smoking status, alcohol use, marital status, province of recruitment and follow up time. And in my second model, I additionally adjusted for chronic conditions and self-rated general health. And the rationale for having the second model is that a lot of researchers and previous literature have debated whether chronic conditions and self-rated general health are mediators of the relationship between obesity and healthcare use. So some researchers argued that we should adjust for it. Some argued that we shouldn't adjust for it. So I wanted to see how my estimates changed if I did adjust for them. And then lastly, to investigate if there were associations, if the associations differed by sex and by age, I stratified my regression models by these two variables. So in terms of results, these are all of the descriptive statistics, weighted descriptive statistics for the variables that I control for in my models. So we can see the proportions for all the different types of variables. We can see that there's an even distribution between males and females. You can see the distribution by the age group. And if we keep going, we can see the distribution of smoking status type of drinking. We could see that for chronic conditions, about 70% of the cohort reported having one or more chronic conditions. And here we can get into the prevalence of obesity when defined by different anthropometric measures. So we could see by indicated by red arrows here that 29% of individuals have obesity by BMI, 42% have obesity by waist circumference, 62% have obesity by waist to hip ratio, and 73% by percentage body fat. So we see a very different prevalence estimates depending on what measure we use to define obesity. And for the most part, the prevalence values are similar between males and females. But I would like to highlight some differences between males and females for waist to hip ratio defined obesity where 37% of females were identified as having obesity by waist to hip ratio versus 88% of males. So we're seeing very different, very different prevalence values between males and females for this anthropometric measure. And then next we can get into looking at the proportions of healthcare use in the cohort where 89% of participants reported having any contact with a general practitioner in the previous 12 months. 48% reported having any contact with a medical specialist, 18% reported being seen in the emergency department, and 7% reported being a patient in a hospital overnight. And for the most part, these estimates are pretty similar between males and females with females having slightly higher proportions for the first three types of healthcare use. So the next thing I did is I computed scatter plots and piercing correlation coefficient, comparing each of the measures to percent body fat. And we can see the scatter plots here demonstrating the correlation between percent body fat on the X axis and BMI on the Y axis. Females are on the left in blue and males are on the right in green. And you'll see black vertical and horizontal lines that represent obesity cut points for each measure so that we can begin to visualize any misclassification. The top right quadrant represents individuals that were correctly classified as having obesity by both BMI and percent body fat. The bottom left quadrant represents correct classification as not having obesity by both measures. And then the top left and the bottom right quadrants contain individuals that were misclassified by either of the measures. So what we can see here for this correlation between BMI and percent body fat is that BMI is strongly correlated with percent body fat. With this correlation being stronger in females having a piercing correlation coefficient of 0.75 and then the value in males is 0.7. And you can also see they're associated 95% confidence intervals. If we look at the correlation between now waist circumference and percent body fat, we can see that there still is a strong correlation between waist circumference and percent body fat. Although now the correlation is stronger in males with a coefficient of 0.75 and then on the left in blue we can see that the correlation is slightly lower in females with a value of 0.7. And then lastly looking at the correlation between waist circumference and percent body fat, we see slightly different results here. We see a weak relation overall with a very weak correlation in females as the coefficient is 0.29. And then still weak but a little bit stronger of a correlation in males with a coefficient of 0.46. So beyond this I also wanted to look at the diagnostic accuracy using sensitivity and specificity of the cut points used for BMI waist circumference and waist tip ratio compared to percent body fat, which was my reference standard here. And what we can see by comparing sensitivity and specificity values is that BMI and waist circumference compared to percent body fat have high specificity at the expense of a lower sensitivity. And this is what we've seen in previous literature, but I think it's important to actually be able to interpret what these numbers mean. So let's look at the sensitivity and specificity of BMI defined obesity in predicting obesity by percent body fat in males. So these are these two values that I've circled in red here on the screen. So the sensitivity value of 40.3%. What that means is that of males who tested positive for obesity by percent body fat. We don't put the project for obesity by BMI. So when we have a lower sensitivity like we do here, this increases our chance of having false negative. And then if we look at the specificity value here of 95.3%. What this value tells us is that of males who tested negative for obesity by percent body fat, 95% of them also tested negative for obesity by body mass index. So when we have a high specificity value like we're seeing here, this decreases our chance of having false positives. So we're pretty sure that when we're letting someone know that they don't have obesity, that it's actually true when comparing that to the reference standard of percent body fat. And I just wanted to highlight the comparison between waist to hip ratio and body fat separately because we're seeing some very long results here as well. Like we did with our prevalence values and with our correlation coefficient and scatter plots. So what we see is that there is high sensitivity and low specificity using waist to hip ratio to predict percent body fat in males. With the opposite trend appearing in females where there's low sensitivity and high specificity in females. So this is not necessarily agreeing with what we're seeing for BMI and waist circumference and these differences in between sex are a little bit concerning. So what I can gather so far is that body mass index, waist circumference, waist to hip ratio and percent body fat may be measuring different aspects of obesity. And this idea is based off of once again the very different prevalence values we're seeing their correlations and also their sensitivity and specificity. This also just makes sense based off of what these anthropometric measures are actually measuring. So now this led me into doing my regression analysis and comparing these associations looking at the association with healthcare use. So adjusted for the variables in model one, the relative risks and the 95% confidence intervals are all greater than one. This is telling me that adults with any definition of obesity were significantly more likely to have used any of the four healthcare services in the previous 12 months compared to adults without obesity. And the relative risks for each type of health care use don't differ when obesity is defined by a different anthropometric measure. We're all we're seeing that there's increased types of health care use for all types of health care use and for all definitions of obesity. And then on this next slide, what I'm showing here is the same association comparing all measures of obesity to all types of health care use, but I've stratified now by age group. And what we can see is that there is attenuated relative risks in the oldest age group age 75 plus compared to the youngest age group age 45 to 54. And this holds true for every relationship except for in percent body fat defined obesity compared to contact with a GP. The red arrows I've placed on this slide show where the attenuations were significantly different in the oldest age group compared to the youngest age group based off of non overlapping confidence. Intervals. And what these attenuated relative risks or these lower relative risks in the smallest and the oldest age group compared to the youngest age group. What this means is that compared to individuals without obesity in the same age group. Individuals with obesity age 75 plus have a smaller but increased risk of health care use than those aged 50 45 to 54 with obesity. And some possible explanation for why we're seeing these lower relative risks in the older age group is that perhaps obesity is just not a strong predictor of health care use in these older adults. And this makes sense when I was looking at the proportions of health care use by age group where older adults in general just had higher proportions of health care use. So it just may be that obesity is just not a strong factor dictating whether older adults use health care services. And another explanation for why we may be seeing these lower relative risks in the oldest age group is that there may be a selection bias within the CLSA cohort. So it's been reported a few times that the CLSA is a predominantly healthy and educated cohort. And this can be held true specifically when looking at this older age group where these older adults are a lot healthier than the general older adult population in Canada. And in addition to these age stratified models I also computed a sex stratified model where I didn't see any differences in relative risks between males and females so I didn't show it in this presentation. So in this next slide here I additionally controlled for chronic conditions and self-rated general health in model two so I'm comparing my relative risk estimates in model one compared to model two. And what I conclude is that after controlling for chronic conditions and self-rated general health that they may be plausible mediators of the relationship between obesity and health care use. And this idea or notion comes from the fact that all of the relative risks were attenuated in model two compared to model one. This supports my hypothesis that they're mediators because a mediator is on the causal pathway between obesity and health care use. So someone with obesity might then lead them to have a chronic condition and that chronic condition might then lead them to use health care services. So if we control for a mediator like chronic condition we're essentially removing part of that association on the causal pathway. And that makes sense why we're seeing some attenuated relative risks after controlling for these two variables. So next step would be to obviously conduct a formal mediation analysis but since it isn't the primary objective of my paper and I just wanted to sort of clarify what previous researchers have been debating. I decided just to control some of them in an extra model. So next I'm presenting my risk differences for the association between the anthropometric measures and health care use. And what we can see is that adjusted for the variables in model one the risk differences and the 95% confidence intervals are all greater than zero. And this risk difference that I'm presenting here is a risk difference for 100 people or you can also interpret these values as a percentage. And what this data is telling me is that there is a significantly increased difference in the risk of health care use between individuals with any definition of obesity and individuals without obesity. We can also interpret this risk difference as there being a significant access risk that we can attribute to obesity. And what we can do for risk differences which is pretty unique and what we can't do for relative risks is that we can compare the risk difference across across each of the types of health care use to determine which type of health care use is most associated with each definition of obesity. And that's what I've done here as you can see in with red circles is I have circled the greatest risk difference for each definition of obesity. So for BMI Waste Reconference and Waste to Hip Ratio Defined Obesity, we see that there is the greatest risk difference for contact with a specialist. And then for percent body fat defined obesity, we see the greatest risk difference for having visited an emergency department. Although the risk difference for a contact with a specialist is in second place and the confidence intervals are still pretty similar and overlap showing that there's not a big difference between those two values. But I think once again it's important to interpret what this risk difference means. So let's look at that first circle that I have, I have with a BMI defined obesity and contact with a specialist with a risk difference of 4.6. The risk difference for having contact with a specialist in the previous 12 months is 4.6% greater in those with BMI defined obesity compared to those without obesity. And in order to determine what the value is of this risk difference or how important it is, is to look at the meaningfulness of the baseline risk. So on average about 48% of participants had reported having contact with a specialist. So does an increase in the risk of having contact with a specialist of 4.6% mean anything? Perhaps if we're looking at the risk differences for being a patient in a hospital overnight where the unadjusted baseline risk of health care use is lower, perhaps a value like 2.6% is meaningful. And once again with this regression model I stratified by sex and I didn't see any risk, any differences in the risk difference comparing between males and females. And I also stratified by age group and we saw similar findings to when we looked at the relative risks where the older age group had lower risk differences than the youngest age group. So in distance strengths and limitations of this work, so some strengths of these findings are that I was able to use the large CLSA data set which enabled me to stratify by age and by sex. It enabled me to apply sample waste to my data to make it more representative of the eligible Canadian population. And it enabled me to control for many different compounding variables. I was able to use percent body fat measured by dual energy X ray observed geometry, which is pretty rare to see in big population studies like this. And I was able to assess obesity, not only just using BMI like previous studies did, but I also used waste circumference, waste to hip ratio and percent body fat. Although there were many strengths to this study, there are some limitations to highlight, including the relatively short prospective follow up period of about 16 months. I used binary self-reported healthcare use instead of being able to link this data to administrative health claims. Those would be next steps yet it's not yet possible within the CLSA cohort, hopefully soon. And there was a moderately long recall period of 12 months for older adults, which may have affected their ability to accurately recall their healthcare use. In terms of public health and research implications, what these findings tell us is that obesity increases the burden on the healthcare system, but future studies are needed to understand if these healthcare settings provide the opportunity for obesity intervention and prevention. And these findings are also based off of commonly accepted cut points that are used in clinical and research settings, and they don't say anything about how valid of a measure they are of obesity or how accurate they are. So future research should aim to discern the best measure and the best cut points for assessing obesity related health risks. In conclusion, and as a summary of all of the work I've just presented, BMI Waste Reconference, Waste to Hip Ratio and Percent Body Fat may be measuring different aspects of obesity. Yet, regardless of these definitions, obesity defined by any anthropometric measure is associated with an increase in all types of healthcare use in the previous 12 months. And although we didn't see any differences in the association of obesity with healthcare use by sex, we did see that older adults with obesity experience a smaller increase in the relative and absolute risk of healthcare use compared to younger adults with obesity. And lastly, I just wanted to acknowledge everyone who made this work possible and my master thesis possible. So firstly, my supervisor, Dr. Laura Anderson, thank you so much to my committee members and my co-authors, Dr. Lauren Griffith and Dr. Emmanuel Green-Daw. I'd also like to thank once again the CLSA, all the staff, researchers and participants who made this possible, and also a big thank you to CIHR, the Canadian Institutes of Health Research for making this possible. And thank you to all of you for listening. You can ask me questions now, but you're also more than welcome to email me with any questions or comments that you may have. Thank you so much, Alessandra. Thanks for the excellent presentation. Like you said, we should now open it up to questions. Just a reminder, the muting will remain on for everyone, but you can enter your questions into the chat box in the bottom right corner of the WebEx menu. So I have to get mine now. I believe there's already questions, a few questions. Where was it? I don't know if you saw it. So, there it is. So first question from Andrew Patterson. Does the prevalence of obesity differ by age group? Yes. So I did look at that as well. And I'm just taking a look at the sheet that I have in front of me which shows that table. It does differ by age group, tends to increase with age group, but not anything really different. And we don't see consistent trends between all of the measures when looking at whether it increases or decreases as age increases. The next question from Natasha. Are there any measures of stigma that can be entered into the models or have you, I guess, considered of, considered doing that also moving forward? That would be interesting. So I didn't, I'm not sure if the CLSA reports any weight related stigma or any measures of weight related stigma. I don't think that they do. But that might be important for future research to look at whether stigma affects the relationship between obesity and health care use because I definitely know that there is stigma associated with both. That would be interesting to look at that effect. I don't think I'm trying to recall the CLSA questionnaires and I almost certain that we don't collect any information on that. So, yeah, I'm positive. Yeah. Yeah. Were you able to consider BMI changes during the 12 months? So we weren't, unfortunately. So the CLSA is administered. It was just that baseline where they had recorded body measures and then approximately 16 months later is when participants were contacted for that maintaining contact questionnaire. And I believe that was done over the phone and there weren't any updates of any of the body measures. But yeah, so there's no way to essentially look at any BMI changes, although when the next round of data is released, then that would make looking at any BMI changes possible. Right. So Cindy now, sorry if I missed it, which I think it's easy to always miss some pieces of information. So don't no need to apologize. Did the absolute number of health care visits differ between age groups to help further understand the attenuating, attenuating increase in health care use for older, for the older, older cohort. Did the absolute number of health care visits differ between age groups? So I'm not sure I'm understanding the question specifically because I'm getting caught up on the number of health care visits. So the outcome here was a binary outcome of any type of health care visit. So in the previous 12 months, have you had any contact with a general practitioner? Unfortunately, we don't have the data for the number of visits. And if you look at some previous literature, when asking people to self report their health care use, it's actually more valid to be using that binary indicator of health care. You said to ask them the number of visits. So I'm not sure if I answered that question that you were asking. But if we're talking about just binary health care use, it did differ by age group. So older adults tended to report higher proportions of any health care use. Okay, so maybe if that didn't answer it, then Cindy can submit a follow up question to that. But thank you for that. Adelene or Adeline asked for if you can give values according to inches. So I don't know if there was a specific part of your presentation that maybe the waste, or maybe if they're available, she might be able to contact you if she wants it in inches. Yeah, you can send me an email. So I didn't, I don't think I reported any means for waste circumference or any mean BMI measures and waste circumference is measured in centimeters. So if you wanted that, we would just have to do a conversion and switch it over to inches, but I don't, I don't have that available in front of me. Next question from Andrew. When you generated the scatter plots of the various measures, I noticed a lot of over plotting using density plots or hex and plots can give a better impression of where the densities are. So maybe the question is, did you consider using these other types of plots? And it is very dense because there's so many participants that we're looking at. So the plots are pretty heavy. I didn't consider using any of the other measures, but I guess the rationale for behind why I just use a scatter plot is I just wanted to see the general relationship, whether it was linear or not. And I actually didn't mention, but the relationship between BMI and present body fat is actually quadratic and that's been reported a few times in previous literature. But it was the notion behind the scatter plots was more of just getting a visual understanding of what the data might look like, although might be able to visualize it better with one of those plots that you mentioned. And actually thinking of the plots that you showed, one of the things that was striking to me was the differences in the sensitivity and specificity when you looked at males versus females and also the differences in correlations. Again, I apologize if I missed it, but was there any reasons for those differences? Like, I'm sure there, there's obviously differences between males and females, but when you looked at them, they're fairly significant between the sexes. And I think you're talking about for waste to hip ratio specifically is where we saw some big differences between males and females in terms of the prevalence of obesity and then in terms of the scatter plots, the Pearson correlation coefficients and also sensitivity and specificity. And my, my, I guess idea behind that is, I think there might be some issues with using waste to hip ratio as a measure of obesity at all. And also, there might be some issues with the cut points that are used for waste to hip ratio. So I didn't dive into that necessarily in this research that would I think be a totally separate question, but it really questions the validity of should we be using waste to hip ratio at all for looking at obesity. But definitely so something very interesting to highlight and to let people know about. Yeah, definitely. I think the questions are slowing down. So just reminder, if you do have any questions, please post them, but it looks like maybe this is the 1 of the last, the last 1 from Andrew. Are you planning similar analyses for self reported height and weight in the tracking cohort. So that would be our telephone. Yeah, that would be interesting. Not necessarily planning for it yet. This work was just finished, freshly finished of like a month ago when I defended my thesis. But it would be interesting to look, comparing self reported weights. So we know how those tend to change based off of measured weight and height, but also just looking at change in those measures after the 1st follow up so that the 1st follow up data will also be released soon. Yeah, that's a good point as we go on and we have more of the follow up data released something else to look at would be the longitudinal nature of the changes. So, Tasha now asks a question, would you recommend would you recommend using percent body fat to measure obesity? Good question. I keep asking myself that question because I don't know the answer and this work definitely doesn't give us an answer to that question either. I guess it depends on like what is what what does having that obesity by percent body fat mean. So what are the health risks associated with having a high percent body fat. And what we see in this cohort is that I think it was about 69 or 70% of individuals identified as having obesity by percent body fat cup points. So, do we think that truly 70% of the cohort is has obesity. I'm not sure so I don't think so, but it depends on how valid of a measure percent body fat is at associate looking at the association with obesity related health risks like cardiovascular disease or diabetes or even mortality. Good point. Okay, another question from Andrew in the type two diabetes field. There are different BMI thresholds used for screening based on ancestry. Is there any literature related to ancestry for obesity? There are different BMI thresholds used for screening based off of I'm not sure that I'm understanding what ancestry means. Are we talking about like hereditary obesity, ethnicity? Okay, I think that it was just follow up. Oh, yes. Okay, I know what you're talking about like cup points for obesity that are ethnic specific. I think that's what you mean. Yes. Okay. And there are I have seen like BMI specific cup points, especially for waste circumference and even there's some literature on using ethnic specific cup points for percent body fat. So the CLSA cohort, I think is predominantly white. So over about 90% self reported their ethnicity to be white, which is why I didn't look into any ethnic specific cup points. But there definitely is a whole abundance of research looking at whether we should be recommending ethnic specific cup points for different body measures. Good point. Sarah to your question. Yes, the presentation will be available on the CLSA website. So you can you'll be able to access that. Okay. I think that's pretty much it. As I go through my last few bits and bytes here. If anybody has any final questions we can, you can try to, we can stay on and address them at the end, or again, we can take note of them and you can email them backwards, backwards afterwards. So again, thank you again for such a great presentation. We really appreciate you doing this as we appreciate all of our presenters every month. I'd like to remind everyone that CLSA data access request applications are ongoing. The next deadline for applications is June 17 of next month. So market in your calendars. Please visit the CLSA website under data access to review available data, further information and details about the application process. I'd also like to remind everyone to complete their survey, which is located under the polling option. It may have also popped up for you. If you don't see it beside the chat button, please click the drop down there. We're finalizing the details for our June webinar, but please check our website next week for more information and to register. We will be providing an update and exploring opportunities for researchers to engage with the CLSA as a as the focus of next month's presentation. So if you are interested in engaging with the CLSA, whether it's as a trainee as a researcher. I think that next month's presentation will be sure to you'll be sure to find that interesting. Hopefully. And remember the CLSA promotes this webinar series using the hashtag CLSA webinar. We invite you to follow us on Twitter at at CLSA underscore. So, thank you again for today's presentation. And yeah, congratulations on your defense. Alessandra. And I see there was a question. Maybe you can stay on and address the last question by some on at the end, but for the most part, we'll consider the webinar complete. Perfect. And thank you all for listening for tuning in for asking questions. It was awesome. It was fun. Yeah, sorry. If you wanted to just address that last question that came in, if Simone is still on, if not, they can email you. Let's go ahead and take a look at that last question. That last question is asking. Do you think it would be possible to use that surface area on CT scans with regards to abdominal fat to also do this sort of population data? So that's a question. I'm not sure I know the answer to. I'm not sure of how valid a measure of fat surface area by CT scans is at affecting obesity. If it is a valid measure of obesity and the health risks associated with excess fat. So, you might be able to find some studies that might tell you a bit more about that surface area. I'm sorry. I don't, I'm not necessarily sure about that one. Thanks. Well, everybody, I hope you enjoy the rest of the week and we'll see you next month. Bye everyone.