 webinar is, again, population norms and prevalence of frailty about middle-aged and older Canadians. So finally, let me introduce our speakers today. Dr. Olga Hsu and Dr. Mario Ulici Perez-Zepeda. Dr. Hsu is an assistant professor of physiotherapy and geriatric medicine at Dell Housing University and holds the Canada Research Chair in Physical Activity, Mobility and Healthy Aging. She is also an affiliated scientist in geriatric medicine and with the Nova Scotia Health Authority and an adjunct senior lecturer of medicine with the University of Adelaide, Australia. Her research interests include aging, frailty, and physical activity. Dr. Mario Ulici Perez-Zepeda is a post-doctoral fellow in the Department of Medicine at Dell as well. His research is focused on aging, in particular the interplay between geriatric conditions and its determinant at the population level. He is specifically interested in frailty in the older adult and has focused in the past in Latin American older adult cohorts and more recently in data from CLSA as part of his postdoctoral work. So with that, I will take myself off the talking so if I need to deal with any more of my children and I will pass it on to our presenters. Thank you so much and thank you for inviting us to present today on our project using the CLSA data that was actually funded by Research Nova Scotia. I will start with first of all wishing everybody that is well and safe while watching us from their home hopefully. I will start with the introduction of our project then Ulicis will talk about the methodology and the results and then I will finish with some discussion around our research findings. We have no disclosures. We know that people age at very different weights and with different rates and some people are very healthy until they reach the oldest old, while some other people are very unhealthy even from middle age. The concept of frailty was introduced to capture this heterogeneity in aging. Frailty is a state caused by cumulative decline across multiple physiological systems and because of this decline people with higher levels of frailty are vulnerable to multiple stressors physiological, psychological and environmental factors. We can think of a heat wave or an infection like the one that's going around and we can even though we don't have evidence yet but we expect that people with higher levels of frailty they are more likely to experience complications from the infection and COVID disease and potentially are more likely to die after having the COVID 19 disease. We need to think of frailty as a continuum with extreme fitness on one side and extreme frailty on the other side. As we're getting older we're moving across this continuum and this has happened across life course but more important we need to stress that frailty is reversible. We know that even in people with higher levels of frailty it's very potential to reduce their level of frailty to a lower level. There are two main views of frailty. The first one is the syndromic approach that sees frailty as the frailty phenotype and another one is the deficit accumulation approach that we will talk today for our presentation. Here we have an example of someone around the age of 50 where they are moderate frail and how they look like 10 years later when they become severely frail. We can think of its circle represents our body system our body parts and how its circle represents a note a different component of our body which when it's healthy represented by a white circle whereas when it's damaged it is represented by a blue circle. We all accumulate some damage as we're getting older and if someone receives gets to the moderate frail state you can see they already have accumulated already some damage this is just an example but some this person might have have some abnormal sodium levels abnormal glucose levels and some impairment in mobility. Ten years later this person might have accumulated even more damage and for example they might have now experienced some cognitive impairments. To quantify how frail this person is we can look at how much damage they accumulated. The way to operationalize frailty based on the deficit accumulation approach is by creating a frailty index which is focusing more on the number of health problems people have rather than the nature of this has problems. The variables that are included in a frailty index are rarely pre-specified and there are some criteria on how we screen these variables however we know that at least 30 variables need to be typically included in a frailty index and to calculate the frailty index score of a person we simply need to divide the number of health problems this person has by the total number of measures. This shows two people person X and person Y have the same age around 65 but they have very different frailty level. One person close to 0.04 the other close to 0.14. The person X has a frailty index level similar to what on average people have at the age of 45 whereas person Y has a frailty index level similar to what on average people have at the age of 85 and because of this the frailty index has been suggested that could be also be used as a marker of biological aging. Research on frailty has increased a lot over the past years especially since the early 2000s and we see an exponential growth of research studies published on frailty from populations of studies to clinical studies and we don't seem to expect that this growth will stop in the near future. This is just a limited list of studies that frailty indexes have been constructed on their data from our group and many other groups worldwide and as I said again this is just a limited list there is a lot more studies that have been made including frailty indices. This just from Canadian data, US data, European data and other studies across the world. In a scoping review we did on frailty studies in acute care we found that the three main frailty tools that are included in the in this studies hospitalized acutely in older hours was the clinical frailty scale. The second that is missing from your slide here is actually supposed to say frailty index and the frailty phenotype and all these three tools were equally used in across the studies that we found. The association of frailty index with various outcomes has been examined widely and typically showing that the higher the frailty level of a person the more likely they are to experience adverse outcomes. The most commonly used outcome and examine outcome is mortality and this meta analysis that was published in 2018 showed that the pooled hazard ratio for the frailty index was 1.04 which we can say that for every 0.01 increase in the frailty index the mortality risk increase was 4% which might be seem slow small but the increment for the frailty index increases only 0.01 so we can expect that if someone increased the frailty level let's say from 0.1 to 0.2 then there would be a major change in the mortality risk. Also we have to remember that frailty is not only something that applies to older age and middle age also it could be useful for studies that examine younger people's health. The Kaplan-Mirkerfs here are showing the survivor probability for different levels of frailty. The lines for the older people are nicely separated from each other showing that people with higher levels of frailty especially above the cut point of 0.4 which is where typically severe frailty level starts having the lowest survival probability. Middle aged people also we can see a nice separation of the lines. For younger people between the ages of 20 to 40 we don't see as high levels of frailty as we see in the other age groups but still there is a significant amount of people who have a frailty index score above 0.2 and for them the survival probability is lower compared to the other frailty groups. Frailty indexes could also be used as outcome measures in the studies and for this study we did with cardiac rehabilitation data we looked at the frailty index scores of the people at admission to the program and also at discharge and we found that on average people who participate in the programs experienced a decline of on average was 0.07 which is considered a significant change in clinical meaningful. We recently published or recently submitted a paper that showed that the minimally important difference in the frailty index is 0.03 so we can see here that the reduction we saw in the cardiac rehabilitation program was above this minimum change. We expect in the frailty index to be meaningful. Frailty indexes are also could be useful for clinical care. Here you can see two people of different frailty levels and how the experience stressor for example a urinary tract infection or it could be other infections. A non frail person would experience a small decline in their function and then quickly recover back to their baseline state. A frail person has a much bigger reduction in their function and a much slower recovery and they may never even go back to their baseline state and frailty is challenging our current medical healthcare system. This map here shows how the divisions are spread across our hospital in Halifax and I'm sure this is similar to other hostels in Canada and worldwide. When our current healthcare system was designed in the 60s the median age of the population was around 25 and less than 10 percent of the population was 65 or over. At that time the typical patients had single problems and our system has worked greatly and very efficiently for treating these single problems. Currently if you have a heart attack the system will work extremely efficiently. There will be an ambulance arriving in your house in a very short time. You'll be going to the hospital very fast. You'll be in an operating room extremely fast and you will possibly have a great recovery and going home soon. However this is not what we currently see in our typical patients. Currently in our system the typical patients have met multiple medical and social interacting problems. So our system has changed a lot since the design of the beginning of the 60s. Now the median age of the population is 41 and almost 20 percent of the people are over the age of 65 and they typically have multiple problems. And we can complain that our patients do not match our current healthcare system. We need to change our health current healthcare system to fit our current patients. There are some advantages on using the frailty index over some other frailty tools. One main advantage is that it can be done by using data from almost any existing comprehensive health data set. It includes multiple domains and because of this it's considered a comprehensive assessment of health. It provides a continuous score from fitness to frailty. It doesn't just provide two levels frail and non-frail. However we can capture a lot of different levels of frailty like mild, moderate, severe and very severe frailty. It could be used as a predictor but also it could be used as an outcome measure. Because it's a sensitive measure it does not have a ceiling or floor effect. Rarely people have zero on the frailty index and it's extremely unlikely that someone has a frailty index score of one. Actually the maximum score we typically see is around 0.7. If we are interested to examine the association of frailty with comorbidity for example we can exclude comorbidities from the frailty index. So the frailty index could be modified based on our research questions. We also can, even the frailty index could be quite different across studies. It allows us to compare different populations in different studies and finally we don't have to depend as much on imputation methods for missing data because it allows us to work with this missing data. The main challenge is that at least 30 items need to be included and it takes some time to collect this data and also some recording has to happen for this data. Our first protocol that our group published in 2008 gave some general guidelines about how to construct a frailty index. We're currently working on an updated guide about creating a frailty index with existing health data based on the evidence that we have from all these years and also to some give some more specific instructions. The five main criteria when deciding which items should be included in the frailty index are that these items have to be health related. The prevalence of the deficit should increase with age. The deficit should never be too rare or too common. We need at least 30 variables and they should cover multiple several organ systems. So we should not create a frailty index including only for example cardiovascular measures then it becomes a cardiovascular index and not the frailty index. So I'm just going to briefly mention the 10 steps that we used to create the frailty index and at least later we'll show how we did that in the CLSA database. First we select all health related variables. For example we exclude demographic and social variables from the frailty index. Then we exclude variables who would have more than five percent missing data and we decode all variables to zero one. For example zero represents absence of a deficit and one the presence of a deficit. We exclude variables that have a proportion of prevalence of deficit that is less than one percent or over 80 percent. We looked only for variables that have a positive association with age. We exclude variables that are extremely highly correlated with each other showing that maybe they will almost measuring the same thing. And then we count how many deficits remain to make sure that we have at least 30 items. In the final steps we calculate the frailty index scores for the participants and we test the characteristics of this index to see if they behave the way we expect. And finally we use the frailty index for our research questions and our analysis. And with this I would like to introduce the three objectives of our research project using the CLC data. Our first objective was to provide a standardized frailty index using the CLC data. We then wanted to describe the frailty levels of Canadians between the ages of 45 and 85 and provide some normative data for frailty for Canadian people. And I will pass these two lists to talk about our methodology and results. Thank you very much. Thank you for the opportunity to share our results. So as you already imagine we used the CLC data of course and this is a cross-sectional analysis from the baseline assessment. We used both cohorts. We tried to have a frailty index that was useful both for the tracking and the comprehensive cohorts that would give us a common frailty index to be used whenever the pool data was needed. So that's why we had this proposal approved a long time ago and these are results of this. I will walk you through as Dr. Thel mentioned through the steps that we followed in order to have our frailty index for the CLSA. So of course selecting the health-related variables, the coding, screening of those coded already coded items, frailty index calculation and some other steps we will walk through through this presentation and methods results. So first of all we needed to went through all the health-related variables that were available in the pool data set. So that's a really big data set. As you can see in this slide you will find that both cohorts have different variables. More rich of course more comprehensive number of variables health-related are available in the comprehensive cohort. But also the tracking cohorts have some other health-related variables that could be used. But we wanted to use those that were shared by both cohorts. So that was our first step going through each and every one of those health-related variables in both cohorts and then choose only those who were common for both cohorts. So this is how our screening went. Since we have 75 items that were common for the tracking and the comprehensive cohort, we started from there. So we included 52 items. We ended on that, that will be more specific on that. And 23 items were excluded basically because of no correlation with age. I will show you some graphs on that. We ended with self-rated health vision and hearing, some a list of chronic conditions, activities of the LED, living instrumental activities, cognitive function measured by test, and mental health items that mainly were from the depression data that we have available in CLSA. So I want to show how this works, how the steps work. So we can go step one. We already did it health-related variables were chosen. Then we have the steps two through four that Dr. Theo talked about. So the first one is about the missing values. So how many of this question in particular was missing for the whole sample? So in this case, we have really, really low number percentage of missing values. So it's included. Then we codify. This is an easy one, since it goes into one and zero binary. It's quite easy. Then we check the prevalence. In this case, it's not too common. It's not too rare. It's 20.8%. So then we go into checking our correlation with age. As you can see, cataract is one of the best correlated variables with age. And we decided this is how we decide into including one variable or not. As you can see here, we have self-rated health. This is a different example for the coding. We all know, or if you have worked with the failed index previously, that you can do these sort of points or these different scoring. According to already established points, like this liquor scale, you can go from zero to one, having some middle points for this one. Again, we check the missing. We check the prevalence. And we have here a slightly not that flashy correlation as the other one, but it's quite still correlated with age. So this one goes in. And this is an iterative process, as I showed in the previous slide. So we went through these 75 slides that were common for both cohorts. But of course, there were variables that didn't make it. So as we can see in this slide, the variable, this is from the data set of the post-traumatic stress symptoms. From this data set, we show the nightmares associated with a traumatic event. As you can see, it is not related. It is negatively related with age. And even that has a nice prevalence, not too common, not too rare, and missing value that is almost good. It's good, too. Correlation with age precluded us from including it into our failed index. So that's how we went and finished with a 52 items for the index. However, we had to do some concessions to our frail index. We have very low prevalence of some items such as the Alzheimer's disease. This was, of course, expected to be the case in this first baseline assessment. Because, as an example here, we have a prevalence of 0.1%. So it's apparently rare. But we expect that this prevalence will grow in the follow-ups. In addition, we also have to make this difficult decision on cognitive tests, since they have more values missing than commonly. So we had to went up to 12% be a little bit permissive in order to fulfill that criteria of having a multi-dimensional frail index. That's why we decided on having also be a little bit loose with this criteria. Of course, we went to this step of calculating the frail index. Since we ended with 52 items in our frail index, we expect that a person, for example, with 15 problems divided by 52 would have a score of 29, as an example depicted here. Keep in mind that we respected this rule. Participants who had the 20% or more missing items or that individual in particular would have to be not included. In our case, we just had only 87 individuals that had more than 20% of their variables in their frail index not available. That is less than 0.1%. So continuing with the steps, we needed to make some tests on this frail index. The descriptive characteristics are quite useful in order to know if our constructed frail index has the properties that usually has. I will show you in a later graph what this means. Of course, assessing the frail index correlation with H, the full frail index with H. We did norms deriving a quantile regression for percentiles and using these fitted values for H and 6. We could locate the score on a particular fitted percentile. I will show you our graphs. We used a strategy with colors in order to pick those in green that could be in better shape and those in red that could be in or are in highest score of frail index. Therefore, having a higher burden of frail over all worst health steps, we will proceed this in a few more slides. Of course, we use the weighted analysis every time. Since I said we didn't follow all the rules for this F5 with 52 items, we had to do a sensitivity analysis in order to assess the agreement between both tools. We needed to know how they correlated a raw correlation. We needed to know also if the groups of intervals changed between the two FIs. Also, if these groups, as I told you with the norms, these fitted percentiles also changed. We did a weighted kappa for that. We did also limits of agreement of the whole score with a deming regression that accounts for the error between the two measurements. Finally, we go into our results. As you can see, of course, as expected, the majority is female. We have a distribution of age that is more percentage into the 50 to 54 group. I would take your attention on the frailty index that we have here. The mean overall frailty index is 0.08 for all the population. As expected, it's a little bit higher for female. Also, as expected, as the age is higher, the frailty index goes higher. Going from a 0.05 mean frailty index to a 0.13 frailty index in those 80 to 85 years. This also can be seen in this, sorry, no, that would be in the next slide. This other slide shows the property of the distribution of the frailty index. This is a typical distribution of the frailty index. In the second column, you can see correlation with age. Of course, as the frailty index increases or as age increases, the frailty index increases too. This was talking about before. You can see here this distribution, how it modifies at the population depicted is older. It tends to normalize somehow as you are older. You can see also in the second column, the graph from the second on the right, how women has this slightly higher frailty index too. All these first analyses are like the proof of how the frailty index behaves as we expect to behave. I mean, higher in the older ones, a little bit higher also in females, and with a clear correlation with age. Then we go into the intervals. As you can see here, we have intervals of categories of frailty index, less than 0.1, 0.11 to 0.2, 0.21 to 0.3, and 0.31 or above. Of course, what we do expect is that as we are older, the fractions of those people, the percentage of people in the worst status in the highest frailty index groups will grow. Of course, as you grow older, those in the best or the fittest group, the green one, it starts to decrease. Finally, I am showing you the graph that is for the norms. This is the graph for the norms. What you are seeing here are a lot of fittest percentiles for age and sex. This graph in particular is for female, and you can just quickly take a look. If you are a 74-year-old and have a point to frailty index, how are you doing? Just having a quick look into this graph, a similar thing that is done like in the settlement. We try to make visual this increase risk. Also, you have the wrong numbers. If you want to go and see exactly what that number means, we have it here. If you are a 50-year-old woman with a 0.03 score on the frailty index, you will be in the 25th percentile for your age group and your sex. That means you are into the best categories to those who are fittest for these groups, CLA, for Canada. Finally, we did the sensitivity analysis, in which you can see we used a 40-item FI. 12 items that were included in the other FI that I showed you were not used because of missing or prevalence. The raw correlation was 0.99. This is really quite a category. The FI intervals were 98 percent agreement with a Kappa of 0.77. When comparing the norms, we also have a high agreement with a 97.9 percent of agreement and a Kappa of 0.93. This is quite good. When you can see this little one with the hand, that means that if the person changed of group, how big was that change? The mean of the change by group was only 0.04. It's not that bad. And the regression shows also a high coefficient between the errors of both. We also can see the plant-almond plot to your left that shows these are really, really small limits of agreements for our two credit index, the one of 52 and 4. Finally, I can say from what I showed you, limitations. Well, of course, these are values, normative values for Canadian people, individuals. And the limitations particular to this dataset is that people from living in home long-term care facilities or with cognitive impairment were not included into our study because that's the way CLSA is. And that will give us maybe a higher FI for our, if we would have included them. And, of course, prevalence would have been also altered by that. So now I pass the mic to Dr. Theo. Thank you, Alyssa. So we try to think about these results. And who else could this be useful? This new standardized frailty index that we created for the CLSA data and this normative data. We had, we believe that this could be quite useful for CLSA researchers, people who already have access to the CLSA data. They could use this standardized frailty index as an independent variable in regression models on other predictive models, also as an outcome measure. But also if their study is not focusing on frailty, a frailty index, the frailty index could also be used as a covariate for their analysis. With permission with CLSA, our paper is actually under review right now for this project. But with permission with CLSA, we will also share our syntax and code for this FI so other researchers can use it. Also, other researchers could be interested to use this frailty index for other studies. New primary data collection studies, for example, someone might be interested to look at the association of frailty with another outcome in different communities. And they might like to test this frailty index in their sample to find how the participants compared to a representative sample of Canadians that was included in CLSA would be limited on the normative data that we have that only between the ages of 45 and 85. But we understand that this is a big part of the studies that being currently done in Canada. Also, the frailty index of the normative data could be useful for clinicians. We know that many clinicians feel that frailty assessment could be useful for clinical practice to inform decision making. And these clinicians could ask the participants to use a questionnaire including this frailty index items that it's approximately three pages long. And after the self-assess themselves, the clinician would know what's the frailty level of their patients. Policy makers could also use this data to understand the frailty levels that we see in different communities in Canada and in the Canadian population and make some projections about the aging of Canadians. Finally, general public, might be interesting to self-assess the frailty levels in order to compare their frailty levels with other Canadians of the same age and sex. Here, for example, this is similar to the figure that Ulysses showed earlier with the normative data and the percentiles. For example, we have four people, two people of the same age, 50-year-olds and two people of 80-year-olds. So the two 50-year-olds, one has a frailty index score of 0.15 and the other has a frailty index score of 0.05. This is just for males. The man who has a frailty score of 0.15 is frailer than 94% of the Canadian males of the age of 50, whereas the 50-year-old with a frailty index score of 0.05 is frailer than 50% of Canadians of the same age. In the two examples of two 80-year-olds, the person with a frailty index of 0.15 is frailer than 75% of Canadians of the same age, whereas the person with a frailty index score of 0.05 is frailer than 90% of Canadians of the same age. So you can see here, similar to how, as Ulysses mentioned, how osteoporosis screening is done, someone could use the three-page frailty index questionnaire that we have and identify what's the frailty level and how it's compared to other Canadians. So we believe that this new standardized frailty tool and the accompanied normative data could be useful for researchers, clinicians, policy makers, but also for the general public. And to summarize some of the data that Ulysses showed, you can see here that among the people of 45 and 85 that were included in CLC, most of them belong to the non-frail category, has a frailty index score less than 0.1, which is quite good. It shows that most of the middle-aged and older Canadians are in a good health state. However, we had a significant amount of people into the very mild state and some people in the mild and moderate, severe categories. And as we mentioned, this is a pyramid that shows how the Canadian population between the age of 45 and 85 would look like if there were only 100 people. And the proportion of the percentages that you see on your left and your right shows the proportion of people with at least a very mild level of frailty. For middle-aged women it's 34% and for older women it's 58%. For males, the numbers are a little lower but still quite high. As you can see for middle-aged men, more than a quarter of the population has at least a very mild level of frailty. Whereas for older males, almost one in two older men have at least a very mild level of frailty. This shows how frailtics could be useful to capture these very early stages of frailty in order to prevent further progression to higher levels of frailty. So before we finish, we would like to acknowledge our trainees and collaborators and our funders, and especially Research Nova Scotia, which funded this project. And we'd be happy to answer any questions here at the webinar or through our emails. Thank you so much. Great. Well, thank you so much for that very interesting presentation. I couldn't help while you were going, especially through your last few slides, thinking about how we have so many different indexes that we use in a clinical setting on a daily basis, such as BMI. And I can imagine that something like this and frailty indexes at some point be used in a common clinical practice. Even though there are some controversies surrounding them, I think there's lots of utility there. So I just wanted to thank you for your excellent presentation. I'd now like to open it up to questions. I don't think anybody has submitted any questions in the chat box yet, so please feel free to do that. Just a reminder that muting will remain on, but you can enter your questions into that chat box in the bottom right corner of the WebEx window. I also just wanted to mention that we'll be putting the pull up about the webinar. It's up now, actually, so if you do need to leave a few minutes before any additional questions get answered, please take a moment to do that now as well. So let's just go back to the chat. I don't think there's still any questions. So I thought maybe while we're in case people are thinking about questions now, one question that I always tend to usually ask in these webinars at the end is if you've already started to engage with any knowledge users, so in this case it could be clinicians or policymakers to actually move the index that you've created into either a clinical tool or its use in policy decision making. I'm just wondering if any of that work has started yet. I can go first. So because this paper is just submitted for publication, hopefully to be published soon, we haven't taken the next steps, but we definitely have plans potentially putting the question as with permission from CLSA, potentially putting the questions in a web app or some electronic format, so it makes it easier for people to implement the assessment in clinical practice. We know it can be a little, this might be answering a three-page question in a doctor's office and then calculate a score, so we're hoping to make an online version for this, and we're also doing a little bit of work around the differences of frailty levels among provinces and making some specific reports on Nova Scotia to understand a little bit better the levels of frailty in Nova Scotia and what it means in order to pass some information more to policymakers about this. We haven't done this yet. We're still in progress, considering we just finished our analysis, but it's in within our next plans. I think there's a few questions now, so maybe we'll move on to those and if you can either just, which one of you will answer them, you can decide amongst yourselves. Sorry, yeah, I have read it, so yeah, I can answer the three questions. Go ahead. So yeah, regarding the question about multi-dimensional approach, the Tilburg and Grenningham frailty indicator from Daniela Anker, so the frailty index is in nature multi-dimensional, so I think we picked a very well multi-dimensional index. The Tilburg, as far as I remember, includes also some social variables. We believe that those should be separate from the frailty index, and there's a social vulnerability index for that that we also are working on that. And that's, I would answer with that. The question from Cheryl Sadoski, can you comment on polypharmacy and medication burden and how that is integrated in the FI? We can't because there is not available data on medication currently from CLSA, but as a clinician my thought is that this is depicted quite well with using the chronic diseases. From Carmen Garcia-Pena, thank you very much. How do I think these normative values will change in countries like Mexico or Latin America? Of course this will be, I mean this needs to be checked in other data sets, and particularly in countries, low and middle income countries, because my hunch is that this will change a lot. That's for now, so I don't know if Olga wants to comment. Yes, so the question about the 80% criteria that we used at the end, calculating the frailty index scores, we have to have a cut point, and I know sometimes we're a little more loose with this, especially for clinical data sets, because it might not be as reliable if only half of the data is collected for some participants. We might say that we don't have enough information for them. It is typical that these are the frailest people, because missing data is an indication usually of higher levels of frailty. So in populations of data sets, this is never rarely a problem. As Ulysses said earlier, it's less than 0.1% of the CLSA participants that we couldn't calculate the frailty index score, so it was, if I remember correctly, a little more than 50 people, so it wasn't a big issue for the CLSA. It might be bigger for other data sets, and some might consider pushing maybe the 80% cut point to a little lower score, 70%, but they still need a cut point that needs to be picked. And that's when people might start considering whether other methods like imputations or other things, if there's a very high number of missing data, but typically we use the 80% cut point for now. And the one thing about the Groningen versus Tilburg from work that we did before, except the social variables, pretty much everything else that would be included in Tilburg or the Groningen would also be included in the frailty index. So definitely the frailty index that we created is very multidimensional and would include more measures than any other frailty multidimensional tool that currently exists, with the exception of a set for the social variables that we include them in a social vulnerability index. Ulysses, do you want to answer about the last question? Yeah, for Monica, Kelly, if someone doesn't identify as male or female, how do you call perhaps not an issue with our older populations? Well, CLFA includes a really nice set on gender and we are looking into it and we'll certainly make changes in the future with that because they do ask a nice set of questions regarding this. So yeah. And to add to that is that's something we're extremely interested because there's an interesting paradox happens with frailty similar to other health variables where even though males are higher mortality risks, females actually having higher frailty index levels. So we need to understand this better what's causing this paradox and whether it's driven by gender-related factors, how this relates with other biological factors. So we're interesting to see how gender might help us understand these relationships better and this is actually interesting in the current social media, the discussion that's happening around the COVID-19 too, showing that males are more likely to die from COVID-19 and some recommending that maybe this is similar to what we see with frailty. I don't know if there's any more questions. I think you've done a really good job at answering all of them. I've just been following along. If there's any last questions, please anyone feel free to post them. Oh, actually we do have one from Daniel Anchor. I have one issue with the frailty index in my understanding. It does not differentiate frail from multi-morbid and disabled individuals. Can you share your thoughts about this? So maybe after this one, if we don't have any more, we'll fully start bringing it to an end. I can start with that. It's as I said, it's a very different view of frailty compared to the frailty phenotype definition. Multimorbidity and disability items are typically included in the frailty index. Similarly, we see with other AIDS-related syndromes and geriatric syndromes, we see that, as someone has said, the problems of old AIDS come as a package. It's really hard to take one domain out from the other. Even so, we have done work before taking out multi-morbidity and disability from the frailty index and showing that if you have enough items, at least 30 items, the levels of frailty index would be similar with or without comorbidity and disability. However, we do understand that multi-morbidity and disability are great markers of frailty and typically are informing us about the frailty level of individuals and that's why we typically include them. However, as I said, if for the purpose of research or clinical practice, we want to keep them separate, you don't have to include them. They can be constructing a frailty index without including them. So, I really understand, Daniela, with these questions, because as a clinician, I get really anxious when not having a clearness and this reductionist vision that we clinicians have. So, yeah, just adding to what Olga said, these are difficult to separate concepts and the older the individual is, the more difficult to differentiate and make differences between categories. So, frailty makes sense because of that. And to understand the overall health state of someone, it might not be the best to ignore their level of chronic conditions and their level of function. It's quite informative about the overall health state. And that's the goal of the frailty index, is to be able to identify the overall health state of the individual in order to use it for clinical practice or for research. So, I encourage if anybody else has any additional questions, you can still submit them and we can ask for our presenters to follow up with you after the webinar. At this time, I'd like to formally start closing the webinar and thank you, all of you, for attending first of all today and to our presenters for such a great participation. We really appreciate your efforts today and for everyone to be here in the CLSA webinar series. I'd like to remind everyone that CLSA data access request applications are ongoing. The next deadline for applications will be June 17 of this year. Please visit the CLSA website under data access to review the available data, further information and details about the application process. I'd also like to remind everyone to complete the survey that's located under polling. If you don't see it beside the chat button, please click the drop down arrow and you should be able to see it. So, our next webinar will take place on Tuesday, April 28th at noon Eastern time. Dr. Emily Rudder, PhD candidate in the School of Public Health and Health System at the University of Waterloo will present and her presentation is entitled social support availability and executive function in the baseline cohort of the Canadian longitudinal study and we'll send the registration will open next week for that. And remember the CLSA promotes this webinar series using the hashtag CLSA webinar. We invite you to follow us on Twitter at CLSA underscore ELCV. And finally, thank you again for attending today's presentation and I hope everybody stays safe and healthy.