 Hello everyone. My name is Jennifer Boyko. I'm the Scientific Operations Manager of the Canadian Longitudinal Study or CLSA for short. Thank you all for attending our February webinar today entitled, Sarcopenia in the CLSA. The impact of diagnostic criteria on the agreement between definitions and the association of sarcopenia with falls. First I want to acknowledge that the CLSA national coordinating center in Hamilton is located on traditional territories of the Mississauga and Haudenosaunee nations and within the lands protected by the dish with one spoon wampoon agreement. And as attendees of this webinar I encourage you on behalf of the CLSA to learn more and to acknowledge the original habitants of the land where we currently do have the privilege to do our research and live and work wherever that may be. So now on to today's webinar again entitled, Sarcopenia in the CLSA. The impact of diagnostic criteria on the agreement between definitions and the association of sarcopenia with falls. I'd like to introduce our speaker to you, Dr. Alexandra Mayhew. Dr. Mayhew is a scientist working with the CLSA in the Department of Health Research Methods Evidence and Impact here at McMaster. Her research interests include physical function, disability, and body composition in aging adults with a particular focus on issues surrounding how muscle wasting or sarcopenia is measured. And so with no further ado, I will pass it on to Alex. Thank you very much for that introduction, Jennifer. So as she mentioned, my presentation today is about sarcopenia in the Canadian Longitudinal Study on Aging. Focusing on evaluating the agreement between definitions and the association of sarcopenia with falls when different diagnostic criteria are used. In my presentation today I'll go through a quick background about what sarcopenia is and what some of the issues with the definitions are, the objectives of my work, the methods and key findings related to those objectives, and some overarching conclusions. So by way of a brief background, sarcopenia refers to loss of muscle mass, strength, and function that occurs with aging. For hundreds of years it's been noted that from infancy to about 30 years of age, muscle mass increases. In the 30s, there begins a slow decline of muscle mass, and then somewhere around the age of 60 to 70 years, the decline becomes more pronounced. And at some point in time when the lack of muscle mass is considered to be impairing the individual, this is when sarcopenia becomes relevant. Even though the phenomena of sarcopenia has been noted for centuries, it wasn't until 1988 that there is actually the term sarcopenia coming into use. A researcher by the name of Rosenberg saw that there is a lot of literature on the topic of muscle mass, but it was very difficult to unite it because people were using all sorts of different terms to describe what they were doing. So he proposed sarcopenia, which literally translates from Greek as loss of flesh. Ten years later, the first definition of sarcopenia was published. This definition used apendiccilarline mass, which is an estimate of muscle mass divided by height squared. Shortly after that, there were two more definitions involving just muscle mass that came out. One of them adjusting muscle mass for weight, and one of them adjusting for both height and fat mass by using regression models. At this point in time, the sarcopenia research community was finding that they weren't getting the associations of sarcopenia with health that they expected. So as it turned out, muscle mass alone wasn't doing a very good job of discriminating people who were at risk for various outcomes. Based on that, when the next definition came out, which was the beginning of these so-called expert group consensus definitions for sarcopenia, low muscle strength was included, and subsequent definitions also used low muscle function. And there have been four different consensus group definitions that have been released since 2010. To summarize what all these different definitions include, essentially sarcopenia can be operationalized as muscle mass alone adjusted either for height squared, weight, body mass index, or height and fat mass using regression. So you can look at muscle mass alone or in combination with muscle strength or muscle function. And generally, muscle strength is measured by group strength. This is what's recommended by most of the expert group definitions and used by the majority of the literature. And similarly, gait speed is the most common way of operationalizing muscle function. Depending on which of these definitions that you use, sarcopenia is estimated to impact somewhere between one and 29% of community-developing older adults. So some studies have found estimates as high as 70%. So this creates a very interesting problem where are you really identifying, are these definitions functioning the same way when you're identifying such a different prevalence of the condition? We wanted to dig into this a little bit further to better understand what was happening with the sarcopenia literature. So several years ago, we conducted a systematic review and meta-analyses of the prevalence of sarcopenia in community dwelling older adults. So we went through and we searched almost 14,000 titles and abstracts and ended up with 109 eligible studies to be included in our analysis. For each of these studies, we extracted which definition they used, which is listed down the left-hand side of the table and the prevalence estimates from that study. As you can see, we had quite a range of different estimates. And the big take-home message for us was that depending on which definition was used, between 9.9 and 40.4% of community-developing older adults were considered sarcopenic. It was also interesting to note that there is a very high between-study heterogeneity within each of these definitions, which is indicating that even though in theory they're supposed to be defining sarcopenia using the same technique, that might not be happening in practice because of those differences in prevalence estimates. So why does this matter? The big change that has happened in the last couple of years is that sarcopenia is now included in the international classification of disease starting in 2016. So clinicians are now able to bill for diagnosing and treating sarcopenia. However, there's no guidance given to clinicians on how to do this. And if all of these different definitions are functioning differently, this is really an essential piece of the puzzle to answer before it gets integrated into clinical practice. So the objectives of our work were to look at how different combinations of muscle mass, muscle strength, and muscle function, as well as different muscle mass adjustment techniques and how they impact the prevalence of sarcopenia, the agreement between sarcopenia definitions, and the strength of the association of sarcopenia with falls. And combined, these three objectives allowed us to evaluate if these definitions were functioning similarly or differently from one another. In order to do this, we use data from the Canadian Longitudinal Study on Aging. I suspect that most of our listeners today are quite familiar with the study, which I will refer to as the CLSA from here on in. But for anyone who's not, it's a national longitudinal research platform. And at baseline, there were around 51,300 participants aged 45 to 85 years recruited into the study. For the purposes of our analyses, we focused on the 30,000 participants that are included in the comprehensive cohort. These people not only answered questionnaires about their health and other variables, but they also came into our data collection sites and had in-home interviews where we're able to assess different physical measures. And this included a measure of dual energy x-ray absorptimetry, which is how you derive the estimate of appendicular lean mass, as well as grip strength and gate speed, which are essential for diagnostic sarcopenia. So we included those participants, but we did further restrict ourselves to people who identified as European. We know from the literature that muscle mass and muscle strength values, what's considered normal in a population, vary quite dramatically by ethnicity. And unfortunately, within the CLSA, we simply didn't have the sample size for each different ethnic group in order to do subgroup analyses. We did also exclude participants that had had their visit, but for whatever reason didn't complete the dual energy x-ray absorptimetry measure, grip strength, gate speed, or height and weight measurements. So we knew what sample we were interested in using for our analyses, and then we had to figure out how we were actually going to define sarcopenia. And at this point in time, I'd like to point out that we are not the first people who have tackled this problem of looking at how these definitions are functioning similarly or differently. However, the previous literature to date has focused on the expert group definitions as written, which introduces a couple of different problems. So I've pulled this a couple of numbers from this paper by Dan and his co-authors published in 2014. And though it's an excellent paper and was a huge contribution to the literature, it very perfectly illustrates why it's so hard to compare between the expert group definition. So they were comparing between the European Working Group on Sarcopenia or EWG SOP definition and the Foundation for the National Institutes of Health definition, or FNIH. So these definitions are operationalized quite differently. For the European definition, appendicular lean mass is adjusted for height squared, and they look at low muscle mass in combination with low grip strength or low gate speed. And for grip strength, they use the cut-off of less than 30 kilograms for males and less than 20 kilograms for females. And for gate speed, the cut-off of less than 0.8 meters per second. For the FNIH definition, they adjust muscle mass by body mass index, and they look at grip strength cut-offs of less than 26 kilograms for males and less than 16 kilograms for females. Based on these definitions, the European Working Group definition estimated that males, the prevalence of sarcopenia was 5.3 percent, and in females, it was 13.3. For the FNIH definition, the prevalence was 1.3 percent for males and 2.3 percent for females. So we can see that there's quite a marked difference in the prevalence estimates, both between males and females, as well as between the definitions within each gender. I've also included the agreement between these definitions at the bottom, measured by Cohen's CAPA. Cohen's CAPA was around 0.53 for males and 0.14 for females. And these are generally considered, certainly for females, very poor agreement, and for males still quite limited. However, we don't know why there are differences in the prevalence estimates or why the agreement is so low. And what I mean by that is that these definitions differ not only by which variables are included, one includes gate speed, the other does not, but they also differ based on which technique was used to adjust muscle mass, height versus BMI, and there's different cutoffs for the variables. So for example, the grip strength cutoffs are different in both males and females. So even though we know that agreement isn't great, we have no idea why this is. So based on this, we wanted to do something different. There's about a dozen or so studies that have been compared between these expert group definitions. And we didn't want to just replicate what's already been done. We wanted to add something more to the literature. So to do this, we came up with a new technique of operationalizing sarcopenia, which looks at all the different combinations of variables and muscle mass adjustment techniques recommended by the expert group definition. So we operationalized sarcopenia as just low muscle mass using each of the four adjustment techniques, as well as in combination with muscle strength, operationalized his grip strength and muscle function, operationalized his gate speed. The next step for us is figuring out which cutoffs to use for these variables. For grip strength and gate speed, it was actually fairly simple. There's good agreement in the literature based on the expert group definitions about what the relevant cutoffs are. So for grip strength, we use the cutoffs of 30 kilograms and 26 kilograms for males, and 20 kilograms and 16 kilograms for females. And for gate speed, we use the cutoff of less than 0.8 meters per second. As any of you work in the physical function literature, you'll know that there's also a gate speed cutoff of 1.0 meters per second, which is commonly used. We did include this in a sensitivity analysis. However, it identifies around 70 percent of our participants as having low gate speed. Therefore, it's not the most practical cutoff to be using. It was far more complicated to figure out how to handle the muscle mass cutoff. There are dozens upon dozens of different cutoffs recommended in the literature. Each of the expert group definitions tends to cite different studies that have come up with cutoffs, but they also recommend that people can choose their own. And this is a reflection of the fact that generally muscle mass cutoffs developed in one study aren't all that applicable to muscle mass in another study. And there's a couple of different measurement issues that causes that. So the expert group definitions say you can use the lowest quintile approach. So that's what we did. But we're specifically interested in determining the lowest sex specific quintile participants age 65 years and older. And the reason for that is with such a broad age range of participants at the CLSA from 45 to 85 years, had we included the 45 to 64 year olds in determining that lowest quintile, we would have found that we had very, very high values for the cutoffs, which would have identified almost all of our older adults as sarcopenics since muscle mass invariably does decrease with age. So this is a pragmatic way of kind of choosing a sample that was reflective of what other people are looking at. It's also worth noting that we included the 10th and 40th percentile values in a sensitivity analysis. And with these values, we captured the entire spectrum of what has been used for low muscle mass in the literature. For the first component of our analysis, we're looking at the proportion of sarcopenic participants. So after we had taken the comprehensive cohort participants and excluded anybody who is missing the data we required, we had around 25,400 participants left in our analyses. We stratified all of our analyses by age and sex using 10 year age bands starting with 45 to 54 years. And this is very important to do because as I just mentioned, the CLSA has a wide age range of participants. And had we pooled everybody together, we likely would have been missing the nuances within the data. And for each of the definitions, we looked at the proportion of people that were considered sarcopenic. So what we found, and these results specifically refer to males, was that it depended. Sometimes the proportion was similar, sometimes it was different. So orient you to this graph quickly. Across the bottom, we have the three different kind of categories of sarcopenia definitions. On the left hand side are the muscle mass only definitions. In the middle are the muscle mass and grip strain definitions. And on the right hand side, the muscle mass and gait speed definitions. Within each of those categories, you can see the 10 year age bands. And within the 10 year age band, you see the different methods of adjusting muscle mass represented by different colored bars. So in dark green, we have appendicular lean mass adjusted for height. In light green, we have appendicular lean mass adjusted for weight. In light blue, appendicular lean mass adjusted for BMI. And the dark blue is appendicular lean mass adjusted for both height and fat mass using the regression technique. So we mostly our results were actually quite expected. We found that sarcopenia prevalence increased with age, regardless of which definition was used, which is exactly what should be happening for an age related condition. We also found that when we use the muscle mass only definitions, the proportion of people that were identified as sarcopenic was much higher than when it was combined with grip strength or gait speed, which again is quite intuitive because you now have to meet two criteria instead of just one. However, what was interesting for us to note was that regardless of which muscle mass adjustment technique was used, within each age band, the prevalence of sarcopenia tended to be quite similar. And it was also quite similar for muscle mass and grip strength versus muscle mass and gait speed. So that was in males and in females we found that the story was quite similar. So just because we knowing how many people are being identified as sarcopenic doesn't necessarily tell us anything about if they're the same individuals being identified as sarcopenic. So to assess this, we looked at the agreement between definitions. So we use the same 25,400 participants as we did for the prior analyses, again stratifying by age and sex. And this time we assessed agreement between all of the definitions using Cohen's Kappa. Cohen's Kappa provides you a chance adjusted agreement. For many of these definitions, we knew that only 5% or so of our participants were being identified as sarcopenic by any definition. So in that context, there's a very high chance of agreement that people will not have sarcopenia based on either definition. So Cohen's Kappa takes that into consideration. When we looked at the agreement between the different muscle mass adjustment techniques, we found that generally speaking, the Cohen's Kappa values did not exceed 0.55. So on this graph here, which is again referring to male, the dark blue bars are showing the muscle mass only definitions and the light blue bars are showing muscle mass and grip strength based definition. I took out the muscle mass and gait speed definitions. They had pretty much identical values for muscle mass and grip strength. So based on this, you can also see that the muscle mass only definitions had much poorer agreement compared to when they were combined with grip strength. And there's no one standardized way of interpreting Cohen's Kappa values. But generally speaking, we would consider all of this insufficient agreement to really say that they're identifying the same individuals. There were two exceptions for combinations of muscle mass adjustment techniques that we looked at, where when muscle mass is combined with grip strength, the Cohen's Kappa values were up to 0.75. So for these two specific comparisons, perhaps there is starting to be a sufficient agreement. The values were similar in females, though slightly attenuated compared to what we observed in males. What was interesting for us to note was when we looked at the agreement between the different definition types, so we were looking at the agreement of low muscle mass and grip strength versus low muscle mass and gait speed. And we did this for the four different techniques of adjusting muscle mass. And in this analysis, we found that the greatest Cohen's Kappa value that we observed was 0.41, which is showing that very much so when you're looking at people with low muscle mass and grip strength, they're by and large not the same people as low muscle mass and gait speed. And this has implications for the comparability of most of the expert group definitions, which tend to recommend one or the other. And it didn't matter how you adjusted muscle mass. So based on all these analyses, we're reasonably confident in saying that in general, the different components of the sarcopenia definitions are not by and large identifying the same individuals as sarcopenic. And it's important to note that we were looking at each of the individual components. When you're comparing between actual expert group definitions, you're not comparing just changes in one component, you're looking at changes in two or three of the components. Therefore, the Cohen's Kappa values that we observed are representing kind of what the maximal values would be. And in reality, they would be far less if we're looking at the expert group definitions as written. However, it's interesting in the sarcopenia literature that generally speaking, it doesn't all the different definitions seem to be reported to be associated with important health outcomes, even though there's been this underlying knowledge that they're probably not functioning the same way. So we wanted to test this out using our list of definitions and look at the association of sarcopenia with falls. We chose falls as an outcome largely because it is the most commonly used outcome in the sarcopenia literature. And the reason for that is because falls biologically are connected to sarcopenia. It's thought that when people start having these decreases in muscle mass strength and function, when it reaches a certain point, then their ability to correct themselves after tripping declines and therefore they're at higher risk of falls. For this analysis, we limited ourselves to participants age 65 years and older with falls data. And the reason for this is that there simply were not enough people with both sarcopenia and falls under the age of 65 to run any sort of stable analyses. So we're left with around 10,000 participants. We stratified all of our analyses by sex and we used a proportional odds model with the outcome of no falls, one fall, or two or more falls as the outcome. This was important for us to do because in an ideal world, we really would have wanted to focus in on those recurrent fallers and compare them to everybody else, recurrent fallers being people with two or more falls. However, we don't have a great prevalence of recurrent fallers in the CLSA. So the proportional odds model allowed us to consider each of the three levels separately and was very statistically efficient. It's also worth noting that we used the falls question from the maintaining contact questionnaire. The maintaining contact questionnaire asked about injurious falls within the past 12 months and the maintaining contact questionnaire generally was conducted about 18 months after baseline data collection. So we were able to establish some temporality and that we knew that we were measuring sarcopenia at baseline before we were asking about the period of risk of falls, which was an advantage to the way that we conducted our analyses. And in addition to the proportional odds model, we conducted area under the receiver operating curve analyses to assess the discriminative ability of each of these different definitions. What we found was that in males, some definitions of sarcopenia were associated with an increased odds of falling. Specifically, definitions where muscle mass was adjusted for weight, body mass index, or using the residual technique, which refers to adjusting for both fat mass and height simultaneously in a regression model. And we found that for these definitions, the odds of falling were about 2.2 times greater than if you were not sarcopenic. And this is true in all of our different sensitivity analyses. It didn't matter which different muscle mass cutoffs we were using or which cutoffs for grip strength or gait speed. So it was a very robust finding in that sense. When we conducted the same analyses in females, we observed that the odds of falling, none of the definitions were associated with a greater odds of falling. And interestingly, this wasn't just due to sample size issues and not having the power to detect a significant association. It was actually because there is really no meaningful association to detect. So if you look down the column of the odds of falling, by and large, a lot of the values are actually hovering around 1.0 and not nearly as strong in magnitude as they were in males. And a lot of the definitions actually trended towards showing a protective effect of sarcopenia on falling, which is very unexpected in our analyses. So it seems like there's a very pronounced difference between males and females. The next step is conducting the area under the curve analyses. And just like with Cohen's Kappa, there's no one way of interpreting what the area under the curve values should are. Generally speaking, we consider a minimum value of 0.7 as a requirement for clinical usefulness. And values below that aren't considered descriptive enough to actually be implemented into practice. And we do know with area under the curve analyses that a value of 0.50 means that knowing if someone is sarcopenic or not tells you nothing more than chance a loan would about their risk of falling. So in males, we found that the area under the curve values were between 0.51 to 0.60, which is certainly under that threshold for what we consider clinically relevant. And in all actuality is not much better than chance a loan. And the values were even lower in females, 0.50 to 0.55. When we first ran the analyses, we actually had to go back and double check everything that we conducted, because it seemed like it was very odd that the area under the curve analyses of saying that in males, even the best sarcopenia definition in terms of what was most strongly associated with falls based on the proportional odds model didn't really make a big difference in this clinical interpretation. And the reason that there is a discrepancy is because odds ratios are exactly that, odds ratios. So we found that the baseline risk of falling in males in the non sarcopenic individuals was so low that even the marginal increase in the odds in the sarcopenic group translated into quite a large odds ratio. But because of the way the area under the curve analyses is conducted, you're not comparing the odds, and therefore we didn't observe the same effect. So that's the analyses that we conducted as part of this project. And we have some overall conclusions from this. First and foremost, our biggest conclusion is that the sarcopenia definition should not be used interchangeably based on what we have done as well as previous studies. There's simply an overwhelming amount of evidence now saying that they're not identifying the same groups of people as sarcopenic. And at this point in time, sometimes I get challenged saying, no one is actually considering these definitions equivalent. They're using different variables, different cutoffs, different muscle mass adjustment techniques. This is more of a theoretical discussion rather than a practical discussion. And my answer to that is to refer people to a recent paper that was published by the International Clinical Practice Guidelines for Sarcopenia about the screening diagnosis and management of sarcopenia. And for anyone that works in the sarcopenia field, if you look at the list of authors, you'll recognize a lot of names. These are some of the biggest people in the sarcopenia world. And within this guidelines paper, under the recommendation for a diagnosis, they say that the task force emphasized the importance of using an objective measurement tool for the diagnosis of sarcopenia, which is all fine and well. But then they continue to go on to say that any of the validated international operational tools, including the European Working Group, the Foundations for the National Institute of Health, International Working Group, and the Asian Working Group on Sarcopenia could be used. These are all the expert group definitions that in our analyses we looked at. So us as well as other authors of other papers have found that these definitions don't have sufficient agreement to be considered equivalent. Yet in this position paper, they're actually saying that any of them can be used interchangeably for the diagnosis of sarcopenia. We also recommend that people use area under the curve analyses for clinical interpretation. It told us a very different story than when we just used the proportional odds model. There have only been a handful of other sarcopenia papers that have used this approach, but they've come to very similar conclusions that we have. And interestingly, one of these papers, they didn't look at the total, they looked at the change in the area under the curve statistic rather than the absolute value when sarcopenia was added to a model versus just age and sex alone. And what they found was that sarcopenia really wasn't telling you anything that age and sex wasn't. We also think that our method of defining sarcopenia is one of the greatest outputs of this project. It can be readily applied to any other study that has the data required for diagnosing sarcopenia and allows people to systematically look through these different components of the sarcopenia definitions and really understand how and why they're functioning differently. And we also found this quite marked difference between males and females that does warrant further exploration. Unfortunately, in the sarcopenia literature, quite a few studies that have been conducted simply don't have sufficient sample sizes to stratify their analyses. And those that have have had some of the differences in males and females that we've observed, but they haven't commented or elaborated upon it. So there's certainly a lot of work to be done to understand why these groups are different. So based on those kind of overarching conclusions, the big question for us is, is sarcopenia ready to be applied to clinical settings? As I mentioned in the beginning of my presentation, sarcopenia is now included in the international classification of disease. So clinicians are very much so being encouraged to start diagnosing it. And our response is not yet. We quite strongly feel that it's very important for there to be a unified definition for sarcopenia. With all these different definitions identifying different subgroups of people as sarcopenic, it's problematic to make the assumption that if there was a treatment proven to be effective for sarcopenia based on one definition, that it would be effective based on sarcopenia diagnosed using a different definition. And it's unlikely that clinicians are going to pick up on all those different nuances. So getting the definition solidified for sarcopenia first is a very, very important component. So in order to do this, or to accomplish this kind of overall objective of finding a unified definition of sarcopenia, we think it would be ideal to look for the sarcopenia definition, which best discriminates between people that will or will not have poor health outcomes. And to do this, there are several steps that should be taken. First and foremost, it would be very helpful to replicate our results and other samples. As I've mentioned before, there are studies that have compared between the expert group definitions as written, but it'd be very interesting to see if other people have something similar when they compare to the components as we did. It would also be helpful to look at other types of samples. The CLSA is a community dwelling sample, but perhaps sarcopenia functions quite differently in a more clinically based sample. And that's something that is still somewhat unanswered. It would also be beneficial to conduct longitudinal analyses. The way that sarcopenia is discussed is this idea that sarcopenia, you can diagnose it, you can manage it, and therefore prevent all these future health outcomes like falls from occurring. However, there isn't really the data supporting that actual longitudinal association between sarcopenia, let's say three years before a fall or disability occurs. On that note, it would also be helpful to look at outcomes other than falls, such as disability. Something that I would love to see kind of the sarcopenia experts come up with is a list of the outcomes thought to be most relevant for sarcopenia to really help direct the literature in terms of what to focus on. Disability is another prime example of a variable that does have some sort of biological association with sarcopenia. So that would be a good place to start, but that doesn't mean that falls and disability, there could still be other more relevant outcomes. It would also, again, be helpful to stratify analyses by sex, pooling males and females together. Specifically, when you're looking at the association of sarcopenia with outcomes, it might be masking the association in males. It might be saying that there is an association in females when there really isn't. It could also be helpful to start exploring other measures in muscle mass. Throughout this presentation, I've been talking about dual-energy x-ray absorptimetry measured appendicular lean mass. And this is a perfect measure for us to be using in our analyses because it's what is recommended by the expert group definitions for sarcopenia. And in fact, you can find papers in the literature which say that appendicular lean mass is the gold standard for muscle mass estimation in sarcopenia. However, the reality is appendicular lean mass doesn't actually measure muscle mass. Instead, it measures lean mass, which includes all components of the body that aren't fat mass or bone mass. And there are some groups which are showing that it really is actually quite a poor estimation of muscle mass and that there are other techniques that perhaps could be applied in larger studies like the CLSA, such as a diluted creatine method that might give you much more accurate measures. And one specific group has shown that when you use these alternative measures, the strength of the association between muscle mass and different health outcomes is actually quite strengthened. So that would be something worth exploring in the future as well. Of course, I did not complete this work alone. So I'd like to thank Dr. Permindarena, Dr. Stuart Phillips, Dr. Nazmul Sohail, Dr. Russell D'Souza, Dr. Paul McNicholas, Dr. Johnny Parise and Dr. Lohanna Tabani for all their feedback and guidance on these projects. And then I will turn it over to Jennifer for any questions. Sorry about that, Alex. I had some technical difficulties there. Thank you very much for your excellent presentation. I'd like to now open it up to questions. I noticed that there weren't any immediately in the chat box, but I have a couple questions to start the ball rolling. Just a reminder that muting will remain on, but you can enter your questions into the chat box in the bottom right corner of the WebEx window. So one question that I had is that, why do you think that muscle mass adjusted for height was not associated with falls in males when all the other muscle mass adjustment techniques were? Thanks, Jennifer. That's a great question. So the reason for that, we actually had to look into more descriptive statistics to understand what was happening. And it's because the people that are identified when you adjust appendicular lean mass for height as sarcopenic actually turn out to be very normal weight individuals, whereas the other adjustment techniques tend to identify either overweight or obese people as sarcopenic. And obesity is a big risk factor for falling it upon itself. So it's quite likely that we didn't observe that association when adjusting for height because it was actually, in fact, a lower risk group of individuals based on their body mass index. And that has some very interesting implications for the operationalization of sarcopenia. We really don't know if one of these muscle mass adjustment techniques is really identifying a more clinically relevant group of people as sarcopenic or not. And it also is part of the reason why I'm very cautious about assuming that a treatment strategy that would work in one definition will necessarily translate over to another. Because if you're looking at managing obesity, certainly your recommendations are going to be quite different than if it was in normal weight individuals. Great. Thanks for that. So we have a couple questions in the chat box now. One is whether this has been published yet? We're working on the publications. It's been a bit of a frustrating exercise to have journals that are interested in what we've been working on so far. So hopefully in the next couple of months we'll actually see these being accepted into journals. Yes, these things take time. So we also have, if you have enough information from the CLSA to calculate frailty, is the Rockwood clinical frailty scale better at predicting risk of falls in this population? We certainly can calculate a frailty index in the CLSA. And I've had some colleagues who have worked extensively in that field. Unfortunately, I don't have the answer off the top of my head about what the direct comparison would be between sarcopenia and the frailty index in terms of the risk of falling. So I would assume that it's quite a lot higher with the frailty index. And that is a stream of research that I'd be interested in first actually focusing on free frailty phenotypes, which if you're the five criteria overlap quite substantially with the three criteria that come up for sarcopenia. So it could just be in some ways I think that sarcopenia may more so be a pre frailty where people are starting to accumulate some issues, but not as many as would be required to be considered frail. So it'd be very interesting to actually look at the association comparing between freed frailty index and sarcopenia as it's been operationalized by the expert group definitions for various outcomes. And then certainly it's always of interest to compare the frailty index to freed frailty phenotype. So Stephanie also asks, I guess everybody to please consider the recent publication on physically derived cut points that they've recently put out. And thanks for posting that reference. She says that many findings are in agreement with what has been presented here. In particular, the lack of agreement and identifying the same individuals when using different sarcopenia components. The FNIH criteria seem to specifically identify obese sarcopenic persons, which does not bear the same associated health conditions. So thanks for that little tidbit. Did you have anything to comment on related to that paper? I've certainly read it with great interest. And I think it would be more of a sidebar conversation to have at another time. But I'm very curious of the foundation for the National Institute of Health definition. In my opinion is actually probably the best one that has been done in terms of trying to actually figure out clinically relevant cut-offs. But I think there's still some discussion. And the way that they did it, they chose a low gait speed cut-off and then used ROC analysis to determine what the low hand grip strength cut-off should be based on that. And then use the low hand grip strength cut-off to determine what the muscle mass cut-off is, which methodologically has some interesting issues because it's a very circular way of defining sarcopenia. But it is better than all the convenience based cut-offs that other people have come up with. And I know that Stephanie's group followed the FNIH definition, which again I think is kind of the best out there in terms of a framework to follow. But as a sidebar, I'd be very curious to hear about that group's thoughts about if there's a better way perhaps that we could be trying to derive these cut-offs for all the different sarcopenia measures. Some more work to be done. Always. Yes. So another question, what do you think about bone quality in its relationship to sarcopenia? And some questions, could bone quality be a co-variable that modifies the relationship between sarcopenia and foals? Admittedly, bone quality is inside my area of expertise. So we have started up a couple of projects starting to look into it. But I do think it would be a very relevant variable to consider. Certainly in the literature, there's a lot of support for an interaction between muscle mass and bone. And if you have stronger muscle mass, then you're putting more tension on the bones and that could be affecting your bone mineral density. And it would be interesting to know also if there is any sort of influence of bone mineral density on muscle mass. So I think that'd be a very interesting stream of research to look into. All right. I have a couple more questions and maybe others feel free to post any more questions in the meantime. So my next question is what are the reasons, what are some of the proposed reasons why the association of sarcopenia with foals differs between males and females? That's a great question. And with that, you have to consider the difference between sex-based and gender-based variables. So we as well as other people have noticed that there is quite a big difference when subgroup analysis has been conducted between the association of foals, disability, and a couple of other outcomes of sarcopenia. But focusing on foals, it seems to be actually more gender-based. So it's variable. Females are at a higher risk of falling over all the males. The prevalence is much higher in them comparatively. So they do have a higher baseline risk of falling. So part of what we're seeing could just be a function of how odds ratios work, where because the baseline risk is higher, the slight increase in risk of sarcopenia doesn't actually get translated over to a clinically relevant odds ratio or statistically significant odds ratio. But it also seems like there's different drivers of foals and males than females. Specifically variables like urinary incontinence are very, very important risk factors for foals and females, but don't necessarily always show up in males. And if they do, the magnitude of association is far lower. So we could be looking at simply that there are other risk factors that are more important than sarcopenia, and therefore we're not observing the association. It's also been proposed that because males have higher absolute muscle mass and muscle strength, that they rely more on those variables in order to present themselves from falling. So again, using the example, if you're, if you trip and you're beginning to fall, males may be more likely to rely on strength to correct their posture, whereas females may have other techniques. So it could be that the loss of muscle mass and muscle strength actually impact males more negatively. There's been quite a lot of work done into looking at sex-based variables as well, specifically looking at different growth hormone influences and anabolic and catabolic effects. But interesting, none of that literature has been consistent or showing anything of great interest at this point in time. So it looks like more of the other more gender-based variables may be at play. Great, very thorough answer. Another question that I had is whether there's much value in measuring muscle mass in addition to grip strength or gait speed, or would it be better to just measure strength and function? We did conduct analyses looking at grip strength and gait speed alone versus in just combination with muscle mass. And interestingly, adding muscle mass didn't really improve the odds ratios all that much. And I think the reason, so the around 2000, the year 2000, there was that big shift in how sarcopenia was conceptualized and it went from being just muscle mass to this combination of muscle mass and either function or strength. And I think the reason why people are still holding on to the muscle mass component, even though it doesn't really actually show up, it's certainly not showing up as being associated with outcomes on its own. And even in combination with other variables, it's not that helpful is because the entire world of sarcopenia was founded on this idea of muscle mass. Again, literally the definition or the word sarcopenia is translated from Greek as loss of flesh. And there are people that would argue that you should only be looking at that muscle mass component and that if you're looking at a decrease in strength and function, it's dyna-penia, it's a separate clinical entity. So I think from a pragmatic point of view, it might not actually be all that helpful, especially when you consider that the dual x-ray absorptimetry measurements for appendicular lean mass are certainly much more expensive than doing grip strength and gait speed would be. So there might not be a huge benefit to including it, but if you don't include it, then are we really measuring sarcopenia anymore or is it this entirely new construct? And interestingly looking at the physical function literature, it certainly seems that if you start counting up physical function deficits, so grip strength and gait speeds are two examples, but even within the CLSA, we also have balance in the timed up-and-go and a chair rise test. And counting up the number of deficits seems to be much more strongly associated with all the different outcomes that we would think for sarcopenia than sarcopenia actually is. And that kind of goes back to the question about the frailty index. Perhaps it really is you just need to have a really good number of measures and you can really precisely say how people are doing relative to their peers. So I don't think that was exactly a straightforward answer, but muscle mass definitely, I think there is an argument for leaving it out, but it just becomes a question of are you still measuring sarcopenia if you don't include it? Okay. Thanks. So one more, another question came in. Would you consider to adjust height somehow due to the impact of aging in this measurement? And what about using a similar approach to that of osteoporosis and taking advantage of the young sample in the CLSA? That's a great question. And a lot of the sarcopenia literature is actually based off of how osteoporosis is conceptualized. So when you look at the low muscle mass cutoff, quite a lot of them are derived by comparing to a young healthy reference population and looking at a value of 2.5 standard deviations or more below that reference population. So that has been done in the literature. And it does become interesting when you're talking about changes in height and other body composition variables depending on which adjustment technique you're using. I actually found an example of a study that used that less than 2.5 standard deviations below a young healthy reference population. And they had recruited their own young people as well as the older adults. And the prevalence of sarcopenia in women went down. And I believe they were adjusting specifically for body mass index. And because the height was very, very different between the younger versus older females, it caused this phenomenon of decreasing sarcopenia prevalence to occur. Because the height was changing more quickly than the muscle mass was. So I think it's choosing what the cutoff should be and exactly which technique you should use to derive them. I think that's still a very open-ended question because body composition is changing in so many different ways during the aging process. And the decrease of height is certainly part of that. So if you're comparing to the young healthy reference population, does it make sense if people are decreasing in height? Do those values actually have the same meaning at that point in time? And as far as I know, we don't have the answer to that yet. Great. Well, thank you, Alex. I think we'll, I don't see any more questions coming up. But just a reminder, if you do have any questions, feel free to post them. And we can always follow up with you after this is to all our participants, everyone attending. We can always follow up with particular answers to your questions afterwards. So thank you again for a great presentation. We appreciate you taking the time to do a webinar for us here at the CLSA. I'd like to remind everyone that CLSA data access request applications are ongoing. The next deadline for applications is June 17th of this year. So please visit the CLSA website under data access to review available data, further information, and other details about this process. I'd also like to remind everyone to complete their survey that's located under the polling option. If you don't see it beside the chat button, please just click the drop down arrow and you should be able to see it. So our next webinar will take place on Wednesday, March 25th at noon, and Dr. Olga Fio and Dr. Mario Ulises Perez-Zepeda of Dalhousie University will present their webinar entitled population norms and prevalence of frailty among middle-aged and older Canadians. And you can register for that. Oh, there's a nice little picture of them. Registration is now open for this to give us an idea of who will be attending. The next thing I wanted to remind everybody about or to tell you about is that graduate students and postdoctoral fellows with an interest in longitudinal studies on aging are encouraged to save the date for our scientific program in aging 2020 called SPA 2020. This innovative five-day training program will take place in June at Hawkely Valley Resort in southwestern Ontario. Applications are now open on CIHR's research nut for this. But just a reminder, if you are interested, the deadline is soon approaching and that's on March 12th, I believe. And remember the CLSA promotes this webinar series using the hashtag CLSA webinar. So we invite you to follow us on Twitter and do your tweeting. So thank you again for attending today's presentation and I hope everybody has a great rest of the day.