 Welcome to our distinguished speaker, Dr. Yukiko Asada. Yukiko Asada is an associate professor in the Department of Community Health and Epidemiology at Dalhousie University in Halifax, Nova Scotia, and a co-local site and principal investigator at the CLSA's Halifax data collection site. Dr. Asada's research involves articulating the concept of health inequity, advancing the measurement of health inequity, and promoting the public dialogue on health inequity. Using the CLSA, she is expanding her health equity research in the context of aging. Thank you very much. Thank you, Carol, for the kind introduction, and thank you everyone for attending the webinar series. So let's start it. I would like to introduce the team involved for this work. It's Jerry Hurley in McMaster University. He is a co-P.I. of the project. Michelle Guagnon and Susan Kirkland, they're co-investigators. And as everybody knows, Susan is the co-P.I. of the CLSA. As Stefan Fitzburton at Dalhousie University is the research analyst. And our funding comes from C.I.H.R. Okay, so based with an aging population, countries around the world strive to foster successful aging. And what constitutes successful aging, sometimes also referred to as healthy aging or optimal aging, is often debated. But health is indisputably an essential component of successful aging. While successful aging is the most commonly framed at the individual level, it is possible to frame it from the population perspective. Successfully aging population has both an overall level of health and an spare distribution of health. So we think that it's not enough to have just a high level of overall health, for example, higher life expectancy or lower instance of some chronic conditions or that's overall indicator. But we would like to make sure that the good health is distributed in a fair manner. So when viewed from a population perspective, a key indicator of successful aging is whether health inequalities, which are differences in health, and inequities, which are unfair differences in the population increase or decrease over the last course. So this is the kind of context that we would like to promote health equity analysis in aging population. The measuring the health inequalities and inequities is quite complex. And there are a couple of reasons that I will just list you. First difficulty is that we have to define health inequity. And some of you know, inequity is an ethical concept. So here we are talking about fairness. And people have different ideas. So defining it is it's not like everybody agrees on and their different views. And you have to decide which one you're looking at. And then also once you define that concept, you have to incorporate it to the measurement. And then also other difficulty is that then deciding what measure of health to use. And from our work, so we're kind of new to this aging context, but we've done some of the health inequality inequity work in more general population context. And then there we learn that if we use a different definition of health inequity, and then also different measure of health, we could have different results. And that's totally could make sense because you are defining equity in different ways. And then maybe the measurement construct is different. So we have different results. But another thing is that then we need to know what we're doing and why we want to pick certain definition and then a certain measurement. So for this presentation, we are using the definition of health inequity, policy amenability. In short, this view says that the differences in health due to factors amenable to policy are unfair. So more succinctly, if we could address some of the differences in health through policy, we could have done something, but we fail to do so. Then that's unfair. And we would count that as an inequity. So that's the definition we take. And then for this presentation, we focus on grip strength. So grip strength is an estimate of isometric strength in the upper body indicating overall muscle strength. And it is quite simple. You're just usually need a handheld device. And then you can do it in a clinical setting and you don't require any big machine. So that's quite good. And then also it's a physical measurement. So people consider that is more objective. Now I'm giving kind of quotation mark here to objective, but compared to, for example, self rated health. So there are a lot of excitement for this measurement. And some people say grip strength can be used as an indicator of frailty or vulnerability. This is based on some of the studies showing that there is a strong association between grip strength and known Asian markers. So Asian markers we are talking about here is such as like a hearing, visual acuity, hemoglobin, number of teeth, or fracture after age 50. So those are things considered to be Asian markers. And when we look at the association with grip strength, we find a good association. And then also some scholars argue that grip strength could be considered as an indicator of well being. So grip strength indicate muscle strength. And muscle strength is needed to perform some basic activity of daily living. And a basic activity of daily living is kind of quite important as a component or the determinant of the well being. So if we know something about grip strength, maybe we can say that then we could say something about well being of that person. Also there are increasingly literature showing the predictive values of grip strength in terms of future disability, mobility, and mortality. And then I am going to show you some study. It's a little blurry, but I hope you can see it. So this is a systematic review and a meta analysis by Cooper and his colleagues published in 2010. And it's examining association between grip strength and mortality in community dwelling populations. And then they focus on the order population. So for grip strength, they had 14 data points. And if you look at the four different groups and summary of the other ratio for mortality is actually quite great. So this is an after adjustment for age sex and body size. So this is just one example. But if you look at the literature, you could see that interestingly somehow, if you know grip strength of the person now, you might be able to predict what might happen in the future in terms of health status of that person. So our group thought that it would be interesting to measure inequalities and inequities in grip strength among older adults using the CELSA. And I say it here, intention. And the reason is that the more we learn about grip strength and try to use that for the analysis of inequalities and inequities, we've discovered a lot of interesting questions. And then I would like to say underscore that this talk is quite exploratory and work in progress, meaning that our group has not really found the complete ways to analyze inequities using grip strength. And then we've not completed that work yet. And we're still thinking about how best to do it. But at this stage, I think it's quite interesting to share some of the more or the aspect of this measure, which we found very different from some other measures of health, such as health status measures like self-rated health or health utility index or even frailty index that are more often used in health inequality in equity analysis. So that's the kind of context of the study. So let's begin. The data we used for this work is the CELSA comprehensive. So a lot of you today participating would know comprehensive data is a subset of CELSA, about 30,000 non-institutionalized persons is 45 to 85 at the time of the recruitment. And then data collection took place in home interviews and then also the data collection sites visit. So we have a physical physical and clinical assessment. So that's why we have a grip strength measure. So for our analysis, we use the baseline data. And then after missing value and outliers and all those data cleaning, typical for any analysis, we have a sample size of a little over 27,000 people. The number is probably a bit different from the poster of this presentation, but this is the number I'm going to give you today about analysis results. So the variables we're going to use, grip strength is measured in kilograms. In CELSA, they measure three times for each individual. We took the maximum values. In the literature, we see both maximum values and average values. And for now, we just pick the maximum value. Other variables are considered in the analysis I'm going to present, age, sex and height, ethnicity and Aboriginal status, sexual orientation, rurality, province, marital status, social participation, household income, medication home ownership, alcohol use, physical activity, and white circumference. These are the typical variables I think we use in many different health inequality inequity analysis. So this is the distribution of grip strength for both sexes. You see that the grip strength in kilograms and in x-axis and in a data vertically is the density. And then I will show you a lot of plots and then graphs today. These are graphs of the like a distribution. Graphs are not weighted, so it doesn't know how the data would look like. So for this distribution, you could say that this inequality is the distribution of grip strength in the population in your data. And you could summarize this distribution by using an index of inequality. So for example, genicoefficient could summarize how dispersed this distribution is. And if you use that, it's 0.188 genicoefficient from 0 to 1. 0 is the most equal, 1 is most unequal. And that mean is 36.83 kilograms. Genicoefficient is an index, so it in itself doesn't mean anything. But if we want to understand what this number might mean, you could say that the twice of the genicoefficient multiplied by mean indicates average expected difference in grip strength between two randomly selected persons in the population. So from the CLSA sample, if you pick two people randomly on average, you see the difference about than 14 kilograms. So that's quite simple, just looking at inequality. But here, we would go into inequity analysis. So we call that the mindset in this analysis, we are using a particular view of inequity, which is policy amenability. So again, we are saying that the differences in grip strength due to factors amenable to policy and unfair. So here, moving on to inequity analysis, we want to say that are there any factors associated with grip strength that we cannot do anything about through policy? So those were that we cannot address by policy, so we don't worry about, but in other parts, we worry about. So for that, the good suspects are age, sex, and height known to be strongly associated with grip strength. So first, we looked at in our data. In our data, do the sex, age, and height, do they have a strong association with grip strength? And for that, we use the method called a regression based inequality to composition analysis. So essentially what it is is that we can say that which factors are explaining variation in grip strength. So the entire variation in grip strength, in the regression we put, you see the lot of variables we put, that's the ones I introduced at the beginning. So we put everything, and then we cannot still explain about 30%. That's something going on that we cannot explain. But sex explains like 40%. Height explains 20%. And age 7%. And then all other factors explain about 3% of variation in grip strength. So this is consistent with the literature, and it's quite the big relationship we see with these three variables. So talking about the sex, so this is the overall distribution I showed you earlier, but it's actually, well, let's go back. You can see two humps. So this is because we are talking about the two quite distinct distributions of male and female in one. So sex, the importance of sex is quite visible here. So now grip strength and height, you see a fair clear relationship here, have a by men and women. Male seems to be a little bit than a more steep woman. And then also one thing it will be quite interesting to notice is that for men, this vertical, it's not quite straight here. This pen doesn't work quite well. But you see the point that in a vertical is that even given the same height, there's some variation in grip strength. And then that variation looks bigger for men than women. And this is h for both sexes. It's a slight decline, you see, and a men and women. The mean is lower for women than men, and then both declining. And then again, you see more variation for given age, bigger variation among men than women. So going back to the prognostic values I talked about earlier in grip strength, they're increasing literature on kind of fascination that the grip strength of today is associated with the future health event. And then as I showed you that Cooper is one of those systematic review and meta-analysis. But it cannot be that just the role grip strength value is predicting because then all women should have worse health event sooner than men, because men are stronger and it's not like that. So there got to be some adjustment. And then it looks like the study is really very and a little bit confusion about what adjustment should be made when estimating prognostic values of grip strength. And it looks like a loose consensus is in some kind of adjustment for age sex and in the body size. And the body size is understood height and or weight. And then they're not, it's a quite variation in there too. So with that, what can we say about what is the unfair distribution of grip strength? So if the second bullet point, if we assume that then a strong association we saw, the differences in grip strength due to age, sex, and height are not something we can address by policy, then we could say that the deviations from the norm said by age, sex, and height are unfair. So that we thought we could focus on as an unfair component of grip strength. So going back to this Cooper meta-analysis results, this is after adjustment of age, sex, and body size. And you still see this much difference. And we could say that, well, this is unfair because we could have done something with policy. So we need to know about this. And that should be the focus for health inequity analysis using grip strength. So how can we do that analytically? We got some hint from the analysis using the standardized distributions. So there's that score, which is the norm is set by the age dependent. So given the age, what would be expected, the grip strength. And the T score is the norm is based on the age of the peak performance. So I'll give you an example. Like child malnutrition literature, they calculate height for age, weight for age. And then there is a reference population that they were the health organization set out. And then it's almost like you're saying that your child height and weight should look like that given the age. And then if you are your child is below the deviation, below the expected value is this much, usually like a two standard deviation, then we would consider that child to be malnutrition. And a bone density uses the T score. And then it usually the peak performance, like a 30 year old of women. That's as a norm. And there is some discussion about whether the norm should be set separately for different ethnicity. And cognition for that score and T score, it's actually there was a presentation by the Dr. Holy Tuco in January. So you could take a look at that. That scores and T scores for grip strength is an immersion interest. And I could, we could find on some of the literature, but it's not quite set yet. There's some discussion about it. The Canadian reference values were produced using Canadian health major survey. And so one study is an age height and weight to define the reference. The Chong study is the peak age of 30 to 39 year old as the reference. And then Pannon and Magasi study is in 20 to 49 year old grip strength as the norm. So there's some kind of moving target. We are not quite sure. And then we don't know in addition that what is the kind of point that indicates clinically problematic or deficient grip strengths that we need to worry about. So it's a evolving field. So given that we decided to calculate the best score within the data. So essentially what we did is set them for men and women separately. We used an age, age square and then height. And then we run the model. And then we had the expected value, which means that for the given the age and then a height and a sex of the participant in the data, what would be their expected grip strength. And that score is just following the standard calculation of that score. We looked at that deviation divided by the standard deviation of the expected grip strength. So when we do that, this is the result you see. So the male and in a female, the red line is that that score for this the the expected the distribution of grip strength. And then that blue is the what you would observe. And then not that much difference between men and women. But as we saw in the earlier scatter plots, I showed you men are slightly more dispersed, flatter, bigger standard deviation. So for that, we also go interested in what factors are associated with them that that score, meaning that the deviation from the norm. So for that, we separately for men and women run regression analysis. And then here the dependent variable is the z score, we calculated. And then all the independent variables I talked about earlier, the other variables are included in the model. And then we used an or less. And then for this analysis, and we waited, sample waited, and then also our standard errors taken into account for the complex survey design of the CLSA. So here I'm going to summarize some of the key results from the regression analysis using graph. So let me first explain how to look at another graph. So the you see that the blue is male, and the red is the female. And the solid circles indicate that the statistical significance p value less than 1%. And then if it is not solid, then it means it's not statistically significant. So that with these are the coefficient of the regression. So the compared, let's look at the rule. So compared to the urban, the rural residency indicates stronger or the better than expected standard deviation. So rural is stronger. Aboriginal status, we didn't have any statistically significant association compared to white. But in a non white non aboriginal is expect the grip strings expected them, the lower than expected grip strings. And a homosexuality and a bisexuality compared to heterosexuality has an opposite association for men. For men, it goes to the weaker side. So it's lower than expected. And then for women, it's a higher than expected. And widowhood single never married not much, but high social participation is a frequency of certain social activities. For men, it doesn't have a statistically significant association, but enough for women, high social participation would result in the grip strings higher than you would expect given age and effect. So this is a social economic status factors, rather than going one by one, so that this is a social economic status relationship, usually, for if you look at the general health status, social gradient of health. And then here you would expect that the social gradient of the health should go like this blue line. It's the relationship should be like this line. But in a year, we only see the typical gradient for education of the higher end. And then actually education only for men. And then for education, not much at all for both sexes. And then actually, the better educated for men indicates that that is the lower than expected grip strength. And then here is the health behavior, not much going on if you just look at the statistically significant relationship, but the regular drinking and the regular heavy drinking compared to no drinking are associated with the higher than expected grip strength. So the standard deviation is a better side. So it's kind of opposite of what you'd expect. Usually the drinking, heavy drinking would be associated with the lower health. And then in this case, what we found is that it's actually going the other direction of that stronger grip strength that you would expect for sex age and height. So those are the results. And we found these results quite interesting and puzzling. So I would like to kind of summarize in that context and contextualize our results. And then I would love to hear what you think in the discussion time. So let me begin. So the first key question is that what really should the hair fair distribution of grip strings look like? So in our analysis, we used each sex and then a height to set the norm. So that means that weaker grip strings associated with being older, female, and shorter, something we cannot do anything about through policy. And we said that we say that the variation due to these factors, we don't worry about it because that's beyond our policy. So the norm is set with that. And then if you're deviating from that norm, that one we would worry about. But age, sex and height, we don't worry about. But is it okay to do that? So let me just show. Firstly, we did the sum of the age. What if we set the norm just by sex and height? And then we didn't use age. And this is as expected, it's kind of flatter because then there are more deviations introduced by age. And you can see that then so the left side is not using age. And then right side is using age as they are setting the norm. So this is what you would expect. And it's not clear really that some of the few studies I talked about that tries to set the Z score or T score for grip strings, they argue that in theory we could expect somebody older to maintain the grip strings of like, I don't know, 30 year old or 40 years old. And that should be the goal. And we can incorporate that, but that's a judgment call. And it is something we cannot change, or maybe we can change, we don't know. And that has implications for that analysis. For the next one is the sex. So the question here is that being female is usually associated with a weaker grip strength compared to men? And is it inevitable? Is it like we are made? I mean, I'm female. So am I made to be weaker than male? And then we can't do, I cannot do anything about it. Well, it's actually quite, that's the usual assumption, but in some of the figures we found, cast out on that. So this is from the World Health Organization Asian Report. And then you see male and female and it's by country. So when you look at like a South Africa, it looks like a women are as equal, equally strong as men. And then that's the case for like a Mexico yellow line and India blue line. So maybe it's not that we should always expect women are weaker than men. And of course, here we are talking about biological sex or gender. So how much of it is that the body composition that the biologically were made differently, or something we encourage or discourage as a society that has a life course impact of how much strength women or men have. And then also, this graph is interesting in that, you know, we just kind of expected like we didn't think anything about country differences. But why? Why do we have country differences? Are they certain, again, body composition differences due to ethnicity? Or something we like nutrition or certain exercise habit or culture? So how much of it is something we can say this is given and we don't worry about and we don't need to aim to be stronger or something we need to address. So these are actually, this graph got quite a bit of discussion in our group. So I think that the needs quite a still not settled this problem. And then here I would like to underscore one thing. So we are using Z score or maybe potentially T score for the analysis of health inequity. And we think that it has a good promise and it goes well with the way we think about group strength. But the way we are setting the norm is particularly for the context of health inequity analysis. These Z scores and T scores can be used for many different purposes, for example, like a clinical utility. And then for that, maybe what how the norm should be set might be different, it depending on the context. And then all the things I said about what should be amenable to policy and that is influenced by the our particular interest to use it to the health inequity analysis. Okay, so the next question is to what extent should we expect that inequity in group strength is associated with social disadvantage. So here more concretely what we are puzzled about is a no clear, less than expected or counterintuitive socio-economic gradient in the deviation from the norm when we look at the group strength. That is quite different from many of the health inequity analysis because in a substantial amount of health inequity analysis went for investigating really typical ubiquitous socio-economic gradient in health. And then here it's not that much really we are observing. What should we think about it? And the other issue is the some of the health behavior, in our case the drinking, the regular heavy drinking seems to be associated with the group strength better than the norm. So it's kind of counterintuitive. And of course these associations and we discovered the analysis could be fails because our model is very simple and then it maybe we didn't, we should have done something a bit more complicated and subtle nuanced and then we might start to see the real story and that's always possible and we should try something different but then at the same time this is typical way we look at the relationship with the some other health measures and then there we have some typical expected relationship and direction and for group strength it is not the case. So if this is really something peculiar about group strength then we should ask the question should a good measure of health for a health equity analysis just as successful aging of a population be sensitive to social disadvantage and an unhealthy behavior. If there's some counterintuitive relationship maybe group strength wouldn't make the cut as a good measure of health to be used for this purpose of health equity analysis. So in summary, group strength has quite a bit of excitement as an objective simple measure of health in aging population and we argue that the non-successfully aging population has both a good overall level of health and a fair distribution of health. So assessing inequality and inequity in group strength appears promising but the more we learn about it the use of group strength requires some caution. So first thing is that it is important to consider what the fair distribution of group strength should look like and so here I'm talking about what the which factors should determine the norm what we would expect for the group strength but as I discussed for the age, sex, and height the answer is not quite straightforward and then also persons with social disadvantage and unhealthy behavior may have a stronger group strength than expected and then this cast a doubt on its use for health equity analysis as a barometer of successful population aging. So as you see I hope it was as interesting as for you as for us in our group we have we never thought that the group strength would produce this much discussion in our research group and we've been talking about it for months now. So I hope it was interesting and I would love to hear about different ideas or something we could try and what you think. So I stopped here Carol and then we can go for a discussion session. Well thank you Dr. Asada that was a really excellent presentation and hopefully you can have a good discussion period now. I'd like to open it up for questions as a reminder remuting remains on but you can enter your questions into the chat window in the bottom right corner of the WebEx window and I'll read through them. So first we have a comment a really interesting presentation I like your ideas and analysis. A question among many that I could ask what about occupation such as blue or white versus white collar as blue collar jobs involve lots of work with hands which might increase grip strength and this is from Harry. Yes I think it's um it's a very good point and literature is interesting too like you one side of the occupation like a manual workers and then you know those are labor occupations that might improve grip strength but in a some literature also shows that then it's actually kind of overused so you use too much that it's actually deteriorate you can be strong for some time but it's not good for you after all kind of thing. So I think that I have a feeling that we kind of did the first round of analysis and it will be really interesting to go back to the list of variables SLSA can offer and we consider very carefully which wants to consider more so that's an excellent suggestion thank you. The question from myself from Carol here um so you know I think there is some discussions coming from the group about the lack of associations with the economic indicators but rather than not focusing on grip strength in the future do you think maybe the longitudinal analysis might play into answering some of the differences between the grip strength that you're seeing here? Yes I think that's a that's a very good thing to look at and then as the SLSA have more data coming in and the longitudinal analysis will be promising and then there's also a bit of confusion in the literature we found that then are we talking about grip strength of today or are we talking about change in grip strength right which is more more important and then I don't see clear answer you know somebody might be very very strong today but in five years later that person somehow dropped quite a bit and maybe that is more significant than where the baseline is so these are the quite fascinating questions. Certainly when you're talking about health inequity it's really the the minute you're in your your environment throughout your lifespan. Exactly. Very interesting. From Cass there's a comment or asking for advice I want to integrate information about grip strength and of course a nutrition assessment what key messages do you recommend? I would love to know so please that would be good too so I have to confess really in our group we thought that this would go as we expected we never thought that the grip strength is this different from any other typical health status measures we looked at and which I mean the general health status measures so I am learning as I go about grip strength and then how the nutrition and then how that goes together that would be very important question too. Thank you. A question from Laura Anderson. Thank you for the nice presentation. Could you clarify what was on the x-axis of your results plot? Was change in standard deviation the beta or the absolute risk difference? Were they adjusted estimates? Yeah sorry so that is the beta so maybe the changes are confusing but it's the beta itself so the dependent variable is the standard deviation the dependent variable is the z-score for each person right so the beta indicates that what's the difference in the z-score between that category compared to the reference group category so I hope that clarifies. I have a question while we wait for a couple of other people to put their questions into the chat box. Can you talk about what you did with the missing data for grip strength? Did you look at the people that didn't do it or did you include those that were contradicted or how did you deal with all that? It's a good question and I think and it's almost like I said I should have a subgroup to see what the difference is. For grip strength and if the participant didn't have the grip strength we didn't impure because it's the dependent variable and then of course as you would expect that then those people are from other health indicators and a typical thing of the less healthy and more health problems and lower socioeconomic status and typical as you expect and then also we needed to clean up few people not many about like I would say less than 20 people who have outlier values and usually they're very short people like 120 centimeters height and they had quite low grip strength and that had a significant impact in our models so we removed them. So that's the missing value of the dependent variable. Independent variables for income there are tons of missing about 10%, 10%, 15%. So we created those people keep them as a missing category and then for others we mostly deleted from them. Thank you for clarifying. Thank you. So do you think that your associations could be generalized to other measures? I mean you use grip strength because it's very interesting but do you really think it could be a proxy for inequities for functional measures for frailty or kind of healthy aging or even for mortality or really solid end points? No. So that's the interesting point. So what we know from we are also doing a constructing frailty index looking at the more number of variables in the body system and when we look use that as the variable the association we observed for grip strength wouldn't be there. So that is usually clear source of socioeconomic gradient and then if you have a healthy behavior that would bump up your health outcome and that's all what we would expect. So the grip strength is quite unusual in that and then this is the first time we observed this unexpected results and now I'm very curious about other physical measures that CLSA has. So for example like a chair rise or the walking speed you know that kind of a similar of physically assessed measures would they behave the same way? I don't know but that's an interesting question. Absolutely to kind of come up with an index of functional health equality and see how they they compare to each other would be very very interesting. So we have our group here a couple questions from Dr. Reina. Why do you think your alcohol variable is showing benefit? We're also seeing this in several of our analyses where it shows a benefit for many health outcomes except for depression. So do you have any feel for for what you're seeing as being beneficial? I would like to know the answer and then I am very interested in hearing that then that you're seeing alcohol is that way in your analysis too because I thought that in other data sets I never observed this. So something might be going on. So we wish we we should you know discuss across across groups since your approach is also what Dr. Reina's group is looking at for the reference values for many performance tests. Yes that would be very nice. Well if we don't have any other questions to address I'd like to thank you again. We appreciate your participation Dr. Asada for being here for our CLSA webinar series. I'd like to remind everyone that CLSA data access requests applications are ongoing. The next deadline for applications is on June 11th 2018. Please visit our CLSA website under data access to review available data for further information and details about our application process. Also our next webinar is scheduled for next month. We'll be welcoming Dr. Jade Law associate professor at the School of Public Health and Health Systems University of Waterloo and Matthew Quitt PhD candidate in the School of Planning at the University of Waterloo to present exploring the geography of cognitive function and social support availability. Please register soon and join us for next month's webinar. Thank you again for attending today's presentation and thank you very much Dr. Asada for such an excellent presentation. Thank you very much.