 Hi, and welcome all. I'm Carol Basim, a scientific manager for the Canadian Longitudinal Study on Aging for CLSA. Good afternoon. Thanks for joining us today for the May installment of the CLSA webinar series for 2017. We will be welcoming Dr. Philip St. John from University of Manitoba to talk about multiple morbidity in Canada. Before I further introduce our presenter, I just wanted to share a few reminders about today's presentation. The webinar for run for about one hour with an opportunity for questions at the end of the presentation. Participants will be muted, but questions or concerns can be posted at any time. If you have a question, you can enter it into the chat window at the bottom right of the WebEx window at any point during the session. These questions will be moderated at the end of the presentation during the question and answer session. Be sure to select all participants from the drop-down menu before you press the send button. Mobile users must select chat with everyone. Remember the questions will be visible to all attendees and this presentation is recorded and will be posted to the CLSA Web site in the future. I'd now like to welcome today's presenter. Dr. Philip St. John is an associate professor and head of geriatric medicine in the Department of Internal Medicine at the University of Manitoba. He is an affiliate of the Center on Aging at the University of Manitoba and is the code lead investigator of the CLSA Manitoba site. His research interests include rural health and epidemiology of cognitive impairment and depression. Now thank you for joining us, Dr. St. John, and I will turn it over to you. Okay, thanks very much. This is the first webinar I've done. So if I get the technology right, it hopefully will go okay. And if I say something cookie, just write something in the chat box. So I would like to acknowledge some of the co-authors. So Lindsay Torbiak is a medicine resident, Suzanne Tyas and Waterloo, Verana Menek here, Bob Tate here in Winnipeg, Lauren Griffith, and in particular Scott Nowicki, who is the analyst who did most of these analyses. So in terms of conflict of interest, first we did get a grant from the Center on Aging, so I would like to thank them for that. And as well I really don't have any conflicts of interest except as we'll come back to it I am in the high income group, which is a bit of a conflict of interest. So being a clinician, I thought we'd start with a case, and it's a true case that we had on our inpatient unit actually a couple of years ago now. She was 81, she was living in the same house in the north end of Winnipeg for 40 years since she immigrated from the former Yugoslavia public, and her daughter was living in the house immediately next door. She was previously ADL independent, but IADL dependent. She had a past history of multi morbidity, so she had type 2 diabetes, hypertension, macular degeneration actually and cataracts, osteoarthritis, ischemic heart disease, congestive heart failure, chronic renal failure, falls, urinary incontinence, and she had cognitive issues noted by both her daughter and her son, but had never had a formal cognitive assessment. She was followed as an outpatient by a family doctor and eight specialists. So she came into one of the peripheral hospitals with an acute stroke. She had left-sided weakness, slurred speech, and falls. She got TPA at that hospital at the major teaching hospital, so she was transferred to the major teaching hospital and then transferred back to the peripheral hospital, and then after four or five days she was transferred for rehab. Day two of rehab, she got worsening shortness of breath and her baseline dose of Lasix was increased, and the thinking at that time was that she was in heart failure. And then she deteriorated over the next two days to the point that she was on 15 liters of oxygen, which is fairly substantial. And she was transferred to a third acute care hospital in her stay, and she was diagnosed with bronchiolitis or Blitterans, which is a complication of pneumonia, and she was treated with high dose steroids. She settled out, and after five or six days in the third acute care hospital, she came to a second rehab site here in Winnipeg. She actually survived the pneumonia, but she was not doing too well in rehabilitation. She remained with very substantial cognitive and functional deficits, and so we had a family conference to talk about future plans, and at the time, I said, you know, she was lucky to survive the pneumonia, and her daughter said, well, no, no, she wasn't. So after a fairly long discharge planning process, the original plan was to go home with the stroke program follow-up, and actually she would have been followed by 14 specialists, had the original discharge plan been put in place. But throughout her rehab stay, she kept dropping her blood pressure and having symptomatic CHF, so she was concomitantly in heart failure, and with the treatment of the heart failure was quite hypotensive. So she actually elected to stop all of the aggressive care and went home on the palliative care program and actually just her family physician and the palliative care program as a follow-up. So I thought this was sort of a good example of a fairly, not an uncommon case, actually, of somebody with multimorbidities and what can happen to people with multimorbidities while they're in the acute care system. And again, I think most people who work in clinical practice can think they've seen a fair number of people like this. Now we think of multimorbidities being kind of something new and it is very important, so I don't want to underestimate the importance of multimorbidity, particularly in the modern era, but I think it's important to stress that the notion of multimorbidity is not at all new. So it was well described in Byzantine texts on aging in the 300 to 400, there's just some messages coming across with no audio. Do people have the audio? So it was well described in Byzantine texts on aging in circa 300 AD or AC after Common Era and it was actually quite well described in many of these textbooks and it was particularly noted that this was a function of aging. And it's actually been noted in medical textbooks throughout the ages that multimorbidity occurs with aging. So it's not a particularly new phenomenon. It was actually well described in many of the clinical papers in the UK in the 1940s. And it is important now because it's getting increasing attention. And just to give one of the older citations, this was a paper in the Lancet. It was one of ex-Nismith papers where they described one of the early geriatric units. So what I'd like to point out here is the fact that two-thirds of the patient have multiple pathologic conditions and about a third of the people, only a third of the people, had acute illnesses and another about a third of the people had acute illnesses in addition to long-term disorders. And the other key point is that the people with multiple interacting problems that were chronic in nature tended to stay in hospital for the longest. So I mean if you think of the current narrative in geriatric medicine, we tend to say that our hospital system was designed in a time when we had single system acute illness and our system was set up to deal with that. But I wonder if that narrative is true and I wonder if we set up our hospital system when we actually had multiple pathological conditions at the time and we just didn't set it up properly to recognize that, which has different policy implications obviously than saying epidemiology's evolved rather than we set up the system perhaps incorrectly in the first place. So I think it's important to point out that it's not new and it's actually been recognized for a very long time. Now another kind of classic paper was a survey of Scottish primary care clinics in 1964. So again this is before, both these are before I was born and again they talk about multiple disabilities but by disabilities they actually meant chronic illnesses. And men had slightly fewer chronic illnesses than women and I'll come back to it but the average number of chronic illnesses was three of which most were unknown to the family doctor. So again the notion that as we get older we accrue diseases is not really a new notion it's got a long history. And there's also a long history of care models so these are all photographs or pictures of what used to be called hospitals for the incurables. So this would be the British home and hospital for incurables entirely dependent on voluntary contributions. This is one in Paris a hospital for incurables. People from Montreal will recognize this one and then Amiens in France as well. So there are lots and lots of hospitals for what we're called people with incurable problems at the time and these are care models for people largely with multiple interacting problems that were progressive in chronic in nature. So neither the notion of multi morbidity nor the care models are particularly novel. They're very important and worth looking at but it's not really a novel idea. So when you come back to definitions these are the definitions of the American Geriatric Society and again these are based upon Alvin Feinstein's definition. So first we talk about a chronic illness. This is a health problem that requires management over a period of years or decades so I think that's fairly straightforward. And then multi morbidity is the coexistence of multiple chronic diseases and multiple conditions in the same individual. So this is where no one diseases the topic of interest or the topic of main focus rather it's the whole sum of all of the issues that's important. And that's differentiated from comorbidity where you have any additional entity that has existed or may occur during the clinical course of a patient who has the index disease under study. So this would be where you have a disease of interest and then you have the comorbidities. So an example would be heart failure with comorbidities of renal failure, macular degeneration, osteoarthritis, osteoporosis but you're predominantly interested in the heart failure not necessarily the whole spectrum of the diseases. So when you see the terms comorbidity and multi morbidity those are the standard definitions. In the American Geriatric Society the definitions will come back to it is a bit variable some people tend to say greater than or equal to three chronic diseases some people put it is greater than equal to two chronic diseases and some people put it is greater than three chronic diseases. The EGS says greater than or equal to three chronic diseases. And the second key point that's very important is that it's the accumulation of all of these diseases that's important so it's the sum total of the diseases is more important than any particular disease in particular and it's a cumulative effect of the build-up of all of these issues that becomes important. And the other point that the American Geographic Society makes is that multi morbidity is associated with death, with disability, with adverse effects such as hospitalization. When people are hospitalized with multi morbidity they're more likely to have complications of hospitalization and it also predicts institutionalization and it predicts increased resource use and it predicts a decreased quality of life. Now that being said there's a fair bit of heterogeneity in people with multi morbidity. So there's the same person is not the same person so there's a fair bit of differences between people with multi morbidity. One is there are differences in disease severity so class four heart failure is obviously different than class two heart failure. A very very high creatinine is obviously different than a moderately high creatinine. So illness severity is part of the heterogeneity. Another part of the heterogeneity is the functional status of the person and we'll come back to that in a little bit. As well there's a fair bit of heterogeneity in the prognosis of people with multi morbidity. Also a fair bit of differences in the values that the person with multi morbidity has. So there's a huge variability in personal priorities both in the person and their family in terms of aggressiveness of treatment and other decisions like that. And there's also quite a bit of heterogeneity in the risk of adverse events. And again the key point is that we need a flexible approach to care with multi morbidity. So we need to acknowledge the heterogeneity in people with multi morbidity in terms of both their health status and also in terms of their preferences and goals and life experience. So it's actually very important of a flexible approach not just between people but also people may change over time so it's important to make sure with people with multi morbidity that we follow them closely over time. So I know I said it's not really a new idea that multi morbidity is there but it is an extremely important notion and I think one of the uses of the notion of multi morbidity is that we do need to acknowledge the importance of subspecialists but also to realize that we have to treat the person and not the disease and the healthcare system needs to acknowledge that. And I think in acknowledging that we do need to move away from chronic disease management for people who have multiple chronic diseases and go to individual patient managed care rather than just simply putting people on an assembly line for heart failure and renal failure and osteoporosis and macular degeneration and leukemia. We need to acknowledge that there's heterogeneity in the person and that we need to realize that there's trade-offs and balances between the different systems and we need to move a bit away from the chronic disease management model if you've got eight or nine chronic diseases. The chronic disease model works very well of course for people with two or three, one or two or three diseases but when you start getting up into extreme multi morbidity you start getting into issues with the lady at the start where she's seeing 12 different people in the community. The other important point is that prognostically we need to consider the cumulative effect of these morbidities as they build up so it's you know obviously not great to heart failure but it's a lot worse to heart failure and renal failure and it's a lot worse to heart failure, renal failure and osteoarthritis and I think that's fairly obvious to most people but we do need to make sure that we consider it in care planning so that's sort of why it's important even if it's not a new idea. Now that begs the next question and we don't have a lot of time to talk about this but it is extremely important about how we measure multi morbidity and there are a host of different multi morbidity tools which we can talk about but we if we have time but the key point is that there's a host of tools and the reason there's a host of tools is there's a host of different measurement issues. The first issue is what's your data source so data sources can be very very generally speaking they can be administrative data so this would be like a provincial claims data. There are clinical data sets so these would be like electronic medical records. You can have self-reported diseases particularly in epidemiologic surveys and then you can have biomedical measures as well so measures of for instance you're a creatinine so you can have all of these different data sources sometimes you can use one or two or three different data sources but most commonly one would choose just one and immediately you can see some issues arising which will give different different multi morbidity indices depending upon what your data source are so administrative data tend to capture the whole population but it does mean that the person has had to have had to contact with the healthcare system so they have to have seen a family physician or other primary care provider or have been admitted to hospital to be captured and then it relies on the recording of the of the clinician entering into the billing claims data so administrative data are often under report compared to the clinical data set and perhaps self-report but immediately you can see that those different data sources have different implications for measurement which are extremely important so when you're looking at multi morbidity the first question you should ask is where did the data come from the second point is what's the time frame so a lot of them will use a period prevalence so this would be the number of people with a condition in a given time frame but of course some diseases vary so for instance your cholesterol if you're counting hypercholesterolemia as a disease in the multi morbidity index cholesterol is variable over time and at some points people can be dislippant have high cholesterol and then it can normalize so if you use a period prevalence of high cholesterol over the course of a year that's different than than a single point in time so that's actually quite important the next important issue that comes is what conditions you include in the multi morbidity measure often multi morbidity measures contain risk factors which technically aren't diseases so osteoporosis and hypertension would be good examples they are adverse and they are a problem but they're not a disease there are risk factors so some multi morbidity indices exclude risk factors and some include them the other issue is whether you count symptoms or somatic complaints so things like swollen ankles sometimes wind up in multi morbidity indices dentition problems that are self reported sometimes wind up in multi morbidity indices as well and that that can alter the index as well or the measure of multi morbidity the issue of double counting often comes up as well so if you have heart failure most people who have heart failure also have ischemic heart disease and also hypertension and other risk factors so you can run into the issue of some diseases being heavily counted in particular vascular diseases can be heavily counted compared to some of the other diseases because you're actually counting the risk factor and the disease and then the other issue is if you don't measure it you don't count it so there are important conditions that may not be measured in a multi morbidity index in particular rare diseases often don't show up in multi morbidity indices just because they're not highly prevalent then the next issue is whether you count for disease severity so obviously there are differences in disease severity that are important and then the last question that I'll come back to is really should we be decodamizing the measure so we want to say this person is multi morbidity and this person doesn't but most of the studies show that it's actually the total count that matters and should we be using the total tally of diseases rather than cutting it and saying these people have multi morbidity and these people don't but rather treating it as a continuous measure because having eight comorbid conditions is different than having seven which in turn is different than having six and five so that becomes a question both the american and the british geriatric society have suggested decodamizing it but i wonder if that's in fact a great idea now the other key point is that some of the depending how you define frailty of course and and that's a whole other talk that we I won't go into but there is a clear overlap between people with frailty people with disability and people with multi morbidity so there are people who are both frail and have multi morbidity particularly if you use some of the american definitions of frailty and disability and comorbidity overlap fairly substantially as well so there is a fair bit of symptom overlap when you look at clinical surveys that that is important now when you look this is the data from olgathu from from from Halifax there is some overlap but most people with the frailty phenotype actually have disability and comorbidity as well and very few are just frail so while there is some theoretical overlap if we just go back to the previous slide it's not quite like this diagram shows and it's more like this diagram shows that most people who are who are the frailty phenotype have have a significant overlap with disability and comorbidity now when you think about how people get to to multi morbidity this is just a gotten pasted from a review in chamda so underlying the whole process is is the aging process and the physiologic changes that occur with aging so we're as we go through life we're exposed to behaviors and then we're exposed to risk factors and then we acquire diseases diseases which are first subclinical so we have loss of reserve in in multiple organs so loss of reserve as we acquire low grade diseases in multiple systems which then progresses on to multi morbidity and then you see the the adverse effects of multi morbidity so the notion is that slowly steadily over we go as we go through life we acquire behaviors then risk factors then subclinical disease then overt clinical diseases in multiple domains and then we have adverse consequences of that so it's important when you see somebody to think of how they got there over the course of their life and a life course approach becomes actually quite important both at an individual and a policy level so one of the quite prominent papers from 2011 12 rather was a paper looking at the administrative data from scotland looking at essentially the descriptive epidemiology of multi morbidity in scotland and again it's not a new idea but this is quite a nice paper that does just a global description of the how multi morbidity progresses with with age so this is just how old the person is and the percentage of people in this in a large scottish data set and again the key point coming back to my point about what multi morbidity is and how we dichotomize it is as we get into our 80s the vast majority of people actually have multi morbidity so that does raise the question of of how we define multi morbidity and it also the key other key point is that it's very common in late life the other thing that they did was they looked at the how the conditions interrelate with one another and how many people have multi morbidity if they have heart failure and the number of chronic conditions that people have with the individual index case so this would be coming back to the notion of comorbidities so people with heart failure most people with heart failure have important comorbidities most people with stroke have important comorbidities conversely most people with asthma do not have a lot of multi morbidities so it's important to think of how diseases cluster together and also how they interact with one another when we're starting to think about multi morbidity because different disease clusters may cluster together and those different clusters may be important and then the other thing that again I don't think came as any surprise except for the magnitude of the effect which is quite quite striking is they looked at the this person's social positions or their socioeconomic status in relation to the development of multi morbidities and age and as you can see if you're in the in the lower deciles you're much more likely to multi morbidity than if you're in the higher deciles of income this was by the region of of residence but nevertheless it's not a bad surrogate for individual social position and you can see there's actually quite a quite a quite a strong difference particularly in mid-life and importantly and we'll come back to it it does tend to attenuate in in late life a little bit so this was data from all of scotland and this is the similar data looking at the association with mental health and with physical health by social position and the effect is consistent across social positions so this is another study from the UK this is carol brains study the CFAS study and again as we get older it's actually it's a bit of a complicated slide but it's actually quite nice and that the key point is that most people go into 65 with a chronic health problem but most people are functionally intact and cognitively intact at 65 so most of us enter our golden years with a chronic illness but functionally intact as we get older and once we get into our 85 years you know most people actually they're still functionally intact but the rates of ADL impairment go up fairly substantially particularly in in women so this goes along with the earlier notion that as we go through life we acquire health problems and then those lead to late life disability over over the course of the life so that's sort of the prevalence there's actually relatively few studies on the incidence of of multimorbidity so this was another British study just looking at the incidence of multimorbidity and again the key point I'd like to make here is that it's highly age dependent so this is incidence of the number of new people that acquire the condition so the number of new people with multimorbidity and again it's highly highly highly age dependent and those of you who kind of like actuarial science is it starting to look like a a gompert's curve where it's just starts taking off quite substantially with age not not linearly so the incidence does seem to well it is very highly age dependent and these are data from the CDC I catch it while it lasts I guess and again it there's a very strong age effect so as we get older we're much more likely to have multimorbidity but there's also a very strong effect of social position so this is the percentage of people below the poverty line and you can see that it's a fairly strong strong strong effective social position and again interestingly it does attenuate in late life and again if you look at what common diseases are this is one of Linda freed's papers the common disease combinations one is arthritis and visual impairment one is visual impairment in high blood pressure a third common combo is arthritis and high blood pressure so the generally speaking common things tend to accumulate commonly with each other so it's not too surprising at these these do these do tend to run together and the other thing that this paper found which is actually quite important is they found that there are a fair number of interactions and between some of these different diseases so in particular things like COPD interacted with a lot of the other things to produce more disability than some of the other conditions so they did start to look for interactions although that does get a little bit complicated so again there's a set number of papers now quite a few of them coming out showing all of the following one is an increased rate of death or reduced quality of life or reduced functional status and a higher risk of institutionalization and hospitalization as well so i'm just going to I guess be a bit parochial and talk about some of the manitoba data now so this is the data from the mantua follow-up study of airman and bob tate in the gerontologist and what they did is they they looked at six common illnesses and just looked at the number of common illnesses they did not find any interactions and they found that if you have no chronic illnesses the survival was obviously quite a bit better than if you had one chronic illness which is better than two chronic illnesses which was better than having three or more chronic illnesses and interestingly they did not find any any interactions but they did find this strong cumulative effect and again those of us who are in clinical practice will note that drug trials often have to change the change the the axis of the of the survival curves and blow them up to show a difference with most of these you don't need to do any messing around with the axis of the survival curve the effect is very large so i think this is just a it's fine to discover cures for chronic conditions are bread and butter so we looked at this in the manitoba study of health and aging so the manitoba study of health and aging was done in conjunction with the canadian study of health and aging so this was data from the mid-90s 1990s and what we did is we just did a simple disease tally and again this was not just diseases this included risk factors and a few subjective complaints like swollen ankles and joint pain and we just did a simple tally so there were a possible 16 conditions and we just tallied them up and the people with the lower tally so people with no chronic illnesses had a very good survival rate people with seven or more chronic illnesses had had a less good survival rate and as i don't think it's any particular surprise but again it's a fairly strong effect however the effect of functional status was much stronger so this was just the older american resource survey the or scale of functional status which can be categorized as good functional status mild impairment or moderate severe impairment and again functional status i don't think it comes as any surprise to people but it is a very strong predictor of poor outcome and actually it's much stronger than the effect of multi morbidity so it's actually i think important to keep in mind that that as well so when you look at the cox regression model this is that for predicting a hazard ratio of dying over the the five-year time interval so this goes from zero to sixteen so each additional home morbidity increases your risk by about ten percent just in layman's terms so the hazard ratio of mortality is about one one point one for each additional condition and that holds true when you adjust for age gender and education and age gender and other factors including the mmsc and depressive symptoms interesting though though once you put functional status in the model multi morbidity is not predictive of dying in this data set so what we thought was going on was that as we go very much consistent with the earlier idea that as we go through life we acquire multiple chronic conditions which progress and then we acquire disability and the disablement is is the step um on on to on to dying which would i think be consistent with these data um so it's not that multi morbidity doesn't predict dying it's that's probably moderated through the functional status there's another couple of interesting things here one is that cognition remains a very strong predictor of dying independent of the effect of functional status and comorbidity so we do need to consider people's cognition in in in in in prognostication so these are data we haven't published but i've just submitted to kag coming up and this goes back to the notion of what happens with functional status um and multi morbidity so this is the question was does does multi morbidity predict functional decline and this is the older american resource survey and this is kept as a continuous variable and this is just the number of health problems from 0 16 but there weren't a lot of people with more than 10 and you can see an effect um at time one and then time two five years later that even five years later multi morbidity predicts disability five years later um and then this shows up a lot better actually when you look at the categories of functional impairment so this comes back to excellent um functional status mild impairment and moderate uh to severe impairment and this is just cross sectional at time one so people with multi morbidity this is the number of health problems that they have and this is their functional status and you can see quite a strong effect um in the cross sectional analyses um of multi morbidity on functional status um again not particularly surprising what is quite surprising is even five years later there's a very strong effect so this is the number of problems at time one um and the more problems at time one um and again this is just a simple disease tally it's nothing fancy schmancy um you can see that this goes up fairly substantially um five years later um by the number of comorbid problems and i'm getting a message you can't see your pointer um so i guess you'll have to think of it in your in your mind i'm trying to point out here what's happening um now one of the issues is that what happens is as people just to be blunt die um they get taken out of the pool um so that may influence the prospective findings um so if you account for death you still see a very strong effect uh on functional status over time so that's sort of the background of things that we did and here's what we're now doing and bear in mind these are very preliminary data so there's nothing fine let remains fairly early analyses one is to describe the prevalence of multimorbidity in canada two is to determine if their gradients across social position um and i won't talk about disease clustering but it is something we've been interested in so we've been looking at prospective data from wave one just in a cross-sectional way and we were using the tracking cohort um so we are only using the data from the tracking cohort which is intended to be as representative as possible um of the canadian population and we've restricted the analyses to 45 to 85 there are a couple of people in the data set who are 44 and there's actually a handful of people who are over 85 just because of the sampling frame um but we're sticking to the 45 to 85 group um we're looking at about 20 20 odd thousand people um we used we excluded people with any missing data on any of the um multimorbidity items so we're looking at age gender and social position so in social position we used education individual income and household income i won't talk about it but we've also been quite interested just because of betty havens in manitoba in the income source that was one of her areas of interest was the income source as a predictor of outcome so we have been looking at that a little bit um the other fascinating point having worked at the manitoba study of health and aging where the median education was nine years um it's quite dramatically different um in the clsa than it was in the csha and msha so the educational attainment of older people has gone up quite dramatically in the last 20 odd years um as probably certainly in manitoba the one room school rooms um um had had quite a strong effect as they came into being so when we look at again i came i talked about the importance of how you measure multimorbidity and we chose a fairly straightforward measure um we simply looked at the list of conditions we chose ones that were physical health so we did exclude the mental health ones and that's just because one of my areas of interest is looking at the effect of physical health on on on depression and other mental health issues so that's an arbitrary decision we chose just the physical health ones we also arbitrarily chose to include risk factors so we did include osteoporosis and hypertension um and that was done intentionally as well i think certainly from a clinical standpoint it's important because the treatments of hypertension sorry the treatments of osteoarthritis can worsen hypertension so we were interested in the interactions in those areas so we did include risk factors as well as diseases and then we also included the the ones that are chronic in nature and these are all self reported by phone so these are the phone interviews and at the time we put in the submission the the clinical cohort was not available so what did we include we included osteoarthritis this was osteoarthritis of any joint we included RA other arthritis back problems osteoporosis TIA peripheral vascular disease stroke again we were double counting a little bit because TI and stroke are essentially different severities of the same disease um similarly MI angina heart disease diabetes hypertension COPD asthma renal problems hyperhypothyroid and so on so the the key point is that the common things are common and it's no surprise osteoarthritis is common back problems are common hypertension is very common and cataracts are common as our migraines so all we did was simply created a tally of these and that was arbitrary but any disease of multiple by any measure multi morbidity is going to be a bit arbitrary and we simply tallied them up um then we took this tally and then we looked at the cohort and we standardized against the Canadian population of 2011 which was the closest um census data that are available to clsa is between the 2006 and 2011 um censuses so we chose the the 2011 population pyramid of canada and we just did direct standardization to find out what the mean number of conditions was for the Canadian population between 45 to 85 and we came when you do that it's actually really quite creepy because it came out very very very close to the the one from scotland in 1964 which was kind of interesting so i think this does beg the question if we're using a cut point of two to three for multi morbidity that's essentially a median split so it's it does beg the question of whether or not that's the right number to dichotomize at if we're going to dichotomize that that's becomes an interesting question so the second thing we looked at was age gender and multi morbidity and again if you think back to the graph in scotland um from the lancet paper it's actually quite jarringly similar as well actually um from 2012 um so this is the mean number of chronic conditions and again it's quite linearly related um at least up to the age of 85 with age um and again women have about a half um about about a half more of a morbidity um than men a lot of that is osteoporosis and osteoarthritis when you look at dichotomizing it we chose three or more conditions just because otherwise everybody would would we went with two or more um everybody would have it um so you can see again quite a strong age effect and again it's not surprising but it's very impressive so the next thing that we were interested in is looking at income and this is the the the total income of the person which is measured by self report um and you can see quite a strong effect um of of income on multi morbidity so we have the mean number of conditions and it rises obviously with age but people in the low income bracket are much more likely to have multi morbidity than people in the high income bracket and the other important point to make is that we actually didn't have a lot of people in some of the cells who were over $150,000 a year in income so we did collapse 150 down to $100,000 a year in personal income um but the two effects one is the very strong effect of income on multi morbidity which which is quite quite strong um and then that does attenuate with age so it's much more apparent in younger people or middle age people I guess than it is in older people and you see a very similar effect with household income um it's a bit more variable um in terms of the measure because there aren't a lot of people with a very low household income so it looks a little bit irregular but it's a it is quite a strong effect um and again that effect does attenuate with age so the other thing we looked at was gender income and multi morbidity and this is the percentage of people with three or more conditions plotted against their personal income and again you can see quite a strong effect quite a very strong effect and the effect is actually somewhat stronger in in women than in men and again with household income a similar effect is observed and there's a quite a strong effect and again it's not surprising but it is a pretty big effect and again the effect is stronger in in women than in men a little bit um where the gender issues become a bit bigger is actually looking at education um and and the effect so the effect is across education is very strong in women and present in men but a bit less strong and so highly educated women you know have you know one and a half fewer comorbidities than people with a low educated women and the effect is much much less in men we can speculate why in the question session I guess now I'll say we interpret this with a bit of caution these are the main effects models only and of course when you have interactions with income the main effects don't really mean much because it does depend upon age so these odds ratios should be interpreted with a bit of caution however they're quite strong they're not surprising in terms of being there but I was quite taken aback by how strong the effect is and if you look people in the low income bracket so under 20 000 a year are about you know have twice as likely to have multimorbidity as those in in the in the high income bracket so that again that's quite a strong effect adjusted for age and gender and again the effect is quite consistent whether he's personal income or or household income now again when you look at the interaction terms the effect is present in people over 65 but it's a lot weaker than under 65 and of course the effect in people 45 to 65 is of course even stronger so when you look at the logistic regression for education those who did not complete high school had about 1.3 uh and odds of 1.4 ish of of of having multimorbidity versus those who who who had some postgraduate education so the effect of education is a little bit uh less than the effective income um and again just to reiterate there is a very strong interaction between age and income on multimorbidity and the effect is attenuated in the older age groups so in conclusion we found that social position was quite a strong effect uh and it was a gradient effect so it's not just very low income people it was a gradient across the income spectrum going from high to low and it was again quite a strong effect so our interpretation of just the social position things one is the measure of social position is a little bit problematic in late life when people retire they obviously change their income status um and that that can have effects secondly there's issues around gender and pension that are quite strong um in this particularly older age group now and then the last issue that becomes problematic in measuring somebody's social position in late life is their income versus their wealth so just the very measurement of social position is complicated as is as i mentioned the measure of multimorbidity is also complicated um the second issue is that when in life does social position matter and i guess the question comes back to this notion of a life course and it's probably consistent across the whole life course and certain people have um a higher allostatic load is the social epidemiologist would call it of marching through time in a lower social position is different than marching through time at a higher social position another complication of course is the fact that we're dealing with survivors so if people in low social position are more likely to die um and that differentially affects the people with multimorbidity who are young um there'll be survivor effects in late life and then the last complicating factor is there's also cohort period and age effects of wealth and education that have um the very very very difficult to disentangle um except without large prospective cohort studies like the clsa but that won't we won't know that until uh you know 15 or 20 years down the line what the age effect period effects and cohort effects on wealth and education are and then the last thing that we can't really measure is the notion of societal inequality um so we're talking about their individual income but we also have to think about the inequality of the society they're living in which may have effects as well so that was sort of our thinking and bear in mind again these are quite preliminary results so some of the limitations to our our our findings one is our definition of multimorbidity again it was like most measures of multimorbidity a little bit arbitrary in a simple disease tally we arbitrarily decided to exclude mental health and risk and include risk factors and also exclude I think understandably actually acute illness um nevertheless I think this is very consistent actually creepily consistent with the paper from Scotland um as well as a host of other papers which demonstrate an effective social position and again the fact that there's a lot of research doesn't negate the importance of this type of research but it does say it's quite quite consistent um and quite important um so the other conclusions I think one my own view is that we shouldn't be talking about multimorbidity as present or absent we should probably be talking about a multimorbidity index or multimorbidity score because it does seem to behave much more as a linear variable than a dichotomous one so you're missing a lot of information by simply arbitrarily doing a cut point um and related to that if you do choose a cut point you wind up choosing a cut point that essentially enables all older people as having multimorbidity which raises questions of what normal aging is um so my own view is we should think about it as a continuous variable rather than a dichotomous one um and then we can discuss that a bit if people want um so the implications um one is at a clinical level for clinicians like myself we need to review how we look after people um and I think we do need to individualize care we obviously need to be aware of guidelines and and disease management but we have to also be cognizant of when to to ignore the guidelines um we have to be very cognizant of drug interactions both drug disease interactions and uh and drug drug interactions and again a classic one would be hypertension and osteoarthritis you give non-steroidal anti-inflammatories for osteoarthritis and you worsen hypertension or worsen heart failure or worsen renal failure so it becomes very important to consider drug disease interactions um it also becomes important in prognostication for screening um and aggressiveness of care decisions we do need to consider how many diseases a person has when we start treating one very aggressively but related to that we actually do need to consider functional status and cognitive status and disease severity in in that prognostication because multimorbidity alone um doesn't capture some of the quite rich um prognostic uh information in functional status and cognitive status so I'll end with my personal thoughts um I I don't think you know people have been trying to get the ideal measure of multimorbidity and I I actually think that's quite difficult because of all of the measurement issues um so I think we do have to choose a multimorbidity measure and recognize the limitations of the measure we're using um and in particular we need to consider the data at hand so that makes it very difficult to compare prevalence of multimorbidity across uh jurisdictions and across different studies um so it does make replicability quite difficult um because each data set will have a different multimorbidity measure so I think we do need to recognize that as a limitation um and we need to be related to that we need to be very explicit about what our measure is so we need to say what our measure included and what the what what how we measured these diseases so is it by self-report um is on a telephone interview as we have or is it by a clinical data set or is it by biomedical data and those will deal very different findings um that said I think the really fascinating thing is just really how similar our our our our findings are to actually both the Scottish study of 1964 in primary health clinics as well as to the 2012 so I was quite struck by how quite quite creepily consistent a lot of these um a lot of these multimorbidity studies are particularly around age which I don't think comes as any surprise and social position so that was sort of my my final thought is that it does seem to be quite quite consistent across time and place um so anyway I thought that was my final thoughts and then I came up with oh Megan's doing the next one okay thank you Dr. St. John for your excellent presentation I really appreciate you being here with us um can you hear me now to my background line I can hear you so we'll go ahead and have a short question answer session now remember that you can type your questions into the into the chat window at the bottom of your screen um so one question is could some of the attenuation with income be because of the the condomization the three or more conditions it's possible that the low income group has you know lots of conditions five or six conditions and the high income group has you know just two or three do you you know you you said during the the talk that you were maybe kind of dive into that that education and income yeah attenuation a little bit more yeah so um no matter how even if you model it as except just for the people without who were listening you can just do a simple disease tally and count it from zero to 31 which would be ours um or you can dichotomize it and say multi morbidity is three or more or two or more wherever you decide to put the cut point so we actually um my own preference was to leave it as a continuous variable um and actually you see the interaction term even if you use it as a continuous variable um one of the what so so it's no matter how you slice up the data there is a you do attenuate late life um even just the raw disease count does seem to to to to the social effect does seem to narrow even if you use it as a continuous measure um the problem if you treat it as a dichotomous measure is it's so common um in the older age group that you start getting to as you mentioned mythologic problems of not having enough people in some of the cells actually um particularly sort of low income people without multi morbidity in late life are unusual um so my own preference is to leave it as a continuous variable even though it's less commonly done and we do see a strong interaction term um even with leaving it as a continuous variable so does that answer the question that makes sense yeah so we have a question from dr manish karn toronto what are the features of evaluating a mortality index especially with regards to its predictive power as well how do we incorporate the synergistic interactive effects of comorbidities you might have like addressed that during some of the the further talk but do you want to right so there yeah i guess i can try and my my answer is predicated with what i think i i i you know i don't think we'll ever have a perfect measure of multi morbidity um and we certainly won't have one that will be able to be used in all settings at all times because um if you look at particularly using administrative data um you are relying upon capture one of primary care contacts so people with multi morbidity may well not be seeking care so they'll just count as zero um then that's different than if you survey which is of course different if you measure survey and biomedical data um so the ideal one would be to incorporate all of the above so administrative data and clinical data and self-reported data and biomedical data and the ideal one would measure interactions but the ideal is the enemy of the of the possible i guess and i so um um my own feeling is that you need to choose a multi morbidity index based upon the data that you have available um and then just acknowledge that that's a big limitation um um so we're actually running out of time so i'm going to pick pick one more from this good list of questions but um so can you talk to kind of the changes and morbidity in the last 50 years and how medical advances change kind of the i'm just reading it now that's funny um yeah well i you know i clearly health has gotten better in the last 50 years i think that that that that that's so so i think part of it might have been measurement issues across time um um and and to scotland is generally speaking a less healthy place than than than canada so i don't want to sound like nothing's changed and we're not progressing because things things are progressing some of the different tally might be because people are getting healthier but we've invented new diseases so for instance osteoporosis was not in like it just didn't exist as a as a as a measurable disease so one of the most common diseases just wouldn't have been captured so wouldn't have been counted in that three three diseases in this in the scottish paper from 1964 um similarly hypertension has changed as well um so um there will be a lot more people now with hypertension because we've lowered the bar for diagnosing it compared to 1964 um so i don't want to sound like nothing's changed and nothing's gotten better health has gotten better um the and and two we still need this research because i think we still have structured care to deal with single system disease and we still have structured um individual care to deal with single system disease rather than multi system disease so i don't want to say that nothing is gotten better because things have gotten better nor do i want to say that research isn't important because it's all been done before because it has i think rather the the message i'd like to have is that this is very important um but it's not novel this type of research builds upon uh a previous uh history and needs to be considered in that context rather than as something novel and new because we have this preoccupation with novelty and innovation um rather than kind of slow steady progress which i think is a little bit um misleading um so i think there has been slow steady progress on multimodality both clinically and in research but it's been slow steady progress not some kind of major major uh uh innovation and also the aging society brings these these issues to the forefront right right i think that's another good well thank you very much dr st john it was an excellent presentation and we just run out of time for the question so um i'd like to thank you and i'd also like to tell everybody that our next webinar is scheduled for thursday june 29th at one p.m eastern time dr megan o'Connell from the university of suscatchewald by speaking on factorial invariance of the center for epidemiological studies depression scale or c e s d please register and join us for next month's webinar registrations will begin um for the next uh webinar soon but also like to remind everyone that cls clsa data access request applications are ongoing the next deadline for applications is on june 12 2017 please visit the clsa website under data access to review available data for their information and details about the application process to gain access to the data sets um thank you again everybody and particularly thank you very much again dr st john thank you