Welcome everyone. My name is Janine Bennett and I'm the Assistant Secretary of Digital, Business and Sector Engagement at the Department of Health,
Disability and Ageing. Today I'm very pleased to be speaking with Professor John Hirdes, a leading voice in the global conversation on ageing health systems and the transformative role played by data in care delivery.
A self described freak of nurture, Professor Hirdes brings a unique blend of curiosity, compassion and scientific rigour to his work. He's a faculty member at the University of Waterloo's School of Public Health
Sciences and has authored over 350 articles in peer reviewed journals and academic literature.
His leadership in the interRAI Consortium has helped implement standardised assessment tools that have improved care for people across Canada and around the world. In 2023, he was appointed a Member of the Order of Canada in recognition of his pioneering contributions to health screening and quality of life assessments for older adults. Welcome John. Thank you so much for joining us from Canada today.
Hey, thank you for having me. Pleasure Your presence here today couldn't be more timely for us here in Australia, we're witnessing first hand the role of data and digital in shaping the future of aged care and healthcare more broadly.
We're introducing measures that include enhanced rights and protections for older people, stronger rules and regulations for providers, higher quality care, easier access and entry to services, and as well as greater choice and transparency and a stronger workforce in the aged care sector. All of these things are enablers for a better aged care system, but they're also enabled and modernised by our digital infrastructure.
So we're very much looking forward to hearing your insights both from your own work and also from the Canadian perspective and to learn how data and digital can continue to drive really meaningful change in aged care and healthcare more broadly.
I understand you've prepared a short presentation for our audience, so I'll give you the floor and and let you go.
Thanks John. Well, thank you again for having me. It was a pleasure to be in Australia recently and to make connections with folks like you and your team. We have a lot of common areas of interest, so I'm delighted to have this opportunity to explore this a bit further today. So what I'd like to do is to first of all, talk about why Australia and Canada should be natural partners, collaborating and trying to understand how to manage some of the challenges dealing with population ageing and the needs of persons with disabilities and other complex populations. I think there's a lot of affinities between our two countries that give us an opportunity to work together in a very productive way. I'll then move on to talk about data standards and artificial intelligence that's emerging in healthcare system. And I'll talk a bit about the use of interRAI systems in Canada.
And I'll give you one example of how we've used big data and advanced analytics to refine the way the Canadian healthcare system works.
So if we think about our two countries, we're comparable in many different ways. So if you look at life expectancy, Australia has a little bit longer male and female life expectancy than Canada does, but basically the same kind of breakdown.
If we were to look at the populations we serve, Canada has a population of around 40 million individuals, Australia 27,000,000 individuals. About 80% of Canadians are live in urban setting. There's more folks that live in urban settings in the in Australia we have about 1. 8 million Indigenous individuals compared to . 8 million Indigenous individuals in in Australia.
We have a similar population age structure at around 19% in Canada, 17% in Australia. Both are countries with very large land masses and sparsely populated areas of the population density is just a little bit greater in in Canada.
But really we, you know, have very large countries with people concentrated in a few urban areas.
In terms of wealth of the GDP per capita is a bit higher in Australia.
And the one thing that's still different, you have a lot more dangerous animals. Are all your animals want to kill you?
You just have to worry about polar bears and maybe grizzly bears in Canada. So there's an awful lot of affinities between the two countries that make us very comparable. I want to take you back to when I was a graduate student. My PhD supervisor was Doctor Bill Forbes.
He was the founding president of the Canadian Association of Gerontology. He had two PhDs.
One of them was in statistics and he was the country's leading researcher on long term care for the elderly in 1986. And so he wrote this book on the future of institutionalisation of the elderly in Canada. And he noted a few things. He said there was no national data beyond age and sex for long term care.
And he said we really should have a standardised assessment system in place. But he was worried in 1986 because he said well you'll have to have computers to do that then nobody could imagine computers coming into long term care. Of course that's a very different world from where we are today.
But the lack of having standardised data available meant that there was also a very limited conceptualization of quality measurement.
Really what he talked about was survey based research methods to look at resident experience, but not really clinical measures of the quality of care being provided. He was also very worried about how good the evidence was for place people into long term from the community and he worried that if we didn't do things to improve the quality of long term care or nursing homes then they would really be considered. The iron lungs of gerontology was a phrase that he coined to describe that.
And so in that 150 page book by Canada's leading quantitative researcher on care of the elderly in 1986, this is the only data table he had in there.
One table with a description of ADL impairment on about 900 individuals in one region of 1 province. That came from a study. That was it. We had no more data than than that to work with.
Well, he had a sign on his door. He loved gathering data sets from statistics candidate and the sign on his door said data, data everywhere, but not a thought to think that quote came from the guy who came up with the term counterculture.
And what was behind it is that you can't just use data to transform a system like a healthcare system.
But really what he was arguing is you have to have big ideas that you can then test using sound analytic methods to drive change.
So data on their own aren't enough. You have to have solid ideas, policy ideas that you want to test out and a solid analytic framework for how you test those data.
So that brings us to interRAI that I became involved in in the mid 1990s. interRAI is a not for profit charitable network of over 150 researchers and clinicians and policy experts from around the world. We've been doing this type of research for over 30 years and what we do is to develop comprehensive assessment and screening tools to look at the strengths and preferences and needs of vulnerable populations of all ages.
We're interested in people with complex needs rather than people with fairly simple, straightforward health needs.
We do this through a multinational collaborative programme of research where we create these new instruments and test them out.
And we're really interested in how do you take data and transform it into evidence that can inform decision making at all levels of healthcare system.
And we're particularly interested in supporting collaborative communities of practise and learning health systems where we use data to drive improvements in in, in the quality of of care across the healthcare setting. These are the countries that are involved in the interRAI network.
There's over 60 countries now that we're collaborating with. The countries with blue text are ones where we have an interRAI fellow appointed in that country. So Tracy Commons and Glenn Grey are two of the senior people in in Australia that we collaborate with very closely. And in green text are the countries that we don't currently have a fellow appointed, but we're very actively engaged in research with those countries. And you can see that recently we've become very heavily involved in low and middle income nations where health, human resources and and healthcare budgets are very limited.
And we've come up with some creative solutions to deal with the needs of older persons and persons with mental health concerns in those countries.
What's exciting with this international network is you have different cultures, different structures of healthcare system, different population age structures, different models of service delivery. And learning from those differences provides us with natural policy
experiments that we can learn from to improve our own health system.
Now, it's worth thinking about why we need to think at the system level in healthcare in an integrated manner.
And it's a fairly simple answer. It's that people who have comparable needs receive services in different parts of the healthcare system. So that's especially true for people who are older persons, people with mental illness or brain injury or physical disabilities, children or youth who have complex medical conditions or mental health aides, or people who are at the end of life. So not all elderly people live in nursing home, not all people with mental health concerns living in psychiatric settings, not everybody who is at the end of their life is in a Hospice setting.
In fact, those persons are in contact with all parts of the healthcare system and so you really need an integrated approach to providing them support. And so these there's also many conditions that start early on in the life course
and then you have a lifetime history of having to deal with things like disability or mental illness.
From a system point of view, what this means is that we should be thinking about what's the best place for people with different types of needs or different intensities of needs to get services. And from the person's point of view, we need to deal with the challenge that, you know, one person has to tell their story 100 times and may get discontinuous care as they move from one setting to another.
And so continuity of care becomes very important for them and having the healthcare system work in a team based manner to support or the care is more important than a silo disaggregated approach to care. Within the intro instruments, we have instruments that cover everything from newborns all the way to centenarian. So we have instruments for children and youth, for general adult population and for older adults, getting Home Care Services, nursing home services, palliative care and primary care services in Canada.
This map shows the implementation of interi instruments in my country.
Now it's, we could talk about the Canadian healthcare system, but the Canadian healthcare system is really 13 different systems and we might even be 14 depending on how you want to count it.
Because each province and territory is responsible for the management and delivery of health services.
The federal government does transfer payments and the federal government manages some aspects of, you know, armed forces, RCMP, the, the police service and federal correction services. But otherwise, these 13 provinces and territories run their own system. So, you know, there's not one Canadian healthcare system, there's actually 14 of them.
And so comparability and data standards are very important in my country. And so it's been exciting that most of these provinces and territories, all except for Quebec, have or plan to implement one or more intro instruments.
In fact, as of last December, we had over 25,000, 000 intro assessments completed on over 6, 000,000 unique Canadians.
So this is the de facto national standard for aged care in Canada and the data are managed nationally by the Canadian Institute for Health Information.
One of the things we can do with that is for nursing homes, we can report at the nursing home level on the quality of care using person level
data in 1500 nursing homes across the country. So from Australia you can go on to the Kai High or Canadian Institute for Health
Information website and look up the name of a nursing home in a city that where you know somebody.
And you can get information on antipsychotic use, on restraint use, on falls, on worsening of depression and worsening of functional ability and and pain as examples. And it's all based on person level risk adjusted data to allow for fair comparisons. That brings us to the advent of artificial intelligence.
Up until 2022, I thought, well, you know, analysing these data is the big challenge and you're moving ahead with with data standards is important. Now we've entered this, you know, brave new world of artificial intelligence that gives us all kinds of new opportunities, but also new challenges in how we think about decision making in the healthcare system. I tend to not want to use the terms data driven decision making or evidence based decision making because those imply first that data speak for themselves. They don't. You need an analytic framework.
You need to run analysis of data to answer questions. And very quickly we learned that decisions in healthcare aren't only based on evidence on one topic where there's exhaustive evidence to say this is the one thing you must do in relation to that one problem. In fact, an evidence informed approach is better where you're looking at evidence, but you're also taking into account the values of the, the person providing services, the value framework for society. And so decision making is much more complex because it tends to be a combination of empirically based evidence and then objective value frameworks that are prevalent in the society. And the question is, how does AI fit into that kind of a framework? It was, it was a simple matter of if I, if I have X, what is the Y that I should do in turn? In fact, it turns out if you have one condition, there may be multiple solutions for how you respond to that condition or multiple outcomes that you may need to be concerned about.
And we need to think about how AI systems can relate to that. So we have the potential to improve our capacity to make good decisions, but we have to think about what safeguards we put in place for AI and healthcare as well.
I tend to think about 3 different types of artificial intelligence, but these are very blurred categories.
There's lots of overlap between these. So on the far left column are predictive analytics.
These are all the things that are sort of fancy analytic tools to basically take the quantitative statistical methods that we had in the 1990s and 2000s and now transition to more machine learning iterative approaches using a variety of different techniques.
I'm not so concerned about that because that's where I see real opportunity to make new discovery and make better use of data to answer more sophisticated questions. But there are some interesting challenges in the generative AI category. So with generative AI, we're talking about artificial intelligence systems that generate stuff that could be text, it could be images or videos or audio files or animations, computer software code or synthetic data. There's tremendous potential use for this.
So we're all familiar with using chat bots. When we go on to an Amazon website or an online service where chat bot says how can I help you? Those have become much more sophisticated and have some really interesting potential healthcare applications. We can use it to automate, to edit and optimise and translate text or computer code. You can use it to write an organisational strategy. If you need to have a brief written for the minister's office or a policy brief for your board of directors, there's a way to get AI to do that in a pretty constant way.
You can use it to create educational content for patients to explain to them in their language and their reading level what is behind a particular health condition that they're they're they may be affected by. And you can also use it to take very complex documents and summarise them into a simpler, more straightforward form. So there's huge potential for generative AI, but there's also risk because where there's error or bias introduced, then that information can become problematic.
I'm more concerned about new emerging technologies around agentic AI, where you have an autonomous AI system that doesn't have a lot of human oversight, that makes independent decisions. It can adapt to change, it can be goal oriented, it can manage complex tasks, it can learn from experience. But that's one where if humans are out of the decision making loop, we could very quickly go off the rails where the AI may make decisions that don't match the value preferences of the individuals in in question.
Now an interesting question is whether AI can solve problems with bad data. As a starting point, we need to recognise that all data in healthcare are subject to measurement error. Everything has error attached to it. Some has more than error than than other sources of data, and we can use AI and conventional statistical techniques to adjust for those errors to deal with with problems around measurement, calibration or data completeness. We can clean data sets to get rid of problematic observations, and we can in fact make inferences from some types of bias data. But what AI cannot do is overcome, you know, fundamental problems associated with narrative notes. If all the electronic health records says is had a good day, ate well, slept well, they're not gonna be able to make inferences that are meaningful from that kind of stuff.
So, you know, really bad data, you can't perform miracles with that. And so good quality standardised data is a foundational building block for any system. And AI can do much better when you have good quality data in that foundation.
A really interesting problem that's emerging for us or challenge is the emergence of many new types of data.
So with new technology, we can now get much more diverse kinds of data for predicting health outcomes.
And we have much more massive volumes of data that are coming our way. So for example, it's imaginable today that a a resident in a long term care home could have a clinical record including both narrative notes and an intra assessment combined with sensor data about movement in the home. Or they could have a wearable wrist worn device, for example, for fall detection combined with genetic data about that person, geospatial data about where the home is and where they came from. And then there's a new type of technology called neurotech that looks at the electron electric signals of the body to get other inferences about what's going on there.
All those data are are potentially available today, but there's lots of complexities to deal with with that.
First of all, as an analyst, I would say it's horrendously complex to think about how I would integrate all those data into one approach to analysis of very challenging work to do. Some sensor data is better than others, and some may be valid, some may have questionable validity. There's challenges in how you link all these data together because they made all all be in one place. If we thought we had privacy concerns five years ago, the presence of these kinds of data magnifies privacy issues a hundredfold.
And some of these measures can be very intrusive. That's one of the really serious concerns about some of the new neurotech measures where they can in fact infringe on human rights as well. So there's great opportunities with these new forms of data, but there's a lot more complexity that we have to deal with in thinking about it from an ethics point of view and from a a pure analytic point of view.
One of the things I'm very interested in is when we create decision support systems is we need to think about whose interest those systems will serve in
healthcare system. And so with any question that that comes up in healthcare, we have to recognise that the decisions can vary depending on whose perspective is being taken. And there are lots of competing interests within healthcare.
We have to decide whose interests take priority whenever we're making a decision.
So in Canada, we're concerned about delayed discharge and acute care. We call that alternate level of care patients because it affects the flow through the hospital system. If you think about a patient who has been declared as having a delayed discharge or is an ALC patient, the patient is interested in getting good care in the hospital, getting better and going home.
The family might be concerned about, well, how am I going to take care of my relative when they get go home? And what about my job and then my kids and my volunteer responsibilities?
I have the hospital may be concerned about reducing their length of stay cuz I could publicly reported they're worried about the flow of patients through their hospital and there may be reputational risk or penalties that they face if their length of stay become too problematic.
And from the society's point of view, well, we want to minimise tax burden on citizens and we want to be sure that people have access to hospitals. We can create AI systems to target each of these outcomes separately, but it's very hard to imagine one that does it all. And we have to think about, well, which of these outcomes get favoured.
We focus on the patient outcomes or the societal outcomes because we may have different answers for the same situation for the same individual. So I'm gonna end with showing you one example of how I've used advanced analytics to create a decision support tool to manage the allocation of home care resources and placement into nursing homes.
We created something called the method for Assigning Priority Levels or the Maple algorithm. There's no Canadian bias there.
Just so you know, just cuz it's Maple doesn't mean there's a bias there. What we're trying to do is to prioritise who gets access to Home Care Services or
who gets placed into a long term care home first. So it's not an eligibility or an automated placement system,
it's really a prioritisation system to say these people have the higher needs. And the way we framed the analysis was on three basic outcomes that we modelled.
We looked at who was actually placed in a nursing home in real data. We looked at caregiver distress rates because we realised, well,
caregivers provide 70% of the care to older persons. And so if caregivers cannot continue, the whole home care system will fall apart. And then we also modelled where people were rated as being better off elsewhere than a different place than the place where they're living today.
So we model 3 different outcomes for this. And this was our original Maple algorithm and looked at physical disability, behaviour disturbance, cognition, falls, wandering, nutritional issues and issues around medication management, pressure injuries and their environmental home. And we could classify people in the five different priority groups.
So somebody with disability and behavioural issues or disability combined with cognitive impairment and falls would be a very high priority for more Home
Care Services or admission to long term care if that was the appropriate target for them.
When we did the original research was just as we were starting to pilot test interRAI system.
So we have 448 hundred clients in 14 different home care regions.
Then we have validation samples from different provinces in different countries to create that decision tree that I just showed you.
We are now updating the Maple algorithm and now we have 4. 4 million intra assessments in Canada.
We have data from many other countries and we have a longer time period to work with.
So that tree that we had before is now elaborated into this much more complicated tree that takes into account many more variables and divides people into seven different groups of resource intensity. If we look at those seven groups, it's very strongly related to caregiver distress. So you can see in Group number one, 10% have distressed caregivers, whereas in Group #7 almost 80% have distressed caregivers. If we look at being rated as better off elsewhere in Group one, it's around 10% in Group 7. That's half of that population. And if we look at 90 day admissions to nursing homes, you go from below 1% in Group 1 to around 15% within the next 90 days in the in the highest group. So this new algorithm is a bit more granular and it's very strongly related to the becomes of interest. But here's what happened in our healthcare system over time.
We started using the old Maple algorithm in 2005 and what you can see in the light green bar at the bottom. When we started, around 25% of home care clients were in that lowest lightest care population and less than 5% were in the dark purple,which that category didn't exist. So we were serving a fairly light care population.
Almost 40% were the bottom 2 categories. After almost 20 years of using the system to prioritise who gets access to Home
Care Services, the dark green group went from 25% to below 5% and the dark purple group went from below 5% to now 25%.
So we've shifted the home care population to serve a much more complex population than we used to serve. Some of those lighter care clients are now served in other less intensive care settings, but home care itself has become much more focused on the most complex patients.
That's a good thing because you'd want your system to focus on persons with higher highest needs.
But the unexpected consequences, the effect on caregivers. So this is from a Kai High report on caregiver distress.
In my province, we now have over 40% of home care clients have distressed caregivers in Yukon territories in the far northwest, that's 47%. But in some provinces like Alberta, it's only 15%. So within our country, we have tremendous variation in caregiver distress rates and now we need to look at what's underlying that and we can risk adjust those caregiver distress rates using that Maple algorithm.
So that's, that's it. It's a bit of a whirlwind tour, but Janine happy to stop the the screen sharing and and have a chat That sounds great. And how fascinating, really, really interesting to see how those Maple scores can so accurately demonstrate how data changes real world's approaches, decisions and outcomes for people. So yeah, really fascinating. And always those unintended consequences are interesting to unpick as well, aren't they? So you talked quite a bit about the integration of the care sector in Canada.
Can you give us a bit of an overview of Canada's efforts to improve integrated care? Yeah. So when we started all this, we really had different silos of the healthcare system. So all the most sectors of the healthcare system had little or no data and there were no data standards there. So what having data standards introduced across different sectors is allowing us to do is to have more comparable information about people across settings, so we can ask better questions about who gets care in what settings.
And we can share information between home care and nursing home so that as people go from home care to a nursing home setting, their information can follow them and you can have that longitudinal view. But because we went from a fragmented system to start with,
we're in the process of making it better integrated. If I look at the Belgian approach to what they're doing, they have started right from beginning with one common portal where all interRAI data set sit and everybody sees those data at the same time. So if I'm in a nursing home, I can see assessments that were done in home care or in palliative care or in acute care at the same time. Whereas in the Canadian context, because we implemented these separately, over time, we're still in that process of trying to get all the kids in the playground to share, you know, to share data from one sector to another. But it's becoming more integrated.
We have something called Ontario Health Teams, which are intended to be integrated teams linking all the sectors together, but they're still dealing with some fragmentation of data systems. We have intro systems in place in home care, in nursing homes, in palliative care and mental health services. But acute hospitals themselves don't use interRAI assessments yet.
Home care will do assessments of acute hospital patients, but they use those data for home care purposes.
So we're still on a journey trying to make things more integrated. Yeah, it's a complex problem to solve for, I know.
In terms of your integrated care approach in Canada, what are some of the key lessons that you've learned?
And I suppose adversely, I'm interested in hearing a little bit more about those really big challenges.
Yeah. So having comparable data is the first thing you need to change the system.
So usually when you turn the lights on the first time you see data, you get surprised by things you didn't expect to see.
So one of the very common things that we saw in the nursing home sector was very high rates of physical restraint use until people saw the data and started to drive that that down.
Similar kind of thing around antipsychotic use. We also saw lots of people that were placed prematurely in nursing homes and people in home care settings who had fairly light care needs. So that we that evidence we used to try to target things to people who had more complex needs that were taking that were not being served as well and took longer time to get them services. What we didn't do as well was just think about the people that were getting those services that still had needs. So having low needs doesn't mean you have no needs.
So some of those people we took off service, but they still had needs. And we didn't adjust the reimbursement levels to the sectors now dealing with a much more complex population. So you saw home care today is much more complex than it was 20 years ago, but the home care budget change hasn't really matched that. And so what's happened is more burden has gone on to on the families.
And you know, it's, it's a, maybe it's an unintended consequence or it's a consequence that nobody thought of because we weren't asking the question in the 1st place. One of the other things that's interesting is because we started with siloed healthcare systems that we're not trying to integrate.
There's a challenge with getting people to share data. Sometimes data become a source of power.
So if I have data, well, why should I share with you? And so getting people to understand the data are really about the person, and it's the person's right to have everybody in their circle of care have the information that's needed about them to provide them good care.
The data don't belong to the provider. The other thing is that we don't do a very good job yet of sharing information with patients and their family. So we very quickly go to privacy.
Barry's to say, oh, well, I have to be careful about privacy concerns. Well, if they're my data, why is there a privacy concern for me accessing my data? There's a tendency, though, to, you know, want to keep this stuff secret.
And, and when we do share information, it's often not done in a way that is accessible, sometimes too technical. You know, if you've looked at a, you know, a laboratory result for blood work, you know, at least you get information, say what's an out of range value? And you can Google it to see, you know, what does it mean when I have high haemoglobin or whatever? We don't really have those systems in place. And clinicians aren't used to talking with nursing home residents about their assessment results because they didn't have to in the past. So that's a culture change that has to happen for sure.
And then it's a matter of getting everybody used to understanding the language. So one of the most difficult groups to deal with are geriatricians who went to university and we're taught the mini mental state exam, the geriatric Depression scale, and the Bartell ADL scale.
They learned that 20 years ago, and they don't want to learn anything new. They still want to have those things, but those aren't part of this integrated system. So there's an educational challenge that you have to deal with to get everybody used to understanding the comparable measures as well.
In Australia, government LED digital transformation has brought significant change to aged care organisations.
I'm interested to seek your thoughts on what opportunities this might create.
So when we're modernising digitally, we're having access to enhanced data sharing, big data analytics, data standards, artificial intelligence.
What sorts of opportunities should we be thinking about as we're on this digital transformation journey?
For me, the biggest opportunity is around quality of care and resident experience.
So I would like to frame any decision making system in terms of the issue for healthcare improvement, the IHI quintupling framework of can I improve outcomes of, of care, Can I improve the experience of care, the work life in there, can I improve equity and can I improve cost effectiveness.
And so I think we can use better data to evaluate the extent to which we're making progress in those areas and we can do more sophisticated analytics to ensure that we understand the underlying mechanisms that can cause variations of quality and yield better health outcomes. I think when we have better data systems in place, we can understand complicated populations better and more effectively respond to their needs. That's really the future of healthcare. Every country around the world has limited budgets.
We can't spend unending amounts of money on healthcare. And so we're trying to do the best we can with limited resources, whatever the limit is set by the by the country. And so when you have these more complex populations that you're serving, you have less room for error. And so you really have to get it right in terms of what's what services are being provided. And you need these more advanced data systems and good quality data to to get there. My example with Maple shows that if you don't pay attention, you're going to run into real trouble with caregivers who could end up in hospital themselves. Yeah, I was really interested, John, in your presentation talking about it's not just about the data. You have to think about the principles, the outcomes you're actually looking to achieve. And it makes me think about the the value proposition and the role of ethics in working with data. How did that play into your development of Maple?
Well, So what we were, what we did with Maple is we talked to different stakeholder groups to say what's important about home care.
And, and part of was understanding the role of home care in the overall health care system and also understanding who actually provides care in that settings and trying to get at what are the things that make home care work and what are the potential failure points.
And I think by concentrating on caregiver outcomes, we've also started to make the point that home care isn't really about a dyad of the service provider and the patient. You also have to take the family into account because the family's got to be part of the unit of intervention. In New Zealand, there was a Maori Health review done last year and they actually expanded that argument further to say that the entire fawn of the, the community around the person has to be considered. So you have to think about the individual and the support networks that they're embedded in, in order to provide them support. Because if 3/4 or 70% of the care comes from them, you got to be sure that the caregivers have the capacity to do what they need to do.
So you don't have to spend that money, you know. And so if I look at home care expenditures in in Canada, you know, $150.00 a week is a bet my home care client gets that's less than one day of staying in a nursing home. And so there's a sector that things are really inexpensive, but it's because 70% of the the labour cost comes from family. And so you've got to do things that really understand what drives caregiver distress, but you also then have to respond to caregiver distress.
Yeah, really good point. A bit of a change of tack. So you talked a bit about the data standards across international jurisdictions and how that's allowed comparisons to be made across countries that give us some really interesting insights.
It was particularly interesting to see the reach of Inter Eye and how well represented you are across the world.
Could you talk about some of the benefits we've seen from the standardisation that's been driven in Inter Eye?
Yeah, the the opportunity with cross national data standards is that you get to learn things about yourself that you didn't know. So if all you do is look at yourself in the mirror, you you come kind of used to it and think, OK, well, that's that's the way things are. But when you see other experiences, you can learn from that to evaluate your own performance. So for me the the most important example was early on in my career looking at physical restraint use. So this was the mid 1990s. I did my first study with intro systems in Canada in Toronto post acute hospitals.
They were called Complex Continuum Care hospitals, which basically took a medically unstable nursing home resident population.
And we did a study with Inter I measures and the accreditation surveyors in my hospital said your rates of physical restraint use were too high.
So I looked at my hospital's data and I looked at other hospitals in the Toronto area and I found the rates were all the same.
So I presented those to our surveyors and say here's our rate, it's 40% and that's what everybody else's rate is.
And they said, well, you don't want to just do what everybody else does, you want to get better than than everybody else. So just having normal performance isn't good enough.
I was invited to join interRAI that summer and I went to my first meeting where they were working on a special issue of age and ageing and they show data on restraint use in four or five other countries. Remember, my rate in in Toronto was 40%.
The rates in Nordic countries was below 10% and the rate in the United States was 15%, one 5%. So all these other countries had much lower rates of restraint use than than
Canada does. And Canadians really get annoyed when the healthcare system in the US looks better than the Canadian system. So it started a whole dialogue and it helped to inform a political discussion to say we can't continue on with this, we have to change this. The national rate now was below 5%, so we went from 40% to 5%.
But when I first saw the data, I thought, what's wrong with 40%? Everybody's got the same rate and it's only by looking at other countries we realised that we were an outlier. Flip side of this is the Canadian guilt complex with Nordic country.
So we think we're nice people, but we always think we're not quite as nice as Swedes and Norwegians and Danes and, and Icelandic folks and Fins.
And so we often will have studies or news reports to say we need to provide the same kind of Home Care Services that they do in those jurisdictions.
And you know, because in in Denmark apparently you can get 6 home care visits a day to keep you in in home.
But when we look at the the data from Nordic countries using the Maple algorithm, we realise they serve a much lighter care population than we do. Our compared is more like the Netherlands and the UK and New Zealand, but in Italy and France they serve a much heavier population. So there's no point in feeling guilty comparing ourselves to the Nordic countries because they're not serving the same population we are. We can ask the question from a policy framework, should we target that life care population, but they're they're not our peer group. The peer group should be the the UK and the Netherlands.
I like the idea of us getting competitive about healthcare outcomes, just not, not only soccer and the football. 1 cricket. Yeah, that's great. You mentioned some compelling applications of interRAI across Canada.
In Australia, we have been working across government on a programme called SPOT, which is advancing national healthcare data exchange and it uses the FIRE standard to do that, given the diversity of standards globally.
So there's FIRE, there's interRAI, there's many others. How would you approach integrating healthcare healthcare data across countries and what sorts of challenges and opportunities should we be thinking about?
Yeah, so the Canadian Institute for Health Information has done a great deal of work of applying FIRE and HL7 standards to inter eye data to allow for that interoperability within the electronic medical record. And so that can definitely be done.
What inter eye's done is to take things a bit further so you know, diagnostic codes, you know, the diagnosis of diagnosis to to some extent. And so having the comparable code about how a diagnosis is handled in computer system makes sense. But we create standards around things like how you measure pain or how you measure depression or pressure injury that's consistent across settings and across country.
So the clinical definitions of, of the concepts that we're looking at is a more detailed and, and more standardised than what you would have with with just fire and, and HL 7 alone.
So, you know, a challenge would be if you were to measure physical disability in terms of, you know, can the person lift themselves independently off of a chair to a standing position versus does the person, you know, a granular measure of, you know, how much help they need is different from do they do it yes or no on a tick box.
So we have data standards that go, that are very fully elaborated and, and defining what we're talking about.
We apply observational time periods. We are, we make a, we do a lot of work to ensure we get good reliability validity with it. So we can use the fire standards for sharing the data.
But the intra, what we call the I code standards is a conventional, A naming convention that we use for the items. So that if we measure pain, we measure it the same way across 13 different kinds of care settings, rather than trying to map it from very disparate settings to say, wow, this is probably about the same. Yeah, you spoke a bit about the quality of data and that presenting a challenge, but also not being a showstopper in terms of extracting some value from the data.
I was wondering if there are or is there an intent to sort of set standards around data quality where we might grade the value and the quality of sets of data?
Yeah, we do a great deal of work with an inter I around reliability and validity of of the data systems in the 1st place.
So we do lots of international testing to see if two people independently do the same assessment, do they come up with the same answer.
And we could do that in research studies and we can show that yes, we could good inter rate a reliability, which is important because you want one clinician to believe that another clinician would get the same answers they would.
We also do lots of validity testing to look at. Do you are you really measuring what you think you're measuring by looking at predicting outcomes or looking at gold standard measures? But the challenge is it's not good enough just to have good research data from a research study done once and then ten years later, nobody's replicated the study. You really want to think about data quality from the real real time use of data, the real world data. And so we have statistical tools that we use to evaluate data quality and that's very helpful as we go from into new settings. So an important challenge that we had was as we went from the post acute hospital settings originally into nursing homes, the hospital folks said, well, I'm not sure I can trust a nurse in a nursing home to do the same assessment.
We showed you get the same data structure, same reliability, same validity from the 2 in home care.
We have clinically based Home Care Services that were usually nurses and social workers are the case managers for that.
But we also have some community support services where it may not be a regulated health professional doing the intro assessment.
We again, use these statistical techniques to show they get the same answers that, you know, they're, they're basically fully interoperable in terms of the, the data quality that are there.
We can also use those tools, though, to detect problems that emerge in data either because of poor coding practises or because of gaming.
So right now, a concern that we have is we're publicly reporting on antipsychotic use, and we set a national standard to drive that down from 23% to 15% as the national rate for potentially inappropriate antipsychotic use. But what we see is there's some homes that are trying to get there by changing their coding practises around exclusion criteria. But I've got 20 years of data and I can say, OK, in this home, for some reason we have for 19 years a rate of 5% of the residents have delusions and suddenly it's gone up, but none of the other things have changed.
So having these data tools lets you do sophisticated analytics that can also protect the integrity of the data system.
But there's also a cultural dimension. Everybody has to value the data system and want to protect it.
Because if we don't, if we allow gaming to happen, then we potentially go back to my PhD supervisor's book and we have no data.
It's impressive to see the scientist scientific method at work there. You spoke about, you know, having the idea,
having the methodology and then testing. And it is that continuous loop, isn't it, with that iterates towards better, better quality. Very last question, you did mention that your professor
William Forbes started building the data before the computers were readily available to utilise it. What should we be doing now in your opinion to prepare ourselves for that next technology horizon? Well, you've got a great advantage over a country like Canada because you can benefit from the 25 or 30 years of experience that intro I's had of going into the stuff. So in many ways you can leapfrog ahead of of a country like Canada because we're now in the middle of converting people from the old inter I standards to the newer standards.
So you can in fact accelerate your, your adoption with a the most important thing to to work on with with this change is the sharing of information and the public use of the information to inform quality and policy. But also working with clinicians that they embed the use of the instruments in
clinical practise and share it with patients and their family as part of routine practise.
Because the closer that the use of the data comes to the clinical decision that happens at at the, the forefront, the better the data quality will be.
Because now it's me as a, as a nurse working with you and your family to try to provide the best care I can for you.
Not about, OK, how can I code what's, you know, your characteristics are so I can get five more dollars for my nursing home.
You know, make it more about patients and their family and you'll get better data quality out of it.
Professor John Hirdes it's been wonderful chatting with you. Thank you so much for giving up your evening to talk with us.
We've really appreciated your insights and experiences and we hope to talk to you again in the future.
Great. Thank you for having me. It's been a pleasure.