Welcome back. In this section we're going to talk more generally about where system thinking helps us. It helps us to understand the types of problems, the theories, methods, and tools that are used. And I know that we've already started talking about this and giving examples of where we use system thinking with problems related to implementation, when we talk about why as well as problems related to scaling up interventons. But 'm going to try and put this into more generalizable terms about the kinds of issues, or problems, or questions that you might ask and what the implications are with systems thinking approaches. So one set of problems where systems thinking can contribute a lot is when we think about how does an intervention program work and how do you intervene when you're working in complex conditions? Again, the problem of growth or scale up of a program, or the problem of sustainability which is also related to this. And this is in contrast to, I guess it's in contrast to the evidence based medicine traditions, which really depend on having fixed interventions, fixed types of agents that you're implementing that are being intervened, whether it be individuals or otherwise. It allows for room for basically for some stochastic types or probabilistic types of differences, but basically follow a very linear model with fixed interventions and unchanging contexts in which they're occurring. So, this is really much more of a real world approach to why you would need to look use systems thinking and how you can do that. So, those are three broad areas. And the question that we talk about in terms of what effect, or does it work, often really is about how things work, is really the problem that we're looking at. And this this is the systems thinking types of implications when you're looking at real world interventions where context is changing, where your intervention is changing. So the first point is that you really want to assess the unintended consequences. And what you do is use systems thinking. Again systems thinking trying to connect the pieces. You would try to identify the key actors of a system or the relationships between them to look for unintended consequences. So, other than the main outcome for example, if you have a TB control program you're obviously going to be looking at things related to TB. Whether it be treatment completion rates, or theft case identification rates, you're going to be looking at TB. But you might want to look at other aspects of a health program, such as change in the use of health worker time, what happens to drug availability, what happens to the quality of other services that aren't part of the projects? What happens to drug resistance if you all of a sudden make things very available? Are there other providers who come in to compete and undercut you? Are there other kinds of treatments that come in that try to mimic or fake an intervention that you're doing because it may all of a sudden be profitable. These are the kinds of things that you would have to look at or should look at when you're looking for unintended consequences. And again there's probably an infinite number of things that can go wrong. But you should be, in terms of using systems thinking, try to look at what are the key actors? What are the key elements or relationships of the actors when looking for these unanticipated types of problems. Sometimes unintended consequences can be anticipated, but more often than not when they occur is because people didn't think about them. So again systems thinking helps for that. In another example of looking at the problem of what is the effect where systems thinking is important, is when you're trying to generalize from a study about what's the CAS effectiveness or what's he effect of certain strategy. I think when you know that you're working in an environment where things are complex, where the context changes, where the elements change, where there are unintended consequences, you really have to be skeptical about interpreting stable effect sizes. Particularly for cost-effectiveness because we know that costs are very variable from place and time. But effectiveness as well for an intervention or strategy any time that things are going to be changing, any time you have a complex adaptive system. So, just because you have been able to do something in one place, a project that has a certain effect for an intervention whether it be a social mobilization campaign or program to again, provide communicase management for childhood ailments. Just because you've been successful in one context doesn't mean that you can do it somewhere else and get exactly the same kinds of results. That's the fallacy of misplace concreteness that we fall upon all the time and we would like to use research to be able to say that this is what we can do. But really it's, research tells us what has been possible in other places and gives you a benchmark. It doesn't mean that you can repeat it and get the same things. There's a lot of reasons mostly related to the complex system that they're working on, why they may not be reputable. So this does mean that you have to have different thinking for how you interpret quality of evidence when you're working in complex systems. And that the kind of assumptions that reproduce ability of an intervention of the components and results where the cause and effect of results that you typically see in evidence based guidelines may not fit very well when you're working in a complex system. If you can repeat the context, then the more reassurance you have that you should be able to try and get similar results. But again this is another reason why you should use system thinking to make sure that things are actually going as planned or as they have in other places. When the problem is how do you scale up? How do you grow? How do you sustain things? There are other implications for systems thinking as well. You really want to make sure that you have the right model for understanding how things change when they're quite complex. Now we did talk about those five different patterns of change and that's part of it. Those are sort of descriptive things about why they change about how they change in different patterns, and some explanations as to why that occurs. For example, if there are feedback loops or if there are emergent properties for example. But when you're trying to understand this and use system thinking, you have to make sure that you're capturing the right types of changes. So in a health system is the strategy, how is that changing. Either in its intensity, or the components of the strategy, where you're doing it. Are you learning as you go along, and you're able to change your program. Whether it'd be a family planning program, an immunization program, or comprehensive primary care program. You know, your strategy is probably not the same from place to place and over time, and you do want to encourage learning. So you have to be able to understand and look for these changes, and again embrace changes. Look at the role of different actors, how they lead, how they influence. The capabilities particularly of implementors, of beneficiaries, of government that oversees and what kind of incentives the different actors have. You want to be able to capture those changes or at least think about them, either if you're doing it as part of research, part of your regular monitoring and evaluation, or if you're trying to explain or influence how your program is being implemented. Of course, you want to know about the environment, whether they're the policies or the social factors that are changing things, the different trajectories that are going on for individuals and groups. And if you remember, there was the different trajectories of the provinces in Afghanistan for every indicator that we looked at. So be able to understand what's going on that causes those changes or maybe seeing the change and therefore going to see what might cause it. And again, I'm repeating myself, but looking for intended and unintended consequences. Is critical in how those things are changing. Now, research designs or monitoring designs, you try and put them in a way that they don't actually interfere, but actually that they can encourage learning processes for program implementation and that they're not just done for the sake of research. But these are some of the challenges and opportunities for using systems thinking when you're looking at change or sustainability. So I think it's also helpful to try to understand what might cause these complex behaviors, these kind of pathways that we've seen. We've describe the pathways that you've seen. You get some idea of what might cause them because of feedback loops, because of, again, adaptation, but what leads us to those points? And there's some thinking that there's different characteristics of a system that caused these kind of phenomenon. And this might also be helpful when you're looking at systems thinking in terms of knowing where to look. So one of the things that causes this complexity is diversity. Diversity in the different kinds of elements. Now this can be different kinds of healthcare providers, different kinds of health services, different kinds of organizations that are in a health system. Even among organizations, you might have multiple pairs in a health system. You might have different kinds of primary care providers, formally trained, informally trained. Those that take insurance, those that are publicly employed, those that pay a use fee for service. So all of these aspects, that having that diversity leads to complexity. But it's also connectedness in terms of how close they are to each other. How does one person influence another person, or how close they are to be able to do that? And closely related to the notion of connectedness, is this notion of interdependence. In other words, which way does one person or entity influence the other? Do people copy each other? We know, for example, that informal providers where we've looked at it tend to copy the prescribing practices of formally trained doctors in their region, and when they change they tend to copy those changes. Doctors in one place of practice tend to be more like each other. That leads to different sort of clusters of activity of practices in terms of quality of care or when somebody gets a certain procedure done. But again it's related to how close one entity is to another and how close they are in terms of copying or doing something opposite to them to try to differentiate themselves. Or this last point, which is about learning again, the ability to adapt, to learn from something, to create new structures. So those are those four different characteristics in a system that create those phenomenon of feedback of new structures of phase transitions and of the scale free types of networks that may cause tipping points for different types of patterns. So if your looking at a model that you can use for research, that you can use for monitoring an evaluation for understanding complexity in a health system, here's one type of approach. On the left we have these type of initiating conditions, which are the context, the kind of epidemiological, the social, political, economic, legal types of arrangements of a place. Then you might have your, what I'm calling a perturbation, but this might be your program, it might be your project, your intervention, and it goes into a health system, which has this features of complexity, these diversity, connectedness, interdependence and learning of those elements. And remember L system has both actors, it's connection, these services that link each other as well as these outcomes which are there on the right, the kind of outcomes. And we show that loop between the program or perturbation in the health system to show that there's actually a kind of learning that goes on. The program should change if it's actually adapting to the complexity. The program itself should be learning. That's something that's a desired attribute. You don't want to be fixed when, in fact, the health system is changing. And that works within an external environment that changes, these initiating conditions don't stay the same and they're working towards some outcomes. Again, you know we talked about intended outcomes in terms of health status, in terms of trying to prevent impoverishment due to health and trying to improve trust in the health system. But there are also unintended consequences. You can get distortions in services. New services often go to the rich before they go to the poor. Hopefully, you get synergy some are working well to improve child health and one area that main true is maternal child health services are in another. Or you might create unsustainable projects that distort the market, create dependency and then can't continue once the program is over, or that they adversely effect another program because all the effort and time are spent in one particular area. And that also interacts back with the way a health system works. So that's sort of the kind of changing environment on the top there. And down below, what we just see is basically those pattern of change over time, which can lead to the non-complex types of changes, which can be total disorganization, or complication as we described it earlier, or more predictable things that are linear. Equilibrium and periodic orbits are again more predictable types of common patterns of change that are not complex, whereas these feedback loops that create cascading, the tipping points, the phase transitions, path dependence and emergent structures are becoming the patterns of change when you're looking at the complex nature. The bottom line shows that these things occur over time because it's dynamic, and that you really need to be observing them at different points of time. Snapshots are rarely enough, you can create histories going back, but you really need that prime element to be able to understand the complex system. So, that's true whether you're creating a research program, or a monitoring and learning system to account for changes over time, and so that's one framework. So what I'd like to do in the next three slides is not go over them in great detail, but basically point to sort of a catalog of selected theories methods and tools that have been used in systems thinking and the broad diversity of systems, thinking really as a reference for how to look at things rather than to go in great detail in any particular slide. But in systems thinking, there are a number of theories. We've talked before about general systems theory with Brook and looking at sort of theory of Theories. And actually, when I just display that model of complexity in the health system, is actually shows the theory around the notions of diversity connecting this into the pins and learning, leading to complexity. But there are other ones, catastrophe theory, chaos theory, both based in mathematics. And different, related to showing how sudden that large changes occur in chao theory. It's because, as we've discussed, it's about small changes in initiating conditions. In catastrophe theory, it may be other parameters that change along the way. Learning organizations comes out of management theories about how do you change organizations, how members learn, and how you, not only how individual members learn, but how they work together as teams to develop team learning through different kinds of approaches. So there's theories in learning organizations. Lots of theories in, anything from economics and other social sciences to physics to explain path dependency. How you get different outcomes from similar starting points based on not just the initial conditions, but the choices you make along the way, which sometimes are irreversible. And then other ones about punctuated equilibrium, we've seen that in, particularly, social theory around policy, but it was taken from developmental or evolution theory as well, and then, apply to a policy change. So again, different theory that have been used. They're also a whole range of research methods and tools, and there's a continuum of methods and tools. There's not always a fine line between what point a method becomes a tool, but basically they're methodologies that involve a series of tools. And one of them is agent-based modeling, you're going to do a little bit of agent-based modeling in this course and again in the introduction. And that's where you have a way of representing a complex system based on basically showing individual agents and how they interact with each other and the environment. Trying to set a set of predefined rules, and watching how they interact and emerge with self organization. Another set of methods in social network analysis, systems dynamics modeling. I think you're actually going to be doing a little bit of systems dynamic modeling in the course as well. And then a set of tools cause of loop diagrams. We've already shown some examples and this is one that you'll get some practice on. To show basically to prove, to show, taking a mental model and to try to show how things are related usually in a causal type of approach. These are often done in a participatory way to try and get the key people involved in a particular program to try and describe what they think is going on. There's a number of other participatory ways of getting people involved to describe or understand a system, such as innovation change or management history where you look at things, reconstruct how things occurred over time or participatory impact pathways analysis. PIPA is another structured workshop approach to try to show a logical pathway of how things actually work in real life, involving stakeholders who are actually involved in a process and an intervention trying to reach consensus then on how you actually get the kind of outcome that you'd actually like and dealing with some of those unintended consequences. Process mapping is very similar but for the very simple thing it's a flow chart about how things work together or process. But you can make them as simple and as flexible into different kinds of work to look at bottlenecks or inefficiencies. And the last one stock and flow diagrams you'll be doing this in class as well, at least in an introductory way. Often they start off with causal loop diagrams, and you take some of the variables, and you create stock and flow diagrams that show how things change over time and look at feedback again to try and capture nonlinear dynamics between multiple parts in a system. So again, this is an exhaustive list, but it shows you a broad range of theories, methods and tools, and you know that they come across many disciplines. Many of them are quantitative. Many of them are qualitative. There's a participatory element, particularly at the beginning that, in defining the system, that's quite common to them, and all of them are used to try to guess create explicit models of how things relate to each other. Which again is the core of what systems thinking is about, how things relate to each other. So as we wrap this up, we're trying to look at how system thinking informs health systems intervention. And what we've seen is a whole range of applications, the theories, the methods and tools, and what they are helpful for is in many kinds of settings. We've talked about them in terms of settings for trying to understand how to improve implementation, how to improve scale-up and sustainability. But underlying it, what it does is it gives us a better understanding of dynamics. Whether it be dynamics of disease transmission, or other parts of change in a health system. Showing how things are related to each other, particularly as they relate to contextual factors. And in the the health system, we're particularly interested with how they relate to population health. It can be helpful in identifying root causes of variations. Variations that are with us all the time. These are variations in behaviors. They can be variations in services, variations in health outcomes, and this gives us the tools. Systems thinking gives us the tools to be able to embrace and understand those kind of variations. It helps us to identify many factors from different sectors that promote the spread of an innovation or an intervention. Again, intended and unintended consequences gives us a better understanding and appreciation for it as well as where to look. And basically gives us new tools and approaches to understand and facilitate decision-making. So at the end of this, what it means if you are working in a complex health system, systems thinking should tell us that you really need to move beyond that cookie cutter. When you're doing policy or planning, you need to have flexibility to address those dynamic and adaptive properties of a health system. And part of that is to use data in frequent cycles of adaptation. We're encouraging adaptation and frequent cycles of experimenting and learning and planning. And again involving key actors, key actors that you can identify in the health system. Not just providers, not just financiers. But also those beneficiaries are people who are actually trying to influence their own health and that of others.
