Skip to 0 minutes and 10 seconds SPEAKER: This is Alma. Alma designs health programmes. And she has started to notice that the interventions that get funded each year in her district seem to have one thing in common– they are mostly simple. Funding for health programmes is tight. And her government wants to spend its limited cash on interventions that don’t have too many components. So they are easy to understand, and manage, and low cost. Alma notices that these interventions are also linear. Managers assume that in whatever context they are implemented, if you do X, you will always get Y as the result. And if you do more of X, you get more of Y.
Skip to 0 minutes and 49 seconds There’s just one problem and it’s the third thing these interventions have in common– they fail, a lot. The more Alma looks at it, the more she sees that trying to run a simple linear intervention in the complex, messy real world seems not to work a lot of the time. For example, after an outbreak of influenza, the government made the flu vaccine freely available in public hospitals and paid nurses to do vaccinations. Simple, thought Alma. But instead of quickly stopping the epidemic, hospitals were flooded with people demanding the vaccine. And doctors and nurses were distracted from their other work, so other diseases weren’t treated.
Skip to 1 minute and 30 seconds Private health insurers stopped paying for the vaccine, so their customers showed up in public hospitals to be vaccinated, making things even worse. Flooding on roads prevented vaccines from getting to some parts of the district. So then Alma had to deal with angry community members demanding to know why they were being left out and angry staff who were missing out on the incentives. It seemed like the people who designed the intervention had just focused on the simple problem of giving a flu vaccine and assumed that the staff, patients, facilities, insurers, and weather wouldn’t affect the programme or react to it. If you assume that health systems are simple and linear, it makes sense to plan simple linear interventions.
Skip to 2 minutes and 14 seconds But health systems aren’t simple and linear. They are examples of complex adaptive systems. Let’s look at what that means. Systems means that you can’t break down health and health services into little parts. Everything is connected to everything around it. Every facility, every programme, every person, and every policy is connected directly, or indirectly, to everything else. And these things connect to everything else in the world– politics, economic systems, culture, climate and weather, infrastructure, everything is connected. Adaptive means that when some part of the system changes, the system itself changes to adapt to the new situation because all the parts are interconnected.
Skip to 3 minutes and 0 seconds Complex means that the way that the parts of the system relate to each other, and the way the system behaves as a whole, is non-linear and somewhat unpredictable. You can do exactly the same thing and cause very different effects each time. There are many interesting ways that parts of a complex adaptive system interact with each other. Understanding these can be very helpful in understanding health systems and in helping Alma do what is called systems thinking. The first feature of complex systems is path dependence. Path dependence implies that decisions made prior to the ones being made now constrain options and affect the likely impact of an intervention.
Skip to 3 minutes and 39 seconds For example, a reform introduced in a health system that has been stable for a long period might have a different effect than if the same reform is the latest in a long line of reform efforts. In the first example, health system actors might be better able to cope with change and implement it to good effect. In the second example, they may suffer from reform fatigue and have learned how to resist change and maintain the status quo. In some ways, path dependence is just another way of saying history matters and context matters.
Skip to 4 minutes and 13 seconds If Alma is thinking about health programmes using systems thinking, then the history of how the system got to its current state is very important in understanding what interventions might work and why. The second feature is feedback loops. Feedback loops imply that after an intervention achieves an impact, the variable impact, it can have a further effect back on the intervention. Feedback loops can be magnifying, the effects get larger and larger over time, or dampening, the effect of the feedback is to reduce the effect of the intervention, possibly even nullifying or reversing over the long term. An example of a dampening effect could be an attempt to improve access to health care by reducing financial costs to patients.
Skip to 4 minutes and 55 seconds The initial impact might be to increase use of health services, increase effectively treated disease, and reduce ill health in households. But the increased use of services might fatigue health workers and exhaust drug supplies. As time goes on, demotivated health staff and drug stock-outs could lead to poor quality care and the long-term impact could be to reduce use of health services again, potentially even to a level below the initial one. A magnifying impact of the same intervention is also possible, though. As households use health care more effectively and cheaply, household economies and community disease transmission could reduce, magnifying health outcomes over time.
Skip to 5 minutes and 37 seconds Systems thinkers spend a lot of time identifying and understanding feedback loops so they can try to create conditions that encourage positive feedback loops and discourage negative ones. A third feature of complex systems is scale-free networks. This idea says that although all people and organisations in a health system are connected, we are not all connected equally. For example, some people are better connected and more influential than others in a network. If an intervention, such as a health education campaign, is targeted at all community members equally, its impact will mostly be affected by whether it happens to reach and change the perceptions of a relatively small number of influential people.
Skip to 6 minutes and 20 seconds So how health system changes happen depends heavily on which people and organisations take up the changes first, or if they refuse to change. Emergent behaviours describes the ways that entities reorganise themselves when disrupted. This is a feature of physical, chemical, and biological processes. For example, the immune system seeks to re-establish the pre-existing order when a disease agent invades a human body, but is particularly observable in social systems in which human beings seek to protect their interests in the face of attempts to redistribute through reform.
Skip to 6 minutes and 56 seconds For example, professional bodies, such as nurses associations, whose existing function may be to maintain quality standards and internally regulate the profession, may become radicalised by a perceived attack, such as the attempt to establish a new cadre of staff seen as competing with nurses or persistent falls in living standards. This reorganisation changes the nature of the system and how interventions can be implemented with what effects. Perhaps every health system reform now becomes a political battle. As interventions are reshaped to reflect this new reality, so the system changes further. And in this way, the system is set to emerge from complex interactions between people and contextual factors. Phase transitions occur when simple uni-directional relationships reach critical points with unpredicted effects.
Skip to 7 minutes and 46 seconds The tipping point idea is the best-known type of phase transition. A prime example is the behaviour of infectious disease. A few isolated cases of the disease have a predictable and linear impact on the number of other people affected. But when prevalence reaches a certain point, an epidemic may occur or a sudden mushrooming of cases. The herd immunity phenomenon is related to this. If a certain percentage of people are immunised, the disease cannot spread beyond that tipping point and most people are safe from the illness, whether or not they are immunised individually. But if immunisation coverage drops below a critical point, epidemics arise and everyone is at risk if not immunised.
Skip to 8 minutes and 29 seconds These kinds of phase transitions can also occur in human behaviour. If most women in a community give birth at home, it may be quite difficult to persuade a woman to deliver in a health facility. But after a certain proportion of deliveries start happening in facilities, there may be a further rapid rise in women coming to facilities to give birth because it has become a social norm. This is very different to the simple linear cause and effect relationships that most health programmes assume. An intervention might have very little impact until coverage reaches a certain level. Or it might have a lot of impact at small scale, but little impact when scaled up. So how can this way of thinking help Alma?
Skip to 9 minutes and 11 seconds Firstly, it can help her understand and describe her health system more accurately. Instead of thinking about it as a small number of simple moving parts, she can get closer to thinking of it as it is– a huge, complex, constantly changing network. This will help her have realistic expectations of what changes might be possible in the system as she thinks about designing or implementing new policies or interventions. She will think a lot bigger about the political, social, historical, and geographical context when planning what needs to change to improve health. Because she understands complexity, she can think better about how to tailor an intervention that works somewhere else so that it has a greater chance of success in her district.
Skip to 9 minutes and 57 seconds And when she monitors, or evaluates, interventions, it will help her explain why things did or didn’t work. Complex systems thinking is hard. We are trained to break everything down into simple linear parts. But if we want to see health systems as they are, and make realistic plans for improving population health, we’re all going to have to become systems thinkers.
Systems thinking in health systems
Systems thinking is easier for some people than others - and some people who work in health systems naturally think in non-linear, complex ways, even if they don’t give formal names like ‘feedback loops’ to the patterns they identify.
In the next few weeks you’ll keep coming back to this idea as we look at case studies of situations where appreciation of complexity helped (or might have helped) in developing health systems strengthening actions.
Think of the last time you attempted to explain a health systems problem to someone else (or to yourself). For example:
Health workers in the public system provide poor quality of care.
Was it a rather linear explanation? For example:
They don’t make the required effort because they are over-worked and underpaid. They are over-worked and underpaid because there are not enough resources in the public health system. There are not enough resources in the public health system because there is a lack of political commitment.
Did you trace the roots of the problem back to just one or two causes? Or did you think about all the aspects of the system that might be contributing to the problem (and might need to change if the problem was to be addressed). Did you think of contributing factors there were ‘outside’ the formal health system (for example, secondary education quality or workplace politics and culture). Did you think about how the different factors might interact in unexpected or unpredictable ways, or did your explanation assume that one factor always led consistently to another?
Learning to be a systems thinker is hard work, particularly in the context of limited time and information that are common for people working with health systems.
Before moving on, think about the health system problem again. Can you think of at least one feature of complex systems (for example, feedback loops) that might have been part of the complex nature of this problem?
© Nossal Institute for Global Health at the University of Melbourne