Hospitals as complex adaptive systems

SPEAKER: Most people recognise that national health systems

are complex: lots of services, facilities, and people interacting in ways that are often unpredictable. But how many people would say that a district-level hospital is complex? Wouldn’t a primary referral hospital with a limited budget and a small range of services be a fairly straightforward part of the health system? We’ve seen that complex adaptive systems often have certain features, including path dependence, feedback loops, emergent behaviour, phase transitions, and scale-free networks. So when Edwine Barasa and colleagues went looking into planning processes in two county hospitals in Kenya, what did they see? County hospitals in Kenya receive a limited amount of cash funding from the Ministry of Health. And their annual planning is also constrained by strict national guidelines and priorities.

For example, the Kenyan Ministry of Health has made the health of children under five a top priority. So county hospitals were required to provide services for these children for free. However, because the national health budget was limited, 80% of the county hospital’s cash income came by charging user fees to most other patients. Over time, some emergent behaviour was observed. When allocating resources to hospital departments, hospital managers started to give more resources to departments which could generate more income from user fees. Departments that weren’t bringing in many user fees got less and less resources from managers. This is called revenue maximisation behaviour. One department that suffered from this behaviour was the paediatrics department.

Because they couldn’t charge fees for children under five, the department couldn’t raise much cash. And so they were not seen as a priority by managers. Even though the national policy was intended to improve access to health care for children under five, the priorities that emerged in county hospitals had the opposite effect. This kind of unexpected result emerging over time is a key feature of complex systems. The reduced resources to the paediatrics department also set up a negative feedback loop. The less resources the department received, the less services they could provide, and the revenue they could generate decreased. In turn, this led managers to divert even more resources away from paediatrics.

In contrast, departments that received more resources were part of a positive feedback loop. The more resources they received, the more services they provided, which led to more revenue, which encouraged managers to give them more resources. This made the inequity in funding between departments even worse, which went against the intention of the policies created at national level. Another feature of complex systems observed in county hospitals was path dependency. This is the idea that history matters. In county hospitals, the previous budget allocations to each department strongly influenced how much those departments received in the next year’s budget. Departments with a history of getting less resources found it very hard to argue for an increase.

This ultimately meant that staff in low-priority departments were demotivated. They had very little money, couldn’t persuade management to give them more, and the number and quality of their services continued to decline. So what can we learn from what was happening in this part of the health system? When the national policy on user fees was designed, policy makers did not appear to have foreseen the counterproductive effect that would be produced at county level. With a deeper understanding of the complex relationships between managers, departments, financing systems, and national policy, they might have been able to better predict the outcomes of their policy.

They might have instead produced policy which encouraged positive feedback loops and positive emergent behaviour, for example, creating incentives for hospitals to prioritise children’s health, reducing the attractiveness of user fees generated in other departments, and acting to motivate staff in under-resourced departments. Understanding that health systems are complex helps us see things for what they are and produce policies and interventions that have a better chance of success. We just have to start learning to look for the complexity around us.