Population health: a population is not the sum of its people

Add up the risk of every person in a town and you get a number. That number is almost never the town’s risk. Two places with the same average age, income and rates of chronic disease can have very different outcomes. In one, the clinic sits on the bus route and its hours match the end of the workday. In the other, everything depends on a hospital two hours away. The people may be similar; what differs is how they connect.

That is the idea behind this article: a population is not the sum of its people. It is a system, and population health makes more sense when you look at it that way.

Population health is not an average

David Kindig and Greg Stoddart defined population health as the health outcomes of a group of individuals, including the distribution of such outcomes within the group. The second half is the part most often forgotten. An average can improve while a neighborhood, an age group or a community falls behind.

That is why we say your population isn’t an average. Looking at the distribution, not just the central number, is the first step to seeing what the average hides.

What happens when the parts meet

There is a field devoted to systems in which many parts interact and adapt: academics call it complexity science. You don’t need its technical vocabulary to use its ideas. Five of them go a long way in understanding a population’s health:

Interactions. What one person does changes the options of others. Habits, information and trust travel through networks of family, work and neighborhood.
Feedback loops. Some effects amplify and others dampen. When emergency rooms are overcrowded, some people delay care, arrive sicker, and the crowding grows.
Emerging patterns. Some patterns live in no single person and appear only in the whole: waiting lines, places with no nearby services, hours when nobody can attend.
Adaptation. People and institutions respond to what is done. A change in one service can shift demand elsewhere instead of meeting it.
Cascading failures. A network that depends on a few points is fragile. When one is overwhelmed, the pressure moves to the next.

The network that holds up a country’s health

If a population is a system, so is its health system: a network of places, services and people, with nodes and connections. Seen this way, the questions change. It is no longer only how many services there are, but what connects one service to another, which territories depend on a few places, and what happens if one of those places fails.

In our Observatory we map Colombia’s hospital capacity network, municipality by municipality. Five municipalities hold 37% of registered beds. In 26 of 32 departments, a single municipality holds more than half. And 176 municipalities have provider sites but no beds or stretchers. A registered bed is not access, but the shape of the network already says a lot about where it holds and where it is fragile.

The question is not only how much there is, but how it is connected.

Why recipes don’t travel

When a population behaves as a system, the same intervention can produce different results in two places, because each place has its own connections, loops and responses. Harry Rutter and colleagues made the case in The Lancet: public health needs a model of evidence that considers the system in which an intervention takes place, not just the intervention in isolation.

Hence how we work: learn from the global, connect at the regional level and act locally. What worked somewhere else is a clue, not a recipe. In our approach model this is the shifting problem: you move forward by testing small, learning and adjusting.

Simple, but not simplistic

Looking at population health as a system does not mean making it complicated. It means not oversimplifying it. In practice, it changes five things:

Look at the distribution, not just the average. Who is at the top, who is at the bottom, and where they are.
Map the network. Who depends on whom, where people move, and which points hold the system up. That is what geographic and service-network analysis does.
Look for loops before intervening. Ask what could amplify or hold back the change you want.
Test small and learn. Especially when the problem shifts with context.
Measure side effects. A change at one point of the network may show up at another.

Not every population health problem needs this lens: many are clear or technical, and a protocol or a careful analysis is enough. Systems models help understand and anticipate scenarios, but they do not predict with precision. And a network of registered capacity shows structure, not real patient flows.

If you want to see how your population, or your service network, is connected, let’s talk.

A population is not the sum of its people