Not every health problem needs the same approach

A program that worked very well in one city is copied to another and never takes off. A flawless protocol is followed to the letter and the results don’t show up. An expert committee debates for months without reaching agreement. Effort is rarely what is missing. Often the problem starts earlier: in how the situation was defined.

Before choosing a tool, whether a protocol, a model, a dashboard or a program, it helps to recognize that there are different types of health problems and to place the one in front of you. It is the first step in our work, and it comes down to four questions.

Two questions that change everything

The first is how much we know about what to do. Sometimes the link between an action and its result is well established; sometimes it is only suspected, and sometimes it is simply unknown.

The second is how much we share. A surgical team in a routine operation shares the goal and the method. A ministry, an insurer, a hospital network and a community can look at the same problem and want different things.

Crossing those two questions is not a new idea. It comes from Ralph Stacey’s agreement and certainty matrix, which Paul Plsek and Trisha Greenhalgh brought to health care in the early 2000s. A related idea is the Cynefin framework, which distinguishes the simple, the complicated, the complex and the chaotic. What makes these tools useful is that they force a pause before acting.

Four types of health problems

The less we know and the less we share, the further a problem moves from the center. We use everyday words for them; the academic term goes in parentheses.

Clear problem (simple)

We know what to do and there is agreement. A protocol, a guideline or a checklist works. For example, following a surgical safety checklist or applying a risk score already validated in the population where it is used. The most common mistake here is overcomplicating it: inventing a new process for something that already has a good answer.

Technical problem (complicated)

The answer exists, but it takes expertise and detailed analysis. Building, validating and calibrating a risk model with an organization’s own data, or designing a regional network for a specialized procedure, are technical problems. Careful analysis, the right experts and planning work. The common mistake is treating them as clear and skipping the analysis.

Shifting problem (complex)

Cause and effect are only visible in hindsight and change with the context. Why do so many people leave a chronic disease program after the first month? What makes a community accept or reject a service? There is no recipe here. You move forward by testing small, learning and adjusting, with several disciplines at the table and local context at the center.

What worked in one place may not repeat the same way in another. With shifting problems, copying the recipe is the most expensive mistake.

Unstable problem (chaotic)

There is no visible order and action has to be fast: the start of an outbreak, an overwhelmed emergency network. First the situation is stabilized: contain the harm, protect people and keep essential services running, with a shared picture of what is happening. Understanding comes after. Wanting to analyze everything before acting is, here, the mistake.

Two more dimensions: disciplines and place

Knowing the type of problem helps choose the method, but it does not say who should be at the table or where the problem plays out. That is why we add two dimensions.

The first is disciplines. A clear problem can be solved by one. As a problem moves away from the center it needs several, then disciplines working as one, and finally the community too: what the literature calls transdisciplinarity.

The second is place. Some problems live in one specific place; others involve the local, the regional and even the global. We like to sum it up in one idea: learn from the global, connect at the regional level and act locally.

What changes in practice

• The method is chosen by the problem, not by the tool at hand. Having a dashboard does not turn a shifting problem into a clear one.
• Measurement changes. For clear problems you measure compliance; for shifting ones you measure how much is learned and how fast things are adjusted.
• The team is built to fit. Not every problem needs everyone, but the ones far from the center cannot be solved by a single discipline.
• We are honest about what is known. Saying something is uncertain is not weakness; it is the condition for deciding well.

A single challenge usually combines several types of health problems. In type 2 diabetes prevention, deciding whom to contact first is a technical problem: it is solved with a well-validated risk model. Getting a person to stay in a prevention program is a shifting problem: it depends on their context and is learned along the way. Confusing one with the other is one of the reasons so many programs stay on paper.

This model is not a recipe or a final classification. It is a way to talk about a problem before acting. The boundaries between types are fuzzy, and the same problem can move from one type to another over time.

To see how we use this approach to choose among our capabilities, visit Capabilities. And if the underlying idea interests you, start with How health happens.

Further reading

1. Plsek PE, Greenhalgh T. Complexity science: the challenge of complexity in health care. BMJ. 2001;323(7313):625–628.
2. Stacey RD. Strategic Management and Organisational Dynamics. London: Pitman; 1996.
3. Snowden DJ, Boone ME. A leader’s framework for decision making. Harvard Business Review. 2007;85(11):68–76.
4. Khoury MJ, Iademarco MF, Riley WT. Precision public health for the era of precision medicine. American Journal of Preventive Medicine. 2016;50(3):398–401.