Introduction to realist review

Content for steps 2.2 and 2.3 have been provided by Dr Geoff Wong – Clinical Research Fellow in Nuffield Department of Primary Care Health Sciences at University of Oxford, and practicing NHS GP.

You can find out more about his work here.

This step will introduce realist reviews, and step 2.3 will further explain the concepts covered in this step.

What is a realist review?

The terms realist review and synthesis are synonymous – generally the term realist review will be used throughout these steps. Realist review refers to a form of secondary research.

Realist reviews:

• are theory driven – the end product is a theory (a coherent explanation of why certain things happen)
• have more of an explanatory rather than judgemental focus (for example, they are unable tell you if drug can or cannot work, but they can explain why it is that people might or not take medication prescribed/recommended by clinicians)
• are based on a realist philosophy of science (Ontology)
• look for mechanisms and develop middle-range theories (both of these will be covered later)
• test (collect data) to develop, and then to confirm, refute, or refine aspects of these theories
• are iterative (i.e. they are more like detective work than following a recipe)
• look to answer ‘how?’, ‘why?’, ‘for whom?’, ‘to what extent?’, and ‘in what circumstances/contexts things happen?’
• are very helpful in making sense of interventions and programmes that are complex and have outcomes that are context dependent

Programmes or interventions in health are complex

Multiple components

Lots of interventions in pharmacy and healthcare will have multiple components – for example, the way we might present our particular intervention, how people are asked to come to them, or how they are sustained within them.

Multiple outcomes

There may be a final desired outcome – for example an individual’s health has improved (lower blood pressure, better diabetes control etc). To get there, it is often necessary to go through many steps; the final desired outcome is not achieved in one amazing leap or bound.

Emergence

This can mean slightly different things to different people. This concept can relate to:

• how things change as an intervention runs
• the overall intervention is more than the sum of its parts; for example, if you have multiple contacts with a particular client, they may build up a strong relationship with you and take your advice more seriously

Non Linear

A lot of the outcomes we get are non-linear. For example, a lot of time and effort may be spent encouraging an elderly person to take important medication, but the number of times they are asked is not necessarily related to how likely they are to take it. Perhaps something completely different will be the main encouraging factor.

Interacting components

Intervening components interact with each other. For example, the act of dispensing medication is followed by a need for the patient to take the medication. If they don’t take the medication, they may build up a large store, which will then impact when they next pick up medications.

Context-sensitive outcomes

It is often claimed by many that outcomes are context sensitive – i.e. the outcomes which occur tend to depend on different factors. If this is the case then it is important that we have a way of explaining how and why outcomes vary because of context.

A simplified diagram of a ‘complex’ intervention

It is important to consider the idea that a desired outcome may not have an easy path to its accomplishment – there are many potential factors which can influence this.

For example, in the image above, one letter (for example E) may represent the desired outcome of patients taking medicine safely. To achieve this, the medication must be consistently taken, at the correct times, any changes to medication are correctly addressed, and more.

The different sized arrows represent the different routes which may be taken to achieve the desired outcome and also how easy each route might be – for example larger arrows may be easier routes. There may also be some unintended outcomes. For example an outcome may seem negative, such as sub-optimal medication intake, but it could lead to a higher quality of life due to a decreased risk of developing a medication side effect.

Realist reviews aim to work out, in as much detail as necessary, why the various outcomes seem to arise.

What is realism?

All research approaches or methods are underpinned by implicit and explicit assumptions. In realist reviews the assumptions are fairly explicit. Realist reviews are based on a form of realism: a particular way of thinking about how the world is constituted in such a way to enable science to be possible.

It is not necessary to understand this concept fully. It is a different way of thinking about the world and how it must work in order for science to be possible, and simply knowing this will make it easier to understand the basis of why particular processes and methodological approaches are used in realist reviews.

An approach to realism

There is no unified school of realism, and there are multiple types of realist.

The particular approach to realism used in realist reviews has been developed by Professors Ray Pawson and Nick Tilley. They have drawn on a range of works by others to develop their approach. What they have developed is a way of explaining causation which they express as:

Context + Mechanism = Outcome

This is a way of thinking about how the world is constituted – in other words outcomes are caused by mechanisms. It is an expression of causation, a way to link context to outcome, and a way of looking at data. It is also an explanation for why lessons are transferable – because of the nature of mechanisms (this is explained in more detail later).

The video in the next step will further explain these, and all of the ideas expressed in this step.