Systematic reviews

On this page:

Synthesis is the process of combining the findings of research studies. A synthesis is also the product and output of the combined studies. This output may be a written narrative, a table, or graphical plots, including statistical meta-analysis. The process of combining studies and the way the output is reported varies according to the research question of the review.

In primary research there are many research questions and many different methods to address them. The same is true of systematic reviews. Two common and different types of review are those asking about the evidence of impact (effectiveness) of an intervention and those asking about ways of understanding a social phenomena.

If a systematic review question is about the effectiveness of an intervention, then the included studies are likely to be experimental studies that test whether an intervention is effective or not. These studies report evidence of the relative effect of an intervention compared to control conditions.

A synthesis of these types of studies aggregates the findings of the studies together. This produces an overall measure of effect of the intervention (after taking into account the sample sizes of the studies). This is a type of quantitative synthesis that is testing a hypothesis (that an intervention is effective) and the review methods are described in advance (using a deductive a priori paradigm).

If a systematic review question is about ways of understanding a social phenomena, it iteratively analyses the findings of studies to develop overarching concepts, theories or themes. The included studies are likely to provide theories, concepts or insights about a phenomena. This might, for example, be studies trying to explain why patients do not always take the medicines provided to them by doctors.

A synthesis of these types of studies is an arrangement or configuration of the concepts from individual studies. It provides overall ‘meta’ concepts to help understand the phenomena under study.  This type of qualitative or conceptual synthesis is more exploratory and some of the detailed methods may develop during the process of the review (using an inductive iterative paradigm).

There are also multi-component reviews that ask broad question with sub-questions using different review methods.

Systematic evidence maps are a product that describe the nature of research in an area. This is in contrast to a synthesis that provides uses research findings to make a statement about an evidence base. A 'systematic map' can both explain what has been studied and also indicate what has not been studied and where there are gaps in the research (gap maps). They can be useful to compare trends and differences across sets of studies.

Systematic maps can be a standalone finished product of research, without a synthesis, or may also be a component a systematic review that will synthesise studies. 

A systematic map can help to plan a synthesis. It may be that the map shows that the studies to be synthesised are very different from each other, and it may be more appropriate to use a subset of the studies. Where a subset of studies is used in the synthesis, the review question and the boundaries of the review will need to be narrowed in order to provide a rigorous approach for selecting the sub-set of studies from the map. The studies in the map that are not synthesised can help with interpreting the synthesis and drawing conclusions. Please note that, confusingly, the 'scoping review' is sometimes used by people to describe systematic evidence maps and at other times to refer to reviews that are quick, selective scopes of the nature and size of literature in an area.

A systematic map may be published in different formats, such as a written report or database. Increasingly, maps are published as databases with interactive visualisations to enable the user to investigate and visualise different parts of the map. Living systematic maps are regularly updated so the evidence stays current.

Some examples of different maps are shown here: