Causal chain analysis as a way of understanding how interventions work


Oral session: Inclusion of non-randomized designs


Wednesday 23 October 2019 - 14:00 to 15:30


All authors in correct order:

Kneale D1, Thomas J1, Bangpan M1, Waddington H2, Gough D1
1 EPPI-Centre, University College London, UK
2 3ie International Initiative for Impact Evaluation, UK
Presenting author and contact person

Presenting author:

Dylan Kneale

Contact person:

Abstract text
Background: understanding the extent to which an intervention ‘works’ can provide compelling evidence to decision-makers, although without an accompanying explanation of how an intervention works, this evidence can be difficult to apply in other settings, which ultimately impedes its usefulness in decision-making. This type of concern is particularly relevant for social interventions, where the context of delivery may have an important influence on how interventions work.

Objectives: this presentation will explore how a Causal Chain Analysis (CCA) approach in systematic reviewing is instrumental in:
1) developing a logic model that incorporates an understanding of how different intervention components and processes effect change in outcomes;
2) hypothesizing the nature and complexity of the causal relationships being depicted;
3) synthesizing evidence that meets decision-makers’ needs around understanding if and how interventions ‘work’.

A further objective of this presentation is to identify best practice for conducting CCA.

Methods: we explore how CCA should be conducted through:
1) examining key papers on the logic of CCA;
2) examining frameworks for assessing and identifying causal relationships;
3) identifying and analysing case studies providing examples of causal chain analyses (CCA) in the literature.

Results: we describe a number of examples of causal chain analyses (CCA) in the literature that exemplify best practice and the challenges of a CCA approach. No set criteria exist for best practice in the conduct of CCA, although we have developed a set of principles that can be developed further into recommendations:
1) reviewers should be familiar with the underpinning assumptions of CCA;
2) all CCA involve development of a logic model;
3) research questions for synthesis should draw on hypothesised causal chains represented in the logic model;
4) synthesis methods should be selected based on the nature of hypothesised relationships that are identified within the logic model;
5) integration of different forms of evidence serve to strength the mechanistic account of how interventions lead to change;
6) logic models should be updated at the end of the review to reflect the review’s findings.

Conclusions: a CCA approach can assist systematic reviewers in producing findings that are useful to decision makers and practitioners, and in turn help to confirm existing theories or develop entirely new ways of understanding how interventions effect change

Patient or healthcare consumer involvement: we describe how patient and consumer involvement can strengthen the conduct of CCA and demonstrate the way in which this involvement can be integrated from the outset within a CCA, ensuring that evidence meets the needs of decision-making.