Using systematic reviews to identify the essential components of interventions: the example of parenting

ID: 

18904

Session: 

Long oral session 12: Improving implementability of evidence

Date: 

Thursday 14 September 2017 - 14:00 to 15:30

Location: 

All authors in correct order:

GARDNER F1, LEIJTEN P2, HUTCHINGS J3, MELENDEZ-TORRES G1, MIKTON C4
1 OXFORD UNIVERSITY, United Kingdom
2 OXFORD UNIVERSITY & UNIVERSITY OF AMSTERDAM, United Kingdom
3 BANGOR UNIVERSITY, United Kingdom
4 UNIVERSITY OF THE WEST OF ENGLAND, United Kingdom
Presenting author and contact person

Presenting author:

FRANCES GARDNER

Contact person:

Abstract text
Background: Most complex interventions in psychosocial care deliver a package of knowledge and skills, but rarely is it clear which of their many components are necessary or effective. Elucidating the essential components of interventions could help develop programmes that are briefer, more effective and efficient, and provide a benchmark for assessing the content of interventions that lack formal evidence. These goals are particularly important in low-resource settings, where effectiveness, cost, scalability and sustainability are paramount considerations, and local evidence may be lacking.

Objectives: This project illustrates multiple approaches to elucidating essential components, using the example of parenting interventions, which are prominent in global policy recommendations and implementation efforts (e.g. WHO, UNICEF, UNODC), for preventing violence and improving child outcomes.

Methods: We systematically reviewed evidence from each of 6 methods to identify effective components of existing parenting interventions. The methodological strategies were:
1: Meta-analysis of associations, to test whether interventions with certain components are less or more effective than those without this component (or combination of components);
2: Meta-analysis of ‘Decomposing’, multi-arm trials, which test different combinations of components of parenting interventions in different trial arms;
3: Meta-analysis of microtrials - focused randomised experiments to test the causal effects of individual intervention components;
4: Secondary mediation analyses within trials, testing changes in parenting that predict child outcomes — mechanisms of change can reflect components that contributed to change;
5: Systematic review of expert opinion/consensus methods for identifying the essential components.
6: Optimisation studies, e.g. factorial trial designs.

Results and conclusions: We describe strengths and weaknesses of each method, in terms of strength of causal inference, and generalisabiity to ‘real-world’ interventions; summarise findings from each strategy (some 200 RCTs included); and, discuss implications for scaling up of these common interventions.