Theorising breastfeeding peer support as intervention in a complex ecological system: lessons for implementation and evaluation in a Welsh context

Background and Aim: By international standards, breastfeeding rates in the UK are low, with social and geographical polarisation in feeding decisions. The evidence for breastfeeding peer support intervention is mixed. As an intervention, peer support is heterogeneously described and poorly theorised. Through this thesis, I explore and articulate theories of breastfeeding peer support and consider their potential to inform intervention design. Methods: I use realist methods and an emergent fit approach to explore understandings about how peer supporters help mothers to breastfeed and what prevents them from doing so. The empirical research proceeds through three phases. I iterate between findings from Phase 1 (face-to-face interviews with 15 policy leads and infant feeding leads), from Phase 2 (realist review of 15 breastfeeding peer support intervention case study experiments), and Phase 3 (focus groups with parents, peer supporters and health professionals) to develop my conclusions. Findings: Stakeholder experiences are consistent with an understanding that ‘low breastfeeding rates’ are a ‘wicked problem’ in a complex system of influences. The implementation landscape is contested and policy rationales shift. Personal feeding journeys have powerful feedback effects. There is partial dissonance between breastfeeding advocates’ own motivation to improve women’s experiences and formal policy goals to increase breastfeeding rates as a mechanism to improve health outcomes. I identified three registers for understanding how breastfeeding peer support works: these were, (i) improving the health care pathway, (ii) creating a sub-community of mothers and sisters, and (iii) diffusing the practice of breastfeeding like ripples in the pond. The realist review showed that the experimental evidence is heterogeneous but almost exclusively relates to interventions that seek to improve the care pathway. From the review, I developed 20 statements and a staged thinking tool to inform future intervention design, highlighting the need to consider a sequence of interactions beginning with interaction with existing social norms and with the existing health care pathway. These statements were extended and nuanced on the basis of discussion with parents, peer supporters and health professionals, resulting in a total of 39 statements to support future intervention design. Conclusions: Theoretical approaches that rely on triggering mechanisms at the interpersonal level are likely helpful as part of intervention theory, but are insufficient, as they tend to be decontextualised. There is a need to explore new ‘registers’ for intervention development and evaluation that consider the potential for peer supporters to make a contextual difference. Furthermore, there is a need to explicate the relationship between maternal experience and health policy goals, to acknowledge the contested quality of the implementation context, to pay attention to the agency of mothers, and to develop a community-level theory of how change in infant feeding behaviour happens with which peer support can cohere. The thesis concludes by highlighting 18 points to support theory development for infant feeding interventions