Patterns in bottlenecks for implementation of health promotion interventions: a cross-sectional observational study on intervention-context interactions in the Netherlands

BACKGROUND: From a complex systems perspective, implementation should be understood as the introduction of an intervention in a context with which it needs to interact in order to achieve its function in terms of improved health. The presence of intervention-context interactions could mean that during implementation particular patterns of crucial interaction points might arise. We examined the presence of - and regularities in - such 'bottlenecks for implementation', as this could create opportunities to predict and intervene in potential implementation problems. METHODS: We conducted a cross-sectional observational study against the background of municipal intersectoral policymaking in the Netherlands. We asked implementers of health promotion interventions to identify bottlenecks by rating the presence and importance of conditions for implementation in a range of intervention systems. We used descriptive statistics to characterize these systems (by their behaviour change method, health theme and implementation setting) and the conditions that acted as bottlenecks. After stratifying bottlenecks by intervention system and the system's characteristics, we tested our hypotheses by comparing the number and nature of the bottlenecks that emerged. RESULTS: More than half of the possible conditions were identified as a bottleneck for implementation. Bottlenecks occurred in all categories of conditions, e.g., relating to the implementer, the intervention, and political and administrative support, and often connected with intersectoral policymaking, e.g., relating to the co-implementer and the co-implementer's organization. Both our hypotheses were supported: (1) Each intervention system came across a unique set of - a limited number of - conditions hampering implementation; (2) Most bottlenecks were associated with the characteristics of the system in which they occurred, but bottlenecks also appeared in the absence of such an association, or remained absent in the presence thereof. CONCLUSIONS: We conclude that intervention-context interactions in integrated health policymaking may lead to both regularities and variations in bottlenecks for implementation. Regularities may partly be predicted by the function of an intervention system, and may serve as the basis for building the capacity needed for the structural changes that can bring about long-lasting health improvements. Variations may point at the need for flexibility in further tailoring the implementation approach to the - mostly unpredictable - problems at individual sites.