Background: There is an increasing awareness among researchers and policymakers of the potential for healthcare interventions to have consequences beyond those initially intended. These so-called “spillover effects” result from complex features of healthcare organization and delivery and can either reinforce or hinder the effectiveness of interventions. Thus, the failure to properly consider them may lead to flawed evaluations and policy making. Despite this, attempts to account for spillover effects remain limited and uncoordinated across different disciplines.
Objective: Our objective is twofold. First, we conduct a scoping review to understand the approaches that have been taken to address spillover effects in the healthcare literature. Second, we develop a comprehensive framework to identify and conceptualize spillover effects.
Findings: We identify a wide range of different spillover effects, which can be experienced by agents not originally targeted by the intervention or be impacts on additional outcomes for targeted agents. We find that spillovers can lead both to multiplier effects and violation of crucial assumptions necessary for evaluation design. Despite being profoundly different in nature, distinctions between genuine spillover effects and sources of control group contamination that invalidate the assumptions of comparative evaluation designs are rarely drawn in the literature. Unanticipated behavioural responses, complementarities and resource sharing in the production of healthcare also emerge as important drivers of spillover effects. We find little use of well-developed theoretical concepts to explain the mechanisms underpinning spillover effects. Building on these findings, we develop a comprehensive INTENTS framework (Intended Non-intended TargEted Non-Targeted Spillovers) to classify types of spillovers. We also found regularities in the reported mechanisms underpinning different type of spillover effects.
Discussion: Our scoping review highlights the need for a thoughtful consideration of the nature and mechanisms of spillover effects in healthcare, in the interest of improved policymaking. We suggest that potential spillovers should be classified at intervention design stage, distinguishing the group of subjects affected and the outcome involved in the spillover, clarifying if they are direct or mediated by specific mechanisms. To this end, we suggest a taxonomy of spillover effects and propose a comprehensive framework suggesting relevant steps for their identification.
Back to Research