Article Text

Download PDFPDF
OP120 Development of a tool to conceptualise and measure agentic demands of population health interventions for diet and physical activity
  1. Kate Garrott1,
  2. Jenna Panter1,
  3. Amanda Sowden2,
  4. Mark Petticrew3,
  5. David Ogilvie1,
  6. Martin White1,
  7. Jean Adams1
  1. 1MRC Epidemiology Unit, University of Cambridge, Cambridge, UK
  2. 2Centre for Reviews and Dissemination, University of York, York, UK
  3. 3Department of Public Health, Environments and Society, London School of Hygiene and Tropical Medicine, London, UK


Background Population health interventions (PHIs) vary by the agentic demands they place on individual recipients. For example, ‘highly agentic’ interventions such as information campaigns rely on recipients noticing, engaging and responding to the intervention, whereas ‘less agentic’ interventions, such as food reformulation require little action from recipients to achieve public health benefit. Highly agentic interventions often require recipients to undertake intermediate steps to engage in intended behaviours and be able, motivated and free to act. These attributes are socio-economically patterned, and theories therefore posit that the agentic demands of different interventions may influence intervention effectiveness and equity. To date, systematic evidence to confirm these associations is lacking. Providing such evidence requires a tool for classifying the agentic demands of interventions, and we aimed to develop such a tool.

Methods Tool development was in four stages. Firstly, we designed a systematic search to identify a broad range of PHIs (n=73). Secondly, we coded the actors involved and the actions required to enable each PHI to achieve its intended effect. We merged coding for similar intervention types to generate higher order intervention schemas (n=8), which were compared to identify key distinctions and generate a draft tool. Thirdly, online workshops (n=4) with purposively sampled academic, policy and practice experts (n=20) assessed the tool’s content validity, and informed revisions. Lastly, we explored inter-rater reliability by recruiting a further experts (n=22) to apply the tool to intervention examples (n=53).

Results The tool divides PHIs into single intervention components and identifies three concepts influencing their agentic demands: (1) exposure to the component (active/passive); (2) mechanism of action (cognitive/social/economic/biomedical/physical environment); (3) engagement with the mechanism of action (active/passive). The tool combines these concepts to categorise intervention components within a 20-category matrix. Users also identify additional actors (macro-environmental; micro-environmental; gatekeepers and secondary recipients) required to exert agency in intervention implementation. Inter-rater reliability varied by concept and was hampered by poor intervention descriptions in primary studies. Cohen’s kappa was 0.35 for exposure, 0.66 for mechanism of action and 0.27 for engagement.

Conclusion Our tool presents a novel method for classifying the agentic demands of PHIs. It identifies key features of intervention design that influence agentic demand, which may be useful to consider when developing multi-component PHIs. We are clarifying user instructions, which may enhance inter-rater reliability. We plan to use the tool within a systematic review to explore whether agentic demands of PHIs are associated with effectiveness and equity.

  • Interventions
  • Diet
  • Physical Activity

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.