Methods for interventions using networks to improve health: A narrative synthesis of methodological research on network data collection, visualisation and intervention

The methods used to study and intervene in networks have evolved over the past 90 years. Network research has gained strength from being an interdisciplinary field, but this has also meant that the innovations occurring within some disciplinary communities may not have disseminated to others. Disciplines focussing on health improvement may not have adopted innovations coming from other disciplines relevant to network science. Through a review of the literature, we focus on the key aspects which can inform community-based networks and health improvement (NHI) projects. The review aims to draw these innovations together to understand the range of methods currently available for NHI research and practice. We conducted a narrative synthesis of published literature that may be relevant for NHI projects, synthesising existing work on methods of network data collection, visualisation and intervention approaches. Searches were conducted between the 8–11th January 2021, within the following databases; ACM Digital Library, EconLIt, ERIC, IEEE Explore, Medline, PsycInfo, Scopus, Social Sciences Citation Index and Sociological Abstracts. The expert community was also consulted to identify relevant research. Searches focused English language papers relating to methods for data collection, visualisation, and implementing interventions, but not statistical analysis. The search was not restricted to studies applied to health. We used a systemic review methodology to identify peer-reviewed articles that met pre-defined inclusion criteria. Data extraction was restricted to 84 papers published since 2018, of which some were included within more than one category (Network Data collection N = 41, Network Visualisation N = 32, Network Intervention N = 30). Analysis uncovered a diverse range of approaches to collecting, visualising and interventions using network data, and based on the included studies (rather than existing typologies) we developed a preliminary threefold typology of network and health improvement methods. We found nine types of network visualisation, eight types of data collection, and six types of network intervention approaches. Data visualisations commonly used node-link (circle-line) diagrams to visualise networks whilst key player interventions mostly used whole network data collection, and interventions using personal reflections of networks mostly used egonets. Visualisation was a feature of eight out of 30 intervention papers. Evaluation of the network methods was highly variable. Our findings suggest potential areas for future methodological research around the use of network methods in community interventions, we propose further integration of data collection and visualisation approaches as part of intervention design, and encourage the network intervention community to integrate methods testing as part of their project to improve the evidence base for network methodology.