Relevance of social contact definitions for use in infectious disease transmission modeling: a systematic review and recommendations

Abstract Background Social mixing studies provide crucial evidence for parameterizing mathematical models of infectious disease transmission, and their validity for use in models is dependent on their study design and how well the contacts they capture map to the transmission routes of the pathogen of interest. This systematic review aims to catalogue contact definitions used in social mixing studies from 2005 to 2024 and evaluate their conceptual alignment to three archetypal pathogens. Methods We searched Ovid Medline, Embase, Scopus, and Global Index Medicus for studies of human-to-human interactions that collected data via survey, diary, or interview published between January 2005 and August 2024. We excluded studies that focused on sexually-transmitted or food/water/vector-borne pathogens or used only GPS or sensor data. We excluded studies of contact tracing as a public health effort. Screening and data collection were conducted in Covidence. Contact definitions were presented verbatim and qualitatively coded to identify key elements. Results were stratified by prospective or retrospective design. We used the Mixed Methods Appraisal Tool to evaluate risk of bias. Results We included 112 eligible studies, half (52.7%) of which began during pandemic periods (H1N1 [2009] or COVID-19 [2020]), commonly in the United States (18%) or United Kingdom (16%). Most (68%) had a retrospective design in which participants were asked to recall contacts that occurred prior to survey administration and 73% used a single-survey cross sectional design using random (16%) or stratified random (44%) sampling. Relevant contacts were often defined by an exchange of words (77%) or physical touch (59%). A minority of studies differentiated between contacts that occurred outdoors vs. indoors (28%) or allowed participants to report large group contacts separately from individually named contacts (28%). Contact definitions and attributes conceptually aligned with the transmission biology of influenza, tuberculosis, and norovirus in 77%, 6%, and 50% of studies respectively. Conclusions Contact studies were limited in their global scope and non-pandemic representativeness. Critical modifiers of transmission risk such as location, household membership, and shared space with large numbers of people were under-measured. Future modeling and social mixing studies should align measured contact rates to the target transmission routes by incorporating these elements.