Social contact patterns derived from an epidemiological survey and GPS-based co-location data – A systematic comparison using parallel data collections during the COVID-19 pandemic in Germany

The parametrisation of contact behaviour is crucial for infectious disease transmission models. Contact information derived from self-reported surveys and from co-location in space and time (GPS-based) may reflect different dimensions of contact behaviour, which might be associated with distinct epidemiological risks depending on the contagion of interest. This study explores whether and how contacts measured using these distinct approaches exhibit similar or complementary contact patterns. We compare the mean number of contacts and the mean excess number of contacts (i.e. the ratio of mean squared contacts to mean contacts) from the COVIMOD survey and NETCHECK GPS co-location data between April 2020 and December 2021. While mean contacts measure contact intensity, mean excess contacts reflect dispersion, which is important for understanding superspreading behaviour. Mean contacts were considerably higher in co-location data (11.04; 95 %CI: 10.90-11.19) than in survey data (3.38; 95 % CI: 3.30-3.47); however, both data sources correlated well with each other. Mean excess contacts were similar during periods of strict non-pharmaceutical interventions (NPIs) but diverged when NPIs were lifted, with co-location data values rising more markedly. Setting-specific contact patterns also differed, potentially due to methodological differences in setting classification and data capture. Furthermore, regional variation was more pronounced in co-location data, with densely populated city-states showing higher contact numbers. Comparative insights from the two data sources demonstrate that GPS-based and survey-based contact data capture complementary and distinct aspects of human interaction. Combining both sources could provide a more comprehensive picture of human interactions relevant to infectious disease modelling.