Novel surveillance design exploiting network analysis for efficient, rapid detection of emerging Clostridioides difficile strains

Whole genome sequencing is being introduced in England to support national surveillance of key hospital-acquired pathogens, starting with Clostridioides difficile, to facilitate novel strain identification and enable timely interventions. However sequencing capacity is limited. Symptomatic and asymptomatic patients attending multiple hospitals can act as inter-facility transmission vectors; here we therefore identify the empirical network of shared patients from analysing national admission data and simulate spread of a hypothetical novel strain. Algorithmically optimising detection, incorporating logistical constraints, we identify sentinel sites which detect a novel strain 27% faster than random sentinel selection, whilst sequencing <15% cases. Sensitivity and scenario analyses using a range of plausible pathogen characteristics and historical networks confirm epidemiologically- and longitudinally-robust sentinel selection and performance. The new surveillance system, established from our findings, benefits from stress-tested sentinel set selection to deliver rapid, efficient identification of novel strains within real world constraints, to inform control interventions, and provides a roadmap for future hospital-acquired pathogen surveillance.