4.1 Representativeness

GLASS recognises that most patient-level data are derived from hospital-based surveillance sites. As hospitals process many or most of their samples from hospital inpatients, samples from inpatients are often over-represented in the AMR surveillance data – this can lead to the assumption that AMR levels are higher than they actually are in the whole population. As national AMR surveillance systems mature (which happens over many years), there is an expectation that the choice of surveillance sites should expand to include outpatient healthcare facilities, for example, general practices, renal dialysis units (which often see the same cohort of patients regularly) and nursing homes. This allows for community AMR trend monitoring. Diversity or representativeness of AMR data is key, as it ensures that policy decisions, strategies and interventions developed at national level are based on data from the widest possible population. Surveillance sites should ensure consistency in procedures for sample collection and reporting to the national AMR surveillance system.