Both the O’Neill review and the first iteration of the Global Research on Antimicrobial Resistance (GRAM) project have significantly heightened global awareness about antimicrobial resistance (AMR) as a critical public health issue, and emphasized the importance of quantifying its burden. Bacteria causing various infections can be isolated from two types of patient samples: non-sterile (e.g., urine) and sterile (e.g., blood). While sterile samples often correspond to more severe infections, non-sterile ones are easier to collect and more readily available. Therefore, an important clinically-relevant question is whether AMR levels in non-sterile samples are representative of those in sterile samples. Most information used to estimate the burden of AMR comes from passively collected hospital surveillance data worldwide. However, industry-led datasets openly available through the Vivli initiative could complement existing surveillance initiatives and support the analysis of extended research questions. Since these industry datasets include information on the sample types of bacterial isolates (e.g., blood, urine, gastrointestinal, respiratory…), we propose to use these datasets to investigate whether AMR levels in different sample types are correlated. Our study will compare bacterial isolates from sterile samples, such as bloodstream infections, to those from non-sterile samples, such as urinary tract infections. We will use correlation and regression methods to estimate whether there is a significant difference in the prevalence of resistance between sample types across various geographies and years available. This analysis aims to determine whether AMR estimates from non-sterile samples can serve as a reliable proxy for those in sterile infections. We will focus on selected bacteria and antibiotic combinations, chosen based on clinical priority and data availability.
Center for Global Research Data