Geospatial Modeling for Antibiotic Resistance in Hospital-Acquired Infections in the United States

Abstract

Antibiotic resistance is a major public health crisis due to the lack of new drugs in development and the continuous rise of resistance mechanisms. On the other hand, hospitalized patients are particularly vulnerable to obtaining a hospital-acquired infection, which increases the risk for death and injury. Unfortunately, the hospital setting often fosters the rise of resistance due to the high usage of antibiotics and the increased density of bacteria and vulnerable patients. Ultimately, this leads to higher risk of antibiotic resistance in hospital-acquired infections. Implementing infection control and antibiotic stewardship policies require sufficient knowledge of local resistance trends. However, data on antibiotic resistance specific to the hospital setting are sparse. This study analyzes point-prevalence data of antibiotic resistance rates in hospital-acquired infections utilizing spatial interpolation for five pathogens. Along with machine learning models and stacked generalization, environmental covariates are combined with unique location data points to make continuous predictions of resistance across the United States. I found that resistance rates were low in hospital-acquired infections in the United States, although high rates of resistance were identified for Acinetobacter spp. The map of predictions highlights potential hotspots of resistance in hospital-acquired infections, allowing for more targeted interventions and research.