2440. Using a geospatially explicit agent-based model of a regional healthcare network to assess varied antibiotic risk on Clostridioides difficile infection incidence

Abstract

Background

Different antibiotic classes are associated with different Clostridioides difficile infection (CDI) risk. The impact of varied antibiotic risk on CDI incidence can be explored using agent-based models (ABMs). ABMs can simulate complete systems (e.g., regional healthcare networks) comprised of discrete, unique agents (e.g., patients) which can be represented using a synthetic population, or model-generated representation of the population. We used an ABM of a North Carolina (NC) regional healthcare network to assess the impact of increasing antibiotic risk ratios (RRs) across network locations on healthcare-associated (HA) and community-associated (CA) CDI incidence.

Methods

The ABM describes CDI acquisition and patient movement across 14 network locations (i.e., nodes) (11 short-term acute care hospitals, 1 long-term acute care hospital, 1 nursing home, and the community). We used a sample of 2 million synthetic NC residents as ABM microdata. We updated agent states (i.e., location, antibiotic exposure, C. difficile colonization, CDI status) daily. We applied antibiotic RRs of 1, 5, 8.9 (original model RR), 15, and 20 to agents across the network to simulate varied risk corresponding to different antibiotic classes. We determined network HA-CDI and CA-CDI incidence and percent mean change for each RR.

Results

In this simulation study, HA-CDI incidence increased with increasing antibiotic risk, ranging from 11.3 to 81.4 HA-CDI cases/100,000 person-years for antibiotic RRs of 1 to 20, respectively. On average, the per unit increase in antibiotic RR was 33% for HA-CDI and 6% for CA-CDI (figure).

Conclusion

We used a geospatially explicit ABM to simulate increasing antibiotic risk, corresponding to different antibiotic classes, and to explore the impact on CDI incidence. The per unit increase in antibiotic risk was greater for HA-CDI than CA-CDI due to the higher probability of receiving antibiotics and higher concentration of agents with other CDI risk factors in the healthcare facilities of the ABM. These types of analyses, which demonstrate the interconnectedness of network healthcare facilities and the associated community served by the network, might help inform targeted antibiotic stewardship efforts in certain network locations.

Disclosures

All authors: No reported disclosures.