JMIR Publications

ABSTRACT

Background:

Rising antimicrobial resistance poses a significant threat to public health yet current tools to direct empiric antibiotic therapy remain limited.

Objective:

To develop more precise solutions that facilitate both patient safety and population health, we identified factors influencing resistance and associated risk of treatment failure.

Methods:

We analyzed five years of susceptibility testing and patient data across a large academic medical center. We developed algorithms that mapped these data to individual cases and implemented a prototype digital empiric antimicrobial support system which we evaluated against actual prescribing outcomes.

Results:

We determined patient-specific factors across syndromes and contexts and identified relevant local patterns of antimicrobial resistance by clinical syndrome. Mortality and length of stay differed significantly depending on these factors and could be used to generate heuristic targets for acceptable risk of under-prescription. Combined with the developed “remaining risk” algorithm, these factors can be employed to inform clinicians’ reasoning. Analysis of the system showed comparable results to clinicians for low-risk diseases such as UTI, while recognizing risk and recommending better coverage in high mortality conditions such as sepsis.

Conclusions:

The application of such data-driven, patient-centered tools may guide empirical prescribing for clinicians to balance morbidity and mortality with antimicrobial stewardship.