Utilizing the Electronic Health Record to Construct Syndrome-Specific Antibiograms for Previously Healthy Children

BACKGROUND: Institutional antibiograms can help guide empiric antibiotic therapy but may miss differences in resistance across patient sub-populations or clinical syndromes. The usefulness of antibiograms that estimate the risk of organism-drug mismatch for community-acquired skin and soft-tissue infections (SSTIs) and urinary tract infections (UTIs) in pediatrics is poorly understood. METHODS: We constructed a complete line-listing of Staphylococcus aureus isolates from skin and soft tissue body sites (October 1(st), 2015 to May 1(st), 2017) and Gram-negative bacilli from urine isolates (October 2(nd), 2016 to May 1(st), 2017) from patients ≤18 years hospitalized at Lurie Children’s Hospital. A cohort of previously healthy patients, defined as those without comorbidities (using ICD-10 diagnostic coding for complex medical conditions), prior admissions (within 1 year), or recent antibiotic use (within 90 days), was constructed using administrative and pharmacy data from the electronic health record (EHR). Syndrome-specific antibiograms were generated for commonly used empiric agents. RESULTS: A total of 767 SSTI and 812 UTI isolates were identified. Compared with UTI isolates from all patients, antibiotic susceptibility rates were significantly higher among previously healthy patients for amoxicillin/clavulanic acid (77.0% vs. 85.7%, P = 0.0003), cefazolin (58.3% vs. 65.4%, P = 0.018), and ceftriaxone (86.2% vs. 92.1%, P = 0.0025), but not for trimethoprim/sulfamethoxazole (70.5% vs.. 76.5%, P = 0.15). A comparison of SSTI isolates did not reveal differences in susceptibility for clindamycin (77.6% vs. 81.7%, P = 0.17), oxacillin (71.3% vs. 73.7%, P = 0.48), and trimethoprim/sulfamethoxazole (89.7% vs. 92.6%, P = 0.24). CONCLUSION: We demonstrated the feasibility of constructing EHR-derived syndrome-specific antibiograms for a cohort of children without chronic medical conditions. Our methodology can be applied to create antibiograms tailored to a variety of clinical presentations. For UTIs, automated antibiograms for previously healthy children can avoid overestimation of resistance and better guide empiric therapy. DISCLOSURES: L. Kociolek, Merck: Grant Investigator, Grant recipient