Methods: This was a prospective observational study of hospitalized patients with UTI between 1/2015-2/2016. Adult patients with community-onset UTIs were included if they had symptoms of UTI, pyuria, and received ≥48h of antibiotics. Patients were grouped by ESBL vs non-ESBL EK UTI and compared for clinical presentation, antibiotic therapy, and outcomes.
Results: 243 patients met inclusion criteria, of which 40% had ESBL EK. Mean age was 54yo, 34% were male, and 90% resided at home prior to admission. 23% of patients met sepsis3 criteria. The most common chief complaint was fever/chills (ESBL 44% vs. non-ESBL 53%, p=0.2) then dysuria (ESBL 37% vs. 32%, p=0.5). Over one-third of patients had pyelonephritis (ESBL 38% vs. 41%, p=0.7) and concurrent bacteremia was seen in 23% of all patients. Upon admission, APACHE II scores were significantly higher in the ESBL group (13 vs 10, p=0.03). Empiric ceftriaxone therapy was more common in the non-ESBL group (73% vs 55%, p=0.02), while a carbapenem was more common in the ESBL group (20% vs 4%, p<0.0001). Of the ESBL group who received empiric carbapenem, 65% had prior history of ESBL UTI. Duration of inpatient antibiotic therapy was similar (ESBL 5 vs 4d, p=0.09). Despite over 2/3 of patients receiving ineffective therapy empirically in the ESBL group, the need for ICU admission (ESBL 34% vs 24%, p=0.1) and 30d mortality (ESBL 2% vs 1%, p=1) were similar. However, less ESBL patients achieved early clinical success at 96h (76% vs 89%, p=0.01) and LOS was 1d longer (p=0.06) primarily due to delay in discharge for receipt of IV antibiotic.
Conclusion: Hospitalized patients for UTI did not significantly differ in clinical presentation regardless of ESBL or non-ESBL cause. A delay in initiation of effective therapy in the ESBL group delayed clinical response and prolonged discharge. Methods for early detection of resistance such as rapid diagnostics will be beneficial for early initiation of effective therapy without overuse of carbapenem therapy.