Background: Delayed or in vitro inactive empiric antibiotic therapy may be detrimental to survival in patients with bloodstream infections (BSIs). Understanding the landscape of delayed or discordant empiric antibiotic therapy (DDEAT) across different patient, pathogen, and hospital types, as well as by their baseline resistance milieu, may enable providers, antimicrobial stewardship programs, and policy makers to optimize empiric prescribing. Methods: Inpatients with clinically suspected serious infection (based on sampling of blood cultures and receiving systemic antibiotic therapy on the same or next day) found to have BSI were identified in the Cerner Healthfacts EHR database. Patients were considered to have received DDEAT when, on culture sampling day, they received either no antibiotic(s) or none that displayed in vitro activity against the pathogenic bloodstream isolate. Antibiotic-resistant phenotypes were defined by in vitro resistance to taxon-specific prototype antibiotics (eg, methicillin/oxacillin resistance in S. aureus) and were used to estimate baseline resistance prevalence encountered by the hospital. The probability of DDEAT was examined by bacterial taxon, by time of BSI onset, and by presence versus absence of antibiotic-resistance phenotypes, sepsis or septic shock, hospital type, and baseline resistance. Results: Of 26,036 assessable patients with a BSI at 131 US hospitals between 2005 and 2014, 14,658 (56%) had sepsis, 3,623 (14%) had septic shock, 5,084 (20%) had antibiotic-resistant phenotypes, and 8,593 (33%) received DDEAT. Also, 4,428 (52%) recipients of DDEAT received no antibiotics on culture sampling day, whereas the remaining 4,165 (48%) received in vitro discordant therapy. DDEAT occurred most often in S. maltophilia (87%) and E. faecium (80%) BSIs; however, 75% of DDEAT cases and 76% of deaths among recipients of DDEAT collectively occurred among patients with S. aureus and Enterobacteriales BSIs. For every 8 bacteremic patients presenting with septic shock, 1 patient did not receive any antibiotics on culture day (Fig. 1A). Patients with BSIs of hospital (vs community) onset were twice as likely to receive no antibiotics on culture day, whereas those with bloodstream pathogens displaying antibiotic-resistant (vs susceptible) phenotypes were 3 times as likely to receive in vitro discordant therapy (Fig. 1B). The median proportion of DDEAT ranged between 25% (14, 37%) in eight <300-bed teaching hospitals in the lowest baseline resistance quartile and 40% (31, 50%) at five ≥300-bed teaching hospitals in the third baseline resistance quartile (Fig. 2). Conclusions: Delayed or in vitro discordant empiric antibiotic therapy is common among patients with BSI in US hospitals regardless of hospital size, teaching status, or local resistance patterns. Prompt empiric antibiotic therapy in septic shock and hospital-onset BSI needs more support. Reliable detection of S. aureus and Enterobacteriales bloodstream pathogens and their resistance patterns earlier with rapid point-of-care diagnostics may mitigate the population-level impact of DDEAT in BSI.
Funding: This study was funded in part by the National Institutes of Health Clinical Center, National Institutes of Allergy and Infectious Diseases, National Cancer Institute (NCI contract no. HHSN261200800001E) and the Agency for Healthcare Research and Quality.