Background: Timely treatment of sepsis is the cornerstone of reducing sepsis-related morbidity and mortality, and a large body of work has demonstrated improved patient outcomes with initiation of early empiric antibiotics. Time to antibiotics after first sepsis recognition has increasingly gained attention as a quality of care metric with Surviving Sepsis Guidelines recommending initiation of antibiotics within one hour of sepsis recognition. While much work has focused on expediting provider identification of sepsis and antibiotic ordering, little is known about factors that may delay antibiotic administration after order placement. We sought to determine 1) time to antibiotic administration after STAT antibiotic order and 2) identify patient and system factors associated with antibiotic delays.
Methods: We conducted a retrospective study of patients presenting to an emergency department (ED) at a large academic medical center with suspected sepsis, (defined as ≥ 18yo, having blood cultures drawn and antibiotics received within 48 hours of ED presentation), between January 1, 2019 and June 30, 2021. Demographic, vital sign, clinical outcome, medication data, process measures, and system factors were obtained from the EHR along with date and clock timestamps (Table 1). In order to limit analysis to patients deemed by the ordering provider most necessary to administer antibiotics without delay, the study sample was limited to those with “STAT” antibiotic order. Our primary outcome was defined as time from antibiotic order to administration (OTA), in minutes. We dichotomized patients as “Delayed OTA,” defined as > 60 minutes between order and administration vs “Timely OTA,” defined as ≤60 minutes. We then used bivariable and multivariable regression models to examine associations between patient characteristics, process and system factors on OTA.
Results: A total of 6,784 patients met inclusion criteria. Overall median time from antibiotic order to antibiotic administration (OTA) was 43 minutes (IQR 25, 70). Of these 4,610 (68%) patients were found to have “Timely OTA,” whereas 2,174 (32%) had “Delayed OTA” . Compared to those with timely OTA, patients with delayed OTA were more likely to be female (p=0.03), less likely to have Limited English proficiency (p=0.03) and had lower comorbidities (Elixhauser index) (p=0.04) and lower severity of illness on presentation (qSOFA on triage, triage priority, admission level of care, Table 1, p-values < 0.001). System factors associated with delayed OTA included more hours spent in the ED (10.5 vs 9.7, p-values < 0.001), weekday order (as opposed to weekend) (p=0.002), and patient changing location in the hospital between order time and administration (p< 0.001), or order occurring after arrival to hospital inpatient unit (p< 0.001). Of all antibiotic types administered, delayed OTA was noted to be associated with patients receiving vancomycin (median minutes = 65 (39, 100), 27% in delayed OTA vs 10.7% in timely OTA). In a fully adjusted multivariable model, independent factors associated with increased risk of delayed OTA were primarily health system factors (Table 2).
Conclusions: Our study demonstrates that time between antibiotic order and administration is commonly delayed, and is an important quality metric in the era of the 1-hour bundle and Surviving Sepsis Campaign. This study demonstrates that several health system factors contribute to these delays and could serve as modifiable targets for quality improvement interventions.
