Improving Time to Stat Intravenous Antibiotic Administration: An 8-Year Quality Initiative Free

Our institution is a tertiary children’s hospital within a large academic medical center. Care settings include a level I PICU, a level IV NICU, a pediatric emergency department (PED) associated with the level I pediatric trauma center, an acute care pediatrics (ACP) unit, and a newborn nursery. The hospital contains 189 inpatient beds with ∼350 NICU admissions, 2500 ACP admissions, 1000 PICU admissions, and 20 000 PED visits per year. Stat IV antibiotics are ordered in the electronic health record (EHR) by attending physicians, residents, fellows, and advanced practice providers. Clinicians determine if a stat IV antibiotic is warranted on the basis of a clinical presentation with concern for serious infection or sepsis.⁹ 

Our QI team made multiple institutional changes and interventions over an 8-year period. Interventions addressed key drivers impacting time to antibiotics (Fig 1). Broad categories of interventions included the shared mental model, pharmacy, EHR utilization, and increasing bedside resources. These interventions were chosen on the basis of previously published QI work, membership in national sepsis collaboratives, and identification of barriers in our specific context.^4–8 Table 1 includes additional institutional-level changes, as well as unit-specific timing of interventions during the study period.

Although some sepsis QI work started on individual units earlier, the formation of the Pediatric Sepsis Committee (PSC) in 2013 was the first institution-wide initiative to improve the treatment of patients with sepsis. The PSC is a multidisciplinary team with representation from inpatient and ambulatory care units. With regular monthly meetings, they planned and executed hospital-wide and unit-based interventions. In January 2016, the PSC started sharing unit-specific time-to-stat IV antibiotic data with individual units with a shared aim of increasing the percentage of stat IV antibiotics administered in <1 hour to >70%. Unit representatives analyzed and presented monthly data at PSC meetings. Unit-level run charts were frequently posted on units to help motivate improvement at the grassroots level.

The implementation of the American Academy of Pediatrics (AAP) Pediatric Septic Shock Collaborative (PSSC) sepsis screening tool targeted the prompt identification of patients with a serious infection but also promoted a shared mental model with triggered huddles between clinicians and nurses. In January 2016, a paper version of the screening tool was initiated in the PED and subsequently rolled out in the ACP unit and PICU. Between February and November 2019, the manual screening process transitioned to an automatic sepsis notification within the EHR.⁷ 

Pharmacy delivery of IV antibiotics was a common reason for the delay of antibiotic administration. The PED had used ADMs for some time, and this was felt to be a significant driver of prompt antibiotic administration. Therefore, in January 2015, IV antibiotics, including ampicillin, cefazolin, cefepime, ceftriaxone, clindamycin, gentamicin, meropenem, piperacillin-tazobactam, and vancomycin were added to the ADMs in the PICU, NICU, and ACP.

In September 2015, sepsis order sets were implemented to improve timely, goal-directed care. Order sets were created with input from the PSC and included unit-specific broad-spectrum antibiotics, fluids, laboratory, and monitoring orders. Automated sepsis alerts were rolled out in February 2019. In the months that followed implementation, screening criteria were refined to reduce false positive and negative alerts.

In April 2017, to increase resources for bedside nursing and increase unit awareness, an automated page was sent to the unit charge nurse when stat antibiotics were ordered. This provided additional support to the bedside nurse, both for the ill patient with new orders and to help shift resources to care for that nurse’s other patients so that they could focus on tasks such as difficult IV access and pulling urgent antibiotics from the ADM.

The Difficult Intravenous Access (DIVA) initiative formalized an escalation pathway for patients with difficult vascular access. This process limited access attempts and promoted early escalation to unit access experts and anesthesia team members who employed additional strategies for access, including ultrasound guidance. The first DIVA guideline created for the ED was published in August 2018, and an inpatient DIVA guideline followed in October 2018.

Our primary outcome measure was the percentage of stat IV antibiotics delivered <1 hour from the time of order. Time to antibiotics was defined as the time from placement of a stat IV antibiotic order to charted medication administration. Time to antibiotics has previously been used as a process marker in QI work.¹⁰  There are limited single-center pediatric data and several adult studies that reveal an association between mortality and delayed antibiotics in sepsis.^1,2,11–13  As a balancing measure, we assessed PED and inpatient antibiotic usage to monitor for unintended overutilization of stat antibiotics and prolonged inpatient length of stay. We obtained data from the her and included stat IV antibiotics orders from 2012 to 2020 for all patients <18 years of age admitted to dedicated pediatric units. We included sepsis-specific antibiotics and specifically excluded cefazolin because of its common use in the perioperative setting. The sepsis-specific antibiotics included were Ampicillin, Cefepime, Ceftriaxone, Clindamycin, Gentamicin, Meropenem, Piperacillin-Tazobactam, and Vancomycin. If an individual patient had multiple stat antibiotics ordered at a single time, only the first administered antibiotic in a 3-hour period was included. We analyzed the impact of interventions on the time to stat IV antibiotics with statistical process control (SPC) charts (Microsoft Excel 2016, Redmond, WA). We defined special cause variation using the rules suggested for the interpretation of Shewhart charts in health care.¹⁴ 

The institutional review board at our institution determined that this project does not meet the criteria for human subject research.

We used 20 months of historical data from January 2012 to August 2013 to establish a baseline. Our baseline frequency of stat IV antibiotics <1 hour was 33% in the 20 months before implementation. After the implementation of QI interventions between August 2013 and July 2020, the frequency increased to 77% (Fig 2 and Supplemental Table 2). A total of 5 centerline shifts occurred during the study period. Special cause variation first occurred in March 2014, 5 months after the chartering of the PSC, with improvement from 33% to 40% of antibiotics administered in <1 hour. The centerline again shifted to 47% in April 2015 after the introduction of antibiotics in ADMs on units and the joining of the AAP PSSC in January 2015. In January 2016, the PSC first began disseminating monthly time to antibiotic administration data, and manual sepsis screening was initiated. These interventions were followed by a centerline shift in May 2016 with an increase to 56% of stat antibiotics administered in <1 hour. Special cause variation was noted in April 2017 with the implementation of automated pages to charge nurses for all stat antibiotic orders with a centerline shift to 67%. The final shift to 77% was in September 2018 after PED DIVA guidelines were published and implemented. From September 2018 to the end of the study period in July 2020, the centerline remained stable with 77% of stat antibiotics administered in <1 hour. SPC charts for individual units all reveal improvement over the study period (Supplemental Fig 3).

Inpatient and PED antibiotic utilization and inpatient length of stay were tracked as balancing measures (Supplemental Table 3). The average length of stay ranged from 4.11 to 4.64 days. Inpatient antibiotic utilization ranged from 5.56 to 7.63 orders per 100 inpatient days, with the highest number of orders per 100 inpatient days in 2012 and 2013 and the lowest in 2017. PED antibiotic orders per 1000 PED discharges ranged from 11.79 to 15.56, with the most orders in 2017 and the fewest in 2012. Overall, the average length of stay and antibiotic utilization in PED and inpatient units did not increase over the study period.

We describe a successful longitudinal QI effort in which interventions over an 8-year period were associated with decreased time to antibiotic administration across all pediatric units at a single institution. The goal of prompt antibiotic administration is supported by the most recent Surviving Sepsis guidelines.³  We improved our antibiotic delivery in <1 hour from 33% to 77% across all units over the 8-year study period.

The formation of the PSC in November 2013 was likely a significant driver in the improvement of timely antibiotic administration as part of work to elevate the care of children with sepsis. The interventions discussed were planned and conducted by this committee. The success of the interventions was made possible by the collaboration of pharmacists, nurses, physicians, microbiologists, clinical informaticists, and performance improvement specialists from across the hospital. The impact of this group is suggested not only by the overall improvement over time but also by the first centerline shift after the group’s formation in 2013 from 34% to 40% of antibiotics delivered in <1 hour.

In addition to the formation of the PSC, other interventions may have helped build a shared mental model among interprofessional team members. The implementation of manual sepsis screens in January 2016 brought bedside nurses and providers together to discuss patient care. In January 2016, the PSC also started sharing unit-specific time-to-stat antibiotic data. These 2 interventions were associated with a centerline shift in the months that followed to 56% of antibiotics administered in <1 hour. Although participation in national sepsis collaboratives such as the AAP PSSC and Children’s Hospital Association Improving Pediatric Sepsis Outcomes (IPSO) Collaborative in 2015 and 2017, respectively, were followed by centerline shifts, these shifts were likely due to other more direct interventions. AAP PSSC and IPSO collaboratives were powerful motivators for the adoption of innovative ideas and a continued drive toward improvement and likely had a more longitudinal impact.

Although pharmacists play a key role in order review and medication distribution, in serious infections, a delay in antibiotic administration may lead to harm. ADMs are a safe way to decrease the time from order to antibiotic administration in emergent situations.⁴  The inclusion of commonly ordered broad-spectrum antibiotics in ADMs in January 2015 was associated with a shift from 40% to 47% in March 2015.

EHRs are established tools for QI efforts. The use of EHR order sets and automated alerts may have improved adherence to guidelines and facilitated bundled care. Although order sets did not directly impact time from order to administration, the use of an order set may have led to increased bedside resources and team awareness of concern for serious infection. After recommendations from the Surviving Sepsis Campaign, the EHR was also used for the implementation of automated sepsis alerts. Data included in this project may not reflect the impact of the introduction of automated sepsis alerts on time from order to administration. However, recently published work supports the efficacy of this intervention and reveals automatic electronic sepsis alerts are useful in the early identification and prompt treatment of pediatric patients with sepsis.^10,15 

Bedside resources were identified as a barrier to timely antibiotics in our setting. We implemented difficult vascular access guidelines and automated charge nurse pages to address this gap. To our knowledge, automated paging of charge nurses for stat IV antibiotics orders is a novel intervention. The purpose of the automated page was to increase situational awareness and bring additional resources to the bedside in the form of additional nurses and support staff. Past QI work suggests that strategies to increase bedside nursing support are effective at reducing the time from recognition to first antibiotics.¹⁶  Special cause variation was noted in the first month of the implementation of this intervention.

Balancing measures of inpatient and PED antibiotic utilization and inpatient length of stay were chosen to assess for potential misidentification of sepsis. We hypothesize that increased sepsis awareness could potentially result in overutilization of antibiotics and longer hospital admissions. Within the study period of January 2012 to July 2020, there was no trend of increased utilization in any of the balancing measures selected.

The significance of this project is the sustained improvement over the course of the 8-year study period. Our most recent data reflects that 77% of stat IV antibiotics were administered in <1 hour. This achievement has been maintained over 22 months and reveals consistent and sustainable improvement. Although it is challenging to identify the single most effective intervention because of the number of institutional changes and interventions, we believe that the formation of a multidisciplinary PSC played an important role in both the overall success of the QI effort and the sustainable change.

Our initiative is limited by our use of time from provider order of stat IV antibiotics as a proxy for time of sepsis recognition. Time from antibiotic order to administration does not necessarily reflect the time from sepsis onset or even detection of sepsis to the administration of antibiotics. In addition, many stat IV antibiotics orders are placed for the treatment of serious infections, which may not meet the criteria for sepsis. There have been efforts to decrease utilization of the stat order for nonemergent situations by direct e-mail feedback to providers on the overuse of “stat”-level priority on routine antibiotics. Notably, our balancing measures did not reveal an increase in antibiotic utilization with these QI efforts (Supplemental Fig 3). Although previous literature has revealed that delayed treatment leads to poor outcomes and care in concordance with national guidelines improves outcomes, our study does not include data on patient outcomes.^1,17  The lack of additional process measures and outcome measures at the time of implementation makes it difficult to determine which interventions were most impactful. In addition, this QI effort spans a long timeframe. The interventions and institutional changes that are discussed here may not be all- inclusive, and there may be unidentified factors that contributed to improvements in timely antibiotic administration. The number of institutional changes and interventions and the overlapping nature of these events limit our ability to attribute improvement to specific interventions. Finally, these interventions were made at a single institution. These interventions in other contexts may not have the same results.

This project reveals how a series of focused QI initiatives were associated with sustained improvement in time from order to the administration of stat IV antibiotics. Our next steps include assessing whether our QI interventions may contribute to improved clinical outcomes in children with sepsis.