BACKGROUND: Children with intestinal failure (IF) on parenteral nutrition (PN) are at high risk for bacteremia, and delays in antibiotic administration have been associated with increased morbidity and mortality. We designed an emergency department (ED) quality improvement (QI) initiative to reduce time to administration of intravenous antibiotics in febrile children with IF on PN.
METHODS: Our aim was to decrease the mean time for febrile children with IF on PN to receive intravenous antibiotics by 50% to <60 minutes over a 12-month period. Secondary outcome measures were ED, hospital, and ICU length of stay (LOS). Our process measure was the rate of ordering recommended antibiotics, and our balancing measure was the rate of hypoglycemia. Interventions included increasing provider knowledge of IF, streamlining order entry, providing individualized feedback, and standardizing the triage process. Results were analyzed by using statistical process control methodology and time series analysis.
RESULTS: We identified 149 eligible ED patients, of which 62 (41.6%) had bacteremia. The mean time to antibiotics decreased after the onset of the QI initiative from 112 to 39 minutes, and the ED LOS decreased from 286 to 247 minutes, but the total length of hospital and ICU stays were unchanged. The rate of hypoglycemia was also unchanged.
CONCLUSIONS: Our QI intervention for febrile children with IF on PN shortened the time to receive antibiotics. Larger studies are needed to demonstrate the impact on overall LOS and mortality.
- CLABSI —
- central line-associated bloodstream infection
- CVC —
- central venous catheter
- ED —
- emergency department
- EHR —
- electronic health record
- IF —
- intestinal failure
- LOS —
- length of stay
- PN —
- parenteral nutrition
- QI —
- quality improvement
- SBS —
- short bowel syndrome
Short bowel syndrome (SBS) is a disease of insufficient intestinal nutritional absorption because of inadequate bowel length.1,2 In a 1992 study, researchers estimated there were 40 000 children nationally with SBS, and although there are no more recent published estimates, the population is thought to be growing in number.3 In its most severe form, SBS may lead to intestinal failure (IF), defined by the necessity for parenteral nutrition (PN), which must be delivered by an indwelling central venous catheter (CVC).4 Less commonly, IF may occur in children with adequate bowel length but inadequate absorption or motility.2
Children with IF are at high risk of bloodstream infection because of poor nutritional status, disrupted intestinal immunity, potential bacterial translocation across the bowel wall, and the presence of the CVC, which provides a locus for bacterial growth. As a result, when children with IF develop fever, they have a high rate of bacteremia. In 1 study, researchers demonstrated that 46% of hospitalized children with SBS had a bloodstream infection, whereas authors of another study reported a bacteremia rate of 53% among children with IF who presented to an emergency department (ED) with fever.5 Because bacteremia can rapidly progress to sepsis, shock, and death, and given that expedited empirical treatment in patients with sepsis has improved mortality rates, rapid administration of antimicrobial therapy is necessary in PN-dependent children presenting to medical attention with fever or other symptoms that are concerning for a bloodstream infection.6
Quality improvement (QI) initiatives to decrease time to antibiotics in febrile patients with oncologic diagnoses have been successful in the ED setting and have been associated with decreased ICU needs on admission in children with fever and neutropenia.7 However, there have been no reports of similar initiatives focused on reducing time to antibiotics in patients with SBS or IF and fever, despite the fact that these children have substantially higher rates of bloodstream infection than febrile oncology patients.5,8
We therefore conducted a QI initiative to improve the timeliness of antibiotic administration for PN-dependent patients with IF and fever presenting to our ED. Our primary aim was to reduce the mean time from ED arrival to antibiotic administration by 50% to <60 minutes within 12 months, with secondary aims to reduce ED and hospital length of stay (LOS).
This QI initiative was implemented within the ED of a freestanding quaternary care children’s hospital with ∼60 000 visits per year. The multidisciplinary IF clinic at the hospital manages ∼95 patients and has a previously reported rate of 1.5 CVC infections per 1000 catheter days.9 Children with IF who present to the ED are cared for primarily by ED staff and trainees. Consultation is obtained from a general surgery resident and attending physician if they have SBS and by a gastroenterology fellow if the etiology of their IF is functional or malabsorptive.
At the onset of the intervention, we targeted all patients with SBS who received either partial or total PN at home through a CVC and presented to the ED with a documented fever ≥38°C in the ED or by history. Although initially we targeted patients specifically with SBS from surgical resection, 2 weeks after initiation of the project, we expanded it to include patients with IF from medical causes as well. This was done both because it was difficult on ED arrival to determine the etiology of a patient’s IF and because patients with IF were at high risk of bloodstream infection regardless of the underlying cause of the disease. Patients were excluded if they had received any intravenous antibiotics in the previous 24 hours.
The initial planning step for the intervention was assembling a multidisciplinary team of physicians, nurses, pharmacists, and clinical assistants from the departments of emergency medicine, general surgery, and gastroenterology. Jointly, a driver diagram (Fig 1) was developed that identified the following key drivers: (1) rapid identification of patients with IF and fever at ED triage, (2) improving staff knowledge and awareness of IF and the risk of bloodstream infection, (3) timely feedback to providers regarding their time to antibiotics after an episode of care, and (4) faster ordering and improved communication with the pharmacy to prepare antibiotics more rapidly.
A Pareto chart was created to identify time periods during the ED visit when delays in the process were most common before our intervention (Fig 2), and the results were used to identify opportunities for improvement.
Our initial plan-do-study-act cycles focused on increasing awareness among frontline staff at the provider level and streamlining the ordering process at the system level, areas that had been identified as key drivers by our group.
Increasing Awareness of Frontline Staff
To improve staff awareness of the high risk of bloodstream infection in the IF population, a series of presentations at meetings were arranged for physicians and nurses, along with face-to-face nursing education. The presentations were prepared and given by 2 of the authors in various settings, including teaching conferences with emergency medicine fellows and monthly physician faculty and nursing meetings. Additional physician colleagues were also enlisted in the initiative, and continuing medical education credits from the American Board of Pediatrics were given for their efforts. Weekly e-mails were sent to the ED staff dispelling SBS “myths” such as a perceived (but incorrect) prohibition on using the patient’s CVC for laboratory draws and antibiotic administration. These e-mails also included reminders not to wait on consultant input before initiation of antibiotics and provided updates on the progress of the initiative. Signage was placed in several locations in the ED to remind staff of the initiative as well as reminders within our daily team huddles at the start of each shift. Scripted nursing education about SBS and reminders of the initiative were provided before and during shifts. Each nurse was educated in person by at least 1 other nurse during the improvement project. These initiatives were implemented over a 2-week intervention period.
Streamlining Order Entry
To address confusion surrounding antibiotic ordering, information technology support staff helped to develop a standardized antibiotic order set that limited antibiotic choices to those preferred for empirical bloodstream infection coverage in the IF population at our institution (piperacillin-tazobactam and vancomycin) and provided recommendations on which medications and laboratory tests should be ordered routinely. Automatic paging to the appropriate consultant was also embedded into the order set.
Modifying the Triage Process
Two months into the improvement project, the triage process was modified by assigning the triage nurse or charge nurse to identify eligible patients presenting to the ED and then to notify the appropriate attending physician to expedite their care. This responsibility was communicated to the nursing staff through monthly nursing meetings and face-to-face education.
Three months after the start of the intervention, provider feedback was initiated. When a patient with IF and fever presented to the ED, the attending physician, fellow, and nurse responsible for the care of the patient were informed via e-mail the next day of their quality metrics. Providers were also asked to identify factors that may have contributed to delays in care. Additionally, any special cause points identified facilitated an in-person meeting with the providers to better identify barriers to successful antibiotic administration. Finally, divisional progress reports were presented monthly at physician faculty meetings.
The initiative began on October 1, 2015, and continued through October 31, 2016, during which time data were reviewed weekly by the project leaders, the core project group met monthly, and subspecialty services were updated through e-mails and formal presentations periodically. At the end of this period the initiative went into a quiescent phase, with ongoing data monitoring but cessation of active promotion.
Study of the Intervention
Eligible patients were identified by querying the hospital data warehouse between January 1, 2014, and September 30, 2015, for ED encounters of patients ≤21 years old with an International Classification of Diseases, Ninth Revision or International Classification of Diseases, 10th Revision diagnosis code for SBS and the presence of either a chief complaint of fever or a temperature at the time of ED check-in of 38°C or higher. Two study authors then reviewed the electronic medical record to determine if the identified encounter met eligibility criteria.
The main process measure was time from ED arrival to the administration of the first parenteral antibiotic. An additional process measure assessed order set adherence by measuring the proportion of encounters in which the recommended antibiotic regimen was ordered. Time to admission order was also assessed in the pre- and postintervention groups. Outcome measures included 30-day mortality and ED, hospital, and ICU LOS. Two balancing measures were assessed: rate of hypoglycemia (defined as glucose <60 mg/dL) (because of concern that after abrupt discontinuation of the PN, patients would have rebound hypoglycemia from high circulating insulin levels) and rate of floor-to-ICU transfer within 24 hours of ED discharge (because of concern that more rapid ED disposition may lead to inadequate evaluation of a patient’s clinical deterioration).
A pre-post cohort design was used, and differences in process, outcome, and balancing measures between the 2-time periods were assessed by statistical process control methodology and time series analysis. Shifts in the mean were made when 8 consecutive points were found either above or below the centerline. To assess for differences in patient characteristics (age, sex, race, ethnicity, reason for IF, history of prematurity, and type of CVC) and as a secondary form of analysis for our quality measures, encounters in the pre- and postintervention periods were compared by using Student’s t tests for normally distributed continuous variables, Wilcoxon rank tests for nonnormally distributed continuous variables, and the χ2 test of proportions for categorical data. P values <.05 were considered significant.
Analysis was performed by using SPSS Statistics 23 (IBM SPSS Statistics, IBM Corporation, Armonk NY) and statistical process control charts were constructed by using CHARTrunner version 3.0 (PQ Systems, Dayton, OH). Study data were collected and managed by using REDCap (Research Electronic Data Capture, Nashville, TN) electronic data capture tools hosted at Boston Children’s Hospital.10
The institutional review board at our institution determined that this study met the criteria for a waiver of informed consent.
We identified 309 encounters, and 149 ED encounters involving 56 unique patients were included in the analysis. Reasons for exclusion are shown in Fig 3.
There were no differences in clinical characteristics of children with encounters from the pre- and postintervention periods (Table 1).
By using statistical process control methodology, time to antibiotic administration decreased by 73 minutes (65.2%) from a mean of 112 minutes (lower control limit, 0; upper control limit, 332) to 39 minutes (lower control limit, 13; upper control limit, 65) (Fig 4). Although no change in slope was detected, interrupted time series analysis showed a significant-level change in time to antibiotics in the postimplementation period (mean 45% decrease, 95% confidence interval 15%–65%).
Antibiotic variability decreased, as indicated by increased use of piperacillin-tazobactam (67% vs 95.1%, P < .01) and vancomycin (67% vs 93.4%, P < .01) after implementation (Fig 5).
Outcome and Balancing Measures
ED LOS decreased from 286 to 247 minutes after implementation of the initiative (Fig 6), although there were no differences in 30-day mortality or hospital or ICU LOS. The time to admission order decreased from 151 to 101 minutes after implementation of the initiative (Supplemental Fig 7). There were no changes in the balancing measures of rate of hypoglycemia or rate of transfer to an ICU within 24 hours of ED discharge (Table 2).
Other Clinical Outcomes
Other clinical outcomes are shown in Table 2. Notably, there was a higher rate of patients experiencing hypotension during their ED stay, despite no change in the rates of central line–associated bloodstream infections (CLABSIs). Only 1 patient died within 30 days of their ED visit during the study period: a 5-year-old who presented to the ED with septic shock and died within 24 hours of their ED admission.
By using a variety of QI methods, the initiative reduced mean time to antibiotics in children with IF, fever, and a CVC from 112 to 39 minutes. Phased interventions were used, initially focusing on physician and nursing knowledge gaps and increasing awareness of high risk of bloodstream infection among patients with SBS and IF. The order entry process was simplified and antibiotic choices were limited in an effort to reduce variation within the process. Additional interventions were aimed at providing feedback to frontline staff on their performance and standardizing the triage process. Each set of interventions contributed to the overall reduction in time to antibiotics.
Improving awareness at a provider level has consistently been shown to be a powerful tool in successful QI interventions.11,12 SBS and IF, although increasing in overall incidence, remain relatively rare.13 Provider unfamiliarity with the complications and recommended management of patients with SBS and IF may lead to adverse outcomes, given the high rates of morbidity and mortality associated with the condition.14,15 Therefore, a concerted effort at improving provider knowledge and awareness was an important first step in improving care for SBS and IF patients with fever. Given the large number of physicians and nurses within our ED, improvements in education and awareness were achieved by using a multimodal approach that included presentations at mandatory staff meetings, e-mail reminders, signage within the ED, and face-to-face education.
Introduction of an electronic health record (EHR) that empowered autonomous ED clinician decision-making and limited antibiotic choices to those recommended by our specialist colleagues was an important part of the success of our project. EHRs and electronic order entry are frequently targeted in QI initiatives. Researchers in multiple studies have shown that implementation of an EHR is associated with an improvement in a variety of outcomes, including reductions in medication errors and reductions in CLABSIs in an ICU setting.16,17 More specifically, encouraging the use of an order set has led to greater adherence to recommended standards of care in both the outpatient and inpatient settings.18 Choosing default settings or limiting provider choice within an order set has also been shown to drive physician ordering habits.19
After the awareness campaign and modification of the order set, interventions occurred on the clinician level by providing individualized feedback to physicians and nurses on their time to antibiotics in our patient population. This feedback process also allowed staff to identify potential barriers to timely care via an electronic survey distributed after the relevant ED encounter. Individualized provider feedback has been shown to improve care in a variety of ways, including improving safety and efficiency and reducing variation.20,21 In addition to identifying barriers to meeting care goals, the feedback process also improved awareness of the initiative among providers.
ED LOS was reduced by almost 40 minutes, likely because of the emphasis on the early ordering of antibiotics without subspecialty consultation and earlier disposition decisions, as illustrated by the reduction in time to the placement of the order to admit. However, no changes in hospital LOS, ICU LOS, or mortality were identified. There are several possible explanations for this lack of improvement. The mortality rate in the preintervention period was 0%; thus, no positive change was possible as a result of the interventions. It is possible that earlier antibiotic administration does not lead to a shortening of hospital or ICU LOS. Another possibility is that increasing complexity and frequency of comorbidities may have contributed to longer ICU LOS for reasons unrelated to our intervention, although no specific markers of severity of ICU admissions over the study period were included in this analysis. Increased average bed occupancy and higher overall hospital census may also have contributed to this lack of improvement in LOS. Additionally, there were only 14 patients in the postintervention period admitted to the ICU, and this small sample size may not have allowed for detection of differences in LOS.
The CLABSI rate did not change from the pre- to the postintervention period. This is likely because previous QI interventions aimed at improving sterile practice around CVC care at home and in the hospital and increasing the use of ethanol locks to prevent CLABSI preceded the study period.
One other notable difference between the pre- and postimplementation periods was that more patients developed hypotension during their ED stay in the postimplementation group. There are several possible reasons for this increase. First, hypotension after antibiotic administration is a well-described phenomenon, hypothesized to be a result of endotoxin release from bacterial cell death.22 Development of hypotension during the ED stay as a result of earlier antibiotic administration (as opposed to later during the admission) allows for more appropriate triage to intensive care rather than admission to the floor followed by potential deterioration in clinical status. Additionally, as awareness of risk of bacteremia and sepsis among patients with IF has increased at our hospital, it is possible that this specific population was monitored more closely as a result, leading to the identification of hypotension more quickly than in patients treated before the intervention.
A QI initiative with interventions targeted at both individual and system-level processes was successful in reducing time to antibiotics and ED LOS in a high-risk patient population. For relatively uncommon diseases like SBS and IF, interventions that minimize both awareness gaps and variation within a system are of particular importance. QI interventions aimed at optimizing care for children with SBS and IF are likely to increase in importance as this population ages and grows in number.
This study has several limitations. Our ED is a quaternary care setting with 24-hour in-house general surgery consultation available, which is not true of most community-based EDs. Although this may limit generalizability, interventions in this study focused on ED staff actions independent of consultant input and enabled ED providers to order appropriate antibiotics without waiting for consultant input. These aspects of the initiative could be replicated in virtually any ED setting. Additionally, most patients with SBS receive care in similar settings because of their complex health needs. A second limitation is the inability to measure order set usage. However, the reduction in antibiotic variation suggests that the order set was used routinely to place orders. Another limitation is the implementation of >1 intervention simultaneously, meaning the assessment of which specific intervention was the largest contributor to the change in time to antibiotics was difficult. However, the initial interventions targeted both individual providers and system-level issues, and replication of this initiative is likely to require similar multimodal efforts. It is likely that there was a synergistic effect between interventions because the interventions aimed at filling knowledge and awareness gaps also encouraged use of the order set. Finally, there was a trend toward improvement in time to antibiotics in the time period immediately before our intervention. The reasons for this are unclear, but it is likely that the Hawthorne effect played some role because the project was being discussed and planned during this period. Those trends did not meet criteria for special cause or an adjustment of the mean.
This QI initiative significantly reduced time to antibiotics and ED LOS for febrile children with IF presenting to the ED. Similar projects will become more important as this medically complex patient population continues to grow.
- Accepted August 7, 2017.
- Address correspondence to Joel Hudgins, MD, Division of Emergency Medicine, Boston Children’s Hospital, 300 Longwood Ave, Boston, MA 02115. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- The Oley Foundation
- Pironi L,
- Arends J,
- Baxter J, et al; Home Artificial Nutrition & Chronic Intestinal Failure; Acute Intestinal Failure Special Interest Groups of ESPEN
- Overdyk FJ,
- Dowling O,
- Newman S, et al
- Copyright © 2017 by the American Academy of Pediatrics