A Quality Improvement Project to Reduce Antibiotic Exposure in Premature Neonates

The quality improvement project took place at a level IV, 60-bed, single-patient room design NICU located in a children’s hospital within an adult academic medical center. Data monitoring took place over 6 years and included 1903 patients, with 292 born <28 weeks’ gestational age. Fifteen percent of the included patients were born at an outlying hospital, and 11.3% of patients underwent surgical procedures (Table 1). All antibiotic doses were included. The current neonatal group is composed of 10 neonatologists and 19 neonatal nurse practitioners, with several of the new providers joining over the last few years.

The quality improvement project did not meet criteria for research by the institutional review board. Institutional review board approval was not necessary because the project was limited to program evaluation, quality improvement, or quality assurance activities designed specifically to assess or improve performance within the department, hospital, or classroom setting.

A multidisciplinary collaborative iNICQ team composed of neonatologists, neonatal nurse practitioners, pediatric infectious disease physicians, infection control experts, pediatric surgeons, pediatric pharmacists, neonatal nurses, and a member of our family advisory board was established. A driver diagram was created (Fig 1) and 3 major groups of interventions took place to address the main drivers.

The iNICQ team reviewed the previous 2016 EOS guidelines, which did not require alteration. They are similar to those that were subsequently published in 2018, with the inclusion of monitoring with a blood culture for low-risk infants (see Supplemental Information for local LOS guidelines).¹⁷  The group chose to monitor with a blood culture because it was felt that frequently patients had a clinical deterioration after the first 24 hours of life and the reassurance of a negative blood culture would be helpful in ruling out infection as the cause. In the spring of 2018, reeducation of the guidelines was performed in a neonatology group monthly meeting and education was incorporated into monthly resident orientation. Printed copies were posted in the NICU workroom for all providers to reference when placing admission orders. Treatment of culture-negative sepsis >5 days was strongly discouraged, with an emphasis on stopping antibiotics when cultures were negative for 36 hours in the absence of a site-specific infection. In 2021, further interventions took place, including nursing education by online education modules, the addition of a “SmartPhrase” for providers to document rationale for antibiotic initiation and planned length of therapy, and the updating of admission order sets to default to 36 hour stop times to aid in timely discontinuation of antibiotics. Multiple sources of feedback were in place, including real-time feedback by neonatal pharmacists on weekdays during rounds, as well as monthly audits and feedback by the iNICQ team that was discussed by e-mail, as well as during monthly provider meetings. In 2022, mass provider education on EOS was done during a departmental grand rounds.

The iNICQ team performed a baseline survey of the local neonatology and infectious disease groups regarding preferred length of therapy for common neonatal conditions (pneumonia, urinary tract infection [UTI], bacteremia, culture-negative sepsis, and meningitis) to determine the extent of group variation. The response rate of the survey was 83%, with 19 of 23 providers responding. All bacterial sepsis and UTIs in the NICU from 2017 were reviewed to evaluate actual practices. Responses from the survey, as well as actual practice review, were shared with the groups so all could visualize the heterogeneity of stated preferences and actual practice. This information was used in the subsequent creation of guidelines in May 2018. LOS guidelines were developed for specific neonatal conditions, including pneumonia, UTI, culture-negative sepsis, meningitis, and NEC (Supplemental Table 2). Small groups were formed to research recommendations for each condition. Published studies through PubMed, national health and safety network manuals, and shared guidelines from other units were reviewed. Recommendations for the LOS categories were compiled and presented to the iNICQ group for initial feedback then subsequently to the neonatology group in a grand rounds for discussion. Finalized versions of the guidelines were created and printed for reference in the workroom, as well as throughout the NICU for easy access (see Supplemental Information for local LOS guidelines). They were e-mailed to the provider, pharmacy, and nursing groups and presented at a nursing council meeting, Rotating pediatric residents were made aware of the guidelines during orientation. Neonatal pharmacists reviewed antibiotic courses during rounds to ensure optimal antimicrobial types and durations. The guidelines, as well as local antibiograms, were used as a reference. Similar to EOS courses, multiple sources of feedback were in place, including real-time feedback by neonatal pharmacists on weekdays during rounds, as well as audits and feedback by the iNICQ team that were discussed by e-mail, as well as during monthly provider meetings.

In the process of reviewing historical bacteremia cases, inconsistencies in methods of blood cultures obtainment were noted. Thus, the group reviewed CDC guidelines and literature regarding methods of obtaining blood cultures for LOS evaluation.^23–25  The findings were discussed during a neonatology group meeting. The group stopped collecting cultures from indwelling lines with fluids in place and instead began obtaining 2 cultures via venipuncture if a central line was present to decrease the chance of culture contamination, which could result in prolonged unnecessary antibiotic administration. Changes were discussed with the nursing practice council and e-mailed to the nursing staff to advise and educate on the rationale for practice change in March 2018. Bacterial culture reports were monitored to ensure compliance.

Another common indication for antibiotic administration in the premature population is for perioperative administration. The pediatric surgery group worked with the iNICQ team in April 2018 to develop an evidence-based set of guidelines for surgical prophylaxis on the basis of wound classification to ensure optimal antimicrobial coverage and duration. Previously, there were no guidelines which resulted in a variety of practices. The new guidelines were distributed to the neonatology, surgical, and pharmacy groups. Antibiotics were subsequently ordered according to the guidelines, primarily by the surgical nurse practitioner. Neonatal pharmacists assisted with reviewing orders, as well. Compliance was reviewed and ongoing group feedback provided.

The main outcome measure for this project was the AUR for infants born <35 weeks’ gestation expressed as antibiotic doses per 1000 patient days. Process measures included the percentage of infants receiving antibiotics for EOS; the percentage of infants receiving a 36-hour rule out or >5 days of antibiotics for culture-negative EOS; the total number of LOS antibiotic courses; and compliance with EOS, LOS, and surgical prophylaxis guidelines. Balancing measures included monitoring of all positive blood and urine cultures to evaluate for inadequate treatment or early recurrence of a disease process.

We reported our data using mean and SD. Outcomes and process measures were plotted as an annotated control chart using excel QI Macros 2019. We used standard Shewart rules to analyze special cause variations.

The main outcome of AUR for infants born <35 weeks’ gestation expressed as antibiotic doses per 1000 patient days decreased from a mean of 524 to 394, for a decrease of 24.8% (Fig 2). Approximately one-third of the antibiotic doses given throughout the time period were during the first 72 hours of life. For the population born <28 weeks’ gestation, EOS usage accounted for 14.4% of total antibiotic usage, and for the population born 28 weeks’ to 34 6/7 weeks’ gestation, EOS accounted for 49.8% of total antibiotic usage. The initial EOS guidelines developed in 2016 resulted in an improvement in antibiotic discontinuation with a negative blood culture at 36 hours from 20.27% to 65.64%, with further improvement to 80.68% with the 2018 changes (Fig 3). The number of infants receiving any antibiotics for EOS did not decrease until the 2018 interventions, with further improvement in 2020 (Fig 4). The 2018 interventions resulted in a decline of prolonged antibiotic courses >5 days for culture-negative EOS in the <28-week population from 17.27% to 8.67%.

Other process measures included monitoring the frequency of LOS antibiotic courses and compliance with the various antimicrobial guidelines throughout the project. The overall number of LOS courses per 100 patient days decreased for the population born <28 weeks’ gestation from 2.1 courses per 100 patient days to 1.5 courses per 100 patient days. The trend was similar for the 28- to 34 6/7-week population, with a decrease from 0.8 courses per 100 patient days to 0.4 courses per 100 patient days. Complete compliance with LOS guidelines was difficult to achieve, with an average of 60%, and the total number of antibiotic doses per LOS course did not change. The group achieved 78% complete compliance with recommendations for surgical prophylaxis. Complete compliance was defined as antimicrobial type and duration within the range as recommended by the guidelines. Typically, noncompliance was because of an extra 1 to 2 doses of antibiotic beyond the range and/or initial antibiotic selection.

We observed several points of special cause variation throughout the project, the first being because of a limited outbreak of Salmonella bacteremia in the unit requiring prolonged antibiotic treatment. Other outlying months included those where new neonatology attending physicians or nurse practitioners joined the group and were not as familiar with the unit guidelines or small clusters of NEC cases. The central line-associated bloodstream infection rate was also low from March 2020 to June 2021, resulting in a decrease in LOS courses. There was also a spike in antibiotic use in July 2021 and January 2022, with the birth of multiple 22- to 24-week neonates who experienced difficulty with skin integrity and need for surgical interventions.

Throughout the project, we monitored all positive blood and urine cultures to ensure there were no cases of inadequate treatment with shortened antimicrobial durations. We also wanted to ensure that the reduction in AUR was not solely attributed to a decrease in the incidence of sepsis. Throughout the monitoring period, the rate of positive true blood or urine cultures remained unchanged. We noted no cases of inadequately treated sepsis resulting in subsequent positive blood or urine cultures.

Standardizing practices and providing ongoing feedback were effective in safely reducing the AUR for infants born <35 weeks’ gestation in our level IV NICU. It is important to note that the establishment of guidelines without ongoing feedback and monitoring of outcome, process, and balancing measures was not very effective. EOS guidelines and 36-hour stop times had been introduced in our unit in the fall of 2016; however, this only resulted in improvement of 36-hour stop times and did not decrease the overall number of infants receiving antibiotics. The EOS interventions starting in 2018 primarily impacted the rate of antibiotic exposure in the population born 28 to 34 6/7 weeks’ gestation because 27% of these neonates were born because of maternal indications. Only 16% of patients born <28 weeks’ gestation were for maternal indications, and the vast majority of those patients still received antibiotics for EOS because of clinical instability. The EOS interventions were successful in decreasing the rate of prolonged antibiotic exposure beyond 5 days for the <28-week population.

Examining the variation in group practices was enlightening to the group overall. We found we had a treatment duration range of 5 to 14 days for both pneumonia and UTI treatment among the group before the development and implementation of the guidelines. Through the LOS interventions, we were able to reduce the overall variation to 5 to 7 days for uncomplicated cases. There was inconsistent compliance with the LOS guidelines; however, most of the deviations were because of an incidental extra 1 to 2 doses of antibiotic. There is an opportunity for creation of an electronic order set to address these issues. We did not expect the surgical prophylaxis guidelines to have a major impact on the AUR; however, we felt this aspect of the project was important to standardize to decrease the risk of postsurgical infections. The guidelines were effective in standardizing the type and duration of perioperative antimicrobial usage.

One element that impacted the use of antibiotics for LOS was the rate of central line-associated bloodstream infections in the unit. During the time period of April 2020 to June 2021, this rate was low and resulted in a decrease in number of LOS courses. This identifies the need for improved infection prevention strategies in the unit, particularly for the <28-week population in which LOS accounted for 86% of the total antibiotic usage.

Much of the previous focus on antimicrobial optimization in the NICU has previously been on the term population for EOS or for reducing specific types of antimicrobial agents such as vancomycin. We believe this article significantly adds to the literature the need for further work on the preterm population.

Some of the major strengths of this project included the engagement of a large multidisciplinary team. This project would not have been as successful without buy-in from the entire group, which was made possible by ensuring all had a stake in the development and implementation of the guidelines. Another strength is the length of time we have monitored the project to ensure sustainment of our initial results. Sustainment is a frequent issue reported in the literature.²⁶  We believe the keys to sustaining our results have been because of the ongoing involvement of our NICU pharmacists who provide reminders in real time on rounds, as well as the education of incoming providers. Between 2018 and 2022, we had 5 new neonatologists and 3 new neonatal nurse practitioners join our group, along with 10 new pediatric and 6 medicine–pediatric residents every year.

Some of the limitations to the generalizability of this work include the availability of resources. At our institution, blood cultures are kept in-house with laboratory personnel around the clock. We have in-house providers who can readily and serially examine a neonate. When developing the LOS guidelines, we used our best judgement, along with pediatric infectious disease and neonatal pharmacy input on the basis of available evidence and local antibiograms to determine a recommended length of treatment and antimicrobial type for various common disease processes. Thus, the guidelines we developed would need to be altered in other locations on the basis of local antibiograms. We will need to review these guidelines on a regular basis to ensure they are up to date.

Future directions for our group include the incorporation of LOS guidelines into the electronic ordering system to include antimicrobial type and duration on the basis of infectious process. We implemented automated stop times for EOS in admission order sets in 2021 and will need to continue monitoring this process for cases of inadequately treated sepsis. Finally, the most important aspect the unit is working to address is improving the rate of hospital-acquired infections.

Creation of sepsis guidelines can be effective in decreasing unnecessary antibiotic exposure in a NICU when combined with ongoing education, monitoring, and feedback in particular. Guidelines appear to be most impactful in EOS, in particular for the preterm population aged ≥28 weeks. The most important aspect for reducing unnecessary antibiotic exposure for the most preterm population aged <28 weeks is through the prevention of hospital-acquired infections.

We thank the NICU, neonatal pharmacy, pediatric surgery, and pediatric infectious disease teams at OSF Children’s Hospital of Illinois for partnership critical to the success of the quality improvement project.