Quality Improvement Through Nurse-initiated Kaiser Sepsis Calculator at a County Hospital

QIKSIC was conducted at University Health, an academic county hospital affiliate of University of Texas Health San Antonio with ∼2000 to 2500 births per year, of which there are ∼100 to 150 chorioamnionitis-exposed infants ≥35 weeks’ gestation. Infants in the level I nursery are primarily attended by pediatric hospitalist medicine and secondarily by family medicine (FM). Phototherapy is managed in the nursery, but intravenous fluid or antibiotics necessitates transfer to the level IV NICU. The division of neonatology plays a role in protocol development and provides overnight nursery coverage. This QI project was done in collaboration with multiple centers as part of the Better Outcomes through Research for Newborns (BORN) network; however, this manuscript focuses solely on our institution with permission from the BORN network for publication.

The core interdisciplinary QI team included pediatric hospitalists, a neonatologist, a pediatric infectious disease specialist, a neonatology fellow, a maternal–fetal medicine fellow, pediatric residents, and nurses from the newborn nursery and NICU. Meetings were held virtually to allow for wider involvement from multidepartmental staff and shift nurses.

In 2017, a hospitalwide practice adjustment had been made from strict adherence to the categorical risk assessment^14,15  (which resulted in nearly 100% antibiotic treatment of chorioamnionitis-exposed infants) to an algorithmic approach (involving similar risk factors, but well-appearing, chorioamnionitis-exposed infants would automatically have a BCx and CBC drawn in addition to 48-hour observation). This improved antibiotic administration from all to approximately one-quarter of exposed infants, but still resulted in universal collection of at least 1 BCx and CBC, often leading to follow-up laboratories. Excessive laboratory draws led to formation of the QI team seeking intervention.

Our initial core QI team, a nurse, resident, fellow, and an attending, conducted a literature review for successful implementation of KSC into the localized standard of care. Nurse-initiated KSC protocols were not described in the literature, and although the focus had been to decrease antibiotic usage, a significant opportunity was identified to decrease laboratory work. The core team grew to include 7 nurses, 5 residents, 2 fellows, and 3 attending physicians. The team identified potential stakeholders in the divisions of pediatric hospitalist medicine, FM, and neonatology, acknowledging that the highlighted interdisciplinary intervention would be a significant institutional culture change given concerns for patient safety and provider comfort (Fig 1).

Early in the planning stages of QIKSIC, we anticipated major barriers to the implementation of a large-scale QI project across multiple departments. Therefore, an interdisciplinary, interdepartmental team was developed with universal leadership support to mitigate challenges.

Education was essential for initiation of QIKSIC. Each discipline led its own education: nurses provided education at huddles in the nursery, NICU, and labor and delivery, with subsequent follow-up; residents provided education in FM and pediatric residency programs; a neonatologist and NICU fellow provided education to neonatologists, advanced-practice providers, and hospitalists; and maternal–fetal medicine and neonatal fellows provided joint education to obstetric providers.

Emphasis was placed on our new EMR to facilitate implementation and data collection. Through collaboration with the BORN network and after reviewing our EMR, it was noted the data required by the KSC could be automatically populated into a flowsheet. Nurses reviewed flowsheet material and notified providers to assure usage of the protocol. A smart phrase was generated and inserted into the standard nursery history and physical note to remind physicians to use the KSC. Lastly, a report was generated identifying all chorioamnionitis-exposed infants to aid in data collection and quality assurance.

The QIKSIC team developed a protocol utilizing the KSC with EMR implementation for infants born ≥35 weeks with the maternal diagnosis of chorioamnionitis (obstetric diagnosis) in our level I nursery and level IV NICU. When a chorioamnionitis-exposed infant was born, the newborn nursery nurse would verify the EMR-generated KSC values and add necessary changes into the EMR flowsheet (maximum maternal antepartum temperature, gestational age [GA] of the infant, rupture of membrane duration, GBS status, and timing and selection of maternal antibiotics). Further clinical intervention was guided by the EMR-integrated KSC and by the nurse’s physical assessment of the infant. The provider team also verified the same steps with the KSC using the Centers for Disease Control and Prevention incidence of 0.5 EOS cases per 1000 births, and performed an immediate physical exam if the KSC indicated elevated risk or infection was clinically suspected. The provider also included KSC score and management recommendations on the admission note via a standardized note template (Fig 2).

To ensure safe and consistent implementation across the multiple teams and rotating resident physicians, members of the core QI team conducted educational sessions each month for the pediatrics and FM teams, involving formal lectures, education at huddles, and individualized over-the-shoulder teaching in clinical spaces (Fig 3). Neonatology and nursing had their own educational sessions conducted by a same-discipline core QI team members.

During the first Plan, Do, Study, Act cycle, as the nurse flowsheet documented the KSC, it was noted that further education was needed on the process of the nurse scoring the KSC directly after delivery and where to enter the data. It was also noted that the KSC was not always documented in the resident admission notes. Supplemental education was provided via monthly PowerPoint slide presentations for new resident teams. The final of 3 total Plan, Do, Study, Act cycles began in July 2021 providing education to new residents and fellows, and reeducation to current staff.

Chart reviews were conducted for chorioamnionitis-exposed infants born 6 months preintervention (August 2020–January 2021) and 7 months postintervention (March 2021–October 2021). Six months was chosen as the preintervention time because of implementation of an EMR, and 7 months as the postintervention time to comply with data collection for the BORN project.

Demographic and clinical information was collected on all infants. All chorioamnionitis-exposed infants born ≥35 weeks’ gestation were reviewed using an EMR-generated report and manual data extraction. Demographics were noted and outcome measures recorded included whether chorioamnionitis-exposed infants had CBCs or other inflammatory markers, BCx, or empirical antibiotic therapy for EOS. The data used by the KSC were also collected. Other measures obtained were the admission unit, method of delivery, or if the infant developed clinical illness, required transfer to another unit, or had an emergency department visit or hospital readmission within 2 weeks of hospital discharge.

A total of 3409 infants aged ≥35 weeks were born during the intervention period (1459 preintervention, 1950 post intervention). Chorioamnionitis exposure occurred in 53 of 1459 (3.6% [95% confidence interval 2.7%–4.7%]) infants in the preintervention period and 51 of 1950 (2.6% [95% confidence interval 2.0%–3.4%]) in the postintervention period. Median GA of chorioamnionitis-exposed infants at birth was 39 6/7 weeks (interquartile range 38 6/7–40 3/7), and 76% self-identified as Hispanic, which is consistent with our general hospital patient population. Baseline demographics including race, infant sex, ethnicity, GA, and mode of delivery were similar in our pre- and postintervention groups.

Assessment of EOS risk factors demonstrated 41% had rupture of membranes ≥18 hours, and 20% GBS positivity, 7% unknown, and 73% negative at time of delivery. All infants evaluated had a maternal diagnosis of chorioamnionitis per inclusion criteria, but 8% did not have a documented maternal temperature ≥38°C before delivery. Only 14% of mothers received broad-spectrum antibiotics >4 hours before delivery; 13% received broad-spectrum antibiotics 2 to 3.9 hours before delivery, 37% received GBS-specific antibiotics >2 hours before delivery, and 36% received antibiotics <2 hours or not at all before delivery.

During the preintervention period, 96% of chorioamnionitis-exposed infants had CBCs drawn, decreasing to 27% in the postintervention cohort. Before the intervention, it was common for follow-up CBCs to be drawn if the first CBC had abnormal values. The number of CBCs per exposed infant declined from 1.35 to 0.27, with infants with >1 CBC drawn decreasing from 32% to 4%. BCx collection also dropped from 98% to 37% after the intervention. Antibiotic use decreased from 23% to 14% during the study period. The number of infants admitted to the NICU decreased from 25% to 16%, correlating with antibiotic administration (Table 1).

Our process measures included evaluating the use of the KSC in the workup of chorioamnionitis-exposed infants, when antibiotics were administered, and when BCx and/or CBCs were drawn in infants not indicated by the KSC. Although project population size was limited, all 51 chorioamnionitis-exposed infants were entered into the KSC during the postintervention period. Antibiotics were administered in 14% (7 infants) of the postintervention cohort. No infants received antibiotics against the recommendation of the KSC, and all symptomatic infants (as defined by the KSC) were treated within 4 hours. BCx were drawn in 37% of postintervention cohort infants, including 8% (4 infants) who had blood cultures drawn against the recommendation of the KSC. CBCs were drawn in 27% of postintervention infants, all of which went against the KSC recommendations because the KSC does not recommend a CBC in an infectious workup.

Balancing measures for our study included:

1. subjective staff time dedicated to cal culating the KSC result and implementing the recommendation;

2. whether infants not started on antibiotics because of KSC recommendation subsequently developed EOS or needed NICU admission; and

3. readmission to the hospital after dis charge for concern for sepsis.

Nurse and provider time was not quantified, but subjective reports suggest less time was spent with the QIKSIC protocol. Finally, no diagnoses of EOS were missed through implementation of the KSC, and no infants were readmitted with sepsis to the hospital within 14 days, acknowledging the low incidence of EOS and small study population.

Our interdisciplinary team safely met the primary aim and markedly reduced both BCx and CBCs. Importantly, no infants with EOS were missed because of protocol utilization. Additionally, QIKSIC modestly reduced the number of infants exposed to antibiotics, likely because of the relatively low baseline rate with our preintervention protocol. Given that this was a QI project with a limited sample size, control charts and descriptive statistics were used rather than more advanced statistical tests (Fig 4).

With regard to balancing measures, subjective nurse reports suggest less time was spent with the QIKSIC protocol because the KSC allowed them time to do other tasks because of the reduction in laboratory draws. Nurses also reported a decrease in mother–infant separation because phlebotomy is performed in newborn nursery. Providers were called more frequently to make earlier decisions regarding infant risk stratification; however, providers reported minimal increase in time spent, because the KSC calculation only had to be verified.

Process measures indicated 100% entry of chorioamnionitis-exposed infants into the KSC, suggesting that the noted improvements were a direct result of the intervention. Reduction in antibiotic utilization with the use of the KSC has been demonstrated in multiple studies and is a primary driver for KSC implementation at many hospitals.^20–23  No antibiotics were prescribed against KSC recommendations. We were unable to determine reasons for drawing a BCx against the recommendation of the KSC (8%) because this was a retrospective chart review. All CBCs in the postintervention group were drawn against the KSC and may represent diagnostic inertia of providers’ habits. CBCs may have also been obtained for noninfectious reasons including petechiae, direct antibody-positive infants, or pale infants; these may not represent deviations from the KSC protocol.

One EOS case occurred in the preintervention period; a screening CBC and BCx were done (although the CBC was reassuring and the BCx was sterile, subsequent BCx after onset of illness was positive). If the KSC would have been used, serial vital signs and close observation would have allowed for potential identification. Subsequently, no cases of EOS occurred within our intervention period. The decrease in BCx is consistent with other studies.^8,20,22  The decrease in CBCs is reported with the KSC,⁸  but many providers and institutions continue to collect a CBC to attempt to predict infants at risk despite reported lack of utility.^10–12 

The success of our QI project can be attributed to several key aspects, most importantly the early involvement of a large interdisciplinary team. Communication regarding the maternal chorioamnionitis diagnosis, the initial nursing evaluation, the evaluation and documentation by newborn providers, and discipline-specific education were all critical for our success. Nursing leadership for group and 1-on-1 education was paramount for this nurse-initiated protocol. The protocol empowered nurses to perform the initial timely exam and assessment of all chorioamnionitis-exposed infants without waiting on provider availability ensuring early identification, diagnosis, and treatment of at-risk infants while still allowing appropriate provider oversight.

The use of physician trainees in QIKSIC allowed for hands-on learning of both QI processes and EOS evaluation. This knowledge was disseminated by trainee-to-trainee and trainee-to-faculty education sessions. The trainee involvement and understanding of the KSC was crucial in the success of QIKSIC. Trainees became more educated and interested in EOS and QI methodology. Given growing interest, the core team grew from 1 resident and fellow at onset to 5 residents and 2 fellows at intervention initiation.

Lastly, the ability to integrate the KSC into the EMR was vital in ensuring compliance and accuracy of the intervention. The hospital’s recently updated EMR allowed for automatic retrieval of KSC criteria and smart phrases, as well as ensured automatic implementation into the provider’s medical decision-making. Similar capabilities exist in most EMRs and could be employed by other institutions instituting similar projects. The ease of translating raw data into a flowsheet or table in the admission note facilitated adherence to the protocol. The templated admission note ensured that the KSC was considered on every patient admitted to the nursery, regardless of risk factors. This prompted the provider to justify the change in practice of collection of laboratory work and choice of whether to initiate antibiotics in all infants admitted to the nursery.

This project has several limitations. First, it is a single hospital intervention, which allows for ease in education and integration of a new protocol. However, we feel that a similar success could be found within a larger system with multiple hospitals by providing multiple education sessions and integrating team members from many different roles and specialties across hospitals. Second, the low incidence of EOS and limited sample size could give a false sense of security with regard to EOS cases, but the safety of the KSC has been repeatedly demonstrated. Additionally, as a large county hospital, not all our infants have primary follow-up within our institution. A child discharged with EOS could have been readmitted elsewhere; however, most infants readmitted within the first weeks of life are readmitted to our hospital as parents return to where they last received care. Lastly, our project was conducted at a large academic county hospital in an urban setting with an established pattern of protocol implementation and may not be generalizable to small or remote hospitals.

The interdisciplinary nurse-initiated KSC QI initiative led to a marked reduction in BCx and CBC utilization among chorioamnionitis-exposed infants. No safety signals were observed; specifically, no cases of EOS occurred. Future directions may include evaluation of the KSC to nonchorioamnionitis infants with other risk factors, expansion of tools like the KSC to premature infants, or developing a similar sepsis calculator for febrile infants presenting in the first 3 weeks of life. This project highlights the feasibility of using an EMR-integrated, nurse-initiated protocol and reinforces the importance of nursing– physician collaboration. The next steps for our center are to work toward 100% protocol adherence and expand the routine use of KSC to all infants ≥34 weeks’ gestation with infectious risk factors.

We thank the entire University of Texas Health San Antonio inpatient pediatrics division, pediatric and family medicine residency programs, and the many nurses for their participation in this project. We also thank the lead QIKSIC nurses Brittany Biddle, RN; Sara Dickinson, RN; Jennifer Nance, RN; Zahra Nejat, RN; Rhonda Ritchey, RN; Alma Saravia, RN; and Azra Sheikh, RN, without whom the project would not have been possible.