A Clinical Monitoring Approach for Early Onset Sepsis: A Community Hospital Experience

Routine screening and intrapartum antibiotic prophylaxis (IAP) for Group B Streptococcus has dramatically reduced the risk of early onset sepsis (EOS) in late preterm and term neonates.^1–3  EOS is now rare in well-appearing neonates even if traditional risk factors (maternal fever or chorioamnionitis, prolonged rupture of membranes, and late preterm birth) are present. Traditional empirical antibiotic treatment strategies for EOS on the basis of risk factors alone result in frequent overtreatment of well-appearing, uninfected neonates.^4,5  As part of neonatal antimicrobial stewardship efforts, updated EOS treatment approaches have recently been advanced that use (1) multivariate risk assessment models incorporating risk factors plus clinical examination (eg, newborn sepsis calculator),^6,7  or (2) serial clinical examination^8,9  to determine the need for antibiotic treatment. These updated approaches have resulted in 40% to 60% reductions in antibiotic use in neonates with no safety concerns. Both approaches are now recommended by the American Academy of Pediatrics as potential options for EOS risk assessment in late preterm and term neonates.^10,11 

In addition to reduction in antibiotic use, a serial clinical examination approach can reduce separation of mother and baby. Many hospitals now keep newborn infants with postpartum mothers as a dyad couplet, by using nurses to care for both the mother and the baby. Studies have shown that keeping babies in the same room as mothers can improve breastfeeding and breast milk production.¹² 

To date, successful implementation of a serial clinical examination–based approach for EOS has been reported in the United States in 1 university hospital in California,^13,14  and there are no similar reports on its implementation in a community hospital setting. An EOS approach based on serial clinical examination might not be easily translated to other centers and will need to be tailored for the individual hospital.

In our community hospital NICU we noticed that many of our admissions were to rule out sepsis in well-appearing late preterm and term neonates. Our sepsis pathway dictated admission to the NICU, laboratory screening with complete blood count (CBC) and C-reactive protein (CRP), and antibiotic treatment based on the presence of certain perinatal risk factors (ie, maternal fever, prolonged rupture of membrane >18 hours, late preterm birth, Group B Streptococcus positive mother without adequate IAP) even in well-appearing infants. As a result, many well-appearing neonates were being separated from their mothers and treated with antibiotics. With the specific aim to reduce unnecessary maternal separation, antibiotic exposure, and laboratory testing in well-appearing late preterm and term neonates born in our hospital, we adapted a previously published^9,13,14  and evidence-based serial clinical examination approach for EOS at our hospital. The current report evaluates the impact of our updated approach after the first 19 months of implementation.

El Camino Hospital is a community hospital that offers obstetric and neonatal services (over 4500 deliveries annually). Well newborns are managed by community pediatricians in the mother baby unit (MBU) where neonates room in with the mother. Sick newborns (term and preterm) are cared for in a level III NICU staffed by in-house neonatologists 24 hours per day. Neonatologists attend all high-risk deliveries and evaluate neonates at birth. Community pediatricians consult neonatologists to comanage symptomatic newborns and transfer care to NICU when necessary.

The preintervention EOS pathway at El Camino Hospital was based on categorical risk factors and included laboratory testing and empirical antibiotic treatment if specific risk factor(s) were present. Neonatologists attended all deliveries with maternal fever ≥ 38°C. Maternal fever ≥38°C with 1 other additional risk factor (prolonged rupture of membranes ≥18 hours, inadequate intrapartum antibiotic prophylaxis for GBS colonization, or late preterm birth) or maternal fever ≥38.6°C alone resulted in direct admission to the NICU for laboratory testing and empirical antibiotic treatment. Sepsis laboratory testing included complete blood count (CBC) and a blood culture on admission, followed by serial CRPs every 24 hours. Antibiotics were discontinued after 48 hours if 2 CRPs were ≤1 mg/dL and blood culture remained negative. Antibiotics were continued past 48 hours for persistently elevated CRPs or persistently abnormal vital signs at the discretion of the attending neonatologist. Lumbar puncture was reserved for symptomatic neonates with neurologic signs of infection such as seizure or when the CRP was >4 mg/dL.

Neonates born to mothers with maternal fever ≥ 38.0 to 38.5°C without other risk factors or neonates born to mothers without maternal fever but who had 2 or more other risk factors as listed above were admitted to the MBU and monitored with serial screening laboratory tests (CBCs and CRPs) for 36 to 48 hours along with vital sign checks every 4 hours by the MBU nurse for the first 48 hours of life. Antibiotics were started if the CRP was ≥ 1 mg/dL and rising or ≥ 4 mg/dL at any time. In addition, any symptomatic neonate (temperature instability, respiratory rate <30 or >70 breaths per minute for over 15 minutes, heart rate <100 or >170 beats per minute for 15 minutes, grunting, flaring, retractions, or CRP was >2) was admitted directly to the NICU for laboratory workup and empirical antibiotics.

Revision of the EOS sepsis pathway was based on the experience by Joshi et al⁹  and was similar to their phase II published approach. One difference in our pathway was that Joshi et al defined maternal diagnosis of chorioamnionitis as an EOS risk factor, whereas we used maternal fever ≥38°C alone. In addition, Joshi et al reported using a level II NICU nursery nurse to monitor neonates born to mothers with chorioamnionitis for the first 2 hours after birth in the delivery room, and then transitioning to a level I nurse in the postpartum unit. Our community hospital was unable to staff a NICU nurse in the delivery room, and thus monitoring of infants was done by a level I MBU nurse.

A multidisciplinary group of neonatologists, nurses, pediatricians, and obstetricians worked together over the course of 3 months to adapt the published guidelines for our community hospital setting. The proposed guideline was presented at a department meeting and revised based on feedback from community physicians. A Maternal Child Health (MCH) Grand Rounds was held to disseminate changes in the revised EOS pathway in which authors from the Joshi et al 2018 paper detailed the authors’ experience and data to support the clinical approach to EOS. The guideline was then brought to the nursing leadership in the MBU to discuss the impact on nurses of increased assessments and responsibility of identifying symptomatic neonates. After a review of the evidence and discussions on the benefit of mothers and babies staying together, agreement was reached on the proposed pathway. Because babies not on the sepsis pathway had vital sign checks every 4 hours for the first 12 hours of life, then every 8 hours until discharge, the MBU nurses suggested modification of vital signs checks to all neonates in the MBU to be every 4 hours regardless of risk for EOS to avoid confusion. The final guideline (Fig 1) was then discussed and approved by all MCH departments including Pediatrics, Obstetrics, and the hospital Medical Executive Committee and the Board of Directors.

Before implementation, all MBU and NICU nurses were educated with an online HealthStream module on the literature supporting a clinical examination approach to EOS screening and the new EOS pathway. This module also reviewed signs and symptoms of sepsis including abnormal vital signs, altered level of consciousness, poor perfusion, and decreased feeding. Videos were provided in the HealthStream to show examples of grunting, head bobbing, stridor, and abnormal tone. There was a quiz at the end of the module with scenarios of different clinical situations. The HealthStream module also reviewed what to do with abnormal vital signs. For temperature <36.5°C or ≥38°C, nurses were instructed to repeat the temperature and report to the pediatrician if the second reading remained abnormal. Pediatricians were also notified for a heart rate of <100 beats per minute or >179 beats per minute for >15 minutes when the baby was quiet during the exam. Similarly, providers were notified for respiratory rates <30 breaths per minute or >70 for 15 minutes in quiet infants. Multidisciplinary teams were also educated through a “Doc Talk” presented by the team revising the EOS pathway. Community physicians were notified of the sepsis pathway change by E-mail as well as in person with Grand Rounds and through department meetings as above. Finally, town halls were held before and after implementation of the new pathway to answer questions from all members of the team including nurses, obstetricians, and community pediatricians. Copies of the updated guideline were emailed to NICU and MBU staff and providers and were also posted in each unit in the work areas.

Under the new EOS pathway which began March 2017, a neonatologist attended all deliveries of mothers with a temperature ≥38.0°C. Well-appearing neonates were admitted to MBU and stayed with the parents. Nurse staffing ratios for all mothers and neonates in the MBU were 1:3 (ie, 1 nurse for every 3 couplets). Vital sign assessment and serial clinical examinations were performed by the nurse every 30 minutes for the first 120 minutes, followed by every 4 hours until discharge. The clinical assessments and vital signs were documented in the electronic health record. In 2009, when the mother baby couplet care was initiated, a “transition nursery” was established in the NICU to monitor symptomatic neonates for the first 6 hours after birth according to “perinatal care” guidelines. This transition nursery was staffed by NICU nurses and neonatologists and had cardio respiratory monitors. Symptomatic newborns were admitted to the transition nursery (up to 6 hours) or to the NICU depending on their respiratory management. If the neonate’s symptoms improved by 6 hours, the neonate would be transferred to the MBU and monitored closely with vital signs and serial clinical assessments. Neonates admitted to the NICU were evaluated and/or treated with antibiotics at the discretion of the neonatologist.

This project was reviewed by the hospital institutional review board and determined to be a local quality improvement project that did not meet the definition of human subjects research.

Neonates ≥35 weeks’ gestation born at our hospital were tracked over the 14-month period before (preintervention) and 19 months after (postintervention) the EOS pathway change via a query of the electronic health record. Prenatal risk factors for sepsis including maternal fever, ROM, GBS status, and intrapartum antibiotics were collected. Sepsis laboratory testing, blood culture results, and antibiotic exposures were collected during the birth hospitalization. Outcome measures included the percentage of neonates who were exposed to any parenteral antibiotic treatment and/or had CRP measured during the birth hospitalization.

Comparisons between the pre and postintervention implementation period in antibiotic treatment and CRP testing were assessed via a chi-squared test (χ²) test. In addition, interrupted time series analyses were performed to model the level and trend in monthly antibiotic exposure and CRP pre and postintervention implementation. Autocorrelation was tested by using the Durbin-Watson statistic. Neonates born to febrile mothers ≥38.0°C during labor were also examined as a subgroup.

Clinical outcomes of all cases of positive blood cultures during the birth hospitalization in neonates born at our hospital over the time period of study were reviewed. In addition, any hospital readmission for sepsis within 30 days after birth was reviewed, including laboratory sepsis screening and blood culture results. Statistical significance was set at P <.05. Data were analyzed by using Stata 13 (StataCorp, College Station, TX).

A total of 10 631 neonates ≥ 35 weeks were born over the evaluation period: 4756 during the preintervention period and 5875 during the postintervention period. Neonates were similar between both time periods (Table 1), except they were of slightly lower gestational age during the postintervention period with more late preterm neonates.

After implementation of the revised EOS pathway, overall antibiotic use in neonates ≥35 weeks decreased from 6.7% (n = 314/4694) to 2.6% (n = 153/5937; P < .001). Similarly, measurement of CRP decreased from 13.3% (n = 626/4694) to 5.3% (n = 312/5937; P < .001). Interrupted time series analysis (Fig 2A) showed a significant change in the level of antibiotic use in the postintervention period compared to the preintervention period (−3.8% [95% CI −5.8 to −1.7%]; P = .001). No significant trend in antibiotic use during the pre or postintervention period was found. Similarly, the measurement of CRP was lower in the postintervention period compared to the preintervention period (−5.2% [95% CI −7.8 to −2.6%]; P = .001) based on interrupted time series analysis (Fig 2B). A small downward trend in the measurement of CRP was found during the postintervention period (−0.27% [95% CI −0.39 to −0.15%] per month; P < .001).

No neonate had culture-positive EOS in the pre or postintervention period. No neonate in the postintervention period was readmitted within our hospital system within 30 days from birth with a positive blood culture. However, we were unable to determine if such readmissions occurred in outside hospitals.

In neonates born to mothers with an intrapartum fever ≥38°C, antibiotic use decreased from 57.9% (n = 139/240) to 6.9% (n = 20/290; P < .0001), and measurement of CRP decreased from 90.0% (n = 216/240) to 44% (n = 44/290; P < .0001). Of the 240 neonates born to mothers with an intrapartum fever ≥38°C in the preintervention period, 135 (56.2%) were admitted to the NICU or transition nursery at birth. In comparison, 18 of 290 (6.2%) neonates born to mothers with an intrapartum fever ≥38°C in the postintervention period were admitted to the NICU or transition nursery at birth, with 4 ultimately returning to the MBU without receiving antibiotics when symptoms resolved within 6 hours of life. The other 14 neonates were admitted to the NICU and received antibiotics. We identified 6 neonates born to mothers with an intrapartum fever that were initially well-appearing in the MBU and later transferred them to the NICU because of symptoms or elevated CRP and started them on antibiotics. Clinical signs in those admitted included respiratory distress, hypoglycemia, hypothermia, and fever.

A serial clinical examination approach to EOS in late preterm and term neonates has previously been shown to reduce antibiotic exposure and separation from mothers at a large academic center.^9,12,13  This study demonstrates similar success in a community hospital setting. In the postintervention period, there was a 61% reduction in antibiotic use and a 60% reduction in sepsis laboratory testing. Our rate of antibiotic use in the postintervention period was 2.6% which is comparable to recent reports that implemented either a serial clinical examination approach or multivariate risk assessment model to guide EOS treatment.^8,15 

Compared with our previous practice, we avoided 157 unnecessary antibiotic exposures and admissions to the NICU in the postintervention period. When hypothetically evaluating neonates in the postintervention group using our previous sepsis pathway, 80 neonates would have been admitted to the NICU and started on antibiotics based on severity of maternal fever alone (>38.6 C). Another 67 neonates would have been treated based on maternal fever and prolonged rupture of membranes >18 hours; 6 neonates for maternal fever and inadequate GBS treatment; and 4 neonates for maternal fever and late preterm status.

Sepsis laboratory testing decreased in the postintervention period, and there was a trend for an ongoing decrease in testing within the postintervention period. Sepsis laboratory testing (8.8%) was higher than antibiotic use (3.0%) in the first 3 months of the postintervention period, which suggests some clinicians continued to perform such testing to “screen” for infection. The difference between sepsis laboratory testing and antibiotic use decreased over time with a rate of 3.3% and 2.4%, respectively, in the last 3 months of the postintervention period. While we cannot know with certainty, we hypothesize that this ongoing decrease likely represents an increase in comfort over time in not utilizing sepsis laboratory testing to screen babies with traditional perinatal risk factors and/or “equivocal” clinical signs not concerning enough to start antibiotics.

Avoidance of separation of mother and baby has implications beyond cost savings and laboratory utilization. Neonates are better able to bond when with their mothers and are more likely to have successful breastfeeding.^16,17  In addition, NICU admission has been shown to impact the psychological health of parents including distress and postpartum depression. According to 1 study, 35% of mothers and 24% of fathers experience acute stress disorder after hospitalization of their neonate.¹⁸  The feeling of helplessness and loss of the parental role can last beyond the NICU stay. Finally, avoiding NICU admissions protects neonates from painful procedures and laboratory draws. Each NICU admission to rule out sepsis can mean several painful pokes for intravenous antibiotics, laboratory draws, and possibly lumbar punctures. Benis et al¹⁸  reported an average of 6 painful procedures per day in the NICU. Repetitive painful procedures can cause long-lasting effects on neonates and affect their pain processing and reactivity long term.¹⁹  This study demonstrates a safe way to protect newborns and their families from unnecessary harm by avoiding admission to NICU.

Although a clinical examination approach to EOS has been reported to be successful in a large university hospital, no community hospital has published a similar experience. Translation of care approaches from one hospital to another can be challenging, and the unique local processes, staffing and resources must be considered. For example, many MBUs in community hospitals are staffed by community pediatricians, often from many different practice groups. Finding consensus and implementing standardized practices can, therefore, be difficult. To overcome such barriers at our hospital, it was critical to have local champions who could build consensus and facilitate implementation. By reviewing the evidence and providing education to our community pediatricians through focused talks, faculty meetings, along with real-time in-person education during the care of neonates, our neonatology group was able to obtain buy-in on our new pathway. Similarly, inclusion of all providers and stakeholders, from obstetricians to labor and delivery nurses, to hospital leadership in townhalls and unit-based education programs, facilitated an overall cultural change around our EOS practices and likely helped support adoption of our new EOS pathway. One advantage of our hospital system that was incorporated into our approach is the ability to monitor neonates for up to 6 hours of life in our transition nursery. Mildly symptomatic neonates can be brought to the transition nursery and observed by NICU nurses and an in-house neonatologist. Depending on the resolution or progress of their symptoms, neonates are either transferred to the MBU or admitted for a higher level of care to the NICU. Four neonates in the postintervention period of this study were transferred to their mother after observation in the transition nursery.

This study has several limitations. First, our center has in-house neonatologists 24 hours per day, and we were able to develop an EOS pathway which required attendance of a neonatologist at all deliveries of febrile mothers. Community hospitals without in-house neonatologists could rely on in-house pediatricians, hospitalists, or NICU delivery attendance team (nurse and respiratory therapist) to attend these deliveries and evaluate neonates at birth. However, hospitals without any such coverage might have difficulty following our pathway. Second, our center had no cases of culture-positive EOS during the pre or postintervention study period. This limited our ability to assess the sensitivity of our new approach in the timely identification of infected neonates. However, our cohort represented over 10 000 births with an upper 95% confidence interval limit in the rate of EOS of 0.5 per 1000. The low rate of EOS is in line with recent previous reports^2,5  and it highlights the rarity of EOS in the current era of routine screening and IAP for Group B Streptococcus. Successful approaches for EOS will need to ensure systems of excellent clinical care and repeated assessments that allow timely identification of the “needle in the haystack.” Our updated approach is rooted in this framework. Lastly, although we did not find any readmissions in our hospital due to sepsis, we were unable to confirm potential readmissions to different hospitals in the area.

Use of a serial clinical examination approach for EOS in neonates was successfully and safely implemented in a community hospital setting. Implementation was associated with substantial reductions in antibiotic use and laboratory testing.