Implementation of a clinical guideline to decrease laboratory tests in newborns evaluated for early onset sepsis

Abbreviations

1 Introduction

Early-onset sepsis (EOS) remains a disease cause of neonatal morbidity and mortality [1]. Infants born to mothers with suspected chorioamnionitis are at higher risk for development of EOS [2, 3]. The Centers for Disease Control and Prevention (CDC) established guidelines to treat these newborns with empiric intravenous antibiotics for a minimum of 48 hours during the evaluation of EOS [4]. This recommendation was endorsed by both the American Congress of Obstetricians and Gynecologists (ACOG) and the American Academy of Pediatrics (AAP) [5, 6]. However, a clinical challenge arises when trying to determine if sepsis can be ruled out at 48 hours.

The evaluation of EOS in newborns remains difficult since early clinical symptoms and diagnostic markers are nonspecific [7, 8]. The gold standard for diagnosis of bacterial sepsis is a positive blood culture [9]. Culture positive EOS occurred in an estimate of 0.98 per 1000 live births [3] and represented approximately 5% of all clinically suspected EOS [10]. Unfortunately, there is no gold standard for ruling out sepsis. Though a negative blood culture is reassuring, it has been shown that a false negative result can occur up to 14% of documented cases of newborn sepsis at autopsy [11]. Furthermore, up to 25% of infants with EOS may have negative blood culture due to low bacterial concentrations [6, 12]. Therefore, physicians often use other clinical indices to determine the ongoing need for antibiotics such as the history, risk factors, the neonate’s clinical signs and symptoms, and other diagnostic blood tests such as a complete blood count with differential (CBC).

The CDC recommends obtaining a CBC at birth and/or at 6 to 12 hours after birth [4], whereas the AAP recommends waiting until 6 to 12 hours after birth to obtain the first CBC [6]. Unfortunately, the CBC has a poor positive predictive accuracy, especially when obtained shortly after birth because the newborn has not yet had enough time to establish an adequate inflammatory response [6]. Studies have found that due to exhaustion of bone marrow reserves, the CBCs shortly after birth will have low band counts leading to a falsely low ratio and will be more likely to be abnormal 6 to 12 hours after birth [13, 14]. The CBCs obtained later will have improved sensitivity and have better predictive values [15].

Despite the aforementioned recommendations, there is still a wide range of clinical practice and many providers obtain a CBC shortly after birth. In an effort to prevent EOS in accordance to the CDC national guidelines, our institution admits all infants born to mothers with suspected chorioamnionitis to the neonatal intensive care unit (NICU). Our initial evaluation of EOS typically includes obtaining a blood culture and CBC on NICU admission before the initiation of empiric antibiotics. There is also a wide range in how many serial CBCs are obtained in our institution.

The aim of the study was to implement a clinical guideline, aligning with the AAP, to obtain the first CBC at 12 hours after birth rather than shortly after birth on NICU admission and minimize serial CBC sampling for healthy term infants admitted for evaluation of EOS secondary to maternal diagnosis of chorioamnionitis. We hypothesized that this initiative would decrease the total number of CBCs drawn without negatively impacting patient care.

2 Methods

This clinical guideline was implemented at three of the Rady Children’s Hospital of San Diego Intermediate Level II NICU sites in La Jolla, Encinitas, and Murrieta and one Community Level III NICU site in Escondido. The study period was from June 1, 2015 to June 30, 2016. Baseline data of our EOS evaluation was collected from June 1, 2015 to October 31, 2015. The new clinical guideline was implemented on November 1, 2015. These changes in practice included that the first CBC should be collected at 12 hours after birth instead of on NICU admission and also discouraged obtaining sequential CBCs if the medical team did not find initial values concerning for EOS.

The inclusion criteria for this project were all term newborns that were evaluated for EOS due to suspected maternal chorioamnionitis. Per our institutional policies, these infants were admitted to the NICU and received empirical antibiotics. Suspected maternal chorioamnionitis was defined as either 1) a diagnosis of chorioamnionitis by the obstetrician, 2) maternal temperature equal to or greater than 101 degrees Fahrenheit during labor or within two hours after delivery, or 3) a maternal temperature between 100.4 to 100.9 degrees Fahrenheit with two other clinical symptoms from the following: fetal tachycardia (fetal heart rate greater than 160), maternal tachycardia (heart rate greater than 100), maternal white blood cell (WBC) count greater than 15,000, uterine tenderness, or foul smelling amniotic fluid. Exclusion criteria were preterm infants less than 37 weeks postmenstrual age, or infants that were not otherwise well and had other diagnoses besides evaluation of sepsis.

Patients were identified through the electronic medical records with admission diagnosis of sepsis evaluation secondary to maternal chorioamnionitis. Data was collected via a chart review and included demographic data (gestational age at birth, date and time of birth, gender, maternal age, gravida, para, mode of delivery, Apgar scores), date of discharge, length of stay, number of CBCs, date and time of CBCs collections, blood culture results, hospital readmissions and number of patients with culture negative sepsis defined as infants receiving seven or more days of antibiotics and having a negative blood culture result. The statistical significance of differences in categorical and discrete variables (i.e. abnormal / normal WBC, number of CBCs) after implementation of our guideline was determined with Fisher’s exact test. The statistical significance of changes in continuous variables (i.e. mean time of first CBC, length of stay) was determined by a two-sample t-test. All analyses were performed using the latest version of the R statistical software. This project was submitted to the University of California, San Diego Institutional Review Board (IRB) and was given an IRB exemption since it was a quality improvement project.

3 Results

There were 451 patients identified during this study period based on admission diagnosis of EOS evaluation secondary to maternal chorioamnionitis in our electronic medical records. We excluded 141 term infants admitted for evaluation of EOS because they did not meet the study criteria. There were 310 healthy term infants eligible for the study, of which 132 infants were born before and 178 infants were born after the implementation of the clinical guideline, respectively. See Table 1 for demographics of eligible infants.

After the implementation of our clinical guideline, the average number of CBCs per patient consistently decreased as demonstrated in Fig. 1. The means were 2.31±0.62 and 1.52±0.65, before and after the initiative respectively (p < 0.001). Significance was determined by Fisher’s exact test. Distribution of patients with different total number of CBC’s ordered for EOS significantly decreased after the protocol (p < 0.001), as determined by Fisher’s exact test. Initially, 93.9% of patients had at least two serial CBCs but decreased to 43.8% after the protocol. In addition, 33.3% of patients had a third CBC but decreased to 9.0% after the initiative (Table 2).

OPEN IN VIEWER

Fig.1

Run chart of the average number of CBCs obtained per patient before and after a clinical guideline implementation date (November 1st, 2015).

The mean time and median time of first CBC obtained were statistically different between the two groups (p < 0.001). Significance was determined by a two sample t-test and then by the Mann-Whitney U test. The mean time was 2.06±1.56 hours and the median (range) time was 1.69 (0.35, 11.38) hours after birth for the infants born prior to the protocol. During the implementation period, the mean time was 11.63±3.53 hours and median (range) time was 12.57 (1.0, 17.0) hours after infant’s birth (Fig. 2).

OPEN IN VIEWER

Fig.2

Time versus age distribution of the first CBC obtained by date. The vertical dashed line indicates a clinical guideline implementation date (November 1st, 2015). The horizontal dashed line represents the time at 12 hours after birth.

The components of the first CBC were analyzed and compared based on timing obtained after birth. There were 145 first CBCs obtained before 4 hours of birth compared to 136 first CBCs obtained 12 hours after birth. We found a significant increase in WBC values expressed in thousands with mean of 19.98±6.24 versus 22.67±7.03, respectively (p < 0.001). The number of CBCs with abnormal WBC count, defined as WBC >30 or <8, obtained before 4 hours after birth were 11/145 (7.6%) compared to 18/136 (13.2%) CBCs obtained after 12 hours of birth. This demonstrated an increase trend in the number of abnormal WBC count with time after birth but was not statistically significant (p = 0.174). Additionally, a trend was seen in an increase percentage mean of neutrophils (57.79±12.98 versus 60.59±11.64, respectively; p = 0.058).

The number of patients diagnosed with culture-negative sepsis decreased from 14 (10.6%) before compared to 13 (7.3%) after the protocol respectively but was not statistically significant (p = 0.317). Significance was determined by Fisher’s exact test. Only one infant had a group B streptococcus positive culture and was born before the clinical guideline implementation date.

The average hospital length of stay decreased from 2.35±1.73 days to 2.22±1.47 days but was not statistically significant (p = 0.488). Significance was determined by a two sample t-test. Two infants in each group were readmitted to the hospital within 30 days after birth (p = 0.764) but none of those four infants were readmitted for sepsis evaluation.

Following the CDC and AAP recommendations to prevent EOS, our community NICUs at Rady Children’s Hospital of San Diego admitted all infants born to mothers with suspected chorioamnionitis. Although the incidence of chorioamnionitis varies from 3–7%, the overall incidence of EOS in infants delivered to mothers with chorioamnionitis is low at 0.24%, or less than 1 in 400 [16]. In our practice, many asymptomatic infants are being admitted and treated with empiric antibiotics for EOS. In order to minimize CBC blood draws on our newborns, we designed a protocol to obtain the first CBC at 12 hours after birth rather than immediately after birth, in accordance with previous supporting literature [6, 15].

Consistent with our hypothesis, our guideline resulted in a significant decrease in mean number of CBC tests per patient. This is likely due to a combination of factors. Some providers in our practice automatically ordered a CBC on admission and a second CBC for the following morning. The natural history of both the WBC and neutrophil count is to rise [13, 17]. Seeing this increased trend on two serial CBCs (even if the counts are still in the normal range) could alarm the provider, resulting in more CBCs being ordered. Whereas, having the initial CBC at 12 hours after birth compels the providers to review the results before deciding if further CBCs are needed. Furthermore, our protocol allowed the providers to have more time to incorporate additional clinical assessments, including a second physical exam, prior to deciding if further CBCs were warranted.

We found a significant increase in WBC count at 12 hours which is consistent with previous studies [13 , 17]. One may speculate that having a higher WBC count will lead providers to order additional CBCs. However, we found the opposite to be true. Our practice was supported by Hornik et al study, which showed that high WBCs were not associated with EOS [18]. Prior to the guideline, only 6.1% of the infants had one CBC obtained whereas, after the initiative, 56.2% had only one CBC obtained. Therefore, the majority of infants (93.9%) had serial CBCs performed confirming our suspicions that unnecessary laboratory tests were being ordered on reflex prior to the clinical guideline implementation. By establishing this clinical guideline at our institution, we encouraged providers to review the laboratory findings and patient’s clinical presentation prior to ordering a subsequent CBC. Our results were reassuring that a decrease in serial CBCs did not negatively affect the hospital course of the infant. We did not find any adverse outcomes that resulted from our clinical guideline. We actually found a trend for shorter length of stay and less culture negative sepsis without an increase in hospital readmissions. Our one reported infant with positive GBS EOS was born before the clinical guideline implementation. This infant was clinically asymptomatic with normal CBCs but the blood culture became positive at 18 hours after birth while the infant was still on empirical antibiotics. Therefore, this infant’s medical management would not have been affected even if the infant was evaluated after the new clinical guideline implementation.

Certain limitations exist in our protocol. Despite the implementation of the guideline, some providers continue to order CBCs before 12 hours of birth, which could have a confounding impact on our study results. In addition, although our institution is the only major children’s hospital in the region, we may underestimate the readmission rate if some newborns were readmitted outside of our region. Given the overall low incidence of EOS, a larger study population may lead to different results.

This clinical guideline had direct benefits to our institution and, more importantly, to our patients as well. The utility of this protocol reduces clinical practice variations and improves compliance with national guidelines regarding the best timing of CBC for EOS. Furthermore, reducing the number of CBCs minimizes infant’s pain and decreases blood loss due to multiple laboratory draws [19]. This guideline also has potential in minimizing parental fear and anxiety during painful procedures [20]. Notably, there was an absent of significant adverse effects such as increased exposure to antibiotics for culture negative sepsis, missed sepsis, readmissions for EOS, or longer hospital stays.

Although, many institutions are still utilizing the AAP and CDC guidelines for EOS evaluation, there is an evolving trend to obtain less laboratory tests in asymptomatic neonates. Some institutions are utilizing the Kaiser sepsis calculator while others are developing their local EOS protocol. As more studies are being published, there will hopefully be some elucidation on future standards for EOS evaluation.

5 Conclusion(s)

Consistent with our hypothesis, the implementation of our clinical guideline resulted in a significant decrease in mean number of CBC tests per patient. We did not find any adverse outcomes due to our protocol. We actually found a trend for shorter length of stay and less culture negative sepsis without an increase in hospital readmissions.

Disclosure statements

None of the authors have any conflicts of interest to disclose. This manuscript was first drafted by Dr. Crystal Le and then reviewed/edited by the co-authors who uniformly approved for submission to The Journal of Neonatal-Perinatal Medicine. No funding was needed for this study. There was no honorarium, grant, or any other form of payment given to anyone to produce this manuscript.