Introduction: Compared with multivariate risk assessment, traditional category-based risk assessment (CRA) approaches for neonatal early-onset sepsis (EOS) screening are usually straightforward to use, do not require electronic devices, but are associated with higher rates of antibiotic use. This study aims to evaluate the performance of a novel enhanced CRA (eCRA) framework on EOS admissions and antibiotic use and to investigate whether a modified version with adjustments in risk factor weighting can allow its performance to match the EOS calculator while remaining easy to implement. Method: This is a prospective, single-center, two-phase observational study. Infants of all gestations delivered in a tertiary hospital in Hong Kong with risk factors or clinical features of EOS were recruited. Phase I: A novel eCRA framework (period 2) was compared with the CDC 2010-based protocol (period 1). Phase II: A modified eCRA framework was compared theoretically with the EOS calculator. EOS-specific admissions and antibiotic use were measured. Results: Phase I: 1,025 at-risk infants were recruited during period 2 and compared with 757 infants of period 1. Admissions and antibiotic use decreased from 45.8% to 29.4% and 41.1% to 28.2%, respectively. Antibiotics among those at-risk but well-appearing infants decreased from 25.3% to 16.3% (p < 0.001 for all). Phase II: antibiotic use was similar (7.3 vs. 6.4%, p = 0.42) between the modified eCRA framework and the EOS calculator. Conclusions: An eCRA framework can effectively and safely provide individualized guidance for EOS screening without the need for tools such as the EOS calculator.

Despite reductions in the incidence of neonatal early-onset sepsis (EOS) after the implementation of universal group B Streptococcus (GBS) screening and intrapartum antibiotic prophylaxis (IAP) [1], overtreatment of infants with risk factors remains common. The American Academy of Pediatrics has suggested three approaches to EOS screening [1], namely category-based risk assessment (CRA), multivariate risk assessment (MRA), e.g., the Kaiser Permanente EOS calculator, and examination-based risk assessment (ERA).

Prior to the development of the MRA and ERA approaches, CRAs, such as the Centers for Disease Control and Prevention (CDC) GBS-specific protocol [2], were the most widely implemented strategies for EOS screening. However, a CRA’s ability to provide individualized risk assessment is limited compared with MRA-based approaches such as the EOS calculator [3, 4]. ERA approaches [5] require frequent clinical assessments to avoid empirical treatment for risk factors alone. Most studies evaluating the real-world performance of these approaches are from high-resource centers where close surveillance is possible, EOS incidence is low, and electronic devices are readily available for MRA. In many localities, access to tools such as the EOS calculator (https://neonatalsepsiscalculator.kaiserpermanente.org) is restricted.

In our unit, universal GBS screening was performed, and IAP was administered according to the CDC 2010 GBS guidelines. For management of infants at-risk of EOS, we followed an algorithm based on this CDC protocol (see online suppl. materials; for all online suppl. material, see https://doi.org/10.1159/000534091). Empirical antibiotics were started in the neonatal unit for babies with high-risk, e.g., chorioamnionitis, while those at lower risk levels were monitored without antibiotics in nurseries or postnatal wards. Despite the incidence of EOS in our unit decreasing from 1.3 to 0.6 per 1,000 births between 2005 and 2018 [6], the number of infants treated for each proven case remained high.

However, transition to an MRA approach using the EOS calculator was not feasible in our unit because it only included infants born ≥34 weeks’ gestation and included fewer risk factors than were mandated by a territory-wide guideline [7]. Apart from gestational age [8], highest maternal temperature [9], duration of membranes ruptured [10], GBS status [11], and adequacy of IAP [12], we also included maternal white-cell count (WCC) [13], and other features of intrauterine infection such as fetal tachycardia [14]. Furthermore, the weighting of each risk factor in the EOS calculator was not modifiable by users. Importantly, the monitoring recommendations were difficult to implement in our setting [15] because “routine care” is assigned to some neonates with “equivocal” presentation. Instead of close monitoring in the nursery or the neonatal intensive care unit, this would mean placing babies with unresolved clinical features in the postnatal ward, where continuous close surveillance would be challenging. By incorporating features from MRA and ERA, we designed an enhanced CRA (eCRA) framework which could be applied to infants of all gestations and did not rely on access to electronic devices.

Hypothesis

We hypothesized that our novel eCRA framework could reduce admission rates and antibiotic use compared with the CDC 2010-based protocol and could also be adjusted to match the performance of the EOS calculator.

Objectives

  • 1.

    Assessing the performance of our eCRA framework compared with a CDC 2010-based protocol in terms of EOS-specific admissions and antibiotic use.

  • 2.

    Determine if our eCRA framework could be adjusted to match the performance of the EOS calculator.

We conducted a prospective, single-center, two-phase observational study at our unit, which serves a population of approximately 1.3 million and includes a level III neonatal intensive care unit. EOS was defined as positive culture of pathogen in blood or cerebrospinal fluid sampled within the first 72 h of life.

Study Periods

Clinical data were collected from period 1 (July 1, 2021, to December 14, 2021), when infants were managed with our CDC 2010-based protocol, and period 2 (December 15, 2021, to July 31, 2022), when infants were managed with our new eCRA framework.

Study Phases

In phase I, the outcomes of subjects in period 2 were compared against those in period 1. In phase II, our eCRA framework was modified to match the risk factors used in the EOS calculator. Data required for both our eCRA and the EOS calculator assessments were collected prospectively and throughout the inpatient stay of each subject. The management outcomes of infants recruited during period 2 were calculated retrospectively using both a modified eCRA (meCRA) framework and the EOS calculator. Although the calculations were conducted retrospectively, the first risk assessment before any treatment had been initiated would have been similar to applying the EOS calculator in a prospective manner.

Subjects

In phase I, infants of all gestations with either risk factors or clinical features of EOS were included (see Fig. 1). Infants were excluded if born outside our hospital. In phase II, infants born at gestation <34 weeks with prolonged rupture of membranes (PROM) of ≥240 h were excluded as the EOS calculator could not be applied with these characteristics.

Fig. 1.

Risk factors and clinical features of EOS.

Fig. 1.

Risk factors and clinical features of EOS.

Close modal

Outcome Measures

Primary outcomes in phase I included EOS-specific admissions and antibiotic use. Secondary outcomes included rates of enhanced monitoring, timing of antibiotic initiation, the number of cases, and the reason for transfer from the postnatal ward and staff compliance. In phase II, the rates of antibiotics and enhanced monitoring were compared.

eCRA Framework

Risk categories such as gestational age, IAP adequacy, duration of membrane rupture, highest maternal temperature, maternal WCC were considered. To emulate the individualized risk assessment offered by the EOS calculator, elements of an MRA approach were incorporated into the new algorithm.

As there was a territory-wide agreement to adhere to the CDC guideline for maternal GBS management, we could only implement our eCRA framework with risk factors based on the original protocol in period 1. Treatment would be warranted in the presence of one red flag, e.g., maternal septicemia, or multiple lower-risk factors. To provide an individualized risk assessment, each risk factor was stratified into risk levels with different EOS points assigned (see Fig. 2). For example, maternal temperature of ≥39°C would carry a higher weight than 38–38.9°C. This was applied to other categories such as gestational age and maternal WCC. The total EOS points were calculated by adding up points from all categories.

Fig. 2.

Management flowchart and EOS point chart of the enhanced category-based risk assessment (eCRA) framework.

Fig. 2.

Management flowchart and EOS point chart of the enhanced category-based risk assessment (eCRA) framework.

Close modal

An exhaustive permutational analysis was conducted to assess the >20,000 risk factor combinations. Risk factors’ weightings were evaluated to ensure that no combinations would result in an undesirable management plan. After examining all the combinations, we set the threshold for empirical antibiotic treatment at 4 points (see Fig. 2). All other at-risk infants with fewer than 4 points would receive enhanced monitoring, i.e., hourly heart rate, respiratory rate, pulse oximetry, temperature in the first 4 h then at least 6-hourly for 48 h or until discharge. This could be carried out either in the neonatal unit or postnatal ward.

Before the start of period 2, C-reactive protein (CRP) and WCC were measured in almost all at-risk infants. However, studies have shown that the validity of using CRP as a screening test in neonatal EOS is low [16, 17], especially in term infants [6]. Therefore, during period 2, routine screening of CRP was no longer performed in term infants. The use of WCC was retained but would be evaluated at 4–24 h of life according to the diagnostic utilities derived from local data [6].

For phase II, we modified our eCRA framework to match the EOS calculator by only considering risk factors evaluated by it and no longer considering factors like maternal WCC. Another permutational analysis was performed to evaluate the combinations. Infants with clinical signs would only be started on antibiotics if they fell into the category of “equivocal” or “clinical illness.” This aligned with the EOS calculator’s recommendation to avoid treating infants with transient features only [18].

Data Collection

Relevant maternal and neonatal data were extracted from an audit form (see online suppl. materials) and the medical records. The EOS calculator was retrospectively applied to infants ≥34 weeks of gestation in period 2. Vital signs charts were reviewed to determine infants’ clinical conditions according to different categories as defined by the EOS calculator. For cases missing the highest maternal intrapartum temperature, it was assumed to be 37.4°C, the maximum possible temperature above which pediatric staff would be notified.

Statistical Analysis

Statistical analysis was performed using IBM SPSS Statistics V26.0 (Armonk, NY, USA: IBM Corp.). Baseline characteristics of the two groups and the study outcomes were compared using the Mann-Whitney U test and χ2 test, as appropriate. Assuming a two-sided test with a 5% significance level and 80% power, the estimated study size required for each period was 315.

Ethics

The study was approved by the Joint The Chinese University of Hong Kong-New Territories East Cluster Clinical Research Ethics Committee.

During period 1, there were 1,850 infants, of whom 757 were screened (see Fig. 3) and 1 was diagnosed with EOS. During period 2, there were 2,258 infants, of whom 1,025 were screened and 2 were diagnosed with EOS. The baseline characteristics of the infants from the two periods were comparable (see Table 1).

Fig. 3.

Summary of patient distribution in phase I and phase II of the study.

Fig. 3.

Summary of patient distribution in phase I and phase II of the study.

Close modal
Table 1.

Baseline characteristics (phase I)

Period 1: CDC 2010-based protocol (N = 757)Period 2: eCRA framework (N = 1,025)p value
Gestationa, weeks 38.9 (38–40) 39 (38.14–40) 0.09 
Gestation, <35 weeks 5.9% (45) 5% (51) 0.802 
Birth weight, kg 3.085 (2.75–3.34) 3.095 (2.8–3.38) 0.12 
Sex     0.65 
 Male 51.9% (393) 50.8% (521)  
 Female 48.1% (364) 49.1% (504)  
Apgar <7 (1 min) 0.1% (57) 0.1% (80) 0.63 
Apgar <7 (5 min) 0.0% (2) 0.0% (10) 0.31 
Mode of delivery     0.34 
 Normal spontaneous delivery 64.5% (488) 67.4% (691)  
 Caesarean section 23.4% (177) 20.6% (211)  
 Instrumental 12.1% (92) 12.0% (123)  
PROM 16.8% (127) 17.2% (177) 0.79 
Liquor     0.34 
 Clear 88.1% (667) 89.4% (917)  
 Meconium-stained 10.4% (79) 9.8% (100)  
 Blood-stained 1.4% (11) 0.8% (8)  
GBS status     0.07 
 Positive 49.7% (376) 52.6% (539)  
 Negative 34.6% (262) 35.4% (363)  
 Unknown 15.7% (119) 12.0% (123)  
IAP adequacyb     0.03 
 Adequate 64.9% (273) 58.0% (359)  
 Inadequate 35.1% (148) 42.0% (260)  
Infection risks category     <0.001 
 Low 41.3% (313) 34.3% (352)  
 Medium 2.8% (21) 3.5% (36)  
 High 17.6% (133) 25.1% (257)  
 GBS with adequate cover 22.2% (168) 25.1% (258)  
 Unclassified 2.1% (16) 6.8% (70)  
 None 14.0% (106) 5.1% (52)  
Maternal condition 
 Intrapartum fever 24.0% (182) 25.7% (263) 0.44 
 Postpartum fever 14.1% (107) 14.7% (151) 0.72 
 Maternal leukocytosis 33.3% (252) 37.0% (379) 0.11 
 Foul-smelling liquor 0.1% (1) 0.0% (0) 0.24 
Fetal/neonatal condition 
 Fetal tachycardia 0.5% (4) 2.6% (27) 0.001 
 Presence of clinical features e.g., respiratory distress, fever 24.3% (184) 15.7% (161) <0.001 
Period 1: CDC 2010-based protocol (N = 757)Period 2: eCRA framework (N = 1,025)p value
Gestationa, weeks 38.9 (38–40) 39 (38.14–40) 0.09 
Gestation, <35 weeks 5.9% (45) 5% (51) 0.802 
Birth weight, kg 3.085 (2.75–3.34) 3.095 (2.8–3.38) 0.12 
Sex     0.65 
 Male 51.9% (393) 50.8% (521)  
 Female 48.1% (364) 49.1% (504)  
Apgar <7 (1 min) 0.1% (57) 0.1% (80) 0.63 
Apgar <7 (5 min) 0.0% (2) 0.0% (10) 0.31 
Mode of delivery     0.34 
 Normal spontaneous delivery 64.5% (488) 67.4% (691)  
 Caesarean section 23.4% (177) 20.6% (211)  
 Instrumental 12.1% (92) 12.0% (123)  
PROM 16.8% (127) 17.2% (177) 0.79 
Liquor     0.34 
 Clear 88.1% (667) 89.4% (917)  
 Meconium-stained 10.4% (79) 9.8% (100)  
 Blood-stained 1.4% (11) 0.8% (8)  
GBS status     0.07 
 Positive 49.7% (376) 52.6% (539)  
 Negative 34.6% (262) 35.4% (363)  
 Unknown 15.7% (119) 12.0% (123)  
IAP adequacyb     0.03 
 Adequate 64.9% (273) 58.0% (359)  
 Inadequate 35.1% (148) 42.0% (260)  
Infection risks category     <0.001 
 Low 41.3% (313) 34.3% (352)  
 Medium 2.8% (21) 3.5% (36)  
 High 17.6% (133) 25.1% (257)  
 GBS with adequate cover 22.2% (168) 25.1% (258)  
 Unclassified 2.1% (16) 6.8% (70)  
 None 14.0% (106) 5.1% (52)  
Maternal condition 
 Intrapartum fever 24.0% (182) 25.7% (263) 0.44 
 Postpartum fever 14.1% (107) 14.7% (151) 0.72 
 Maternal leukocytosis 33.3% (252) 37.0% (379) 0.11 
 Foul-smelling liquor 0.1% (1) 0.0% (0) 0.24 
Fetal/neonatal condition 
 Fetal tachycardia 0.5% (4) 2.6% (27) 0.001 
 Presence of clinical features e.g., respiratory distress, fever 24.3% (184) 15.7% (161) <0.001 

aMean and interquartile range of gestation (weeks) of both groups were presented; they did not follow a normal distribution as in the Kolmogorov-Smirnov test. See online supplementary materials for detailed distribution of gestation.

bPercentage (%) calculated based on the total number of cases indicated for IAP prophylaxis (n = 421 and 619, respectively).

Phase I – CDC 2010-Based Protocol versus eCRA Framework

With the eCRA framework, EOS-specific admissions decreased from 45.8% to 29.4% (p < 0.001) (Table 2). Overall antibiotic use decreased from 41.1% to 28.2% (p < 0.001), pre-emptive treatment in at-risk, but well-appearing infants decreased from 25.3% to 16.3% (p < 0.001). Subgroup analysis showed similar results for the group ≥35 weeks (Table 3). For <35 weeks, EOS admission did not differ, but there was an increase in empirical treatment for well-appearing babies (44.4 vs. 87.5%, p = 0.021). Enhanced monitoring was carried out on almost all infants during period 2 (98.2%). Significantly fewer cases were transferred from postnatal wards to the neonatal unit under the eCRA framework (17.2 vs. 4.1%, p < 0.001). There was no increase in babies developing EOS signs on the postnatal ward.

Table 2.

Outcomes (all gestations)

Phase I
period 1: CDC 2010-based protocol (N = 757)period 2: eCRA framework (N = 1,025)p value
Primary outcomes 
EOS-related admissions 
 Among at-risk infants 45.8% (347) 29.4% (301) <0.001 
 Among all deliveriesa 18.7% (347) 13.3% (301) <0.001 
Total antibiotic use 
 Among at-risk infants 41.1% (311) 28.2% (289) <0.001 
 Among all deliveriesa 16.8% (311) 12.8% (289) <0.001 
Antibiotic use in at-risk but well-appearing infantsb 25.3% (145) 16.3% (141) <0.001 
 Risk factors alone 17.6% (101) 15.7% (136)  
 Abnormal CRP 7% (40) 0.2% (2)  
 Abnormal WCC 0.7% (4) 0.3% (3)  
Antibiotic use among those with clinical featuresc 90.2% (166) 91.9% (148) 0.56 
Secondary outcomes 
Enhanced monitoring 88.8% (672) 98.2% (1,007) <0.001 
Time of antibiotics initiation, h 4.25 (2.23–17.83) 3.27 (2–8.65) 0.001 
Case transferred from postnatal ward to neonatal unitd 17.2% (82) 4.4% (30) <0.001 
Reason for transferal     <0.001 
 Risk features 8.0% (38) 3.4% (23)  
 Abnormal CRP 8.6% (41) 0.1% (1)  
 Abnormal WCC 0.8% (4) 0.9% (6)  
Compliance 80.3% (608) 91.2% (935) <0.001 
Phase I
period 1: CDC 2010-based protocol (N = 757)period 2: eCRA framework (N = 1,025)p value
Primary outcomes 
EOS-related admissions 
 Among at-risk infants 45.8% (347) 29.4% (301) <0.001 
 Among all deliveriesa 18.7% (347) 13.3% (301) <0.001 
Total antibiotic use 
 Among at-risk infants 41.1% (311) 28.2% (289) <0.001 
 Among all deliveriesa 16.8% (311) 12.8% (289) <0.001 
Antibiotic use in at-risk but well-appearing infantsb 25.3% (145) 16.3% (141) <0.001 
 Risk factors alone 17.6% (101) 15.7% (136)  
 Abnormal CRP 7% (40) 0.2% (2)  
 Abnormal WCC 0.7% (4) 0.3% (3)  
Antibiotic use among those with clinical featuresc 90.2% (166) 91.9% (148) 0.56 
Secondary outcomes 
Enhanced monitoring 88.8% (672) 98.2% (1,007) <0.001 
Time of antibiotics initiation, h 4.25 (2.23–17.83) 3.27 (2–8.65) 0.001 
Case transferred from postnatal ward to neonatal unitd 17.2% (82) 4.4% (30) <0.001 
Reason for transferal     <0.001 
 Risk features 8.0% (38) 3.4% (23)  
 Abnormal CRP 8.6% (41) 0.1% (1)  
 Abnormal WCC 0.8% (4) 0.9% (6)  
Compliance 80.3% (608) 91.2% (935) <0.001 
Phase II
meCRA framework (N = 971)EOS calculator (N = 971)p value
Total antibiotic use 7.3% (71) 6.4% (62) 0.42 
Enhanced monitoring 100% (971) 14.8% (144) <0.001 
Phase II
meCRA framework (N = 971)EOS calculator (N = 971)p value
Total antibiotic use 7.3% (71) 6.4% (62) 0.42 
Enhanced monitoring 100% (971) 14.8% (144) <0.001 

aPercentage calculated based on the total number of all deliveries (n = 1,850 and 2,258).

bPercentage calculated based on the total number at-risk but well-appearing infants (n = 573 and 864, respectively).

cPercentage calculated based on the total number of infants with clinical features (n = 184 and 161, respectively).

dExcluding cases transferred merely due to an update of the infection risk profile or non-EOS-specific reasons.

Table 3.

Subgroup analysis of phase I

≥35 weeks<35 weeks
period 1: CDC 2010-based protocol (N = 712), %period 2: eCRA framework (N = 974), %p valueperiod 1: CDC 2010-based protocol (N = 45), %period 2: eCRA framework (N = 51), %p value
Primary outcomes 
EOS-related admissions 
 Among at-risk infants 43.1 (307) 25.9 (252) <0.001 88.9 (40) 96.1 (49) 0.18 
 Among all deliveriesa 17.2 (307) 11.6 (252) <0.001 63.5 (40) 62.3 (49) 0.93 
Total antibiotic use 
 Among at-risk infants 38.2 (272) 24.6 (240) <0.001 86.7 (39) 96.1 (49) 0.10 
 Among all deliveriesa 15.2 (272) 11 (240) <0.001 61.9 (39) 62.8 (49) 0.91 
Antibiotic use in at-risk but well-appearing  infantsb 25 (141) 15 (127) <0.001 44.4 (4) 87.5 (14) 0.02 
 Risk factors alone 17.2 (97) 14.4 (122)  44.4 (4) 87.5 (14)  
 Abnormal CRP 7.1 (40) 0.2 (2)  (0) (0)  
 Abnormal WCC 0.7 (4) 0.4 (3)  (0) (0)  
Antibiotic use among those with clinical  featuresc 88.5 (131) 89.7 (113) 0.76 97.2 (35) 100 (35) 0.32 
Secondary outcomes 
Enhanced monitoring 88.2 (628) 98.2 (956) <0.001 97.8 (44) 100 (51) 0.29 
≥35 weeks<35 weeks
period 1: CDC 2010-based protocol (N = 712), %period 2: eCRA framework (N = 974), %p valueperiod 1: CDC 2010-based protocol (N = 45), %period 2: eCRA framework (N = 51), %p value
Primary outcomes 
EOS-related admissions 
 Among at-risk infants 43.1 (307) 25.9 (252) <0.001 88.9 (40) 96.1 (49) 0.18 
 Among all deliveriesa 17.2 (307) 11.6 (252) <0.001 63.5 (40) 62.3 (49) 0.93 
Total antibiotic use 
 Among at-risk infants 38.2 (272) 24.6 (240) <0.001 86.7 (39) 96.1 (49) 0.10 
 Among all deliveriesa 15.2 (272) 11 (240) <0.001 61.9 (39) 62.8 (49) 0.91 
Antibiotic use in at-risk but well-appearing  infantsb 25 (141) 15 (127) <0.001 44.4 (4) 87.5 (14) 0.02 
 Risk factors alone 17.2 (97) 14.4 (122)  44.4 (4) 87.5 (14)  
 Abnormal CRP 7.1 (40) 0.2 (2)  (0) (0)  
 Abnormal WCC 0.7 (4) 0.4 (3)  (0) (0)  
Antibiotic use among those with clinical  featuresc 88.5 (131) 89.7 (113) 0.76 97.2 (35) 100 (35) 0.32 
Secondary outcomes 
Enhanced monitoring 88.2 (628) 98.2 (956) <0.001 97.8 (44) 100 (51) 0.29 

aPercentage calculated based on the total number of all deliveries (n = 1,787 and 2,180, respectively, for ≥35 weeks, n = 63 and 78, respectively, for <35 weeks).

bPercentage calculated based on the total number at-risk but well-appearing infants (n = 564 and 848, respectively, for ≥35 weeks, n = 9 and 16, respectively, for <35 weeks).

cPercentage calculated based on the total number of infants with clinical features (n = 148 and 126, respectively, for ≥35 weeks, n = 36 and 35, respectively, for <35 weeks).

Phase II – meCRA Framework versus EOS Calculator

971 infants in period 2 were further assessed (see Fig. 3). Assuming an EOS incidence of 0.6 per 1,000 live births, the rates of antibiotic use were comparable between our meCRA framework and the EOS calculator (7.3 vs. 6.4%, p = 0.42). Enhanced monitoring was suggested in only 14.8% of screened infants by the EOS calculator compared to 100% under our meCRA framework (p < 0.001). Two infants with EOS were diagnosed during period 2. Both cases presented with signs of EOS and would have been treated regardless of the protocol used.

Implementation of our eCRA framework led to a reduction of EOS-specific admissions from 18.7% to 13.3% and antibiotic use from 16.8% to 12.8%. After matching the risk factors and management strategies of our eCRA framework with the EOS calculator, their performance was similar in terms of antibiotic use for those screened for EOS (7.3 vs. 6.4%).

The use of conventional CRAs such as the CDC 2010 guideline [2] and NICE guideline CG149 [19] is associated with high rates of empirical antibiotic use [20, 21]. This is likely due to the oversimplification of risk factors which are considered to be dichotomous. In our eCRA framework, we combined elements of an MRA and assigned weighting to these variables to enable a more nuanced approach. For example, isolated maternal fever of 39°C is assigned 4 EOS points and would warrant empirical treatment even in isolation; however, for lower temperatures, other risk factors such as maternal leukocytosis, fetal tachycardia would be required. For infants <35 weeks gestation, there was an increase in empirical treatment use. This could be explained by the lower gestational age and higher risk categories among infants in period 2 compared with period 1 (see online suppl. materials). In addition, a relatively higher weighting on gestational age and a wider coverage of risk factors in the eCRA framework may lead to more empirical treatment in preterm infants. However, as the numbers of infants included were small, the results should be interpreted with care.

Regarding safety, our protocol identified both EOS cases in a timely fashion. There was no increase in transfers from the postnatal wards for EOS management. Significantly fewer cases (41 vs. 1) were transferred to the neonatal unit for raised CRP. Despite a decreased reliance on CRP, we did not see an increase in cases developing signs of sepsis.

As reported in a meta-analysis [22], the use of the EOS calculator was associated with a significant reduction in empirical antibiotic treatment when compared with older management protocols. Similarly, a much lower antibiotic rate was found (6.4%) using the EOS calculator in period 2 compared with the CDC 2010-based protocol or eCRA framework. One of the reasons may be a wide list of risk factors used in CRAs, which also explained a high proportion of infants that were considered at-risk (40.9 and 47.5% in the two periods). However, in phase II, we demonstrated that by simplifying the eCRA framework and only consider risk factors listed in the EOS calculator, the performance on antibiotic use becomes equivalent. This is an important finding, suggesting that these promising outcomes are reproducible without requiring a transition to the EOS calculator.

The use of an eCRA framework carries a few advantages over the EOS calculator. First, it is easily accessible in the form of a single-page chart. While the EOS calculator is freely available in many Western countries, its access is limited in some localities around the world. In addition, our eCRA framework provides a transparent assessment of risk factors that contribute to EOS risk. Furthermore, these risk factor weightings can be readily adjusted after considering local data and/or a permutational analysis. In contrast, the data that the EOS calculator is based on [3] may not be applicable to the locality where it is being used. With only the incidence being user-modifiable, further studies may be required to evaluate its applicability in local settings. Finally, infants with delayed clinical features may not be identified promptly by the EOS calculator. In our study, “routine care” was assigned to 65.2% of infants born to mothers with clinical chorioamnionitis and 92.3% with inadequate IAP by the EOS calculator.

There are several limitations to our study. First, the number of proven EOS cases was limited owing to a low incidence of EOS (0.7 per 1,000), limiting the safety assessment of each protocol. Nevertheless, our protocol achieved significant reduction in admissions and antibiotic use without missing any infants with EOS. Secondly, subgroup analysis for preterm infants <35 weeks was limited by a small sample size. Third, the meCRA framework and EOS calculator were not used to prospectively influence clinical management. Thus, it is possible that the safety of these approaches may have been inflated. Future multicenter, randomized controlled trials would be required to further evaluate these protocols. Our study demonstrates that in units where a direct implementation of the EOS calculator is not feasible, safe reductions in EOS treatment are possible with our eCRA framework, which provides a broad coverage of risk factors, structured clinical surveillance, and a transparent risk assessment that is readily accessible and easily adaptable.

Our eCRA framework has successfully reduced admissions and antibiotic use in our unit without delaying antibiotic initiation or missing EOS cases. With further adjustment, it can match the performance of the EOS calculator.

This study protocol was reviewed and approved by The Joint The Chinese University of Hong Kong-New Territories East Cluster Clinical Research Ethics Committee, approval number [2022.118]; it has also been granted an exemption from requiring written informed consent.

The other authors have no conflicts of interest to disclose.

No funding was secured for this study.

Lam, Hugh Simon conceptualized and designed the study, supervised the study, liaised with the various team members of the research team, and critically reviewed and revised the manuscript. Lau, Hoi Ying Sharon designed the study, collected, and analyzed the data. She liaised with various team members of the research team, wrote the first draft, and critically revised the manuscript. Wang, Xuelian participated in data interpretation and critically revised the manuscript for important intellectual content. Wong, Ho Tsun Michelia and Lam, Ka Hei Catherine participated in data collection and interpretation and critically revised the manuscript for important intellectual content. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

All data generated or analyzed during this study are included in this article and its supplementary material files. Further inquiries can be directed to the corresponding author.

1.
Puopolo
KM
,
Benitz
WE
,
Zaoutis
TE
,
Cummings
J
,
Juul
S
,
Hand
I
.
Management of neonates born at ≥35 0/7 Week gestation with suspected or proven early-onset bacterial sepsis
.
Pediatrics
.
2018 Dec
142
6
e20182894
.
2.
Verani
JR
,
McGee
L
,
Schrag
SJ
Division of Bacterial Diseases, National Center for Immunization and Respiratory Diseases, Centers for Disease Control and Prevention CDC
.
Prevention of perinatal group B streptococcal disease: revised guidelines from CDC
.
MMWR Recomm Rep
.
2010 Nov 19
59
RR-10
1
36
.
3.
Puopolo
KM
,
Draper
D
,
Wi
S
,
Newman
TB
,
Zupancic
J
,
Lieberman
E
.
Estimating the probability of neonatal early-onset infection on the basis of maternal risk factors
.
Pediatrics
.
2011 Nov
128
5
e1155
63
.
4.
Kuzniewicz
MW
,
Puopolo
KM
,
Fischer
A
,
Walsh
EM
,
Li
S
,
Newman
TB
.
A quantitative, risk-based approach to the management of neonatal early-onset sepsis
.
JAMA Pediatr
.
2017 Apr 1
171
4
365
71
.
5.
Berardi
A
,
Buffagni
AM
,
Rossi
C
,
Vaccina
E
,
Cattelani
C
,
Gambini
L
.
Serial physical examinations, a simple and reliable tool for managing neonates at risk for early-onset sepsis
.
World J Clin Pediatr
.
2016 Nov 8
5
4
358
64
.
6.
Wang
X
Big data for quality improvement in neonatal clinical care[D]. Hong Kong SAR
China
The Chinese University of Hong Kong
2021
. https://repository.lib.cuhk.edu.hk/tc/item/cuhk-3121903?solr_nav%5Bid%5D=2a4081feb9e5f51a8740&solr_nav%5Bpage%5D=0&solr_nav%5Boffset%5D=0.
7.
Wang
X
,
Chan
PHY
,
Lau
HYS
,
Tsoi
K
,
Lam
HS
.
Epidemiologic changes of neonatal early-onset sepsis after the implementation of universal maternal screening for group B Streptococcus in Hong Kong
.
Pediatr Infect Dis J
.
2023 Jul 3
42
10
914
20
.
8.
Mukhopadhyay
S
,
Puopolo
KM
.
Risk assessment in neonatal early onset sepsis
.
Semin Perinatol
.
2012 Dec
36
6
408
15
.
9.
Committee Opinion No. 712
.
Summary: intrapartum management of intraamniotic infection
.
Obstet Gynecol
.
2017 Aug
130
2
490
2
.
10.
Pryles
CV
,
Steg
NL
,
Nair
S
,
Gellis
SS
,
Tenney
B
.
A controlled study of the influence on the newborn of prolonged premature rupture of the amniotic membranes and/or infection in the mother
.
Pediatrics
.
1963 Apr
31
4
608
22
.
11.
Benitz
WE
,
Gould
JB
,
Druzin
ML
.
Risk factors for early-onset group B streptococcal sepsis: estimation of odds ratios by critical literature review
.
Pediatrics
.
1999 Jun
103
6
e77
.
12.
Puopolo
KM
,
Madoff
LC
,
Eichenwald
EC
.
Early-onset group B streptococcal disease in the era of maternal screening
.
Pediatrics
.
2005 May
115
5
1240
6
.
13.
Gibbs
RS
,
Blanco
JD
,
St Clair
PJ
,
Castaneda
YS
.
Quantitative bacteriology of amniotic fluid from women with clinical intraamniotic infection at term
.
J Infect Dis
.
1982 Jan
145
1
1
8
.
14.
Romero
R
,
Chaemsaithong
P
,
Korzeniewski
SJ
,
Kusanovic
JP
,
Docheva
N
,
Martinez-Varea
A
.
Clinical chorioamnionitis at term III: how well do clinical criteria perform in the identification of proven intra-amniotic infection
.
J Perinat Med
.
2016 Jan
44
1
23
32
.
15.
Achten
NB
,
Plötz
FB
,
Klingenberg
C
,
Stocker
M
,
Bokelaar
R
,
Bijlsma
M
.
Stratification of culture-proven early-onset sepsis cases by the neonatal early-onset sepsis calculator: an individual patient data meta-analysis
.
J Pediatr
.
2021 Jul
234
77
84.e8
.
16.
Khan
F
.
C-Reactive protein as a screening biomarker in neonatal sepsis
.
J Coll Physicians Surg Pak
.
2019 Oct
29
10
951
3
.
17.
Dhudasia
M
,
Puopolo
K
,
Mukhopadhyay
S
,
Dysart
K
Impact of C-reactive protein results on prolonged early antibiotic use in infants with negative cultures
.
2019
.
18.
Kuzniewicz
MW
,
Walsh
EM
,
Li
S
,
Fischer
A
,
Escobar
GJ
.
Development and implementation of an early-onset sepsis calculator to guide antibiotic management in late preterm and term neonates
.
Jt Comm J Qual Patient Saf
.
2016 May
42
5
232
9
.
19.
NICE
Neonatal infection (early onset): antibiotics for prevention and treatment CG149 2012
.
20.
Gievers
LL
,
Sedler
J
,
Phillipi
CA
,
Dukhovny
D
,
Geddes
J
,
Graven
P
.
Implementation of the sepsis risk score for chorioamnionitis-exposed newborns
.
J Perinatol
.
2018 Nov
38
11
1581
7
.
21.
Kimpton
JA
,
Verma
A
,
Thakkar
D
,
Teoh
S
,
Verma
A
,
Piyasena
C
.
Comparison of NICE guideline CG149 and the sepsis risk calculator for the management of early-onset sepsis on the postnatal ward
.
Neonatology
.
2021
;
118
(
5
):
562
8
.
22.
Achten
NB
,
Klingenberg
C
,
Benitz
WE
,
Stocker
M
,
Schlapbach
LJ
,
Giannoni
E
.
Association of use of the neonatal early-onset sepsis calculator with reduction in antibiotic therapy and safety: a systematic review and meta-analysis
.
JAMA Pediatr
.
2019 Nov 1
173
11
1032
40
.