Early-onset neonatal sepsis (EOS) is a bacterial infection of the bloodstream or cerebrospinal fluid that occurs within 72 h of birth, with a 3% mortality rate among late preterm and term neonates (1). Key risk factors for EOS include premature rupture of membranes (ROM), maternal fever, prematurity, prolonged labor, and colonization with Group B Streptococcus (GBS) or other pathogenic bacteria (2). GBS and E. coli are the leading organisms (2). The global incidence of EOS varies widely, ranging from 0.5–0.79 per 1,000 live births in high-income countries (3, 4) and 3–24 per 1,000 in low- and middle-income countries (5). Developed countries, such as the United States, have reported declining EOS rates, primarily due to the widespread use of maternal screening and intrapartum prophylaxis for GBS (1). The implementation of universal GBS screening and the Centers for Disease Control and Prevention (CDC) guidelines in the USA resulted in 14%–24% of newborns undergoing blood tests and 5%–12% receiving antibiotics, while the EOS incidence remained below 0.1% (6, 7).

Neonatal antibiotic use has been associated with a higher risk of asthma, allergy, and autoimmune diseases (8–10). Other potential drawbacks include longer hospital stays, separation from the mother, and increased costs. The American Academy of Pediatrics (AAP) recommends three evidence-based strategies for the secondary prevention of EOS (2). These are the multivariate risk prediction model, categorical risk assessment, and reserving antibiotics for infants showing clinical signs. The National Institute for Health and Care Excellence (NICE) UK suggests using a combination of clinical indicators and risk factors, recommending antibiotics if multiple criteria or red flag signs are present, while advocating for close monitoring in lower-risk cases (11). Other organizations, such as the Canadian Pediatric Society, employ similar risk- and sign-based algorithms, structured assessments, and ongoing clinical evaluations (12).

The Early-Onset Sepsis risk Calculator (EOSCAL), designed by the Kaiser Permanente (KP) group in the US, is a multivariate model for predicting the EOS risk in neonates (13). It is accessible online at https://neonatalsepsiscalculator.kaiserpermanente.org/InfectionProbabilityCalculator.aspx.

EOSCAL derives the Sepsis Risk Score (SRS) at birth by modifying the population risk with intrapartum risk factors specific to the mother-infant pair; it then incorporates the infant's clinical status during the first 12 h to compute the final SRS. Although the SRS computation algorithm is evidence-based, the clinical management recommendations (based on the SRS) are based on the consensus of the original researchers. The original model (2017 version) was based on data collected prior to the implementation of universal maternal GBS screening across KP hospitals. The authors of EOSCAL have recently updated the calculator (2024 version) based on more recent prospective data and recommend using the updated calculator in the context of universal GBS screening.

They recommended intravenous antibiotics for SRS ≥ 3, blood culture for SRS between 1 and <3, and routine care for SRS < 1. The implementation of EOSCAL in KP hospitals resulted in a 50% reduction in antibiotic use without increasing the incidence of missed EOS or readmissions (13). Multiple centers have consistently reported a 45%–50% reduction in antibiotic use following the implementation of the calculator (6, 14).

The AAP policy has endorsed EOSCAL for managing EOS in infants greater than 34 weeks' gestation (2). However, some reports have raised concerns about the risk of delayed and missed treatment for babies with EOS. Scott et al. (15) reported that the use of EOSCAL missed 35% of EOS cases, compared to 3% with the Australian guidelines. In a systematic review including 75 EOS cases identified by the NICE recommendation, Pettinger et al. (2019) (16) estimated that at least 14 (18%) cases would be missed if EOSCAL was used.

Since mid-2018, Qatar has employed the universal GBS screening strategy for pregnant mothers. The overall incidence of GBS EOS in Qatar is 0.58/1,000, and no significant trend has been observed since 2015 (17). To manage neonates at risk for EOS, the hospitals in the country currently follow the categorical approach proposed by the CDC (18), resulting in high antibiotic use. A recent quality improvement study from Qatar (19) observed that 11% of all liveborn babies received early antibiotics. The same study reported a 60% reduction in antibiotic use among term, well-appearing chorioamnionitis-exposed babies when EOSCAL was applied, without missing any actual EOS cases. Although EOSCAL use significantly impacts patient care, flow, and cost reduction, the possibility of missing and delaying treatment is a concern. Our study aimed to estimate the proportion of EOS cases that would have been correctly identified by retrospectively applying EOSCAL to a cohort of culture-proven EOS cases in Qatar.

Our study aimed to estimate the proportion of culture-proven EOS cases in Qatar that would have been correctly identified by retrospectively applying EOSCAL.

Table 1 shows the baseline data. Between July 2015 and December 2022, among 179,147 live births, there were 157 cases of early-onset sepsis (EOS), resulting in an incidence rate of 0.88 per 1,000 live births. The study included 105 EOS cases born at 34 weeks of gestation or later. Of these, 44.9% were male and 58.1% were female; 40% were Qatari and 60% non-Qatari. The mean birth weight was 3,203 grams (±460), and the mean gestational age was 38.4 weeks (±1.6). Most infants were between 37 and 40 weeks' gestation. Group B Streptococcus was the predominant organism (69.5%), followed by E. coli (7.6%). The mortality rate among the included cases was 7.6%.

The categorization and timing of management, according to the CDC algorithm (18), are shown in Figure 1.

Peripartum risk factors and neonatal clinical status are detailed in Table 2. Premature rupture of membranes beyond 18 h before delivery occurred in 13.3% of cases. The GBS carrier status was unknown for 54.3% of mothers, and 12.4% were confirmed carriers. Approximately 22% of mothers had an intrapartum temperature ≥38°C, and 28.6% had chorioamnionitis. Although antibiotics were indicated for 41.9% of mothers, only 6.7% received adequate intrapartum antibiotics. Among 17 mothers who received intrapartum antibiotics more than 2 h before delivery, 15 received broad-spectrum antibiotics, which included a combination of ampicillin or ceftriaxone with gentamicin.

During the first hour of life, 56.2% of infants were well-appearing, 1.9% had equivocal status, and 41.9% showed clinical illness. Symptom onset was gradual, occurring in 41.9% within the first hour, rising to 67.6% by 12 h, and 91.5% by 72 h. About 11.4% exhibited symptoms such as hypoactivity or poor sucking and were not classified as clinical or equivocal illness. Notably, 8.5% of babies remained asymptomatic throughout.

By 12 h of age, 58.1% were categorized as clinically ill, 40% as well-appearing, and 1.9% as equivocal.

At birth, 54.3% of neonates had an SRS below 1, indicating routine care, while 37.1% had an SRS of 3 or higher, warranting blood cultures and antibiotics (Table 3). By 12 h, the proportion with SRS ≥ 3 increased to 58%. Correspondingly, those with SRS < 1 decreased to 33%.

Table 4 shows the comparative efficacy of EOSCAL and CDC recommendations in identifying EOS cases at birth and at 12 h of age. At birth, the EOSCAL recommended antibiotics in 37.1% (95% CI, 27.8%–46.4%) of cases, significantly fewer than the CDC guidelines, which recommended antibiotics in 52% (95% CI, 42.9%–62.0%) (p < 0.01). This resulted in the delayed identification of 16 additional cases by EOSCAL, compared to the CDC recommendation. By 12 h, EOSCAL's detection rate improved to 58.1% (95% CI 48.6%–67.5%) but remained significantly lower than the CDC recommendation of 72.5% (95% CI 63.9%–80.9%) (p < 0.01), resulting in missed or delayed antibiotic treatment in 15 cases. However, considering 9 cases for whom SRS recommended blood culture, the overall number of missed cases would have been 6.

There were eight deaths (7.6%), and all deaths occurred after 12 h of age. Among the eight babies who expired, EOSCAL recommended antibiotics for two and blood culture for another two at birth. CDC recommended antibiotics for all four of them at birth. By 12 h, neither EOSCAL nor CDC had recommended antibiotics for three infants, who subsequently died between 18 and 32 h of age.

Logistic regression was used to identify predictors for the missed cases by EOSCAL but not by the categorical approach of CDC (Table 5). All the evaluated variables—including gender, gestational age, birth weight, GBS EOS, chorioamnionitis, mode of delivery, ROM > 18 h, being symptomatic by 1 h, or being asymptomatic during the first 12 h—were not significant.

This study observed an EOS incidence of 0.88/1,000 live births, with a predominance of GBS, aligning with reports from high-income countries (3, 4). The observed mortality of 7.6% among late preterm and term babies with EOS exceeded the reported rate of 3% in developed countries (20).

Notably, 40% of babies had no identifiable prenatal risk factors other than unknown GBS status (term gestation, no indication for intrapartum antibiotics, no chorioamnionitis, no maternal fever or rupture of membranes beyond 18 h). These babies were identified based on clinical signs or symptoms at various time points, from birth to 72 h. If the CDC recommendation (18) was applied, 42% of mothers would require intrapartum antibiotics. However, more than 75% of them did not receive adequate intrapartum antibiotics, mostly due to the late appearance of risk factors, such as fever.

More than 50% of the babies were asymptomatic during the first hour. Even at 12 h of age, 40% of the babies continued to appear well. Moreover, some babies with symptoms like poor sucking or hypoactivity would have been misclassified as well-appearing. Consequently, EOSCAL would have recommended antibiotics to only 37% (95% CI, 27.8%–46.4%) during the first hour and to 58% (95% CI, 48.6%–67.5%) by 12 h. This would be significantly lower (p < 0.01) than the cases identified by the CDC's recommendation of 52% (95% CI 42.9%–62.0%) during the first hour and 72.5% (95% CI 63.9%–80.9%) by 12 h of age. Compared to the CDC recommendation, EOSCAL would have delayed antibiotics in 16 additional cases during the first hour and 15 cases by 12 h of age. Excluding 9 babies for whom the calculator recommended blood culture without antibiotics, six additional cases would have been missed by EOSCAL at 12 h.

Following the publication of the original recommendation (13), several studies have confirmed the efficacy of EOSCAL in reducing the blood culture rate, antibiotic use, and hospital stay (6, 14). The AAP has also recommended it as an evidence-based tool for risk stratification and management of EOS. The recently updated EOSCAL model (21) has a sensitivity of 0.80 (95% CI 0.68–0.89). They also reported that the use of EOSCAL did not increase the incidence or mortality of EOS. Yoshida et al. (2022) (22) conducted a prospective observational study in 26 NHS neonatal units over a 12-month period, comparing EOSCAL with NICE guidelines. There was no statistically significant difference in the number of culture-proven EOS cases missed by EOSCAL compared to the NICE guidelines.

However, some recent reports have raised concerns about the safety of EOSCAL. In a systematic review, Pettinger et al. (2019) (16) compared EOSCAL to NICE guidelines and observed that EOSCAL would have missed or delayed treatment in 14–22 cases out of 75 culture-proven EOS cases. Scott et al. (2021) (15) reviewed 31 consecutive cases of EOS among babies greater than 34 weeks' gestation who were admitted to the Royal Brisbane and Women's Hospital in Australia. They compared the efficacy of EOSCAL and the Queensland Health Guidelines in identifying EOS cases by 24 h of age. Using the neonatal calculator, 11 neonates (35%, 95% confidence interval 19.2%–54.6%) would not have received antibiotics by 24 h of age. In comparison, only one neonate (3%, 95% confidence interval 0.1%–16.7%) would not have received antibiotics by 24 h of age using the local guidelines. Similarly, Achten et al. (2021) (23) published a systematic review involving individual patient data meta-analysis of 234 cases of culture-proven EOS. The calculator recommended antibiotics for 40.6% (95% CI, 34.2%–47.2%) upon initial assignment and for 60.1% (95% CI, 54.5%–67.4%) by 12 h of age. However, this review included studies from different countries with varying baseline EOS risks and clinical practices, which introduces bias.

Our findings align with those of the above systematic reviews reported by Pettinger et al. (2019) (16) and Achten et al. (2021) (23). The finding that 40% of cases had no underlying risk factors and over a quarter of babies were asymptomatic at 12 h emphasizes the importance of continued postnatal observation.

Notably, the CDC recommendations performed better in the early identification and treatment of four of the eight babies who later expired. However, three of the expired babies were not identified by both methods, even at 12 h; despite that, these deaths occurred between 18 and 32 h of age.

While some studies suggest that babies exposed to chorioamnionitis have the highest risk of being missed by EOSCAL (16), our logistic regression found no significant predictors of missed cases, including chorioamnionitis. Although the proportion of chorioamnionitis was higher among the missed cases (46%) compared to the entire EOS cohort (28.5%), this difference did not reach statistical significance (p = 0.23), likely due to the limited sample size. Early use of intrapartum antipyretics is common in our settings. A previous study by our team observed that nearly half of the mothers with suspected chorioamnionitis had a peak intrapartum temperature below 38°C, stressing the possibility that antipyretics may affect the peak temperature (24). The developers of EOSCAL have also cautioned about the accuracy of peak intrapartum temperature in settings with routine antipyretic use in the delivery units.

In our team's prior quality improvement project using EOSCAL among asymptomatic term babies born to mothers with chorioamnionitis, none of the three EOS cases were missed (19). However, this protocol had additional safety nets, like excluding symptomatic and preterm babies and observing all babies for a minimum period of 48 h. Moreover, a higher baseline risk of 2 in 1,000 was used.

In summary, we observed that the direct application of EOSCAL to our setting would have been inferior to the existing practice. Antibiotic administration would have been delayed or missed in a significant proportion of EOS cases. Implementing the EOS Calculator should consider these proportions in local circumstances. Clinical vigilance remains essential for all newborns. Future studies should compare multiple strategies in local settings, involving careful monitoring and follow-up.

In this study, the incidence of early-onset sepsis was 0.88 per 1,000 live births, predominantly caused by Group B Streptococcus, consistent with trends reported in high-income countries. Notably, 40% of the affected neonates lacked identifiable prenatal risk factors, underscoring the critical need for vigilant postnatal clinical observation, as a substantial proportion of cases presented with delayed symptoms.

When comparing the EOSCAL to CDC guidelines, our findings indicate that EOSCAL would have delayed or missed antibiotic treatment in a significant number of cases, highlighting limitations of direct application in our setting. These results align with systematic reviews that question the safety of EOSCAL in certain populations. Therefore, local baseline risk, clinical practices, and rigorous monitoring protocols should inform the decision to implement EOSCAL alongside continued clinical vigilance for all newborns.

Future prospective studies comparing multiple risk stratification strategies within local contexts are needed to optimize EOS management while minimizing missed cases and unnecessary antibiotic use.