Abstract
Objective:
To determine delivery risk phenotype-specific incidence of early-onset sepsis (EOS) among preterm infants.
Study Design:
Retrospective cohort study of infants born <35 weeks’ gestation at four perinatal centers during 2017–2021. Infants were classified into one of six delivery risk phenotypes incorporating delivery mode, presence of labor, and duration of rupture of membranes (ROM). The primary outcome was EOS incidence within the overall cohort and each risk phenotype.
Results:
Among 2,937 preterm infants, 21 had EOS (0.7%, or 7.1 cases/1000 preterm infants). The majority of EOS cases (13/21, 62%) occurred in the setting of prolonged ROM ≥18 hours, with a phenotype incidence of 23.8 cases/1000 preterm infants. There were no EOS cases among infants born by cesarean section without ROM (with or without labor), nor via cesarean section with ROM <18 hours without labor.
Conclusion:
Delivery risk phenotyping may inform EOS risk stratification in preterm infants.
Keywords: prematurity, infant, neonate, early-onset sepsis, bacteremia
INTRODUCTION
Neonatal early-onset sepsis (EOS) remains a significant cause of morbidity and mortality among very preterm infants, with a contemporary incidence of approximately 13–18 cases per 1000 preterm births <29 weeks gestation (1,2) and associated all-cause mortality as high as 35% (3). Given the high EOS burden in this population, 80–90% of very-low birth weight infants (<1500 grams) are administered empiric antibiotics at birth (4). Clinicians must balance decisions to treat preterm infants at risk of EOS against emerging evidence suggesting potential adverse impacts of early antimicrobial exposure, including effects on growth (5,6) as well as on later preterm morbidities (e.g., late-onset sepsis, necrotizing enterocolitis, brain injury, bronchopulmonary dysplasia) and mortality (7–12).
Among term infants, multivariable EOS risk-stratification tools have been associated with decreased empiric antibiotic utilization (13). In contrast, risk stratification is challenging in preterm infants and validated approaches are limited to those at lowest risk of EOS(14–17). Multivariable risk models for term infants leverage the markedly different prevalence of established EOS risk factors among infected and uninfected newborns, and the relatively low (approximately 1/2000) prior probability of EOS. However, prematurity is the single strongest predictor of EOS (1,2,18). Risk factors such as preterm premature rupture of membranes and concern for intra-amniotic infection (IAI) are commonly present among preterm infants, and EOS prevalence ranges from approximately 1/20 to 1/100 preterm births, depending on gestational age.
Because EOS pathogenesis is most often ascribed to intrapartum ascending infection with maternal flora, EOS risk is also modulated by delivery characteristics defining the mechanism of preterm birth (PTB) (18). Current American Academy of Pediatrics (AAP) guidance for EOS management in infants <35 weeks gestation focuses on the indication for and delivery characteristics of preterm birth to define a narrow category of preterm infants at lower EOS risk (19). This category is characterized by cesarean section birth (CB) for non-infectious indications, absence of labor, and rupture of membranes (ROM) at the time of delivery(14–17,19).
A more comprehensive application of delivery characteristic-based risk phenotypes may further discriminate preterm EOS risk. We hypothesized that EOS incidence among preterm infants is unequally distributed and may cluster by delivery risk phenotypes. The objective of this study was to categorize preterm infants <35 weeks gestation by delivery risk phenotype, and to assess whether EOS incidence differs by phenotype.
METHODS
Study Design and Study Population
We performed a retrospective cohort study of preterm infants born at one of four perinatal centers affiliated with the University of Pennsylvania Health System. The cohort consisted of all preterm infants who were inborn, <35 0/7 weeks gestation, and admitted to the neonatal intensive care unit (NICU) between January 1, 2017 and December 31, 2021. Exclusion criteria included: (1) outborn status; (2) provision of palliative care in the delivery room; (3) survival <6 hours after birth; or (4) transfer out of the study site prior to day 3 after birth. Participant data were extracted from the health system’s common electronic medical record, with manual chart review performed to verify preterm birth indications and characteristics. This study was approved by the Institutional Review Board at the University of Pennsylvania and granted a waiver of informed consent.
Definitions
We defined novel delivery risk phenotypes based on a combination of objective characteristics that are related to the birth mechanism, are known at the time of birth, and are associated with EOS risk. Characteristics composing the six prespecified delivery risk phenotypes included the mode of delivery, presence or absence of labor, and rupture of membranes (ROM) duration: (1) CB with ROM at delivery and no labor, (2) CB with ROM at delivery and any labor, (3) CB with ROM <18 hours and no labor, (4) CB with ROM <18 hours and any labor, (5) vaginal birth (VB) with ROM <18 hours, and (6) any delivery mode with ROM ≥18 hours (Supplemental Table 1) (19). We categorized all infants born in the setting of prolonged ROM (defined as duration ≥18 hours) in a single phenotype, given existing data suggesting that these infants have elevated EOS risk regardless of delivery mode or presence of labor(1,3,17). As a secondary exposure, we identified the primary clinical indication prompting preterm birth which we defined by leveraging obstetric data defining spontaneous versus medically-indicated preterm birth(20–25).
The primary outcome was EOS, defined as isolation of a bacterial or fungal pathogen from blood or cerebrospinal fluid cultures within the first 72 hours after birth. Common commensal organisms (e.g., coagulase-negative Staphylococcus, Corynebacterium, and Micrococcus) were not considered EOS-defining pathogens. Secondary outcomes included death occurring during days 0–3 after birth, and a composite outcome of EOS or death during days 0–3.
Data Collection
Maternal characteristics included age; race and ethnicity (self-reported on hospital admission); receipt of any prenatal care (defined as attendance at least one prenatal visit); presence of any hypertensive disorder during pregnancy (chronic or gestational hypertension, preeclampsia, or eclampsia, as defined by International Classification of Disease [ICD]-9/10 codes)(26); presence of any diabetes during pregnancy (type 1, type 2, or gestational diabetes, as defined by ICD-9/10 codes)(26); maternal group B Streptococcus colonization status at delivery; documented obstetric diagnosis of IAI at delivery (e.g., clinical chorioamnionitis); maternal medication administration (e.g. antenatal corticosteroids, intrapartum magnesium sulfate, antibiotics); and mode of delivery. Race and ethnicity are acknowledged as social and not biologic constructs, and included in this analysis due to evidence of variable rates of preterm birth, mode of delivery and EOS incidence by race/ethnicity(3,27). Data were also collected to address delivery characteristics (as defined in Supplemental Table 1). Infant characteristics included gestational age at birth (GA), birth weight, and sex; use of surfactant therapy, administration of vasopressor medications and maximum level of respiratory support during days 0–3 after birth; final disposition (home, transfer to another medical facility, died); and total NICU length of stay. For the purpose of assessing outcome timing after birth, “day 0” was defined as the calendar day of birth.
Statistical Analysis
We summarized demographic and clinical data within the overall cohort using standard descriptive statistics. We used Pearson’s chi-squared test and the Wilcoxon rank-sum test to compare infants who did and did not receive EOS evaluations, and infants with and without EOS. We assessed the frequency of each delivery risk phenotype in the overall cohort, the incidence of EOS within each phenotype, and the frequency of EOS when combining delivery risk phenotypes with categorical indications for PTB. We calculated confidence intervals for these estimates based on the exact binomial distribution. We compared EOS incidence between phenotypes using the Kruskal-Wallis test. Because providers may initiate EOS evaluations more frequently in some birth scenarios compared to others, we performed a sensitivity analysis of EOS incidence restricted to only those infants who received an EOS evaluation (defined as any blood culture sent <72 hours after birth). Two-sided p-values <0.05 were considered statistically significant. All analyses were performed using Stata version 17 (College Station, TX, USA).
RESULTS
Study Population Characteristics
During the 2017–2021 study period, there were 2,937 infants born <35 weeks GA eligible for analysis (Figure 1). Clinical characteristics of the study cohort are summarized in Table 1. Median GA was 33 weeks (interquartile range [IQR] 30, 34) and median birth weight was 1,795 grams (IQR 1,325, 2,170).
Figure 1:
Flow diagram of cohort construction.
Table 1:
Cohort Demographic and Clinical Characteristics, by EOS Evaluation Status
| Overall n = 2,937 |
EOS Evaluation Performed n = 1,907 |
No EOS Evaluation n = 1,030 |
p-value | |
|---|---|---|---|---|
| n (%) or median (interquartile range) | ||||
| Maternal Characteristics | ||||
| Maternal Age (years) | 31 (27 – 35) | 31 (27 – 35) | 32 (27 – 35) | <0.001 |
| Maternal Race | 0.70 | |||
| White | 1,019 (35%) | 666 (35%) | 353 (34%) | |
| Black | 1,327 (45%) | 845 (44%) | 482 (47%) | |
| Asian | 181 (6%) | 122 (6%) | 59 (6%) | |
| Multi-Racial | 361 (12%) | 242 (13%) | 119 (12%) | |
| Unknown | 49 (2%) | 32 (2%) | 17 (2%) | |
| Maternal Hispanic/Latina Ethnicity | 270 (9%) | 179 (10%) | 91 (9%) | 0.59 |
| Any prenatal care | 2,837 (97%) | 1,827 (96%) | 1,010 (98%) | 0.001 |
| Hypertensive disorder during pregnancy | 1,309 (45%) | 614 (32%) | 695 (68%) | <0.001 |
| Any diabetes during pregnancy | 537 (18%) | 319 (17%) | 218 (21%) | 0.003 |
| Perinatal characteristics | ||||
| Multiple Gestation | 809 (28%) | 500 (26%) | 309 (30%) | 0.03 |
| Delivery via Cesarean section | 1,765 (60%) | 977 (51%) | 788 (77%) | <0.001 |
| Labor Onset Mechanism | <0.001 | |||
| None | 1,033 (35%) | 385 (20%) | 648 (63%) | |
| Induced | 659 (22%) | 370 (19%) | 289 (28%) | |
| Spontaneous | 1,245 (42%) | 1,152 (60%) | 93 (9%) | |
| Any induction of labor1 | 801 (27%) | 498 (26%) | 303 (29%) | 0.06 |
| Membrane Rupture Mechanism | <0.001 | |||
| None | 1,224 (42%) | 539 (28%) | 685 (67%) | |
| Artificial | 560 (19%) | 295 (16%) | 265 (26%) | |
| Spontaneous | 1,153 (39%) | 1,073 (56%) | 80 (8%) | |
| Duration of ROM (h) | 0.1 (0 – 9.9) | 2.4 (0 – 21.9) | 0 (0 – 1.5) | <0.001 |
| Prolonged ROM (≥18 hours) | 548 (19%) | 522 (27%) | 26 (3%) | <0.001 |
| Maternal receipt of magnesium | 1,465 (50%) | 890 (47%) | 575 (56%) | <0.001 |
| Maternal receipt of antenatal corticosteroids2 | 2,699 (92%) | 1,721 (90%) | 978 (95%) | <0.001 |
| Maternal receipt of intrapartum antibiotics | 1,422 (48%) | 1,166 (61%) | 256 (25%) | <0.001 |
| Maternal GBS status at delivery | 0.50 | |||
| Negative | 617 (21%) | 394 (21%) | 223 (22%) | |
| Unknown | 1,942 (66%) | 1,275 (67%) | 667 (65%) | |
| Positive | 378 (13%) | 238 (13%) | 140 (14%) | |
| Highest maternal intrapartum temperature (°F) | 98.7 (98.3, 99.1) | 98.7 (98.3, 99.1) | 98.7 (98.4, 99.0) | <0.001 |
| Obstetric diagnosis of intra-amniotic infection | 122 (4%) | 120 (6%) | 2 (0.2%) | <0.001 |
| Infant Characteristics | ||||
| Gestational age (completed weeks) | 33 (30 – 34) | 32 (29 – 34) | 33 (32 – 34) | <0.001 |
| Birth weight (grams) | 1795 (1325 – 2170) | 1760 (1210 – 2170) | 1836 (1450 – 2175) | <0.001 |
| Male Sex | 1,518 (52%) | 1,001 (53%) | 517 (50%) | 0.23 |
| 5-minute Apgar score | 9 (8 – 9) | 8 (7 – 9) | 9 (8 – 9) | <0.001 |
| Maximum Respiratory Support in first 72 hours3 | <0.001 | |||
| None | 997 (34%) | 530 (28%) | 467 (45%) | |
| Non-invasive ventilation | 1,295 (44%) | 840 (44%) | 455 (44%) | |
| Invasive ventilation | 645 (22%) | 537 (28%) | 108 (11%) | |
| Surfactant receipt in first 72 hours | 713 (24%) | 575 (30%) | 138 (13%) | <0.001 |
| Vasopressor receipt in first 72 hours4 | 145 (5%) | 138 (7%) | 7 (1%) | <0.001 |
| Death on day 0–3 after birth | 43 (2%) | 24 (1%) | 1 (0.1%) | <0.001 |
| Death prior to NICU discharge | 54 (2%) | 52 (3%) | 2 (0.2%) | <0.001 |
EOS, early-onset sepsis; ROM, rupture of membranes.
Includes use of pharmacologic or mechanical cervical ripening, use of oxytocin, and/or amniotomy
Administration of betamethasone or dexamethasone.
Non-invasive ventilation defined as receipt of >2 liters high flow nasal cannula, continuous positive airway pressure, or bi-level positive airway pressure. Invasive ventilation defined as mechanical ventilation, regardless of ventilator type.
Dopamine, epinephrine, norepinephrine, phenylephrine, and/or vasopressin
EOS evaluation was performed for 1,907 infants (65%) within the first 72 hours after birth. Infants born at lower gestation were significantly more likely to undergo EOS evaluation, occurring among 97% of infants born between 22–25 weeks GA, 80% of those born 26–28 weeks GA, 72% of those born 29–31 weeks GA, and 57% of those born 32–34 weeks GA (p<0.001). Clinical characteristics of infants who did and did not receive EOS evaluations are presented in Table 1. Infants evaluated for EOS had lower birth weight; more frequent birth in the setting of vaginal delivery, spontaneous onset of labor and/or ROM; and higher initial illness severity (receipt of mechanical ventilation, surfactant, and vasopressors during days 0–3), compared to non-evaluated infants. Among the AAP-endorsed low-risk delivery phenotype (CB, no labor, and ROM at delivery), 249/860 infants (29%) underwent EOS evaluation.
Distribution of Delivery Risk Phenotypes and Preterm Birth Indications
The most commonly-occurring delivery risk phenotypes were: (1) CB without labor and ROM at delivery (860 infants, 29%), and (2) VB with ROM <18 hours (851 infants, 29%) (Figure 1, Table 2). There were 547 infants (19%) born in the setting of prolonged ROM ≥18 hours, regardless of the delivery mode or presence of labor. Within the prolonged ROM phenotype, 59% (321/547) infants were born via VB, and 84% (462/547) were born in the setting of any labor.
Table 2:
EOS Incidence by Delivery Risk Phenotype and Indication for Preterm Birth
| Number of Infants, n (%) | Number of EOS Cases | EOS Case Frequency (%) | EOS Incidence per 1000 preterm infants (95% CI) | |
|---|---|---|---|---|
| Delivery Risk Phenotypes | ||||
| Prolonged ROM ≥18h (any delivery mode) | 547 (19%) | 13 | 2.4 | 23.8 (12.7, 40.3) |
| CB - ROM <18h - any labor | 248 (8%) | 3 | 1.2 | 12.1 (2.5, 34.9) |
| VB - ROM <18h | 851 (29%) | 5 | 0.6 | 5.9 (1.9, 13.7) |
| CB - ROM <18h - no labor | 87 (3%) | 0 | 0.0 | 0 (0, 41.5) * |
| CB - ROM at delivery - any labor | 344 (12%) | 0 | 0.0 | 0 (0, 10.7) * |
| CB - ROM at delivery - no labor | 860 (29%) | 0 | 0.0 | 0 (0, 4.3) * |
| Total | 2,937 (100%) | 21 | 0.7 | 7.2 (4.4, 10.9) |
| Indications for Preterm Birth | ||||
| Premature ROM | 747 (25%) | 15 | 2.0 | 20.1 (11.3, 32.9) |
| Preterm Labor | 786 (27%) | 3 | 0.4 | 3.8 (0.8, 11.1) |
| Maternal Indication | 938 (32%) | 2 | 0.2 | 2.1 (0.3, 7.7) |
| Fetal Indication | 280 (10%) | 1 | 0.4 | 3.6 (0.1, 19.7) |
| Other Indication | 186 (6%) | 0 | 0.0 | 0 (0, 19.6) * |
| Total | 2,937 (100%) | 21 | 0.7 | 7.2 (4.4, 10.9) |
CB, Cesarean birth; VB, vaginal birth; ROM, rupture of membranes.
One-sided, 97.5% confidence interval
Maternal health conditions were the most frequently-occurring primary indication for PTB, accounting for 32% (938/2,937) of births. Preterm labor and premature ROM were the next most frequent PTB indications, occurring in 27% and 25% of preterm births, respectively. Among infants born via CB without labor and ROM at delivery, the majority (556/860, 65%) were born for a maternal indication. Among infants born via VB with ROM <18 hours, the most common PTB indication was preterm labor (415/851, 49%), followed by maternal indications (205/851, 24%) (Supplemental Table 2).
EOS Incidence by Delivery Risk Phenotype
Over the five-year study period, 21 infants developed EOS (0.7%, or 7.2 EOS cases per 1000 preterm infants born <35 weeks GA [95% CI 4.4, 10.9]) (Table 2, Figure 2). EOS occurred among 13 of 504 infants born 22–28 weeks GA (2.6%, or 26 EOS cases per 1000 infants born 22–28 weeks GA). Among infants born 22–25 weeks GA, 12/205 developed EOS (6%, or 59 EOS cases per 1000 infants born 22–25 weeks GA). Among all preterm infants born <35 weeks GA, there were 25 deaths during days 0–3, most of which (22/25, 88%) occurred among infants without EOS. A composite outcome of EOS or death during days 0–3 occurred in 43/2,937 infants (1.5%), with an incidence of 14.6 deaths or EOS cases per 1000 preterm infants born <35 weeks GA (95% CI 10.6, 19.7).
Figure 2:
Heatmap displaying distribution of PTB phenotypes and EOS cases by gestational age and birth weight. Delivery phenotypes are denoted by shaded hexagons, with individual EOS cases denoted by open circles.
The 21 EOS cases were differentially distributed across the six delivery risk phenotypes (p <0.001, Table 2). The majority of cases (13/21, 62%) occurred in the setting of prolonged ROM ≥18 hours, with a phenotype-specific EOS incidence of 23.8 cases per 1000 infants. Five EOS cases (24%) occurred in infants born via VB with ROM <18 hours, with a phenotype EOS incidence of 5.9 cases per 1000 infants. The remaining three EOS cases occurred in infants born via CB with ROM <18h and any labor, with a phenotype incidence of 12.1 EOS cases per 1000 preterm infants. There were no EOS cases among infants born by CB with ROM at delivery (with or without labor), nor among infants born via CB with ROM <18 hours in the absence of labor. Analyses restricted only to those infants receiving EOS evaluations resulted in similar risk incidence estimates (Supplemental Table 3). Among the three delivery risk phenotypes with identified EOS cases, the number of infants needed to treat (NNT) with empiric antibiotics in order to capture one case of EOS were 42 (if born in setting of prolonged ROM ≥18h), 83 (if born via CB with ROM <18 hours and any labor), and 170 infants (if born via VB with ROM <18h).
Detailed characteristics of the 21 infants with EOS are presented in Table 3 and Supplemental Table 4. The primary underlying PTB indication among infants with EOS was premature ROM (15/21 infants, 71%) (Table 3). Preterm labor was the primary PTB indication in three infants (14%). Two infants with EOS were born in the setting of a maternal indication (preeclampsia) after induced labor: one was born via VB with ROM immediately prior to delivery, and the other was born via CB after spontaneous ROM occurring 3 hours prior to delivery. The final infant was born for a fetal indication (intrauterine growth restriction) after induced labor and artificial ROM occurring 19 hours prior to delivery via CB.
Table 3:
Characteristics of 21 Preterm Infants with EOS
| Infant | GA (weeks) | Birth weight (g) | EOS organism | Preterm Birth Indication | Delivery Mode | Labor Type | ROM type | ROM duration (hours) | Intra-amniotic Infection |
|---|---|---|---|---|---|---|---|---|---|
| 1 | 22 | 465 | E. coli | PROM | VB | Induced | Spontaneous | 56 | Yes |
| 2 | 22 | 581 | GBS | PROM | VB | Induced | Spontaneous | 15 | Yes |
| 3 | 23 | 415 | Klebsiella | Maternal | VB | Induced | None | 0 | No |
| 4 | 23 | 490 | Klebsiella | PROM | VB | Spontaneous | Spontaneous | 216 | No |
| 5 | 24 | 545 | E. coli | PROM | CB | None | Spontaneous | 162 | Yes |
| 6 | 24 | 547 | H. influenzae | PTL | VB | Spontaneous | Spontaneous | 5 | No |
| 7 | 24 | 615 | H. influenzae | PROM | CB | None | Spontaneous | 149 | Yes |
| 8 | 24 | 627 | E. coli | PROM | CB | Spontaneous | Spontaneous | 94 | No |
| 9 | 24 | 655 | E. coli | PTL | VB | Spontaneous | Spontaneous | 1 | Yes |
| 10 | 24 | 696 | GBS | PROM | CB | Induced | Spontaneous | 11 | Yes |
| 11 | 24 | 785 | Prevotella | PROM | CB | Spontaneous | Spontaneous | Unknown, presumed >18 hours | Yes |
| 12 | 25 | 710 | H. influenzae | PROM | VB | Spontaneous | Spontaneous | 93 | No |
| 13 | 28 | 1370 | E. coli | PROM | CB | None | Spontaneous | 551 | Yes |
| 14 | 30 | 1366 | GBS | Maternal | CB | Induced | Spontaneous | 3 | No |
| 15 | 30 | 1431 | E. coli | PROM | CB | Spontaneous | Spontaneous | 49 | No |
| 16 | 31 | 1758 | Alpha-hemolytic Streptococcus | PROM | CB | None | Spontaneous | 372 | Yes |
| 17 | 31 | 2129 | E. coli | PTL | VB | Spontaneous | Spontaneous | 0 | No |
| 18 | 33 | 1100 | Morganella | Fetal | CB | Induced | Artificial | 19 | Yes |
| 19 | 33 | 1490 | E. coli | PROM | VB | Spontaneous | Spontaneous | 184 | No |
| 20 | 33 | 1960 | E. coli | PROM | CB | Spontaneous | Spontaneous | 1 | No |
| 21 | 34 | 2580 | E. coli | PROM | VB | Induced | Spontaneous | 100 | No |
CB, Cesarean birth; GA, gestational age; GBS, Group B Streptococcus; PROM, premature rupture of membranes; PTL, preterm labor; ROM, rupture of membranes; VB, vaginal birth
While IAI was not a primary indication for preterm delivery, we identified that 10 of 122 preterm infants born in the setting of IAI (8%) developed EOS (Supplemental Table 4). Among 71 infants born in the setting of IAI who were categorized in the prolonged ROM delivery risk phenotype, 7 infants (10%) developed EOS. Among these subgroups, the NNT to capture one case of EOS were 12 and 10, respectively. In contrast, among all infants born in the setting of prolonged ROM without concern for IAI, 1% (6/470) developed EOS (NNT 78).
DISCUSSION
Within this contemporary preterm cohort, we sought to determine the incidence of EOS among infants born in the setting of specific delivery risk phenotypes. In doing so, we hoped to provide clinically-relevant risk discrimination among infants traditionally viewed at uniformly high risk for infection at birth. The majority of EOS cases occurred among infants born in the setting of prolonged ROM ≥18h. There were no EOS cases identified within three phenotypes, including one previously identified as a “low-risk” by the AAP. These findings suggest that differential EOS risk exists across the spectrum of delivery phenotypes. Recognition of differential risk estimates may expand opportunities for EOS risk stratification in preterm infants.
Numerous efforts have been made to categorize preterm birth, though no schema has yet gained broad traction for use in estimating associations with neonatal outcomes. Prior efforts have most frequently categorized PTB into spontaneous (e.g., preterm labor or premature ROM) versus medically-indicated subtypes (due either to maternal or fetal health indications) (20–25,28–30). Spontaneous PTB, compared to medically-indicated PTB, is associated with higher neonatal morbidity and mortality (21,28). However, this approach to delivery phenotyping likely cannot effectively distinguish EOS risk among infants with disparate birth mechanisms. Other proposed obstetric-focused phenotyping approaches include significantly more detail regarding putative maternal and fetal factors associated with preterm delivery (31,32), but these phenotypes are focused toward maternal care and may not be suitable for neonatal risk stratification. For example, one proposed obstetric phenotyping approach combines significant maternal, fetal, or placental conditions preceding delivery; signs of labor onset; and the pathway to delivery (e.g., caregiver-initiated versus spontaneous). This specifically excludes delivery mode (considered a management decision potentially driven by delivering clinician preference), though this mechanistic information may alter neonatal management considerations related to EOS risk stratification (33).
The delivery risk phenotypes utilized in this study encompass comprehensive mechanistic pathways to preterm delivery, composed of perinatal factors that are relevant to EOS pathogenesis. These phenotypes are constructed from basic objective components (mode of delivery, presence of labor, and duration of ROM prior to delivery), which increases generalizability and the feasibility of delivery risk phenotype identification at birth. Phenotype refinement and validation are needed; while these delivery risk phenotypes could be further partitioned to increase precision, this may further limit power to calculate phenotype-specific EOS incidence.
Similar delivery mechanism-based approaches to PTB phenotyping are reported, largely in the context of identifying infants at low EOS risk (1,10,14–16). On the basis of very low EOS rates among preterm infants born in the setting of a “low-risk” delivery phenotype (obstetric indication for PTB, via CB without labor, and ROM at delivery) the AAP now states that these infants may be observed clinically without mandated empiric antibiotic administration at birth (19). The low-risk delivery phenotype accounted for 30% of our study cohort, within the reported 15–37% prevalence in other preterm cohorts (10,14,16), representing a substantial preterm subpopulation for whom antibiotic stewardship is indicated. Complex maternal-fetal factors impact an individual infant’s ultimate delivery mechanism; we do not endorse any delivery phenotype for the sole purpose of lowering EOS risk. Rather, we highlight that identification of the delivery mechanism and its associated EOS risk may inform clinical decision-making regarding empiric antibiotic utilization at birth.
We also evaluated delivery risk phenotype distribution and EOS incidence according to the underlying PTB indication, given that infants born with a specific preterm birth indication may have disparate delivery mechanisms (and vice versa). While PTB indications commonly clustered within certain delivery risk phenotypes, this overlap was not uniform. As such, risk stratification based solely on PTB indication likely cannot accurately reflect the mechanistic underpinnings of PTB contributing to EOS pathogenesis. However, consideration of the PTB indication in conjunction with the delivery risk phenotype may offer additive granularity to preterm EOS risk stratification. Prior studies of PTB indication-associated EOS risk in preterm infants have identified that EOS occurs more frequently in the setting of premature ROM (17,34–36) and preterm labor(17), and less frequently in the setting of maternal hypertensive disorders (37,38) - findings reflected in our cohort. However, three EOS cases occurred in infants with a maternal PTB indication (preeclampsia, in two cases), or a fetal indication (intrauterine growth restriction, in one case) – highlighting the limitations of risk stratification on PTB indication without consideration of the mechanistic delivery phenotype.
Although not a primary PTB indication, delivery in the setting of IAI was associated with an 8% EOS rate (increasing to 10% if prolonged ROM also present) and very low NNT to identify one case of EOS. The associations of suspected IAI and prolonged ROM with EOS risk are described (3,39,40), and preterm birth in the setting of both conditions yielded the highest EOS delivery risk profile in this cohort. A prior study of infants born 30–34 weeks gestation concluded that prolonged ROM was not associated with higher EOS risk in the absence of IAI (35). However, our study suggests that prolonged ROM remains a substantial EOS risk factor: 1/3 of our identified EOS cases occurred in the setting of prolonged ROM without IAI, and the EOS rate among all preterm infants with isolated prolonged ROM without IAI was 1%.
This exploratory study identified delivery phenotype-specific EOS incidence within a contemporary cohort of almost 3,000 preterm infants. Furthermore, management across our hospital system largely reflected the AAP’s most recent preterm EOS management recommendations, with empiric antibiotic administration deferred among 71% of infants born in the setting of the “low-risk” delivery phenotype (19). The delivery mechanism and PTB indication were confirmed via direct chart review, enhancing the granularity of the dataset while minimizing data missingness and reducing potential misclassification bias.
However, we also acknowledge limitations. EOS incidence among preterm infants ≤28 weeks gestation in our cohort (26 cases/1000 births) is comparable to national analyses of contemporary EOS epidemiology(1,2). Nevertheless, a low number of total EOS cases (21) were identified over the five-year study period. We were underpowered to estimate population-level EOS incidence within each phenotype, and further underpowered to make more granular assessments of EOS risk incorporating both the delivery phenotype and the indication for PTB. Further assessment in larger preterm infant cohorts is required to validate delivery phenotype-specific EOS risk. Implementation of these delivery risk phenotypes in clinical practice would further require decision analyses to identify the optimal EOS prior probability threshold to prompt empiric antibiotic administration. Our study excluded infants who did not receive delivery room resuscitation, which was most relevant to infants born at periviable gestation (22–23 weeks GA). It is possible that some of these infants would have developed EOS and/or died within the first three days after birth had resuscitation been attempted, and our estimates of EOS/death incidence may thus be underestimated. There is a potential risk of ascertainment bias in EOS identification, given that EOS evaluations were initiated more frequently within some delivery phenotypes compared to others. However, a sensitivity analysis demonstrated similar EOS incidence within the overall cohort and when restricted only to infants receiving EOS evaluations. Neonatal clinicians tend to liberally initiate EOS evaluations in the setting of clinical instability, making it unlikely that occult EOS cases were undetected among infants not evaluated for EOS.
The primary objectives of EOS risk stratification are to identify infants at highest EOS risk to facilitate prompt empiric antimicrobial administration, while simultaneously identifying infants at lowest risk for whom antimicrobial exposure may be avoided in the appropriate clinical scenario. Alternatively stated, the primary goal is not antibiotic non-initiation, but rather targeted identification of stewardship opportunities among infants for whom the prior probability of EOS is so low that the potential risks of empiric administration outweigh the benefits. Differential risk estimates can inform research into both EOS prevention and empiric treatment.
CONCLUSION
Within a cohort of preterm infants, we identified differential EOS incidence across a spectrum of six comprehensive, mechanism-based delivery risk phenotypes. Incorporation of delivery phenotyping may inform EOS risk stratification in preterm infants, and ultimately expand opportunities for antimicrobial stewardship and optimization among infants deemed at lower EOS risk based on their PTB indication and mechanism of delivery.
Supplementary Material
Conflicts of Interest/Funding Sources:
Dr. Coggins reports receiving research funding from the National Heart, Lung and Blood Institute of the National Institutes of Health (T32HL007891). Dr. Triebwasser receives research funding from the Michigan Department of Health and Human Services. Dr. Downes is supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (K23HD091365). Dr. Christie is supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health (K24HL115354). Dr. Puopolo reports receiving research funding from the National Institutes of Health, from two contracts with the Centers for Disease Control and Prevention, and from the Children’s Hospital of Philadelphia. None of the authors have conflicts of interest to declare relevant to this study.
Abbreviations:
- AAP
American Academy of Pediatrics
- CB
Cesarean section birth
- EOS
early-onset sepsis
- GA
gestational age
- IAI
intra-amniotic infection
- NICU
neonatal intensive care unit
- NNT
number needed to treat
- PTB
preterm birth
- ROM
rupture of membranes
- VB
vaginal birth
DATA AVAILABILITY:
A de-identified dataset may be made available upon reasonable request.
REFERENCES
- 1.Stoll BJ, Puopolo KM, Hansen NI, Sánchez PJ, Bell EF, Carlo WA, et al. Early-Onset Neonatal Sepsis 2015 to 2017, the Rise of Escherichia coli, and the Need for Novel Prevention Strategies. JAMA Pediatr. 2020;174(7):e200593–e200593. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Flannery DD, Edwards EM, Puopolo KM, Horbar JD. Early-onset sepsis among very preterm infants. Pediatrics. 2021;148(4):2021052456. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Schrag SJ, Farley MM, Petit S, Reingold A, Weston EJ, Pondo T, et al. Epidemiology of Invasive Early-Onset Neonatal Sepsis, 2005 to 2014. Pediatrics. 2016;138(6):e20162013. [DOI] [PubMed] [Google Scholar]
- 4.Flannery DD, Ross RK, Mukhopadhyay S, Tribble AC, Puopolo KM, Gerber JS. Temporal Trends and Center Variation in Early Antibiotic Use Among Premature Infants. JAMA Netw Open. 2018;1(1):e180164. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Mukhopadhyay S, Bryan M, Dhudasia MB, Quarshie W, Gerber JS, Grundmeier RW, et al. Intrapartum group B Streptococcal prophylaxis and childhood weight gain. Arch Dis Child - Fetal Neonatal Ed. 2021;106(6):649–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Koebnick C, Sidell MA, Getahun D, Tartof SY, Rozema E, Taylor B, et al. Intrapartum Antibiotic Exposure and Body Mass Index in Children. Clin Infect Dis. 2021;73(4):e938–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kuppala VS, Meinzen-Derr J, Morrow AL, Schibler KR. Prolonged initial empirical antibiotic treatment is associated with adverse outcomes in premature infants. J Pediatr. 2011;159(5):720–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Cantey JB, Huffman LW, Subramanian A, Marshall AS, Ballard AR, Lefevre C, et al. Antibiotic Exposure and Risk for Death or Bronchopulmonary Dysplasia in Very Low Birth Weight Infants. J Pediatr. 2017;181:289–293.e1. [DOI] [PubMed] [Google Scholar]
- 9.Ting JY, Roberts A, Sherlock R, Ojah C, Cieslak Z, Dunn M, et al. Duration of Initial Empirical Antibiotic Therapy and Outcomes in Very Low Birth Weight Infants. Pediatrics. 2019;143(3):e20182286. [DOI] [PubMed] [Google Scholar]
- 10.Letouzey M, Lorthe E, Marchand-Martin L, Kayem G, Charlier C, Butin M, et al. Early Antibiotic Exposure and Adverse Outcomes in Very Preterm Infants at Low Risk of Early-Onset Sepsis: The EPIPAGE-2 Cohort Study. J Pediatr. 2022;243:91–98.e4. [DOI] [PubMed] [Google Scholar]
- 11.Cotten CM, Taylor S, Stoll B, Goldberg RN, Hansen NI, Sanchez PJ, et al. Prolonged Duration of Initial Empirical Antibiotic Treatment Is Associated With Increased Rates of Necrotizing Enterocolitis and Death for Extremely Low Birth Weight Infants. Pediatrics. 2009;123(1):58–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Cantey JB, Pyle AK, Wozniak PS, Hynan LS, Sánchez PJ. Early Antibiotic Exposure and Adverse Outcomes in Preterm, Very Low Birth Weight Infants. J Pediatr. 2018. Dec 1;203:62–7. [DOI] [PubMed] [Google Scholar]
- 13.Kuzniewicz MW, Puopolo KM, Fischer A, Walsh EM, Li S, Newman TB, et al. A Quantitative, Risk-Based Approach to the Management of Neonatal Early-Onset Sepsis. JAMA Pediatr. 2017;171(4):365–71. [DOI] [PubMed] [Google Scholar]
- 14.Flannery DD, Mukhopadhyay S, Morales KH, Dhudasia MB, Passarella M, Gerber JS, et al. Delivery Characteristics and the Risk of Early-Onset Neonatal Sepsis. Pediatrics. 2022;149(2):e2021052900. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Garber SJ, Dhudasia MB, Flannery DD, Passarella MR, Puopolo KM, Mukhopadhyay S. Delivery-based criteria for empiric antibiotic administration among preterm infants. J Perinatol. 2021;41(2):255–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Puopolo KM, Mukhopadhyay S, Hansen NI, Cotten CM, Stoll BJ, Sanchez PJ, et al. Identification of Extremely Premature Infants at Low Risk for Early-Onset Sepsis. Pediatrics. 2017;140(5):e20170925. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Mukhopadhyay S, Puopolo KM. Clinical and Microbiologic Characteristics of Early-onset Sepsis Among Very Low Birth Weight Infants: Opportunities for Antibiotic Stewardship. Pediatr Infect Dis J. 2017;36(5):477–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Mukhopadhyay S, Puopolo KM. Risk assessment in neonatal early onset sepsis. Semin Perinatol. 2012;36(6):408–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Puopolo KM, Benitz WE, Zaoutis TE. Management of neonates born at ≤34 6/7 weeks’ gestation with suspected or proven early-onset bacterial sepsis. Pediatrics. 2018;142(6):e20182896. [DOI] [PubMed] [Google Scholar]
- 20.Manuck TA, Esplin MS, Biggio J, Bukowski R, Parry S, Zhang H, et al. The Phenotype of Spontaneous Preterm Birth: Application of a Clinical Phenotyping Tool. Am J Obstet Gynecol. 2015;212(4):487.e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Stout MJ, Demaree D, Merfeld E, Methodius Ã, Tuuli G, Wambach JA, et al. Neonatal Outcomes Differ after Spontaneous and Indicated Preterm Birth. Am J Perinatol. 2018;35:494–502. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Henderson JJ, McWilliam OA, Newnham JP, Pennell CE. Preterm birth aetiology 2004–2008. Maternal factors associated with three phenotypes: spontaneous preterm labour, preterm pre-labour rupture of membranes and medically indicated preterm birth. J Matern Neonatal Med. 2012;25(6):642–7. [DOI] [PubMed] [Google Scholar]
- 23.Stout MJ, Busam MR, Macones GA, Tuuli MG. Spontaneous and Indicated Preterm Birth Subtypes: Inter-observer Agreement and Accuracy of Classification. Am J Obstet Gynecol. 2014;211(5):530.e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Goldenberg RL, Culhane JF, Iams JD, Romero R. Epidemiology and causes of preterm birth. Lancet. 2008;371(9606):75–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Gimenez LG, Krupitzki HB, Momany AM, Gili JA, Poletta FA, Campaña H, et al. Maternal and neonatal epidemiological features in clinical subtypes of preterm-birth. J Matern Fetal Neonatal Med. 2016;29(19):3153. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Flannery DD, Gouma S, Dhudasia MB, Mukhopadhyay S, Pfeifer MR, Woodford EC, et al. SARS-CoV-2 seroprevalence among parturient women in Philadelphia. Sci Immunol. 2020;5(49):eabd5709. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Osterman MJK, Hamilton BE, Martin JA, Driscoll AK, Valenzuela CP. Births: Final Data for 2021. Natl Vital Stat Reports. 2023;72(1):15–44. [PubMed] [Google Scholar]
- 28.Kamath-Rayne BD, DeFranco EA, Chung E, Chen A. Subtypes of Preterm Birth and the Risk of Postneonatal Death. J Pediatr. 2013;162(1):28–34.e2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Lynch AM, Wagner BD, Hodges JK, Thevarajah TS, McCourt EA, Cerda AM, et al. The relationship of the subtypes of preterm birth with retinopathy of prematurity. Am J Obstet Gynecol. 2017;217(3):354.e1–354.e8. [DOI] [PubMed] [Google Scholar]
- 30.Kimberlin DF, Hauth JC, Owen J, Bottoms SF, Iams JD, Mercer BM, et al. Indicated versus spontaneous preterm delivery: An evaluation of neonatal morbidity among infants weighing </=1000 grams at birth. Am J Obstet Gynecol. 1999;180(3 Pt 1):683–9. [DOI] [PubMed] [Google Scholar]
- 31.Villar J, Papageorghiou AT, Knight HE, Gravett MG, Iams J, Waller SA, et al. The preterm birth syndrome: A prototype phenotypic classification. Am J Obstet Gynecol. 2012;206(2):119–23. [DOI] [PubMed] [Google Scholar]
- 32.Kramer MS, Papageorghiou A, Culhane J, Bhutta Z, Goldenberg RL, Gravett M, et al. Challenges in defining and classifying the preterm birth syndrome. Am J Obstet Gynecol. 2012;206(2):108–12. [DOI] [PubMed] [Google Scholar]
- 33.Mukhopadhyay S, Underwood MA. Phenotyping preterm infants at birth to predict infection risk. Pediatr Res. 2021;90(3):508–9. [DOI] [PubMed] [Google Scholar]
- 34.Klinger G, Levy I, Sirota L, Boyko V, Reichman B, Lerner-Geva L. Epidemiology and risk factors for early onset sepsis among very-low-birthweight infants. Am J Obstet Gynecol. 2009;201(1):38.e1–38.e6. [DOI] [PubMed] [Google Scholar]
- 35.Ofman G, Vasco N, Cantey JB. Risk of Early-Onset Sepsis following Preterm, Prolonged Rupture of Membranes with or without Chorioamnionitis. Am J Perinatol. 2016;33:339–42. [DOI] [PubMed] [Google Scholar]
- 36.Pinto S, Malheiro MF, Vaz A, Rodrigues T, Montenegro N, Guimarães H. Neonatal outcome in preterm deliveries before 34-week gestation – the influence of the mechanism of labor onset. J Matern Neonatal Med. 2019;32(21):3655–61. [DOI] [PubMed] [Google Scholar]
- 37.Gagliardi L, Rusconi F, Bellù R, Zanini R. Association of Maternal Hypertension and Chorioamnionitis With Preterm Outcomes. Pediatrics. 2014;134(1):e154–61. [DOI] [PubMed] [Google Scholar]
- 38.Lu C-Q, Lin J, Yuan L, Zhou J-G, Liang K, Zhong Q-H, et al. Pregnancy induced hypertension and outcomes in early and moderate preterm infants. Pregnancy Hypertens. 2018;14:68–71. [DOI] [PubMed] [Google Scholar]
- 39.Giannoni E, Agyeman PKA, Stocker M, Posfay-Barbe KM, Heininger U, Spycher BD, et al. Neonatal Sepsis of Early Onset, and Hospital-Acquired and Community-Acquired Late Onset: A Prospective Population-Based Cohort Study. J Pediatr. 2018;201:106–114.e4. [DOI] [PubMed] [Google Scholar]
- 40.Wortham JM, Hansen NI, Schrag SJ, Hale E, Van Meurs K, Sánchez PJ, et al. Chorioamnionitis and Culture-Confirmed, Early-Onset Neonatal Infections. Pediatrics. 2016;137(1):e20152323. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
A de-identified dataset may be made available upon reasonable request.