OBJECTIVE: To define a quantitative stratification algorithm for the risk of early-onset sepsis (EOS) in newborns ≥ 34 weeks' gestation.

METHODS: We conducted a retrospective nested case-control study that used split validation. Data collected on each infant included sepsis risk at birth based on objective maternal factors, demographics, specific clinical milestones, and vital signs during the first 24 hours after birth. Using a combination of recursive partitioning and logistic regression, we developed a risk classification scheme for EOS on the derivation dataset. This scheme was then applied to the validation dataset.

RESULTS: Using a base population of 608,014 live births ≥ 34 weeks' gestation at 14 hospitals between 1993 and 2007, we identified all 350 EOS cases <72 hours of age and frequency matched them by hospital and year of birth to 1063 controls. Using maternal and neonatal data, we defined a risk stratification scheme that divided the neonatal population into 3 groups: treat empirically (4.1% of all live births, 60.8% of all EOS cases, sepsis incidence of 8.4/1000 live births), observe and evaluate (11.1% of births, 23.4% of cases, 1.2/1000), and continued observation (84.8% of births, 15.7% of cases, incidence 0.11/1000).

CONCLUSIONS: It is possible to combine objective maternal data with evolving objective neonatal clinical findings to define more efficient strategies for the evaluation and treatment of EOS in term and late preterm infants. Judicious application of our scheme could result in decreased antibiotic treatment in 80,000 to 240,000 US newborns each year.

<p><em><strong>Background: </strong></em>A comorbidity summary score may support early and systematic identification of women at high risk for adverse obstetric outcomes. The objective of this study was to conduct the initial development and validation of an obstetrics comorbidity risk score for automated implementation in the electronic health record (EHR) for clinical use. <em><strong>Methods: </strong></em>The score was developed and validated using EHR data for a retrospective cohort of pregnancies with delivery between 2010 and 2018 at Kaiser Permanente Northern California, an integrated health care system. The outcome used for model development consisted of adverse obstetric events from delivery hospitalization (<em>e.g.</em>, eclampsia, hemorrhage, death). Candidate predictors included maternal age, parity, multiple gestation, and any maternal diagnoses assigned in health care encounters in the 12 months before admission for delivery. We used penalized regression for variable selection, logistic regression to fit the model, and internal validation for model evaluation. We also evaluated prenatal model performance at 18 weeks of pregnancy. <em><strong>Results:</strong></em> The development cohort ( = 227,405 pregnancies) had an outcome rate of 3.8% and the validation cohort ( = 41,683) had an outcome rate of 2.9%. Of 276 candidate predictors, 37 were included in the final model. The final model had a validation c-statistic of 0.72 (95% confidence interval [CI] 0.70-0.73). When evaluated at 18 weeks of pregnancy, discrimination was modestly diminished (c-statistic 0.68 [95% CI 0.67-0.70]). <em><strong>Conclusions:</strong></em> The obstetric comorbidity score demonstrated good discrimination for adverse obstetric outcomes. After additional appropriate validation, the score can be automated in the EHR to support early identification of high-risk women and assist efforts to ensure risk-appropriate maternal care.</p>

<p>In 2007, we began a National Institutes of Health–funded study of neonatal early-onset sepsis (EOS) whose goal was to develop multivariate predictive models that could be used by clinicians to evaluate a newborn’s risk of EOS. We approached this study with 3 objectives. First, we would evaluate the statistical association of individual, established risk factors for neonatal EOS with the outcome of culture-confirmed infection. Second, we would use routinely captured, objective data that could be found in an electronic medical record. Third, we would determine whether we could develop accurate multivariate predictive models without using the clinical diagnosis of chorioamnionitis. We took a Bayesian perspective that quantifies the value of incremental information explicitly. This approach begins with a prior probability (the incidence of EOS in the population as baseline risk). This prior probability is then modified as more information becomes available. We originally intended to modify this initial risk estimate using the likelihood ratios from 3 models: 1 model based on risk factors known at the moment of birth, 1 model based on the infant’s clinical condition, and 1 model based on the complete blood cell count. Performance of the complete blood cell count components was poor, so we ultimately decided not to include it in our final approach. We first published the model based on risk factors at birth, and later published the model based on the evolving clinical condition along with a proposed clinical management algorithm. The management algorithm was developed during a series of discussions with clinicians who practiced in the birth hospitals of Kaiser-Permanente Northern California (KPNC) that focused on the number-needed-to-treat associated with different levels of estimated risk. To facilitate clinical use of the models, we developed the web-based neonatal EOS calculator (<a href="https://neonatalsepsiscalculator.kaiserpermanente.org">https://neonatal…;). We did not recommend use of this calculator in a vacuum; its deployment at KPNC hospitals was part of a carefully designed implementation program that included reassessment of the calculator with more recent KPNC data. We performed a large prospective validation study at the KPNC birth hospitals and a smaller study in Philadelphia, Pennsylvania. Although these studies used slightly different clinical management algorithms, both demonstrated significant declines in the use of empirical antibiotics among infants born at 35 or more weeks’ gestation. Neither study identified any short-term safety issues; most important, the KPNC study demonstrated no change in the very low incidence (approximately 1 in 20 000 live births) of readmission to the hospital with EOS after initial discharge from the birth hospital. Recently, as reviewed by Achten <em>et al </em>in this issue of <em>JAMA Pediatrics</em>, several other investigators have performed retrospective medical record analyses or prospective implementation studies to address the clinical performance of the calculator models. Achten and colleagues concluded that the use of this quantitative approach to neonatal EOS risk is associated with decreased antibiotic use compared with prior approaches, without concern that the approach fails to identify significant numbers of newborns with EOS.</p>

<p><strong>Importance: </strong>Current algorithms for management of neonatal early-onset sepsis (EOS) result in medical intervention for large numbers of uninfected infants. We developed multivariable prediction models for estimating the risk of EOS among late preterm and term infants based on objective data available at birth and the newborn's clinical status.</p>

<p><strong>Objectives: </strong>To examine the effect of neonatal EOS risk prediction models on sepsis evaluations and antibiotic use and assess their safety in a large integrated health care system.</p>

<p><strong>Design, Setting, and Participants: </strong>The study cohort includes 204 485 infants born at 35 weeks' gestation or later at a Kaiser Permanente Northern California hospital from January 1, 2010, through December 31, 2015. The study compared 3 periods when EOS management was based on (1) national recommended guidelines (baseline period [January 1, 2010, through November 31, 2012]), (2) multivariable estimates of sepsis risk at birth (learning period [December 1, 2012, through June 30, 2014]), and (3) the multivariable risk estimate combined with the infant's clinical condition in the first 24 hours after birth (EOS calculator period [July 1, 2014, through December 31, 2015]).</p>

<p><strong>Main Outcomes and Measures: </strong>The primary outcome was antibiotic administration in the first 24 hours. Secondary outcomes included blood culture use, antibiotic administration between 24 and 72 hours, clinical outcomes, and readmissions for EOS.</p>

<p><strong>Results: </strong>The study cohort included 204 485 infants born at 35 weeks' gestation or later: 95 343 in the baseline period (mean [SD] age, 39.4 [1.3] weeks; 46 651 male [51.0%]; 37 007 white, non-Hispanic [38.8%]), 52 881 in the learning period (mean [SD] age, 39.3 [1.3] weeks; 27 067 male [51.2%]; 20 175 white, non-Hispanic [38.2%]), and 56 261 in the EOS calculator period (mean [SD] age, 39.4 [1.3] weeks; 28 575 male [50.8%]; 20 484 white, non-Hispanic [36.4%]). In a comparison of the baseline period with the EOS calculator period, blood culture use decreased from 14.5% to 4.9% (adjusted difference, -7.7%; 95% CI, -13.1% to -2.4%). Empirical antibiotic administration in the first 24 hours decreased from 5.0% to 2.6% (adjusted difference, -1.8; 95% CI, -2.4% to -1.3%). No increase in antibiotic use occurred between 24 and 72 hours after birth; use decreased from 0.5% to 0.4% (adjusted difference, 0.0%; 95% CI, -0.1% to 0.2%). The incidence of culture-confirmed EOS was similar during the 3 periods (0.3% in the baseline period, 0.3% in the learning period, and 0.2% in the EOS calculator period). Readmissions for EOS (within 7 days of birth) were rare in all periods (5.2 per 100 000 births in the baseline period, 1.9 per 100 000 births in the learning period, and 5.3 per 100 000 births in the EOS calculator period) and did not differ statistically (P = .70). Incidence of adverse clinical outcomes, including need for inotropes, mechanical ventilation, meningitis, and death, was unchanged after introduction of the EOS calculator.</p>

<p><strong>Conclusions and Relevance: </strong>Clinical care algorithms based on individual infant estimates of EOS risk derived from a multivariable risk prediction model reduced the proportion of newborns undergoing laboratory testing and receiving empirical antibiotic treatment without apparent adverse effects.</p>

<p><strong>PURPOSE OF REVIEW: </strong>Neonatal early-onset sepsis (EOS) is a very low-incidence, but potentially fatal condition among term and late preterm newborns. EOS algorithms based on risk-factor threshold values result in evaluation and empiric antibiotic treatment of large numbers of uninfected newborns, leading to unnecessary antibiotic exposures and maternal/infant separation. Ideally, risk stratification should be quantitative, employ information conserving strategies, and be readily transferable to modern comprehensive electronic medical records.</p>

<p><strong>RECENT FINDINGS: </strong>We performed a case-control study of infants born at or above 34 weeks' gestation with blood culture-proven EOS. We defined the relationship of established predictors to the risk of EOS, then used multivariate analyses and split validation to develop a predictive model using objective data. The model provides an estimation of sepsis risk that can identify the same proportion of EOS cases by evaluating fewer infants, as compared with algorithms based on subjective diagnoses and cut-off values for continuous predictors.</p>

<p><strong>SUMMARY: </strong>An alternative approach to EOS risk assessment based only on objective data could decrease the number of infants evaluated and empirically treated for EOS, compared with currently recommended algorithms. Prospective evaluation is needed to determine the accuracy and safety of using the sepsis risk model to guide clinical decision-making.</p>

<p><strong>OBJECTIVE: </strong>To develop a quantitative model to estimate the probability of neonatal early-onset bacterial infection on the basis of maternal intrapartum risk factors.</p>

<p><strong>METHODS: </strong>This was a nested case-control study of infants born at ≥34 weeks' gestation at 14 California and Massachusetts hospitals from 1993 to 2007. Case-subjects had culture-confirmed bacterial infection at &lt;72 hours; controls were randomly selected, frequency-matched 3:1 according to year and birth hospital. We performed multivariate analyses and split validation to define a predictive model based only on information available in the immediate perinatal period.</p>

<p><strong>RESULTS: </strong>We identified 350 case-subjects from a cohort of 608,014 live births. Highest intrapartum maternal temperature revealed a linear relationship with risk of infection below 100.5°F, above which the risk rose rapidly. Duration of rupture of membranes revealed a steadily increasing relationship with infection risk. Increased risk was associated with both late-preterm and postterm delivery. Risk associated with maternal group B Streptococcus colonization is diminished in the era of group B Streptococcus prophylaxis. Any form of intrapartum antibiotic given &gt;4 hours before delivery was associated with decreased risk. Our model showed good discrimination and calibration (c statistic = 0.800 and Hosmer-Lemeshow P = .142 in the entire data set).</p>

<p><strong>CONCLUSIONS: </strong>A predictive model based on information available in the immediate perinatal period performs better than algorithms based on risk-factor threshold values. This model establishes a prior probability for newborn sepsis, which could be combined with neonatal physical examination and laboratory values to establish a posterior probability to guide treatment decisions.</p>

<p><strong>BACKGROUND: </strong>The absolute neutrophil count and the immature/total neutrophil ratio (I/T) provide information about the risk of early onset sepsis in newborns. However, it is not clear how to combine their potentially overlapping information into a single likelihood ratio.</p>

<p><strong>METHODS: </strong>We obtained electronic records of blood cultures and of complete blood counts with manual differentials drawn &lt;1 hour apart on 66,846 infants ≥ 34 weeks gestation and &lt;72 hours of age born at Kaiser Permanente Northern California and Brigham and Women's Hospitals. We hypothesized that dividing the immature neutrophil count (I) by the total neutrophil count (T) squared (I/T) would provide a useful summary of the risk of infection. We evaluated the ability of the I/T to discriminate newborns with pathogenic bacteremia from other newborns tested using the area under the receiver operating characteristic curve (c).</p>

<p><strong>RESULTS: </strong>Discrimination of the I/T (c = 0.79; 95% confidence interval: 0.76-0.82) was similar to that of logistic models with indicator variables for each of 24 combinations of the absolute neutrophil count and the proportion of immature neutrophils (c = 0.80, 95% confidence interval: 0.77-0.83). Discrimination of the I/T improved with age, from 0.70 at &lt;1 hour to 0.87 at ≥ 4 hours. However, 60% of I/T had likelihood ratios of 0.44-1.3, thus only minimally altering the pretest odds of disease.</p>

<p><strong>CONCLUSIONS: </strong>Calculating the I/T could enhance prediction of early onset sepsis, but the complete blood counts will remain helpful mainly when done at &gt;4 hours of age and when the pretest probability of infection is close to the treatment threshold.</p>

<p><strong>BACKGROUND: </strong>A complete blood count (CBC) with white blood cell differential is commonly ordered to evaluate newborns at risk for sepsis.</p>

<p><strong>OBJECTIVES: </strong>To quantify how well components of the CBC predict sepsis in the first 72 hours after birth.</p>

<p><strong>METHODS: </strong>For this retrospective cross-sectional study we identified 67 623 term and late-preterm (≥ 34 weeks gestation) newborns from 12 northern California Kaiser hospitals and 1 Boston, Massachusetts hospital who had a CBC and blood culture within 1 hour of each other at &lt;72 hours of age. We compared CBC results among newborns whose blood cultures were and were not positive and quantified discrimination by using receiver operating characteristic curves and likelihood ratios.</p>

<p><strong>RESULTS: </strong>Blood cultures of 245 infants (3.6 of 1000 tested newborns) were positive. Mean white blood cell (WBC) counts and mean absolute neutrophil counts (ANCs) were lower, and mean proportions of immature neutrophils were higher in newborns with infection; platelet counts did not differ. Discrimination improved with age in the first few hours, especially for WBC counts and ANCs (eg, the area under the receiver operating characteristic curve for WBC counts was 0.52 at &lt;1 hour and 0.87 at ≥ 4 hours). Both WBC counts and ANCs were most informative when very low (eg, the likelihood ratio for ANC &lt; 1000 was 115 at ≥ 4 hours). No test was very sensitive; the lowest likelihood ratio (for WBC count ≥ 20 000 at ≥ 4 hours) was 0.16.</p>

<p><strong>CONCLUSION: </strong>Optimal interpretation of the CBC requires using interval likelihood ratios for the newborn's age in hours.</p>