Prediction of obstetrical and fetal complications using automated electronic health record data.

An increasing number of delivering women experience major morbidity and mortality. Limited work has been done on automated predictive models that could be used for prevention. Using only routinely collected obstetrical data, this study aimed to develop a predictive model suitable for real-time use with an electronic medical record. We used a retrospective cohort study design with split validation. The denominator consisted of women admitted to a delivery service. The numerator consisted of women who experienced a composite outcome that included both maternal (eg, uterine rupture, postpartum hemorrhage), fetal (eg, stillbirth), and neonatal (eg, hypoxic ischemic encephalopathy) adverse events. We employed machine learning methods, assessing model performance using the area under the receiver operator characteristic curve and number needed to evaluate. A total of 303,678 deliveries took place at 15 study hospitals between January 1, 2010, and March 31, 2018, and 4130 (1.36%) had ≥1 obstetrical complication. We employed data from 209,611 randomly selected deliveries (January 1, 2010, to March 31, 2017) as a derivation dataset and validated our findings on data from 52,398 randomly selected deliveries during the same time period (validation 1 dataset). We then applied our model to data from 41,669 deliveries from the last year of the study (April 1, 2017, to March 31, 2018 [validation 2 dataset]). Our model included 35 variables (eg, demographics, vital signs, laboratory tests, progress of labor indicators). In the validation 2 dataset, a gradient boosted model (area under the receiver operating characteristic curve or c statistic, 0.786) was slightly superior to a logistic regression model (c statistic, 0.778). Using an alert threshold of 4.1%, our final model would flag 16.7% of women and detect 52% of adverse outcomes, with a number needed to evaluate of 20.9 and 0.455 first alerts per day per 1000 annual deliveries. In conclusion, electronic medical record data can be used to predict obstetrical complications. The clinical utility of these automated models has not yet been demonstrated. To conduct interventions to assess whether using these models results in patient benefit, future work will need to focus on the development of clinical protocols suitable for use in interventions.

Antenatal and Intrapartum Risk Factors for Hypoxic-Ischemic Encephalopathy in a US Birth Cohort.

OBJECTIVE: To identify risk factors for hypoxic-ischemic encephalopathy (HIE) within a recent US birth cohort. STUDY DESIGN: In a retrospective cohort study of 44 572 singleton infants ≥36 weeks of gestation born at Kaiser Permanente Northern California in 2008-2015, we identified all infants with HIE based on the presence of 3 inclusion criteria: clinical signs of neonatal encephalopathy, NICU admission, and either a 10-minute Apgar of ≤5 or a base excess of ≤-15 mmol/L. Neonatal acidemia was defined as a base excess of ≤-12 mmol/L. We ascertained antenatal and intrapartum complications from electronic records. Multivariable analysis was performed using logistic regression. RESULTS: There were 45 infants (1.0 per 1000) with HIE and 197 (4.4 per 1000) with neonatal acidemia. Of the infants with HIE, 64% had an intrapartum complication consisting of a sentinel event (36%), clinical chorioamnionitis (40%), or both (11%). Risk factors for HIE on multivariable analysis were sentinel event (relative risk [RR], 16.1; 95% CI, 8.4-33) and clinical chorioamnionitis (RR, 5.2; 95% CI, 2.7-9.9). After removing the 16 infants with HIE who were exposed to a sentinel event from multivariate analysis, maternal age of ≥35 years (RR, 2.5; 95% CI, 1.1-5.6) and a urinary tract infection during pregnancy (RR, 2.6; 95% CI, 1.0-6.5) emerged as potential antenatal risk factors for HIE. CONCLUSIONS: A significant proportion of HIE is preceded by a sentinel event, emphasizing the importance of developing improved methodologies to predict and prevent this perinatal complication. Strategies focused on reducing other complications such as clinical chorioamnionitis and/or maternal pyrexia may also improve our ability to prevent HIE.

Predicting the Need for Phototherapy After Discharge.

BACKGROUND AND OBJECTIVES: Bilirubin screening before discharge is performed to identify neonates at risk for future hyperbilirubinemia. The American Academy of Pediatrics recommends using a graph of bilirubin levels by age (the Bhutani Nomogram) to guide follow-up and a different graph to determine phototherapy recommendations. Our objective was to evaluate predictive models that incorporate the difference between the last total serum bilirubin (TSB) before discharge and the American Academy of Pediatrics phototherapy threshold (Δ-TSB) to predict a postdischarge TSB above the phototherapy threshold by using a single graph. METHODS: We studied 148 162 infants born at ≥35 weeks' gestation at 11 Kaiser Permanente Northern California facilities from 2012 to 2017 whose TSB did not exceed phototherapy levels and who did not receive phototherapy during the birth hospitalization. We compared 3 logistic models (Δ-TSB; Δ-TSB-Plus, which included additional variables; and the Bhutani Nomogram) by using the area under the receiver operating characteristic curve (AUC) in a 20% validation subset. RESULTS: A total of 2623 infants (1.8%) exceeded the phototherapy threshold postdischarge. The predicted probability of exceeding the phototherapy threshold after discharge ranged from 56% for a predischarge Δ-TSB 0 to 1 mg/dL below the threshold to 0.008% for Δ-TSB >7 mg/dL below the threshold. Discrimination was better for the Δ-TSB model (AUC 0.93) and the Δ-TSB-Plus model (AUC 0.95) than for the Bhutani Nomogram (AUC 0.88). CONCLUSIONS: The use of Δ-TSB models had excellent ability to predict postdischarge TSB above phototherapy thresholds and may be simpler to use than the Bhutani Nomogram.

Hyperbilirubinemia, Phototherapy, and Childhood Asthma.

OBJECTIVES: Our aim was to quantify the associations of both hyperbilirubinemia and phototherapy with childhood asthma using a population-based cohort with total serum bilirubin (TSB) levels. METHODS: Retrospective cohort study of infants born at ≥35 weeks' gestation in the Kaiser Permanente Northern California health system (n = 109 212) from 2010 to 2014. Cox models were used to estimate hazard ratios (HRs) for a diagnosis of asthma. RESULTS: In the study, 16.7% of infants had a maximum TSB level of ≥15 mg/dL, 4.5% of infants had a maximum TSB level of ≥18 mg/dL, and 11.5% of infants received phototherapy. Compared with children with a maximum TSB level of 3 to 5.9 mg/L, children with a TSB level of 9 to 11.9 mg/dL, 12 to 14.9 mg/dL, and 15 to 17.9 mg/dL were at an increased risk for asthma (HR: 1.22 [95% confidence interval (CI): 1.11-1.3], HR: 1.18 [95% CI: 1.08-1.29], and HR: 1.30 [95% CI: 1.18-1.43], respectively). Children with a TSB level of ≥18 mg/dL were not at an increased risk for asthma (HR: 1.04; 95% CI: 0.90-1.20). In propensity-adjusted analyses, phototherapy was not associated with asthma (HR: 1.07; 95% CI: 0.96-1.20). CONCLUSIONS: Modest levels of hyperbilirubinemia were associated with an increased risk of asthma, but an association was not seen at higher levels. No dose-response relationship was seen. Using phototherapy to prevent infants from reaching these modest TSB levels is unlikely to be protective against asthma.