Prediction of In-Hospital Mortality After 24 Hours in Very Low Birth Weight Infants

Risk Factors for Mortality or Major Morbidities of Very Preterm Infants: A Study from Thailand

OBJECTIVE

Very preterm neonates have high rates of composite outcomes featuring mortality and major morbidities. If the modifiable risk factors could be identified, perhaps the rates could be decreased especially in resource-limited settings.

STUDY DESIGN

We performed a prospective study in a Thai neonatal intensive care unit to identify the risk factors of composite outcomes between 2014 and 2021. The inclusion criterion was neonates who were born in our hospital at a gestational age (GA) of less than 32 weeks. The exclusion criteria were neonates who died in the delivery room or had major congenital anomalies. The composite outcomes were analyzed by multivariable logistic regression with adjusted odds ratios (aORs) and a 95% confidence interval (CI).

RESULTS

Over the 8-year study period, 555 very preterm inborn neonates without major birth defects were delivered. The composite outcomes were 29.4% (163/555). The medians (interquartile ranges) of GA and birth weights of the neonates were 29 (27-31) weeks and 1,180 (860-1,475) grams, respectively. By multivariable analysis, GA (aOR: 0.65; 95% CI: 0.55-0.77), small for GA (aOR: 4.93; 95% CI: 1.79-13.58), multifetal gestation (aOR: 2.23; 95% CI: 1.12-4.46), intubation within 24 hours (aOR: 5.39; 95% CI: 1.35-21.64), and severe respiratory distress syndrome (aOR: 5.00; 95% CI: 1.05-23.89) were significantly associated with composite outcomes.

CONCLUSION

Very preterm infants who had a lower GA were small for GA, twins or more, respiratory failure on the first day of life, and severe respiratory distress syndrome were associated with mortality and/or major morbidities.

KEY POINTS

· In very preterm neonates, the composite outcomes and mortality rate were 29.4 and 12.3%.. · Composite outcomes were associated with lower GA, SGA, multifetal gestation, intubation, and severe RDS.. · Mortality was associated with lower GA or Apgar score at 5 minutes, SGA, and PPHN..

Predicting early mortality and severe intraventricular hemorrhage in very-low birth weight preterm infants: a nationwide, multicenter study using machine learning

Our aim was to develop a machine learning-based predictor for early mortality and severe intraventricular hemorrhage (IVH) in very-low birth weight (VLBW) preterm infants in Taiwan. We collected retrospective data from VLBW infants, dividing them into two cohorts: one for model development and internal validation (Cohort 1, 2016-2021), and another for external validation (Cohort 2, 2022). Primary outcomes included early mortality, severe IVH, and early poor outcomes (a combination of both). Data preprocessing involved 23 variables, with the top four predictors identified as gestational age, birth body weight, 5-min Apgar score, and endotracheal tube ventilation. Six machine learning algorithms were employed. Among 7471 infants analyzed, the selected predictors consistently performed well across all outcomes. Logistic regression and neural network models showed the highest predictive performance (AUC 0.81-0.90 in both internal and external validation) and were well-calibrated, confirmed by calibration plots and the lowest two mean Brier scores (0.0685 and 0.0691). We developed a robust machine learning-based outcome predictor using only four accessible variables, offering valuable prognostic information for parents and aiding healthcare providers in decision-making.

Predicting risk of severe neonatal outcomes in preterm infants born from mother with prelabor rupture of membranes

BACKGROUND

Perinatal complications are common burdens for neonates born from mother with pPROM. Physicians and parents sometimes need to make critical decisions about neonatal care with short- and long-term implications on infant's health and families and it is important to predict severe neonatal outcomes with high accuracy.

METHODS

The study was based on our prospective study on 1001 preterm infants born from mother with pPROM from August 1, 2017, to March 31, 2018 in three hospitals in China. Multivariable logistic regression analysis was applied to build a predicting model incorporating obstetric and neonatal characteristics available within the first day of NICU admission. We used enhanced bootstrap resampling for internal validation.

RESULTS

One thousand one-hundred pregnancies with PROM at preterm with a single fetus were included in our study. SNO was diagnosed in 180 (17.98%) neonates. On multivariate analysis of the primary cohort, independent factors for SNO were respiratory support on the first day,, surfactant on day 1, and birth weight, which were selected into the nomogram. The model displayed good discrimination with a C-index of 0.838 (95%CI, 0.802-0.874) and good calibration performance. High C-index value of 0.835 could still be reached in the internal validation and the calibration curve showed good agreement. Decision curve analysis showed if the threshold is > 15%, using our model would achieve higher net benefit than model with birthweight as the only one predictor.

CONCLUSION

Variables available on the first day in NICU including respiratory support on the first day, the use of surfactant on the first day and birthweight could be used to predict the risk of SNO in infants born from mother with pPROM with good discrimination and calibration performance.

Growth Pattern With Morbidities From Birth to 5 Years of Age in Very Low Birth Weight Infants: Comparison of the Korean National Network and National Health Insurance Service

BACKGROUND

Long-term growth data of very low birth weight (VLBW) infants are currently collected in the Korean Neonatal Network (KNN) and National Health Insurance Service (NHIS) database. However, variance in the number of infants, check-up time, and check-up parameters led to decreased credibility of cumulated data. We aimed to compare the data on serial growth outcomes by major morbidities from birth to 5 years in VLBW infants between the KNN and NHIS databases.

METHODS

We combined the NHIS and KNN data of VLBW infants born between 2013 and 2015. The check-up times in the NHIS database were at 4-6, 9-12, 18-24, 30-36, 42-48, and 54-60 months of age, whereas in the KNN were at 18-24 months of corrected age and at 36 months of age.

RESULT

Among 8,864 VLBW infants enrolled based on the birth certificates from the Statistics Korea, 6,086 infants (69%) were enrolled in the KNN, and 5,086 infants (57%) participated in the NHIS health check-up. Among 6,068 infants, 3,428 infants (56%) were enrolled at a corrected age of 18-24 months and 2,572 infants (42%) were enrolled at a chronological age of 33-36 months according to the KNN follow-up registry. However, based on the national birth statistics data, the overall follow-up rate of the KNN at 36 months of age was as low as 29%. The NHIS screening rate was lower at first (23%); however, it increased over time to exceed the KNN follow-up rate. Growth failure (weight under 10th percentile) at corrected ages of 18-24 months and 36 months were more common in the NHIS than KNN (42% vs. 20%, 37% vs. 34.5%). Infants with bronchopulmonary dysplasia and periventricular leukomalacia showed similar rates of growth failure at 2 years but varying rates at 3 years between the KNN and NHIS.

CONCLUSION

By integrating the KNN and NHIS data indirectly at continuous time points according to morbidities, we found that there are discontinuities and discrepancies between the two databases among VLBW infants. Establishing an integrated system by patient level linking the KNN and NHIS databases can lead to better understanding and improved neonatal outcomes in VLBW infants in Korea.