Poster Session 2
Natalie Suder, MD
Resident
Lenox Hill Hospital, NY, United States
.jpg "Ashley Zimmermann, MD (she/her/hers) photo")
Ashley Zimmermann, MD (she/her/hers)
Resident
Northwell Lenox Hill Hospital
New York, New York, United States
Rachel Kim
University of Chicago
Chicago, Illinois, United States
Karen Huang
University of Chicago
Chicago, Illinois, United States
Maria Teresa Benedetto, MD
Attending Physician
Lenox Hill Hospital, New York, United States
Mio Sawai, MD
Attending Physician
Lenox Hill Hospital, New York, United States
Teresa Cheon, MD
Attending Physician
Lenox Hill Hospital, New York, United States
Yuzuru Anzai, MD
Attending Physician
Lenox Hill Hospital, New York, United States
Identifying small for gestational age (SGA) and large for gestational age (LGA) neonates is imperative to manage potential neonatal complications. The Hadlock growth curve, the most widely used standardized growth curve in the United States, overlooks maternal factors such as race, height, BMI, and pregnancy weight gain, which are known to impact fetal weight. This study aims to create a personalized birth weight percentile chart to improve accuracy of diagnosing SGA and LGA neonates.
Study Design:
Using the 2022 CDC natality dataset, a least squares quantile regression model was developed using Python to predict the 10th and 90th percentile birth weights based on maternal factors. To improve model accuracy, gestational diabetes, gestational hypertension, plurality, tobacco use, infant sex, and gestational age (GA) were included. Polynomial regressions determined the best-fit degrees. The model focused on pregnancies from 27 to 42 weeks GA, excluding extreme birth weights (< 200 grams or > 5000 grams) and unknown parameters (n = 3,428,570). Validity of the model was confirmed using an 80:20 split for testing and training, respectively.
Results: Among the factors in the model, plurality had the strongest effect on determining the fetal weight, followed by GA, maternal race, cigarette smoking, fetal gender, and maternal height, even though GA should not alter the percentile. Pre-pregnancy maternal BMI and pregnancy weight gain showed much smaller effects in this model. The quantile loss at the 10th percentile was 66 grams and 90th percentile was 72 grams.
Conclusion:
The model was trained using a large, national dataset and shows high accuracy with low quantile losses; therefore, this model can supplement physicians’ diagnoses of SGA and LGA and aid in appropriate delivery preparations. However, the model results should be interpreted with caution because medical complications and tobacco smoking status used for the model are known factors that can affect fetal growth. In the next model, we aim to determine the SGA threshold that is associated with adverse fetal outcomes.
