Skip to Content

Arnold School of Public Health

Banner Image

Arnold School researchers develop sex- and race-specific growth charts for preterm infants using novel statistical model

November 30, 2016 | Erin Bluvas, 

In collaboration with scientists* from the Vermont Oxford Network (VON) and the University of Vermont, Arnold School researchers Nansi Boghossian and Marco Geraci have developed anthropometric charts for preterm infants that provide more detailed information and are derived from a larger sample than those previously available. The study, which was published in Pediatrics, benefited from the application of newly statistical methods and software developed by Geraci with support from a University of South Carolina’s Office of the Vice President for Research ASPIRE grant.

According to the authors, previous birth weight and head circumference growth charts for preterm infants are limited by small study samples. This is particularly the case for infants born at the lower end of extreme prematurity. However, reliable and representative growth charts are critical for guiding nutritional decisions and medical care.

The present study used a large, nationwide, racially diverse dataset (provided by VON) of more than 155,000 infants born preterm (between 22 weeks 0 days and 29 weeks 6 days) between 2006 and 2014, but it also included additional characteristics. The researchers were able to organize the growth charts according to sex and race. Prior to this study, there were no recent race-specific growth charts available for preterm infants.

“It is important to have data on sex and race to accurately classify which newborns are growth-restricted for secondary and tertiary prevention of mortality and adverse outcomes,” says Boghossian, an assistant professor in the Department of Epidemiology and Biostatistics (EPID/BIOS) and lead author on the paper. Birth weights vary in boys as compared to girls and among different racial groups for infants of the same gestational age. For example, female infants are usually smaller than male infants. African American babies are usually smaller than Caucasian babies.

“Previous studies, particularly for those at periviable gestation—which means 22-25 weeks gestation or born 15 to 18 weeks early—have had a limited sample size or had no data on race,” Boghossian says. “Large samples are important for obtaining more precise estimates of the birth weight centiles and, thus, for improved ranking of a newborn’s birth weight against a reference population.”   

The authors’ newly released growth charts are ready for immediate clinical use, serving as a reference for assessing the size of preterm infants at birth. Other applications could include the ability to classify babies to better predict morbidity and mortality.

“We are usually interested in the lower 10th centile of birth weight for gestational age, and infants who are born below the 10th centile (small for gestational age – SGA) are usually at a higher risk of morbidity and mortality,” says Boghossian. “These charts will allow a better classification based on sex and race to accurately identify if a newborn is growth-restricted. Our next step will be to examine data on SGA babies as defined by our charts to see if we are able to improve on prediction of mortality and morbidity outcomes.”

The quantile regression methods used for the charts were developed at the Arnold School by Geraci, an experienced innovator in the field of model and software development. These novel statistical models and software outperformed other, well-known traditional methods by providing the best fits of the data with faster computational times.

“These growth charts are the result of a successful interdisciplinary collaboration,” says Geraci, who is an associate professor in EPID/BIOS. “This work also shows the importance of funding from the Office of the Vice President for Research—which helped support the new software, Qtools, we used to create the growth charts.”

In the same issue of Pediatrics, other researchers offered a commentary on the Boghossian et al.’s proffered growth charts. “Birth weight distributions specific for GA (gestational age), sex, race/ethnicity, and region are better at identifying the infants at higher risk for neonatal morbidity,” the authors wrote in their commentary of this important step forward. “We applaud the important observations made by Boghossian et. al. and agree with the authors that clinicians tasked with assessing the nutrition, growth, and well-being of a fetus, a newly born infant, or the growth of a premature infant in the NICU need tools that correct for differences in sex, race/ethnicity, heritage, and genetics.”

*Co-authors include Nansi Boghossian (Epidemiology and Biostatistics), Marco Geraci (Epidemiology and Biostatistics), Erika Edwards (Vermont Oxford Network; University of Vermont), Kate Morrow (Vermont Oxford Network; University of Vermont), Jeffrey Horbar (Vermont Oxford Network; University of Vermont)