A Revised Birth Weight Reference for the United States

Duryea, Elaine L. MD; Hawkins, Josiah S. MD; McIntire, Donald D. PhD; Casey, Brian M. MD; Leveno, Kenneth J. MD

Obstetrics & Gynecology:
doi: 10.1097/AOG.0000000000000345
Contents: Original Research

OBJECTIVE: To generate birth weight curves based on the obstetric estimate of gestational age as specified in the revised 2003 U.S. birth certificate.

METHODS: Using National Center for Health Statistics data from 2011, we constructed birth weight curves for neonates between 24 and 42 weeks of gestation. Curves were developed using the obstetric estimate of gestational age that is included in the revised 2003 U.S. birth certificate, which, when available, incorporates ultrasound dating information. Live-born singleton neonates between 500 and 6,000 g without malformations were included. These curves were compared with curves we generated using 1991 data on which the current national reference of Alexander and colleagues is based, a reference that used only last menstrual period to establish gestational age.

RESULTS: The 1991 curves were based on 3,684,778 U.S. live births and the 2011 on 3,252,011 births. Birth weight percentile values were greater from 28 to 36 weeks of gestation in the 1991 data set. That is, the birth weights for preterm neonates were overestimated when 1991 reference curves were used compared with the proposed 2011 reference. For example, in 1991, a birth weight of 2,000 g was at the 50th percentile between 31 and 32 weeks of gestation, whereas in 2011, a birth weight of 2,000 g now corresponds to the 50th percentile between 33 and 34 weeks of gestation.

CONCLUSIONS: Our revised reference curve for the United States provides an updated national reference for birth weight.


In Brief

Revised birth weight reference curves for the United States take into account current obstetric methods of assigning gestational age.

Author Information

Department of Obstetrics and Gynecology, University of Texas Southwestern Medical Center, Dallas, Texas.

Corresponding author: Elaine L. Duryea, MD, Department of Obstetrics and Gynecology, University of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Boulevard, Dallas, TX 75235-9032; e-mail: Elaine.Duryea@UTSouthwestern.edu.

Presented at the 34th Annual Meeting of the Society of Maternal-Fetal Medicine, February 3–8, 2014, New Orleans, Louisiana.

Financial Disclosure The authors did not report any potential conflicts of interest.

Article Outline

Birth weight percentile is an important clinical measurement used in the prediction of newborn morbidity and mortality. Extremes of birth weight are associated with specific neonatal risks, and many reference curves have been constructed to classify newborns based on their birth weight. Lubchenco1 published the first widely used birth weight curve in 1963, which was based on a group of live-born white neonates delivered in a single Denver hospital. This was a ground-breaking publication that created a tool by which both small-for-gestational-age as well as large-for-gestational-age neonates could be more precisely identified. However, this study has limited application given the well-described phenomena of decreased third-trimester weight gain at higher altitudes.2 Multiple subsequent birth weight curves have been constructed from various populations, many stratified by parity, race, and neonate sex.3–6 These reference curves have been limited by their localized populations and inexact criteria for gestational age. In 1996 Alexander et al7 published a national fetal growth curve based on birth weights of all single live-born neonates reported in 1991 by the National Center for Health Statistics. Gestational age for this curve was calculated using the last menstrual period reported on the birth certificate.

The majority of contemporary obstetric practitioners use ultrasonography to evaluate for fetal abnormalities and confirm or refute gestational age. It has been estimated that more than 90% of women in the United States undergo ultrasound examination in pregnancy.8 The objective of this study was to reanalyze birth weight curves based on the obstetric estimate of gestation age as specified in the revised 2003 U.S. Certificate of Live Birth, which, when available, incorporates ultrasound dating information.

Back to Top | Article Outline


Publically available data sets from the National Center for Health Statistics for live births from 1991 to 2011 were used to construct birth weight curves for neonates between 24 and 42 weeks of gestation. This study was institutional review board-exempt. Live-born singleton neonates without known malformations and with recorded birth weights between 500 and 6,000 g were included. In construction of the curve for 1991, gestational age was based on last menstrual period only as was done by Alexander et al. In the 2003 revision of the birth certificate, a new component entitled “obstetric estimate” was added. The instruction manual for health care providers completing the birth certificate describes this as “the obstetric estimate of the infant's gestation in completed weeks based on the birth attendant's final estimate of gestation which should be determined by all perinatal factors and assessments such as ultrasound, but not the neonatal exam.”9 Additional instructions prohibit completing this field based solely on the neonate’s date of birth and the mother's date of last menstrual period. Thirty-six states had adopted the 2003 revision by 2011, which accounted for 86% of reported live births in 2011.10 In the construction of the reference curves now reported for 2011, the obstetric estimate was used with 3,252,011 births meeting inclusion criteria.

For both 1991 and 2011, data were stratified by gestational age and curves for the 10th, 50th, and 90th percentiles were prepared using quantile regression with gestational age entered as a cubic smoothing spline with knot selection as data-specific. Quantile regression is a method used to estimate the curve under which a targeted percentage (quantile) of the data is expected to be present.11 As a regression function, the neighboring gestational age percentiles affect the estimation of a particular gestational age. Using a cubic smoothing spline allows a nonparametric smooth figure relating the outcome (birth weight percentile) to the independent variable (gestational age). We estimated the birth weight reference curves for both the 1991 and 2011 data using the same method to ensure that any differences between the curves were data-related rather than influenced by the statistical method. The curves published by Alexander et al were also compared with our 1991 estimate and were virtually identical, assuring again that differences are not the result of the method of estimation (Fig. 1).

Back to Top | Article Outline


Maternal demographics for 1991 compared with 2011 are shown in Table 1. The distribution has changed over the past 20 years with an increasing proportion of total live births to Hispanic women. The average maternal age has increased over time with fewer teenage pregnancies and more births to women of advanced maternal age.

Shown in Figure 2 are birth weight curves for 1991 compared with those for 2011. The percentile curves diverge between 28 and 36 weeks of gestation with significantly greater birth weight values in the 1991 data set. For example, in 1991, a birth weight of 2,000 g was at the 50th percentile between 31 and 32 weeks of gestation, whereas in 2011, a birth weight of 2,000 g now corresponds to the 50th percentile between 33 and 34 weeks of gestation. Shown in Table 2 is a birth weight percentile distribution chart for neonates in 2011. When comparing male and female neonates at all weeks of gestation and percentiles, female neonates were consistently smaller except at the 95th percentile. This is demonstrated in Table 3, a birth weight percentile chart for male and female neonates. Of note, shown in Figure 3 are birth weight curves for 1991 and 2011 both with gestational age estimate based on last menstrual period only. In an effort to clinically apply our 2011 birth weight curves, we analyzed the proportions of small-for-gestational-age, appropriate-for-gestational-age, and large-for-gestational-age neonates in 2011 that would be identified with the 1991 reference. Small for gestational age, appropriate for gestational age, and large for gestational age were defined, respectively, as birth weights less the 10th, 10th to 90th, and greater than the 90th percentile based on the 2011 reference. A table of these values during various weeks of the third trimester was constructed (Table 4). Use of the 1991 reference curves in the current population overestimates small for gestational age in the preterm population and underestimates its prevalence at term.

Back to Top | Article Outline


The 2011 revised birth weight reference now proposed for the United States differs substantially from that in current use, which was based on 1991 births dated by last menstrual period. Birth weight percentiles were greater from 28 to 36 weeks of gestation in the 1991 data set compared with 2011 with the magnitude of the difference greatest at 32 weeks of gestation. Specifically, the difference was 886 g for large-for-gestational-age neonates, 378 g for appropriate-for-gestational-age neonates, and 122 g for those classified as small for gestational age. We are of the view that this difference is attributable to use of the recently (2003) revised U.S. birth certificate, which specifies assignment of gestational age based on obstetric criteria to include ultrasound dating when available rather than the last reported menses alone. Put another way, gestational age was estimated using only the last menses in the 1991 data set, whereas in 2011, the last menstrual period was augmented by ultrasound examination. Importantly, the reference curves we derived for our analysis of the 1991 data are virtually superimposable onto those reported for the same year by Alexander and colleagues, supporting that it was not our method of analysis that accounted for the differences, but rather the underlying data.

When the last menses is certain and the menses regularly occur approximately every 28 days, it has been shown that reliance on the last menses alone is an acceptable method of pregnancy dating. However, there are many factors that may contribute to inaccurate determination of gestational age. For example, women may have difficulty with recall of the exact day of the onset of their last menses, have a longer than average intermenstrual interval, or have irregular menses as a result of anovulatory bleeding. Today, the obstetric estimate for assignment of gestational age as described in the 2003 birth certificate is the most common and clinically accepted method for dating pregnancy.12

Many authors have previously described the possibility of excessive birth weights in preterm neonates reflecting systematically underestimated gestational ages.13 This is presumably caused by larger neonates at later gestational ages that are misclassified as an earlier gestational age. Therefore, given that term live births greatly outnumber the number of neonates born prematurely, this has a greater effect at lower gestational ages in which a few misclassified term neonates can significantly skew birth weight curves.14 Alexander et al noticed this effect in what they described as a bimodal distribution for each gestational age group. They attempted to control for this effect by setting a specific birth weight “cut point” for each gestational age based on the second larger mode of the bimodal distribution. This was likely insufficient because the effect of inaccurate dating would be continuous and skew the entire distribution. On analysis of the 2011 data by obstetric estimate, a much less pronounced bimodal distribution at 27–32 weeks of gestation was found, which was significantly improved from that seen with the 1991 data and did not necessitate any exclusion of neonates, as was done by Alexander et al.

Although demographics of pregnant women in the United States have changed over the past 20 years, the difference in our new reference curve can be most attributed to the use of new obstetric dating. This is demonstrated by Figure 3, in which the 90th percentile curve for 2011 with dating by last menstrual period alone is superimposed on those for 1991 without much visible difference.

Our study has limitations. Although the birth certificate is usually completed by a professional medical provider, the process is subject to collection errors. For example, the history provided that includes the last menstrual period or other dating criteria can be recorded based solely on maternal recall when medical records are unavailable at the time of delivery. Although not necessarily a limitation, it is important to note that the cohort of women used for our analysis includes women with known maternal disease and fetal complications other than anomalies. Therefore, our curves must be viewed as a reference and not a standard with standard meaning a limited cohort of only healthy and uncomplicated pregnancies. That is, the 2011 cohort includes neonates with abnormal growth.

There has been a long-standing controversy over the merits of routine ultrasonography during pregnancy. This was truer in the early days of ultrasound use in obstetrics compared with more recent years when assessment of fetal anatomy has become of paramount interest for prenatal diagnosis. Perhaps an unexpected advantage of routine ultrasonography can be proposed based on the results now reported, which suggest that routine use of ultrasonography coupled with menstrual history has facilitated more accurate reference birth weight curves.

Back to Top | Article Outline


1. Lubchenco LO, Hansman C, Dressler M, Boyd E. Intrauterine growth as estimated from liveborn birth-weight data at 24 to 42 weeks of gestation. Pediatrics 1963;32:793–800.
2. Gonzalez GF, Tapia V. Birth weight charts for gestational age in 63,620 healthy infants born in Peruvian public hospitals at low and at high altitude. Acta Paediatr 2008;98:454–8.
3. Brenner WE, Edelman DA, Hendricks CH. A standard of fetal growth for the United States of America. Am J Obstet Gynecol 1976;126:555–64.
4. Williams RL, Creasy RK, Cunningham GC, Hawes WE, Norris FD, Tashiro M. Fetal growth and prenatal viability in California. Obstet Gynecol 1982;59:624–32.
5. Blidner IN, Mcclemont S, Anderson GD, Sinclair JC. Size-at-birth standards for an urban Canadian population. Can Med Assoc J 1984;130:133–40.
6. Hoffman HJ, Start CR, Lundin FE Jr, Ashbrook JD. Analysis of birth weight, gestational age, and fetal viability in U.S. birth, 1968. Obstet Gynecol Surv 1974;29:651–81.
7. Alexander GR, Himes JH, Kaufman RB, Mor J, Kogan M. A United States national reference for fetal growth. Obstet Gynecol 1996;87:163–8.
8. O'Keeffe D, Abuhamad A. Obstetric ultrasound utilization in the United States: data from various health plans. Semin Perinatol 2013;37:292–4.
9. Centers for Disease Control and Prevention. Birth edit specifications for the 2003 proposed revision of the U.S. standard certificate of birth. Available at: http://www.cdc.gov/nchs/data/dvs/birth_edit_specifications.pdf. Retrieved March 26, 2014.
10. Centers for Disease Control and Prevention. User guide to the 2011 natality public use file. Available at: http://www.cdc.gov/nchs/data_access/Vitalstatsonline.htm. Retrieved March 26, 2014.
11. Koenker R, Bassett G. Regression quantiles. Econometrica 1978:146:33–50.
12. Ultrasonography in pregnancy. ACOG Practice Bulletin No. 101. American College of Obstetricians and Gynecologists. Obstet Gynecol 2009;113:451–61.
13. Joseph KS, Huang L, Liu S, Ananth CV, Allen AC, Sauve R, et al.. Reconciling the high rates of preterm and Postterm birth in the United States. Obstet Gynecol 2007;109:813–22.
14. Zhang J, Watson W A Jr. Birth-weight-for-gestational-age patterns by race, sex, and parity in the United States population. Obstet Gynecol 1995;86:200–8.
© 2014 by The American College of Obstetricians and Gynecologists.