Ultrasound Obstet Gynecol 2018; 52: 44–51

Published online 30 May 2018 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/uog.19073

Fetal Medicine Foundation fetal and neonatal population

weight charts
K. H. NICOLAIDES1 , D. WRIGHT2 , A. SYNGELAKI1 , A. WRIGHT2 and R. AKOLEKAR3
1
Kings College Hospital, London, UK; 2 University of Exeter, Exeter, UK; 3 Medway Maritime Hospital, Gillingham, UK

K E Y W O R D S: birth weight; estimated fetal weight; fetal biometry; reference range; small-for-gestational age

ABSTRACT a Bayesian approach to inference, combining information

from the two datasets using Markov Chain Monte-Carlo
Objective To develop fetal and neonatal population
sampling. The fitted model assumed that the mean log
weight charts. The rationale was that, while reference
transformed measurements of EFW and BW are related
ranges of estimated fetal weight (EFW) are representative
to gestational age according to a cubic equation and that
of the whole population, the traditional approach of
deviations about the mean follow a bivariate Gaussian
deriving birth-weight (BW) charts is misleading, because
distribution.
a large proportion of babies born preterm arise from
pathological pregnancy. We propose that the reference Results In the case of EFW in Dataset 2, there was a
population for BW charts, as in the case of EFW charts, good distribution of values < 3rd , < 5th , < 10th , > 90th ,
should comprise all babies at a given gestational age, > 95th and > 97th percentiles of the reference range
including those still in utero. of EFW according to gestational age throughout the
Methods Two sources of data were used for this study. gestational age range of 20 + 0 to 36 + 6 weeks. In the
For both, the inclusion criteria were singleton preg- case of BW, there was a good distribution of values
nancy, dating by fetal crown–rump length at 11 + 0 to only for the cases delivered > 39 weeks’ gestation. For
13 + 6 weeks’ gestation, availability of ultrasonographic preterm births, particularly at 27–36 weeks, BW was
measurements of fetal head circumference (HC), abdom- below the 3rd , 5th and 10th percentiles in a very high
inal circumference (AC) and femur length (FL) and proportion of cases, particularly in cases of iatrogenic
live birth of phenotypically normal neonate. Dataset 1 birth. The incidence of small-for-gestational-age fetuses
comprised a sample of 5163 paired measurements of and neonates in the respective EFW and BW charts was
EFW and BW; ultrasound examinations were carried higher in women of black than those of white racial origin.
out at 22–43 weeks’ gestation and birth occurred within Conclusion We established a BW chart for all babies
2 days of the ultrasound examination. EFW was derived at a given gestational age, including those still in utero,
from the HC, AC and FL measurements using the for- thereby overcoming the problem of underestimation of
mula reported by Hadlock et al. in 1985. Dataset 2 growth restriction in preterm birth. BW and EFW charts
comprised a sample of 95 579 pregnancies with EFW have a common median but differ in the levels of spread
obtained by routine ultrasonographic fetal biometry at from the median. Copyright © 2018 ISUOG. Published
20 + 0 to 23 + 6 weeks’ gestation (n = 45 034), 31 + 0 by John Wiley & Sons Ltd.
to 33 + 6 weeks (n = 19 224) or 35 + 0 to 36 + 6 weeks
(n = 31 321); for the purpose of this study we included INTRODUCTION
data for only one of the three visits per pregnancy. In the
development of reference ranges of EFW and BW accord- There is an apparent contradiction in the relationship
ing to gestational age, the following assumptions were between the ultrasonographic estimation of fetal weight
made: first, that EFW and BW have a common median, (EFW) and birth weight (BW). Although the EFW
dependent on gestational age; and second, that deviations recorded within a few days before birth correlates
from the median occur in both EFW and BW and these strongly with BW and, for a given gestational age,
deviations are correlated with different levels of spread for these two measurements have essentially the same
EFW and BW, dependent on gestational age. We adopted median1 , in reported reference ranges, the median

Correspondence to: Prof. K. H. Nicolaides, Fetal Medicine Research Institute, King’s College Hospital, 16–20 Windsor Walk, Denmark Hill,
London SE5 8BB, UK (e-mail: kypros@fetalmedicine.com)
Accepted: 23 April 2018

Copyright © 2018 ISUOG. Published by John Wiley & Sons Ltd. ORIGINAL PAPER
EFW and birth-weight charts 45

BW according to gestational age for babies born prediction of BW, having the lowest Euclidean distance
preterm is substantially lower than the median EFW2–5 . and highest proportion of pregnancies with an absolute
This difference is likely to be the consequence of mean error of <10.
pathological fetal growth in a high proportion of preterm Dataset 2, from which the reference ranges were
births. Reference ranges of EFW are representative established, comprised a sample of 95 579 pregnan-
of the whole population, whereas in reference ranges cies (not included in Dataset 1) with EFW obtained
of BW, particularly for gestational ages < 37 weeks, by routine ultrasonographic fetal biometry at 20 + 0
there is overrepresentation of pathological pregnancies. to 23 + 6 weeks’ gestation (n = 45 034), 31 + 0 to
One-third of preterm births are iatrogenic, due mainly 33 + 6 weeks (n = 19 224) or 35 + 0 to 36 + 6 weeks
to hypertensive disorders and/or suspected fetal growth (n = 31 321). In the participating hospitals, all women
restriction; there is also evidence of impaired placentation with singleton pregnancy are offered routine ultra-
in a substantial proportion of spontaneous preterm sound examinations at 11 + 0 to 13 + 6 and 20 + 0 to
births6–10 . 23 + 6 weeks’ gestation. During the period 2011 to 2014,
In this study, we propose that the reference population an additional scan was offered at 31 + 0 to 33 + 6 weeks,
for BW charts, as in the case of EFW charts, should and subsequently (between 2014 and 2017) this was
comprise all babies at a given gestational age, including offered at 35 + 0 to 36 + 6 weeks. For the purpose of
those still in utero. Development of these charts was based this study, we included data for only one of the second-
on the assumptions that, first, for a given gestational or third-trimester visits; we used all data obtained at
age, the median BW is the same as the median EFW 31 + 0 to 33 + 6 or 35 + 0 to 36 + 6 weeks and used
in the reference population, and, second, deviations the data of the visit at 20 + 0 to 23 + 6 weeks only for
from the median occur in both BW and EFW and these pregnancies that did not have a third-trimester scan.
deviations follow a bivariate Gaussian distribution, with In the selection of patients, care was taken to include
different levels of spread for BW and EFW, dependent routine scans and not follow-up scans for maternal med-
on gestational age. These assumptions enable data on ical conditions or a suspected problem in fetal growth.
EFW derived from routine scans early in gestation to be Since the objective of the study was to establish ref-
combined with BW at term to produce reference charts erence ranges, rather than normal ranges, we included
for BW and EFW for gestational ages from 20 + 0 to all pregnancies undergoing these routine ultrasound
42 + 6 weeks. examinations. In the case of BW, we restricted data to
pregnancies delivering at 39 + 0 to 41 + 6 weeks because
deliveries at earlier gestations constitute a fraction of the
METHODS total and many of these arise from pathological preg-
nancies; consequently, they are not representative of the
Study population
whole population, including those still in utero.
Two sources of data were used for this study, based on
the same inclusion criteria: singleton pregnancy, dating by Statistical analysis
fetal crown–rump length at 11 + 0 to 13 + 6 weeks’ ges-
tation, availability of ultrasonographic measurements of Measurements of EFW and BW were log transformed
fetal head circumference (HC), abdominal circumference to make the deviations from median close to Gaussian
(AC) and femur length (FL), and live birth of phenotyp- in distribution and the variation about the median
ically normal neonate. The pregnancies were examined at more stable across the range of gestational ages. A
King’s College Hospital, London and Medway Maritime Bayesian approach to inference was adopted, combining
Hospital, Kent, UK, between January 2006 and Decem- information from Datasets 1 and 2 using Markov
ber 2017. The ultrasound scans were carried out by Chain Monte-Carlo sampling. The fitted model assumed
sonographers who had received The Fetal Medicine Foun- that the mean log-transformed measurements of EFW
dation (FMF) Certificate of Competence in ultrasound and BW were related to gestational age according
examination. to a cubic equation and that deviations about the
Dataset 1 comprised a sample of 5163 paired measure- mean followed a bivariate Gaussian distribution. Gross
ments of EFW and BW. The ultrasound examinations outliers were identified from an initial model and
were carried out at 22–43 weeks’ gestation and birth observations with standardized residuals beyond ± 3.89
occurred within 2 days of the scan1 . This dataset, in which (the 0.00005th percentile of the Gaussian distribution)
pathological pregnancies were inevitably overrepresented, were excluded from the final model. A range of model
was used to examine the relationship between EFW and diagnostics was produced to assess the goodness-of-fit
BW; it was not used to establish the reference ranges. of the model. This included summary statistics and
EFW was derived from the HC, AC and FL measurements Gaussian probability plots of z-scores for data on EFW
using the formula of Hadlock et al.11 . The selection of and BW. Non-parametric quantile regression was used
this formula was based on a previous systematic review for direct estimation of percentiles of the EFW and
of the literature1 , in which we identified all models for BW data for comparison with the parametric model.
EFW and found that the formula of Hadlock et al.11 was Details of the analysis and model diagnostics are given
the most accurate among 70 published models for the in Appendix S1.

Copyright © 2018 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2018; 52: 44–51.
46 Nicolaides et al.

6000

5000

4000

3000

2000
BW (g)

1500

1000

700

500

400
400 500 700 1000 1500 2000 3000 4000 6000
EFW (g)

Figure 1 Association between estimated fetal weight (EFW), derived from model of Hadlock et al. using measurements of head
circumference, abdominal circumference and femur length11 , and birth weight (BW) in Dataset 1. Regression line is blue and line EFW = BW
is red.

Statistical software package R was used for data

analyses12 . The R packages mvtnorm13 and quantreg14 4500
were used for multivariate Gaussian statistics and quantile
regression. 4000

RESULTS 3500

Pregnancy characteristics of the two datasets are

3000
summarized in supplementary Tables S1 and S2. The
EFW or BW (g)

association between EFW and BW in the 5163 pregnancies

of Dataset 1, in which birth occurred within 2 days of 2500
the ultrasound examination, is shown in Figure 1. For a
given gestational age, the median EFW is essentially the 2000
same as the median BW. Further evidence to support the
assumption of equivalence in means of EFW and BW is
1500
provided in Appendix S1.
The median, 3rd , 10th , 90th and 97th percentiles of EFW
and BW according to gestational age are shown in Figure 2 1000
and the median, 3rd , 5th 10th , 25th , 75th , 90th , 95th and
97th percentiles of EFW and BW for each gestational week 500
between 20 + 3 and 41 + 3 weeks are shown in Table 1.
The standard deviation and percentiles of EFW and BW 0
for each gestational day between 20 + 0 and 42 + 6 weeks 20 22 24 26 28 30 32 34 36 38 40 42
are shown in Table S3. Gestational age (weeks)
The distribution of EFW according to gestational age
in our chart is compared with those of the World Health Figure 2 Median for both estimated fetal weight (EFW) and birth
Organization (WHO)5 and INTERGROWTH-21st 4 weight (BW) according to gestational age (solid black line), and
charts in Figure 3 and Table S4. The median and 10th 10th and 90th percentiles (solid colored lines) and 3rd and 97th
percentiles (dashed lines) for EFW (blue) and BW (red).
percentiles of the WHO chart5 , and more so those of

Copyright © 2018 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2018; 52: 44–51.
EFW and birth-weight charts 47

Table 1 Median, 3rd , 5th , 10th , 25th , 75th , 90th , 95th and 97th percentiles of estimated fetal weight (EFW) and birth weight (BW) in second
half of gestation

Percentile
Gestational
age 3rd 5th 10th 25th 75th 90th 95th 97th
Weeks Days EFW BW EFW BW EFW BW EFW BW Median EFW BW EFW BW EFW BW EFW BW

20 143 300 283 306 290 314 301 329 322 346 364 372 381 398 392 414 399 424
21 150 358 337 364 345 375 359 392 384 413 435 445 455 475 468 494 477 507
22 157 424 399 432 410 444 427 466 456 491 517 528 542 564 557 587 567 603
23 164 501 472 510 484 525 504 550 538 580 611 624 641 668 659 695 671 713
24 171 588 554 599 568 616 592 646 633 682 719 734 754 785 776 818 791 839
25 178 686 646 699 664 719 691 755 739 797 841 859 883 919 909 957 926 982
26 185 796 750 811 771 835 803 877 859 926 978 999 1027 1069 1058 1113 1078 1143
27 192 918 866 936 889 964 926 1013 992 1070 1131 1154 1188 1236 1224 1288 1248 1322
28 199 1052 993 1072 1020 1105 1062 1162 1138 1228 1299 1326 1365 1420 1407 1480 1435 1520
29 206 1197 1130 1221 1161 1258 1210 1324 1297 1400 1481 1512 1558 1620 1606 1689 1638 1735
30 213 1353 1278 1380 1313 1423 1369 1498 1468 1586 1678 1713 1767 1836 1822 1914 1858 1967
31 220 1518 1435 1549 1475 1598 1538 1683 1649 1782 1888 1926 1988 2066 2051 2154 2092 2213
32 227 1691 1599 1725 1643 1780 1714 1876 1839 1988 2107 2150 2221 2307 2291 2406 2338 2472
33 234 1868 1768 1907 1817 1968 1895 2075 2034 2201 2334 2381 2461 2555 2540 2666 2593 2740
34 241 2048 1938 2091 1993 2159 2079 2277 2233 2416 2564 2615 2705 2808 2793 2930 2851 3012
35 248 2226 2108 2273 2167 2347 2262 2478 2430 2631 2793 2849 2948 3060 3045 3194 3110 3284
36 255 2398 2272 2449 2336 2531 2439 2672 2621 2839 3017 3076 3186 3306 3292 3451 3362 3549
37 262 2561 2427 2616 2496 2704 2607 2857 2802 3037 3229 3292 3412 3539 3526 3696 3602 3801
38 269 2709 2568 2768 2642 2862 2760 3026 2968 3219 3424 3490 3620 3754 3742 3921 3824 4034
39 276 2839 2692 2901 2770 3001 2894 3174 3114 3379 3596 3665 3804 3944 3934 4120 4021 4239
40 283 2945 2795 3011 2876 3115 3006 3297 3236 3512 3740 3811 3959 4103 4095 4288 4187 4412
41 290 3025 2872 3094 2956 3201 3090 3390 3328 3613 3851 3923 4078 4225 4220 4417 4315 4546

Both EFW and BW are given in g. It was assumed that EFW and BW have a common median, dependent on gestational age.

the INTERGROWTH-21st chart4 , are substantially lower

4000
than the respective ones in our FMF chart between 24 and
38 weeks of gestation.
The proportion of cases in Dataset 2 with EFW 3500
and BW < 3rd , < 5th < 10th , > 90th , > 95th and > 97th
percentiles of the appropriate reference range according
3000
to gestational age is shown in Table S5. The distribution
of EFW values was well balanced throughout the
gestational-age range of 20 + 0 to 36 + 6 weeks, whereas 2500
a good distribution of BW values was observed only
EFW (g)

in cases delivered >39 weeks’ gestation. For preterm

births, particularly at 27–36 weeks, BW was below the 2000
3rd , 5th and 10th percentiles in a very high proportion
of cases (Figure 4, Table S6). This was particularly 1500
marked in cases delivered preterm due to iatrogenic
causes (40.3%, 45.1% and 52.5% for BW < 3rd , < 5th
and < 10th percentile, respectively); this is not surprising 1000
because, in 1200 (67.0%) of the 1790 cases of iatrogenic
preterm birth, the indication for delivery was hypertensive 500
disease and/or fetal growth restriction. However, a high
proportion of small-for-gestational-age (SGA) neonates
was also observed among spontaneous preterm births; the 0
20 22 24 26 28 30 32 34 36 38 40
proportion of spontaneous preterm births with BW below
Gestational age (weeks)
the 3rd , 5th and 10th percentiles was 8.6%, 12.4% and
19.8%, respectively (Table S6).
Figure 3 Comparison of 50th (solid lines) and 10th (dashed lines)
The proportion of pregnancies in women of white
percentiles of estimated fetal weight (EFW) according to gestational
and black racial origin in Dataset 2 with EFW and BW age, between World Health Organization (green)5 , INTER-
< 3rd , < 5th and < 10th percentiles of the appropriate GROWTH-21st (red)4 and our Fetal Medicine Foundation (blue)
reference range according to gestational age is shown in charts.

Copyright © 2018 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2018; 52: 44–51.
48 Nicolaides et al.

55

50

45

40

35
Percent

30

25

20

15

10

0
24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43
Gestational age at birth (weeks)

Figure 4 Percentage of cases in Dataset 2 with birth weight below 3rd ( ), 5th ( ) and 10th ( ) percentiles of reference range of birth weight
according to gestational age.

60 the levels of spread from the median. The BW charts

rely on the principle that, at a given gestational age,
especially < 37 weeks, the reference population should
50
include not only those babies that have been born, but all
also, because preterm births are inherently pathological,
40 babies still in utero.
The study has demonstrated that a very high proportion
of preterm births are SGA; this should not be surprising
Percent

30
because, in many such cases, there is iatrogenic birth
for hypertensive disease and/or fetal growth restriction.
20 A high proportion of SGA neonates is also observed
among spontaneous preterm births, providing further
support to histological and uterine artery Doppler findings
10 suggesting that in many such births there is impaired
placentation6–10 . Consequently, to varying degrees, all
0
preterm births arise from pathological pregnancy and it is
≤ 30 31 32 33 34 35 36 37 38 39 40 ≥ 41 misleading to use data from such pregnancies to establish
EFW Gestational age at birth (weeks) reference ranges of BW according to gestational age.
In our heterogeneous unselected population, arising
Figure 5 Percentage of pregnancies of white women ( ) and black from two maternity hospitals in England, about 20%
women ( ) in Dataset 2 with estimated fetal weight (EFW) (left) or of the women were of black racial origin and in these
birth weight (right) below 10th percentile of appropriate reference women the incidence of SGA fetuses and neonates was
range.
higher than in white women. This finding is compatible
with the results of a previous study which reported that
Table S7 and the proportions of cases with EFW and fetal growth is affected by several maternal characteristics:
BW < 10th percentiles are illustrated in Figure 5. The BW increased with maternal weight, height and parity
data demonstrate that the incidence of SGA fetuses and and, after adjustment for these variables, BW was lower
neonates in the respective EFW and BW charts was higher in black than in white women15 . It could therefore
in women of black than in women of white racial origin. be assumed that it is physiological for black women
to produce smaller babies than white women and that
different reference ranges for these racial groups should
DISCUSSION be provided16 . An alternative view is that, in black
Principal findings of the study women living in England, the delivery of smaller babies
is a consequence of pathological influences that would
In this study we have established reference ranges be masked by customized BW percentiles. We have
according to gestational age for BW and EFW. These reported that, in black women, after adjustment for
two charts have a common median but they differ in other demographic and pregnancy characteristics, there

Copyright © 2018 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2018; 52: 44–51.
EFW and birth-weight charts 49

is increased risk for several adverse pregnancy outcomes, preterm births, the EFW was substantially higher than
including miscarriage, stillbirth, pre-eclampsia, fetal the BW; they recommended that the EFW of preterm
growth restriction and both iatrogenic and spontaneous fetuses should not be compared with the distribution of
preterm birth; it is uncertain whether such increased BW, because fetal growth restriction is over-represented in
risks are a consequence of genetic predisposition, preterm births, but rather they should be compared with
socioeconomic deprivation or both17 . We have also shown EFW charts. In our study, we have taken this observation
that BW according to gestational age is reduced in further, highlighting that there is an inherent problem
antepartum stillbirths and there is no significant difference in the traditional construction of BW charts, especially
in the proportion of antepartum stillbirths that are SGA for preterm births, and recommend that they should be
when BW is corrected for maternal characteristics15 . revised based on data from all babies at a given gestational
age, including those still in utero.
Marsál et al.2 recognized that BW charts do not
Strengths and limitations of the study
represent the intrauterine population and proposed that
The main strength of our study is the production of BW it would be preferable to use EFW charts to assess the
reference charts for all babies at a specific gestational growth of both fetuses and neonates. They performed a
age, including those still in utero. This avoids bias and longitudinal study of ultrasonographic fetal biometry at
underestimation of SGA in the assessment of BW in babies 10–41 weeks’ gestation in 86 uncomplicated pregnancies
born preterm. Additional strengths include the large study that delivered at term; they then combined the data from
population of women undergoing routine ultrasound 759 EFWs and 86 BWs to derive an intrauterine growth
examination in pregnancy for Database 2 and use of chart using a fourth-degree polynomial equation. We
their data only once to avoid the potential correlation of agree with Marsal et al.2 on the need to revise BW charts
measurements from different visits, the close proximity and have demonstrated that EFW and BW charts have
of the ultrasound examination to birth for Database 1, a common median, but they differ in the levels of spread
the dating of pregnancies based on fetal crown–rump from the median.
length, the fact that trained sonographers carried out Two international multicenter studies have recently
fetal biometry according to a standardized protocol and reported the construction of charts of EFW according
the utilization for calculation of EFW of a widely used to gestational age4,5 . In the INTERGROWTH-21st
model11 which has been shown to be the most accurate of project, data were derived from 2404 live babies without
70 previously reported models1 . In the establishment of congenital abnormality, who were born within 14 days
reference ranges, we included all pregnancies undergoing of an ultrasound scan; women were recruited from urban
routine ultrasound examination and did not attempt to areas in several countries (Brazil, China, England, India,
select only uncomplicated pregnancies in women thought Italy, Kenya, Oman, Pakistan, South Africa, Thailand and
to be healthy and well-nourished. USA) and had serial ultrasound scans and fetal biometry
A potential limitation of this study is our assumption throughout pregnancy4 . Two cohorts of women were
that, for a given gestational age, EFW and BW have examined; one was unselected and the other was selected
the same median. This was based on the findings from to include healthy, well-nourished, pregnant women who
Dataset 1, but was not possible to investigate further for were at low risk of adverse maternal and perinatal out-
the whole population. Another limitation is the extent comes. The authors reported that the data from different
of extrapolation and interpolation resulting for use of centers were similar and they therefore pooled all data
EFW and BW data. We wanted to include data on EFW and used fractional polynomial models to construct an
arising from routine screening of the whole population international optimal fetal growth chart that would be
and this inevitably restricted the data to the three narrow appropriate for healthy pregnancies in all countries of the
gestational age ranges of 20 + 0 to 23 + 6, 31 + 0 to world4 .
33 + 6 and 35 + 0 to 36 + 6 weeks. In the case of BW, In the WHO study, data were derived from
we restricted data to pregnancies delivering at term to 1387 healthy women with low-risk pregnancies and
avoid bias from inclusion of preterm births, many of unrestricted nutritional and social background who had
which arise from pathological pregnancies. Despite the serial ultrasound scans throughout pregnancy; women
extensive extrapolation and interpolation of data, the were recruited from 10 countries (Argentina, Brazil,
model diagnostics demonstrated a satisfactory fit of the Democratic Republic of the Congo, Denmark, Egypt,
model. France, Germany, India, Norway and Thailand) and
a total of 7924 sets of ultrasound measurements were
Comparison with previous studies analyzed by quantile regression to establish longitudinal
reference intervals for EFW5 . The authors reported that
Salomon et al.3 used the Hadlock formula11 to construct there were significant differences in fetal growth between
EFW charts from biometric data obtained during routine countries.
ultrasound examination at 20–36 weeks’ gestation in In our study, in comparison with the INTER-
18 959 normal fetuses. The authors compared the EFW GROWTH-21st and WHO studies, the population was
to BW charts obtained during the same study period and unselected and considerably larger, the data were derived
in the same single health authority and noted that, for from two centers in the same country and the scans were

Copyright © 2018 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2018; 52: 44–51.
50 Nicolaides et al.

carried out by sonographers with extensive training in ACKNOWLEDGMENT

ultrasound examination in pregnancy. Our approach, like
the one of INTERGROWTH-21st, used a parametric This study was supported by a grant from The Fetal
model. This differs from the approach of the WHO Medicine Foundation (Charity No: 1037116).
study, which used non-parametric quantile regression. A
benefit of the parametric models is that they can be used REFERENCES
easily to obtain z-scores and percentiles for individual
1. Hammami A, Zumaeta AM, Syngelaki A, Akolekar R, Nicolaides KH.
measurements. A drawback is the imposition of a specific Ultrasonographic estimation of fetal weight: development of new model and
parametric relationship. We used a cubic polynomial assessment of performance of previous models. Ultrasound Obstet Gynecol 2018;
52: 35–43.
to represent the relationship between median level of 2. Marsál K, Persson PH, Larsen T, Lilja H, Selbing A, Sultan B. Intrauterine growth
log-transformed weight assuming the same median for curves based on ultrasonically estimated foetal weights. Acta Paediatr 1996; 85:
843–848.
both EFW and BW. Our model assumed that deviations 3. Salomon LJ, Bernard JP, Ville Y. Estimation of fetal weight: reference range at
around the median for BW and EFW followed a bivariate 20–36 weeks’ gestation and comparison with actual birth-weight reference range.
Ultrasound Obstet Gynecol 2007; 29: 550–555.
Gaussian distribution. 4. Stirnemann J, Villar J, Salomon LJ, Ohuma E, Ruyan P, Altman DG, Nosten F, Craik
The 10th percentile of our EFW chart was considerably R, Munim S, Cheikh Ismail L, Barros FC, Lambert A, Norris S, Carvalho M, Jaffer
YA, Noble JA, Bertino E, Gravett MG, Purwar M, Victora CG, Uauy R, Bhutta Z,
higher than those of the INTERGROWTH-21st and Kennedy S, Papageorghiou AT, for The International Fetal and Newborn Growth
WHO charts4,5 . For example, at 36 weeks’ gestation, the Consortium for The 21st Century (INTERGROWTH-21st). International estimated
fetal weight standards of the Intergrowth-21st project. Ultrasound Obstet Gynecol
10th percentile according to our chart was 2531 g, whereas 2017; 49: 478–486.
the respective values in the WHO and INTERGROWTH 5. Kiserud T, Piaggio G, Carroli G, Widmer M, Carvalho J, Neerup Jensen L,
Giordano D, Cecatti JG, Abdel Aleem H, Talegawkar SA, Benachi A, Diemert
-21st charts are 2352 g and 2144 g. Such differences are A, Tshefu Kitoto A, Thinkhamrop J, Lumbiganon P, Tabor A, Kriplani A,
likely to be a consequence of underlying differences Gonzalez Perez R, Hecher K, Hanson MA, Gülmezoglu AM, Platt LD. The World
Health Organization fetal growth charts: a multinational longitudinal study of
in the study populations and demonstrate that the ultrasound biometric measurements and estimated fetal weight. PLoS Med 2017; 14:
desire for a single international standard for all e1002220.
6. Arias F, Rodriguez L, Rayne SC, Kraus FT. Maternal placental vasculopathy and
countries is not appropriate; a single standard would infection: two distinct subgroups among patients with preterm labor and preterm
underestimate growth restriction in countries with normal ruptured membranes. Am J Obstet Gynecol 1993; 168: 585–591.
7. Arias F, Rodriguez L, Victoria A, Cho K, Kraus F. Placental histology and clinical
big babies, such as Norway, and overestimate growth characteristics of patients with preterm premature rupture of membranes. Obstet
restriction in countries with normal small babies, such as Gynecol 1997; 89: 265–271.
8. Kim YM, Chaiworapongsa T, Gomez R, Bujold E, Yoon BH, Rotmensch S, Thaler
India. HT, Romero R. Failure of physiologic transformation of the spiral arteries in the
placental bed in preterm premature rupture of membranes. Am J Obstet Gynecol
2002; 187: 1137–1142.
9. Kim YM, Bujold E, Chaiworapongsa T, Gomez R, Yoon BH, Thaler HT, Rotmensch
Conclusions S, Romero R. Failure of physiologic transformation of the spiral arteries in patients
with preterm labor and intact membranes. Am J Obstet Gynecol 2003; 189:
This study has highlighted the necessity for construction 1063–1069.
10. Fonseca E, Yu CK, Singh M, Papageorghiou AT, Nicolaides KH. Relationship
of BW charts to be based on all babies at a particular between second-trimester uterine artery Doppler and spontaneous early preterm
gestational age, including those still in utero. Within a delivery. Ultrasound Obstet Gynecol 2006; 27: 301–305.
11. Hadlock FP, Harrist RB, Sharman RS, Deter RL, Park SK. Estimation of fetal weight
given country there are variations in BW that depend with the use of head, body, and femur measurements–a prospective study. Am J
on maternal characteristics, such as racial origin, but Obstet Gynecol 1985; 151: 333–337.
12. R Development Core Team. R. A language and environment for statistical
adjustment for such characteristics may be inappropriate computing. R Foundation for Statistical Computing, Vienna, Austria. 2011; ISBN
because such adjustments could result in underestimation 3-900051-07-0, URL http://www.R-project.org/
13. Genz A, Bretz F, Miwa T, Mi X, Leisch F, Scheipl F, Hothorn T. mvtnorm:
of the increased perinatal risk of a disadvantaged group. Multivariate Normal and t Distributions. R package 2016; version 1.0-5. http://
The value of adjustment for maternal weight, height and CRAN.R-project.org/package=mvtnorm.
14. Koenker R. quantreg: Quantile Regression. R package 2016; version 5.29. https://
parity remains controversial. CRAN.R-project.org/package=quantreg.
If our charts are to be used in different countries it 15. Poon LC, Volpe N, Muto B, Syngelaki A, Nicolaides KH. Birthweight with gestation
and maternal characteristics in live births and stillbirths. Fetal Diagn Ther 2012; 32:
would be necessary to ensure that the distribution of 156–165.
values is appropriate, otherwise adjustments would be 16. Buck Louis G, Grewal J, Albert P, Sciscione A, Wing DA, Grobman WA, Newman
RB, Wapner R, D’Alton ME, Skupski D, Nageotte MP, Ranzini AC, Owen J, Chien
necessary either to tailor the charts for a specific setting EK, Craigo S, Hediger ML, Kim S, Zhang C, Grantz KL. Racial/ethnic standards for
or to change the cut-offs for defining SGA or large-for- fetal growth: the NICHD Fetal Growth Studies. Am J Obstet Gynecol 2015; 213:
449.e1–41.
gestational age. In the latter case, these charts could 17. Khalil A, Rezende J, Akolekar R, Syngelaki A, Nicolaides KH. Maternal racial origin
be considered as a benchmark rather than a reference and adverse pregnancy outcome: a cohort study. Ultrasound Obstet Gynecol 2013;
41: 278–285.
chart.

SUPPORTING INFORMATION ON THE INTERNET

The following supporting information may be found in the online version of this article:
Appendix S1 Statistical analysis and justification
Table S1 Characteristics of study population of 5163 pregnancies with paired measurements of estimated fetal
weight and birth weight separated by a maximum of 2 days (Dataset 1)

Copyright © 2018 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2018; 52: 44–51.
EFW and birth-weight charts 51

Table S2 Characteristics of study population of 95 579 pregnancies in Dataset 2

Table S3 Median, 3rd , 5th 10th , 25th , 75th , 90th , 95th and 97th percentiles of estimated fetal weight (EFW) and
birth weight (BW) (in g) for each gestational age (GA) day between 20 and 42 weeks
Table S4 Comparison of 10th , 50th and 90th percentiles of estimated fetal weight (in g) according to
gestational age (GA) between our FMF chart and WHO5 and INTERGROWTH-21st 4 charts
Table S5 Number and percentage of cases in Dataset 2 with estimated fetal weight (EFW) < 3rd , < 5th < 10th ,
> 90th , > 95th and > 97th percentiles of reference range of EFW, according to gestational age (GA), and
respective values for birth weight (BW)
Table S6 Number and percentage of cases in Dataset 2 with birth weight below 3rd , 5th and 10th percentiles of
reference range of birth weight according to gestational age (GA), for iatrogenic and spontaneous births
Table S7 Number and percentage of pregnancies of white and black women in Dataset 2 with estimated fetal
weight (EFW) and birth weight (BW) below 3rd , 5th and 10th percentiles of the appropriate reference range
according to gestational age (GA)

This article has been selected for Journal Club.

A slide presentation, prepared by Dr Maddalena Morlando,
one of UOG's Editors for Trainees, is available online.

Copyright © 2018 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2018; 52: 44–51.
Ultrasound Obstet Gynecol 2018; 52: 44–51
Published online 30 May 2018 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/uog.19073

Gr áficas de la Fetal Medicine Foundation del peso de la poblaci ón fetal y neonatal
RESUMEN
Objetivo Desarrollar gráficas de peso de la población fetal y neonatal. El razonamiento es que, aunque los rangos de referencia
del peso fetal estimado (PFE) son representativos de toda la población, el enfoque tradicional de derivar las gráficas de peso al
nacer (PN) es engañoso, porque una gran proporción de bebés pretérmino provienen de un embarazo con patologı́as. Se propone
que la población de referencia para las gráficas de PN, como en el caso de las gráficas de PFE, deberı́a incluir a todos los bebés en
una edad gestacional determinada, incluidos los que aún están en el útero.
Métodos Para este estudio se utilizaron dos fuentes de datos. Para ambas, los criterios de inclusión fueron el embarazo con feto
único, la fecha estimada mediante la longitud céfalo-caudal del feto entre las semanas 11+0 a 13+6 de gestación, la disponibilidad
de mediciones ecográficas de la circunferencia de la cabeza (CC), la circunferencia abdominal (CA) y longitud del fémur (LF) del
feto, y el nacimiento vivo de un neonato(a) fenotı́picamente normal. El Conjunto de Datos 1 incluı́a una muestra de 5163 medidas
pareadas de PFE y PN; las ecografı́as se realizaron a las 22–43 semanas de gestación y el nacimiento tuvo lugar durante los 2 dı́as
posteriores a la ecografı́a. El PFE se derivó de las mediciones de CC, CA y LF, mediante la fórmula propuesta por Hadlock et al.
en 1985 El Conjunto de Datos 2 incluı́a una muestra de 95 579 embarazos con PFE obtenido mediante biometrı́a fetal ecográfica
rutinaria entre las semanas 20+0 a 23+6 de gestación (n = 45 034), 31+0 a 33+6 semanas (n = 19 224) o 35+0 a 36+6 semanas
(n = 31 321); para el propósito de este estudio, tan solo se incluyeron datos de una de las tres visitas por cada embarazo. En el
desarrollo de los rangos de referencia de PFE y PN en función de la edad gestacional, se hicieron dos suposiciones: primero, que
el PFE y el PN tienen una mediana común, que depende de la edad gestacional; y segundo, que las desviaciones de la mediana
ocurren tanto en el PFE como en el PN y que estas desviaciones están correlacionadas con diferentes niveles de dispersión para
el PFE y el PN, dependiendo de la edad gestacional. Para la inferencia se adoptó un enfoque bayesiano qué combinó los dos
conjuntos de datos mediante un muestreo de Monte-Carlo basado en cadenas de Markov. El modelo ajustado supuso que las
medidas transformadas logarı́tmicamente de las medias del PFE y el PN están relacionadas con la edad gestacional de acuerdo
con una ecuación cúbica y que las desviaciones de la media siguen una distribución gaussiana bivariante.
Resultados En el caso del PFE en el Conjunto de Datos 2, hubo una buena distribución de los valores de los percentiles < 3◦ ,
< 5◦ , < 10◦ , > 90◦ , > 95◦ y > 97◦ del rango de referencia del PFE, de acuerdo con la edad gestacional en todo el rango de edad
gestacional entre las semanas 20 + 0 a 36 + 6. En el caso del PN, se observó una buena distribución de valores tan solo para
los nacimientos con > 39 semanas de gestación. Para los partos pretérmino, particularmente a las 27–36 semanas, el PN estuvo
por debajo de los percentiles 3◦ , 5◦ , y 10◦ en una proporción muy alta de casos, particularmente en los de parto iatrogénico. La
frecuencia de fetos y neonatos pequeños para la edad gestacional en las respectivas gráficas de PFE y PN fue mayor en las mujeres
de raza negra que en las de raza blanca.
Conclusión Se establece una gráfica del PN para todos los bebés a una edad gestacional dada, incluidos los que aún están en el
útero, con lo que se supera el problema de la subestimación de la restricción del crecimiento en el parto pretérmino. Las gráficas
del PN y el PFE tienen una mediana común, pero difieren en los niveles de dispersión de la mediana.

Copyright © 2018 ISUOG. Published by John Wiley & Sons Ltd. ORIGINAL PAPER