Original Article

Gynecol Obstet Invest 2015;80:7884

DOI: 10.1159/000365814

Received: October 29, 2014

Accepted after revision: July 9, 2014
Published online: July 10, 2015

Finding the Best Formula to Predict the

Fetal Weight: Comparison of 18 Formulas
Deniz Esinler Oya Bircan Sertac Esin Elif Gulsah Sahin Omer Kandemir
Serdar Yalvac
Department of Obstetrics and Gynecology, Division of Maternal and Fetal Medicine, Etlik Zubeyde Hanim Maternity
Hospital, Ankara, Turkey

Abstract
Background: To compare the accuracy of 18 formulas in predicting fetal weight and also to make a comparison of these
formulas in low-birth-weight fetuses (<2,500 g) and in fetuses weighing >4,000 g. Methods: Four-hundred-and-ninetyfive pregnant patients were enrolled. The estimated fetal
weight was calculated using 18 different formulas. The mean
percentage error, the mean absolute percentage error and
reliability analysis were used to compare the performance of
the formulas. Results: The Cronbachs alpha was the highest
in the formulas Hadlock I (0.977 (95% CI = 0.9720.980)),
Hadlock III (0.977 (95% CI = 0.9720.980)) and Ott (0.975
(95% CI = 0.9700.979)) in all fetuses. It was the highest in
formulas Ott (0.383 (95% CI = 0.0910.581)), Hadlock IV
(0.371 (95% CI = 0.0740.572)) and Combs (0.369 (95% CI =
0.0710.571)) in fetuses >4,000 g. It was the highest in formulas Coombs (0.957 (95% CI = 0.9400.569)), Ott (0.956
(95% CI = 0.9390.968)) and Hadlock IV (95% CI = 0.956
(0.9380.968)) in fetuses <2,500 g. Conclusion: We noted
that formulas Hadlock I, Hadlock III and Ott may be used to

2015 S. Karger AG, Basel

03787346/15/08020078$39.50/0
E-Mail karger@karger.com
www.karger.com/goi

predict the estimated fetal weight accurately in all fetuses in

our study. Formulas Ott, Hadlock IV and Coombs may be preferred to predict EFW in fetuses <2,500 g and >4,000 g. Better
formulas should be developed to predict the fetal weight in
fetuses >4,000 g.
2015 S. Karger AG, Basel

Introduction

Parallel to the development of ultrasonography machines, antenatal fetal weight estimation became an indispensable part of an obstetric examination. The estimated
fetal weight (EFW) meant physicians could predict an intrauterine growth retardation or macrosomia, and lead to
change of clinical management involving induction or
delay of labor in deciding the method of delivery. There
are several mathematical formulas that use different fetal
structures to predict the fetal weight [115]. The accuracy
and reliability of these formulas may differ and low values
can adversely affect physicians decisions. The performance of formulas may also be different in macrosomic
fetuses (>4,000 g) or low-birth-weight fetuses (<2,500 g)
[1520]. Therefore, it is very important to use the most
reliable and accurate formula when predicting the fetal
weight. Our aim was to compare the accuracy of 18 forDeniz Esinler, MD
Department of Obstetrics and Gynecology, Division of Maternal and Fetal Medicine
Etlik Zubeyde Hanim Maternity Hospital, Kecioren Mah
06010 Ankara (Turkey)
E-Mail denizesinler@gmail.com

Downloaded by:
HINARI Bolivia
198.143.58.33 - 10/20/2015 3:59:18 PM

Key Words
Fetal weight assessment Fetal weight estimation
Birth weight Macrosomia Low birth weight
Prenatal ultrasonography

Table 1. Eighteen formulas to predict the fetal weight

Authors

Parameters

Formula

Campbell [1]

AC

e^(4.564 + 0.282 * AC 0.00331 * AC^2) [kg, cm]

Hadlock VI [3]

HC, AC, FL

10^(1.5662 0.0108 * HC + 0.0468 * AC + 0.171 * FL + 0.00034 * HC^2 0.0003685 *

AC* FL) [g, cm]

Hadlock V [3]

BPD, AC

10^(1.1134 + 0.05845 * AC 0.000604 * AC^2 0.007365 * BPD^2 + 0.000595 * BPD*

AC+ 0.1694 * BPD) [g, cm]

Hadlock II [4]

AC, FL

10^(1.304 + 0.05281 * AC + 0.1938 * FL 0.004 * AC * FL) [g, cm]

Hadlock III [4]

BPD, AC, FL

10^(1.335 0.0034 * AC * FL + 0.0316 * BPD + 0.0457 * AC + 0.1623 * FL) [g, cm]

Hadlock IV [4]

HC, AC, FL

10^(1.326 0.00326 * AC * FL + 0.0107 * HC + 0.0438 * AC + 0.158 * FL) [g, cm]

Hadlock I [4]

BPD, HC, AC, FL

10^(1.3596 + 0.0064 * HC + 0.0424 * AC + 0.174 * FL + 0.00061* BPD * AC 0.00386 *

AC* FL) [g, cm]

Higginbottom [5]

AC

0.0816 * AC^3 [g, cm]

Merz I [6]

BPD, AC

3,200.40479 + 157.07186 * AC + 15.90391 * BPD * BPD [g, cm]

Merz II [6]

AC

0.1 * AC^3 [g, cm]

Ott [7]

HC, AC, FL

10^(2.0661 + 0.04355 * HC + 0.05394 * AC 0.0008582 * HC * AC + 1.2594 * FL/AC)

[kg, cm]

Shepard [9]

BPD, AC

10^(1.7492 + 0.166 * BPD + 0.046 * AC 0.002546 * AC * BPD) [kg, cm]

Thurnau [10]

BPD, AC

(9.337 * BPD * AC) 229 [g, cm]

Warsof [12]

BPD, AC

10^(1.599 + 0.144 * BPD + 0.032 * AC 0.000111 * BPD^2 * AC) [kg, cm]

Vintzileos [11]

BPD, AC

10^(1.879 + 0.084 * BPD + 0.026 * AC) [g, cm]

Combs [2]

HC, AC, FL

0.23718 * AC^2 * FL + 0.03312 * HC^3 [g, cm]

Weiner I [14]

HC, AC, FL

10^(1.6961 + 0.02253 * HC + 0.01645 * AC + 0.06439 * FL) [g, cm]

Weiner II [14]

HC, AC

10^(1.6575 + 0.04035 * HC + 0.01285 * AC) [g, cm]

A retrospective cohort study design was used. The Study Review Board of our hospital approved the study protocol. Four-hundred-and-ninety-five pregnant patients were enrolled retrospectively through our medical records in our maternal fetal medicine
unit from January 2011 to January 2012. Inclusion criteria were:
(1) singleton pregnancies that resulted in a live-born fetus without
congenital anomalies; (2) ultrasound examination <7 days before
the delivery; (3) birth weight >500 g, gestational age >24 weeks.
Women in active labor were excluded.

All the fetal biometric measurements were performed by senior

physicians in the maternal fetal medicine unite of our hospital. The
examinations were performed transabdominally using high-quality ultrasound systems (Voluson 730 Pro, GE Medical Systems,
Milwaukee, USA).
The biparietal diameter (BPD) was measured in a fetal headplane where the cavum septum pellucidum and falx cerebri could
be seen. The cursors were placed from leading edge to leading edge
of the skull bones (outer to inner). The head circumference (HC)
was measured at the outer perimeter of the calvarium, not including the fetal skin, in the same plane as the BPD. An abdominal
circumstance (AC) was measured at the level where the umbilical
vein passes through the liver the image was circular, with symmetric rib images. Femur length (FL) was measured as the entire
bone from metaphysis to metaphysis, perpendicular to the axis of
the bone and not including the epiphyses. Once we had the fetal
biometric measurements (BPD, HC, AC, FL), we calculated the
estimated fetal weight using different formulas (table 1). Gesta-

Finding the Best Formula to Predict the

Fetal Weight

Gynecol Obstet Invest 2015;80:7884

DOI: 10.1159/000365814

mulas in predicting fetal weight and also make a comparison of these formulas in low-birth-weight fetuses
(<2,500 g) and in fetuses weighing >4,000 g.

Materials and Methods

79

Downloaded by:
HINARI Bolivia
198.143.58.33 - 10/20/2015 3:59:18 PM

AC= Abdominal circumference; HC= head circumference; FL= femur length; BPD= biparietal diameter.

Variable
Maternal age, years
Nulliparous, n (%)
Gestational age at delivery, weeks
Time from fetal weight estimation
to delivery, days
Birth weight, g

28.46.0
241 (48.7)
38.23.2
2.31.8
3,044.1984.1

tional age was calculated by the last menstrual period (LMP). LMP
was confirmed by the crown-rump length (CRL). The gestational
age at the delivery and the actual birth weights were obtained from
the hospitals medical records. The newborns were put on an electronic weighing machine after the delivery by senior nurses working in the delivery ward. The results given by the electronic weighing machine were recorded to the patients files.
A subgroup analysis for the fetuses weighing <2,500 g (105
patients) and fetuses weighting >4,000 g (144 patients) was also
performed.
The statistical analyses were performed using Statistics Package for Social Sciences version 21.0 (SPSS, Chicago, III). The percentage error (PE) was calculated using the formula (PE= Estimated birth weight actual birth weight/actual birth weight100).
The absolute percentage error (APE) was calculated using the formula (APE= Estimated birth weight actual birth weight/actual
birth weight 100). Reliability analysis was used to calculate the
Cronbachs alpha value, which measures the correlation between
birth weight and estimated birth weight. The Cronbachs alpha
value shows the degree or the power of the correlation between
the actual birth weight and the predicted birth weight ( 0.9 excellent correlation, 0.7 <0.9 good correlation, 0.6 <0.7 acceptable correlation, 0.5 <0.6 poor correlation, <0.5 unacceptable correlation) [21]. If the Cronbachs alpha value is lower
than 0.5, it is considered unacceptable correction. p values of
<0.05 were considered statistically significant. Values were expressed as the mean standard deviation or mean standard error, unless stated otherwise.

Results

The baseline characteristics of the patients are given in

table2. The mean absolute percentage error (MAPE) and
Cronbachs values of the formulas are summarized in
table 3. The lowest three MAPE values were associated
with Hadlock I (7.7 0.3 (95% CI= 7.08.4)), Ott (7.9
0.3 (95% CI= 7.38.5)) and Hadlock IV (7.9 0.4 (95%
CI= 7.98.5)). Reliability analysis revealed that all formulas estimated the birth weight with significant correlation
with actual birth weight (p < 0.01; table 3). The Cronbachs alpha value showing the degree of correlation was
80

Gynecol Obstet Invest 2015;80:7884

DOI: 10.1159/000365814

highest in Hadlock I (0.977 (95% CI = 0.9720.980)),

Hadlock III (0.977 (95% CI = 0.9720.980)) and Ott
(0.975 (95% CI= 0.9700.979)).
Subgroup analysis of the patients (105 patients) with
birth weights higher than 4,000 g revealed that the lowest
three MAPE values were associated with Merz II (4.8 0.4
(95% CI= 3.95.7)), Hadlock III (5.6 0.5 (95% CI= 4.5
6.6)) and Hadlock II (5.8 0.5 (95% CI= 4.76.9)). The
Cronbachs value was highest in Ott (0.383 (95% CI =
0.0910.581)), Hadlock IV (0.371 (95% CI= 0.0740.572))
and Combs (0.369 (95% CI= 0.0710.571)) (table4).
Subgroup analysis of the patients (144 patients) with
birth weights lower that 2,500 g revealed that the lowest
three MAPE values were associated with Hadlock IV
(9.1 0.7 (95% CI= 7.810.4)), Hadlock I (9.2 0.7 (95%
CI= 7.910.6)) and Ott (9.8 0.7 (8.411.2)). The Cronbachs value showing the degree of correlation was highest in Coombs (0.957 (95% CI= 0.9400.569)), Ott (0.956
(95% CI= 0.9390.968)) and Hadlock IV (95% CI= 0.956
(0.9380.968)) (table5).

Discussion

In our study, Hadlock I (Cronbachs alpha = 0.977

(95% CI= 0.9720.980)) and Hadlock III (Cronbachs alpha= 0.977 (95% CI= 0.972/0.980) and Ott (0.975 (95%
CI = 0.9700.979)) formulas were associated with the
perfect performance in predicting the fetal birth weight
in all patients. Kurmanavicius et al. [22] compared the
Campbell and Wilkin and Shepard formulas, as well as
the two Hadlock and Merz formulas in 5,612 pregnant
women (ranging from 500 to 5,000 g). They noted that the
highest interclass correlation coefficient and the most stable results in all birth weight (BW) groups were generated
with the Hadlock formulas. Both the Hadlock and Campbell formulas had the lowest percent errors in BW groups
between <1,500 and 3,500 g. The Shepard and Merz formulas had lower percent errors in BW groups between
3,501 and >4,000 g. However, in BW groups under 3,500g
they were imprecise. The authors concluded that the
Hadlock formulas showed the most stable results in all of
the weight groups. In the group of fetuses weighing
4,000g (n= 516), the lowest MPE of <5% was obtained
with the Merz formula.
Burd et al. [23] compared the performance of 14 different formulas [24, 712, 14] for prediction of fetal
birth weight. They reported that Hadlock III [4] showed
the best performance according to the bias and precision
method. The sensitivity for detection of small-for-gestaEsinler/Bircan/Esin/Sahin/Kandemir/
Yalvac

Downloaded by:
HINARI Bolivia
198.143.58.33 - 10/20/2015 3:59:18 PM

Table 2. The baseline characteristics of pregnant women included

in the study (n= 495)

Table 3. The MPA, MAPA and the Cronbachs value of the formulas

Formula

Mean percentage error, %

(95% CI)

Mean absolute percentage

error, % (95% CI)

Cronbachs value
(95% CI)

p value*

Campbell [1]
Hadlock VI [3]
Hadlock V [3]
Hadlock II [4]
Hadlock III [4]
Hadlock IV [4]
Hadlock I [4]
Higginbottom [5]
Merz I [6]
Merz II [6]
Ott [7]
Shepard [9]
Thurnau [10]
Warsof [12]
Vintzileos [11]
Combs [2]
Weiner I [14]
Weiner II [14]

1.00.7 (2.3 to 0.3)

0.10.6 (1.2 to 1.4)
29.52.6 (24.2 to 34.7)
0.20.5 (0.8 to 1.2)
0.60.5 (0.4 to 1.6)
2.60.5 (3.6 to 1.7)
1.30.5 (2.3 to 0.4)
4.70.7 (6.0 to 3.3)
16.80.8 (15.1 to 18.4)
5.80.8 (4.2 to 7.4)
1.50.5 (2.4 to 0.6)
0.50.6 (0.6 to 1.6)
13.20.7 (14.6 to 11.8)
4.30.5 (5.3 to 3.2)
6.00.6 (4.7 to 7.3)
3.00.5 (4.0 to 2.0)
20.70.5 (21.7 to 19.8)
12.60.5 (13.5 to 11.7)

10.30.5 (9.3 to 11.2)

9.80.5 (8.8 to 10.7)
37.62.4 (32.9 to 42.3)
8.30.4 (7.6 to 9.0)
8.00.4 (7.2 to 8.7)
7.90.4 (7.9 to 8.5)
7.70.3 (7.0 to 8.4)
11.30.5 (10.3 to 12.3)
20.60.6 (19.3 to 21.8)
11.60.6 (10.3 to 12.9)
7.90.3 (7.3 to 8.5)
9.20.4 (8.5 to 10.0)
18.30.4 (17.3 to 19.1)
9.40.4 (8.6 to 10.2)
12.30.4 (11.4 to 13.1)
8.80.3 (8.1 to 9.5)
21.00.4 (20.2 to 21.9)
13.50.4 (12.7 to 14.3)

0.961 (0.953 to 0.967)

0.967 (0.960 to 0.972)
0.481 (0.380 to 0.565)
0.975 (0.970 to 0.979)
0.977 (0.972 to 0.980)
0.976 (0.972 to 0.980)
0.977 (0.972 to 0.980)
0.961 (0.953 to 0.967)
0.940 (0.928 to 0.949)
0.970 (0.964 to 0.975)
0.975 (0.970 to 0.979)
0.971 (0.966 to 0.976)
0.909 (0.891 to 0.924)
0.971 (0.966 to 0.976)
0.960 (0.952 to 0.966)
0.971 (0.965 to 0.976)
0.946 (0.936 to 0.955)
0.972 (0.976 to 0.976)

<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01

MPA= Mean percentage error; MAPA= mean absolute percentage error; CI= confidence interval.
*p value for reliability analysis.
Cronbachs value= inter-class reliability was calculated between the estimated fetal weight of the formula and actual birth weight.

Table 4. The MPA, MAPA and the Cronbachs value of the formulas (birth weight >4,000 g)

Formula

Mean percentage error, %

(95% CI)

Mean absolute percentage

error, % (95% CI)

Cronbachs value
(95% CI)

p value*

Campbell [1]
Hadlock VI [3]
Hadlock V [3]
Hadlock II [4]
Hadlock III [4]
Hadlock IV [4]
Hadlock I [4]
Higginbottom [5]
Merz I [6]
Merz II [6]
Ott [7]
Shepard [9]
Thurnau [10]
Warsof [12]
Vintzileos [11]
Combs [2]
Weiner I [14]
Weiner II [14]

9.40.5 (10.5 to 8.3)

4.80.7 (6.1 to 3.5)
15.00.4 (15.9 to 14.2)
2.20.8 (3.8 to 0.7)
1.60.7 (3.0 to 0.1)
5.00.7 (6.3 to 3.6)
3.50.7 (4.9 to 2.0)
0.10.9 (2.0 to 1.8)
22.41.2 (19.9 to 24.8)
1.80.6 (3.1 to 0.6)
6.10.7 (7.4 to 4.8)
1.30.8 (2.9 to 0.2)
25.50.5 (26.4 to 24.6)
5.30.8 (6.7 to 3.8)
9.41.0 (7.3 to 11.5)
9.00.6 (10.3 to 7.7)
22.90.8 (24.4 to 21.3)
11.40.8 (13.0 to 9.3)

9.40.5 (8.4 to 10.6)

6.20.5 (5.1 to 7.3)
15.10.4 (14.2 to 15.9)
5.80.5 (4.7 to 6.9)
5.60.5 (4.5 to 6.6)
6.70.5 (5.5 to 7.8)
6.00.5 (4.9 to 7.0)
7.50.7 (6.1 to 8.7)
23.01.1 (20.8 to 25.2)
4.80.5 (3.9 to 5.7)
7.30.5 (6.2 to 8.4)
5.90.5 (4.8 to 7.0)
25.50.5 (24.6 to 26.5)
7.30.6 (6.2 to 8.3)
11.90.7 (10.4 to 13.4)
9.50.6 (8.4 to 10.6)
22.90.8 (21.3 to 24.5)
12.20.7 (10.8 to 13.5)

0.305 (0.023 to 0.523)

0.339 (0.027 to 0.551)
0.105 (0.318 to 0.392)
0.362 (0.061 to 0.567)
0.350 (0.043 to 0.588)
0.371 (0.074 to 0.572)
0.353 (0.047 to 0.560)
0.321 (0.001 to 0.539)
0.291 (0.043 to 0.519)
0.337 (0.025 to 0.550)
0.383 (0.091 to 0.581)
0.311 (0.015 to 0.532)
0.261 (0.088 to 0.498)
0.314 (0.009 to 0.534)
0.271 (0.073 to 0.505)
0.369 (0.071 to 0.571)
0.248 (0.107 to 0.489)
0.320 (0.001 to 0.538)

<0.05
<0.05
NS
<0.05
<0.05
<0.05
<0.05
<0.05
<0.05
<0.05
<0.05
<0.05
NS
<0.05
NS
<0.05
NS
<0.05

Finding the Best Formula to Predict the

Fetal Weight

Gynecol Obstet Invest 2015;80:7884

DOI: 10.1159/000365814

81

Downloaded by:
HINARI Bolivia
198.143.58.33 - 10/20/2015 3:59:18 PM

MPA= Mean percentage error; MAPA= mean absolute percentage error; CI= confidence interval.
*p value for reliability analysis.
Cronbachs value= inter-class reliability was calculated between the estimated fetal weight of the formula and actual birth weight.

Table 5. The MPA, MAPA and the Cronbachs value of the formulas (birth weight <2,500 g)

Formula

Mean percentage error, %

(95% CI)

Mean absolute percentage

error, % (95% CI)

Cronbachs value
(95% CI)

p value*

Campbell [1]
Hadlock VI [3]
Hadlock V [3]
Hadlock II [4]
Hadlock III [4]
Hadlock IV [4]
Hadlock I [4]
Higginbottom [5]
Merz I [6]
Merz II [6]
Ott [7]
Shepard [9]
Thurnau [10]
Warsof [12]
Vintzileos [11]
Combs [2]
Weiner I [14]
Weiner II [14]

7.31.5 (4.4 to 10.3)

5.41.5 (2.6 to 8.4)
98.35.6 (87.3 to 109.4)
2.81.1 (0.5 to 5.0)
2.91.0 (0.8 to 5.1)
1.81.0 (2.2 to 1.8)
0.91.0 (1.1 to 2.9)
7.71.3 (10.3 to 4.9)
13.11.7 (9.8 to 16.4)
13.91.9 (10.0 to 17.6)
3.71.0 (1.6 to 5.7)
3.81.2 (1.4 to 6.3)
4.21.4 (1.4 to 7.0)
1.71.2 (4.0 to 0.6)
5.91.3 (3.3 to 8.4)
4.61.0 (2.6 to 6.7)
16.91.0 (18.9 to 14.9)
13.00.9 (14.8 to 11.3)

14.81.0 (12.8 to 16.9)

14.20.9 (12.3 to 16.1)
98.45.6 (87.5 to 109.0)
10.60.7 (9.1 to 12.0)
10.00.7 (8.6 to 11.4)
9.10.7 (7.8 to 10.4)
9.20.7 (7.9 to 10.6)
14.00.9 (12.1 to 15.9)
18.91.2 (16.4 to 21.3)
20.11.5 (17.2 to 23.0)
9.80.7 (8.4 to 11.2)
11.90.8 (10.2 to 13.5)
12.90.9 (11.0 to 14.9)
10.70.8 (9.1 to 12.2)
12.80.8 (11.2 to 14.5)
10.30.7 (8.9 to 11.7)
17.70.9 (15.9 to 19.5)
14.30.7 (12.9 to 15.7)

0.929 (0.901 to 0.949)

0.927 (0.898 to 0.947)
0.652 (0.517 to 0.750)
0.949 (0.930 to 0.964)
0.952 (0.933 to 0.966)
0.956 (0.938 to 0.968)
0.955 (0.937 to 0.967)
0.925 (0.896 to 0.946)
0.907 (0.871 to 0.933)
0.934 (0.909 to 0.953)
0.956 (0.939 to 0.968)
0.937 (0.913 to 0.955)
0.925 (0.895 to 0.946)
0.938 (0.914 to 0.955)
0.935 (0.910 to 0.953)
0.957 (0.940 to 0.569)
0.920 (0.889 to 0.943)
0.949 (0.929 to 0.963)

<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01

tional-age (SGA) fetuses ranged from 72 to 100%, and

specificity was 4188%. Hadlock III showed the optimal
sensitivity/specificity trade-off for detection of SGA.
However, the results of this study should be carefully
interpreted, since the study had a small sample size (81
patients) and patients included in the study had
preeclampsia.
It is well known that fetuses with macrosomia are associated with higher rates of shoulder dystocia, brachial
plexus, fetal asphyxia, prolonged labor, surgical delivery
and postpartum hemorrhagia [24, 25]. Therefore, the
precise prediction of fetal weight in macrosomic fetuses
is extremely important to prevent these complications.
Combs et al. [16] compared 31 sonographic fetal weightestimation models for the prediction of fetal macrosomia
within two weeks of delivery in a group of 165 diabetic
women. They noted that the model of Ott [7] yielded the
best results with a sensitivity (45%) and positive predictive value (81%) for identifying macrosomia (>4,000 g).
However, none of the formulae was superior to others in
assessing weight in fetal macrosomia.
Hart et al. [18] recently developed a new formula including AC, FL, HC and maternal weight to predict the
EFW in macrosomic fetuses, and compared this formula
82

Gynecol Obstet Invest 2015;80:7884

DOI: 10.1159/000365814

with seven different formulas (Hadlock I [4], Hadlock II

[4], Hadlock III [4], Hadlock IV [4], Warsof [13], Campbell [1] and Merz [6]). They concluded that the new formula (the lowest mean absolute percentage error 3.69%,
range 0.0513.57) allows better weight estimation in the
macrosomic fetus. The main drawback of our study was
that we could not use Hart formulas to calculate fetal
weight because this needed the maternal weight at USG
examination. Unfortunately, the maternal weight during
USG examination was not available in our study.
Recently, Hoopmann et al. [19] compared the performance of 36 different weight estimation formulae in fetuses with macrosomia. They noted that among all formulae the mean detection rates for fetuses with a BW
4,000, 4,300 and 4,500 g were low (29, 24 and 22%,
respectively) for impressively high false positive rates (12
and 7%). Interestingly, they noted that the detection rate
for fetuses with a birth weight (BW) 4,000 and 4,500 g
was 100 and 0% with the Hart formula. MAPE was smallest with the Hart formula (MAPE 3.9%).
Faschingbauer et al. [17] compared the accuracy of 10
commonly used weight estimation formulas in a group
of fetuses with extreme macrosomia (4,500 g). They
reported that MPE showed the largest deviation from
Esinler/Bircan/Esin/Sahin/Kandemir/
Yalvac

Downloaded by:
HINARI Bolivia
198.143.58.33 - 10/20/2015 3:59:18 PM

MPA= Mean percentage error; MAPA= mean absolute percentage error; CI= confidence interval.
*p value for reliability analysis.
Cronbachs value= inter-class reliability was calculated between the estimated fetal weight of the formula and actual birth weight.

zero with the Schild formula (MPE 15.43%) and the

Shepard formula (MPE +6.08%) and was closest to zero
with the Hadlock II formula (MPE 5.34%). They noted
that the lowest MAPE was found for the Merz formula
(7.23%).
Interestingly, in our study Hadlock V, Thurnau, Vintzileos and Weiner I failed to predict the fetal weight accurately in macrosomic fetuses (p> 0.5; table4). However,
all formulas except the above-mentioned ones predicted
the fetal weight significantly in macrosomic fetuses (p<
0.5; table4). In our study, Ott (0.383 (95% CI= 0.091
0.581)), Hadlock IV (0.371 (95% CI= 0.0740.572)) and
Combs (0.369 (95% CI= 0.0710.571)) produced the best
three performance in predicting the fetal weight of fetus
weighting >4,000 g. However, we noted that all formulas
produced Cronbachs alpha values lower than 0.5 in fetuses >4,000 g. These results show us that the better formulas should be developed to predict the fetal birth
weight in fetuses weighting >4,000 g.
It is obvious that low-birth-weight infants (<2,500 g)
are at an increased risk of perinatal morbidity and mortality [2628]. Therefore, accurate EFW estimation is
also very important in fetuses with low birth weight. In
our study, we noted that the Coombs (0.957 (95% CI=
0.9400.569)), Ott (0.956 (95% CI= 0.9390.968)) and
Hadlock IV (0.956 (95% CI= 0.956 (0.9380.968)) formulas produced the best three performances in predicting EFW in low-birth-weight fetuses (table5). In the literature there is a paucity of data comparing the performance of fetal weight estimation formulas in
low-birth-weight fetuses. Lalys et al. [20] developed a
two-step method to predict the fetal birth weight in
<2,500 g fetuses. In the first step, small fetuses are identified using Hadlocks formula [4] based on third-trimester sonogram measurements. Hadlock EFW values equal
to or lower than 2,800 g define small fetuses. In the second step, a new formula is used to predict the fetal weight
for small fetuses. The authors noted that their model gave
better accuracy of EFW due to the fact that both systematic and random errors were smaller than those obtained
using only the Hadlocks formula (2.3 9.6 vs. 7.2
11.4). Woo and Wan [15] developed a new formula
(EFW= (1.4 FL (cm) BPD (cm) BPD (cm)) 200
g) to predict birth weight in small-for-gestational-age
(SGA) fetuses. They noted that the formula produced a
standard deviation of 98 g/kg in SGA fetuses. We applied
this formula to our patients and found that the MPA,
MAPA and Cronbachs values were 6.3 1.0 (95% CI=
8.3 to 4.3), 10.6 0.7 (95% CI= 9.112.0) and 0.955
(95% CI= 0.9380.968), respectively. The Woo formula

produced good performance but could not enter among

the highest three performances (table4).
Another important issue is the time interval between
the date of USG and the date of birth. If the time increases, the MAPA value of the formulas increases. Basha et al.
[29] investigated the effect of time interval on MAPA values. They reported that MAPA of ultrasound fetal weight
estimation (Hadlock I formula) for all infants was 6.5%;
however, it was 6.1 and 8.2% if the fetus was delivered
within seven days or 814 days of sonography, respectively. In our study, the maximum time interval between
the date of USG and the date of birth was less than seven
days. Additionally, the mean time interval between the
date of USG and the date of birth was 2.3 1.8 (days).
We did not calculate the detection rates of formulas for
macrosomic or low-birth-weight fetuses. We believe that
this is not an accurate method to compare the performance of the formulas. To detect macrosomic or lowbirth-weight fetuses alone is not enough; the ideal formula should detect macrosomic or small fetuses but
should also precisely predict the actual birth weight.
Therefore, instead of reporting detection rates, we preferred to calculate reliability scale analysis. This analysis
produces a Cronbachs alpha value representing whether
the prediction of the formula is far from or close to the
actual birth weight.
The main problem in developing an accurate formula
is that the data retrieved from a population may not be
suitable for other populations. From our point of view,
the best method for clinicians is to encourage evaluating
the performance of their chosen formulas in their populations. The use of EFW formulae should not be random
[30].
In conclusion, formulas Hadlock I, Hadlock III and
Ott produced the best three performances in predicting
the estimated fetal weight (EFW) accurately in all fetuses
in our study. The formulas Ott, Hadlock IV and Coombs
produced the best three performances in predicting the
EFW in fetuses <2,500 g. Although the formulas Ott,
Hadlock IV and Coombs produced the best three performances in predicting the EFW in fetuses >4,000 g, the
Cronbachs values were below 0.5, which showed that the
correlation was not good. Therefore, better formulas
should be developed to predict fetal weight in fetuses
>4,000 g.

Finding the Best Formula to Predict the

Fetal Weight

Gynecol Obstet Invest 2015;80:7884

DOI: 10.1159/000365814

None declared.

83

Downloaded by:
HINARI Bolivia
198.143.58.33 - 10/20/2015 3:59:18 PM

Disclosure Statement

References

84

12

13

14

15

16

17

18

19

20

21

femur, and thigh circumference measurements. Am J Obstet Gynecol 1987; 157: 410
414.
Warsof SL, Gohari P, Berkowitz RL, Hobbins
JC: The estimation of fetal weight by computer-assisted analysis. Am J Obstet Gynecol
1977;128:881892.
Warsof SL, Wolf P, Coulehan J, Queenan JT:
Comparison of fetal weight estimation formulas with and without head measurements.
Obstet gynecol 1986;67:569573.
Weiner CP, Sabbagha RE, Vaisrub N, Socol
ML: Ultrasonic fetal weight prediction: role of
head circumference and femur length. Obstet
Gynecol 1985;65:812817.
Woo JS, Wan MC: An evaluation of fetal
weight prediction using a simple equation
containing the fetal femur length. J Ultrasound Med 1986;5:453457.
Combs CA, Rosenn B, Miodovnik M, Siddiqi
TA: Sonographic EFW and macrosomia: is
there an optimum formula to predict diabetic
fetal macrosomia? J Matern Fetal Med 2000;
9:5561.
Faschingbauer F, Beckmann MW, Goecke
TW, Yazdi B, Siemer J, Schmid M, Mayr A,
Schild RL: A new formula for optimized
weight estimation in extreme fetal macrosomia (4500 g). Ultraschall Med 2012;33:480
488.
Hart NC, Hilbert A, Meurer B, Schrauder M,
Schmid M, Siemer J, Voigt M, Schild RL: Macrosomia: a new formula for optimized fetal
weight estimation. Ultrasound Obstet Gynecol 2010;35:4247.
Hoopmann M, Abele H, Wagner N, Wallwiener D, Kagan KO: Performance of 36 different
weight estimation formulae in fetuses with
macrosomia. Fetal Diagn Ther 2010; 27: 204
213.
Lalys L, Pineau JC, Guihard-Costa AM: Small
and large foetuses: identification and estimation of foetal weight at delivery from thirdtrimester ultrasound data. Early Hum Dev
2010;86:753757.
George D, Mallery P: SPSS for Windows Step
by Step: A Simple Guide and Reference 11.0
update, ed 4. Boston, Allyn & Bacon, 2003.

Gynecol Obstet Invest 2015;80:7884

DOI: 10.1159/000365814

22 Kurmanavicius J, Burkhardt T, Wisser J,

Huch R: Ultrasonographic fetal weight estimation: accuracy of formulas and accuracy of
examiners by birth weight from 500 to 5000 g.
J Perinat Med 2004;32:155161.
23 Burd I, Srinivas S, Pare E, Dharan V, Wang E:
Is sonographic assessment of fetal weight influenced by formula selection? J Ultrasound
Med 2009;28:10191024.
24 Weissmann-Brenner A, Simchen MJ, Zilberberg E, Kalter A, Weisz B, Achiron R, Dulitzky M: Maternal and neonatal outcomes of
macrosomic pregnancies. Med Sci Monit
2012;18:PH77PH81.
25 Saucedo R, Zarate A, Basurto L, Hernandez
M, Puello E, Campos S, Moreno E: Women
with gestational diabetes develop glucose intolerance with high frequency within one year
postpartum. Gynecol Obstet Invest 2012; 73:
5862.
26 McCormick MC, Litt JS, Smith VC, Zupancic
JA: Prematurity: an overview and public
health implications. Annu Rev Public Health
2011;32:367379.
27 Mutz-Dehbalaie I, Scheier M, Jerabek-Klestil
S, Brantner C, Windbichler GH, Leitner H,
Egle D, Ramoni A, Oberaigner W: Perinatal
mortality and advanced maternal age. Gynecol Obstet Invest 2014;77:5057.
28 Fukuda M, Tanaka T, Kamada M, Hayashi A,
Yamashita Y, Terai Y, Ohmichi M: Comparison of the perinatal outcomes after laparoscopic myomectomy versus abdominal myomectomy. Gynecol Obstet Invest 2013; 76:
203208.
29 Basha AS, Abu-Khader IB, Qutishat RM, Amarin ZO: Accuracy of sonographic fetal weight
estimation within 14 days of delivery in a Jordanian population using hadlock formula 1.
Med Princ Pract 2012;21:366369.
30 Benavides-Serralde A, Hernandez-Andrade
E, Fernandez-Lara A, Figueras F, MorenoAlvarez O, Camargo-Marin L, AcevedoGallegos S, Gallardo-Gaona J, VelazquezTorres B, Guzman-Huerta M: Accuracy of
different equations for estimating fetal
weight. Gynecol Obstet Invest 2011; 72: 264
268.

Esinler/Bircan/Esin/Sahin/Kandemir/
Yalvac

Downloaded by:
HINARI Bolivia
198.143.58.33 - 10/20/2015 3:59:18 PM

1 Campbell S, Wilkin D: Ultrasonic measurement of fetal abdomen circumference in the

estimation of fetal weight. Br J Obstet Gynaecol 1975;82:689697.
2 Combs CA, Jaekle RK, Rosenn B, Pope M,
Miodovnik M, Siddiqi TA: Sonographic estimation of fetal weight based on a model of
fetal volume. Obstet Gynecol 1993; 82: 365
370.
3 Hadlock FP, Harrist RB, Carpenter RJ, Deter
RL, Park SK: Sonographic estimation of fetal
weight. The value of femur length in addition
to head and abdomen measurements. Radiology 1984;150:535540.
4 Hadlock FP, Harrist RB, Sharman RS, Deter
RL, Park SK: Estimation of fetal weight with
the use of head, body, and femur measurements a prospective study. Am J Obstet Gynecol 1985;151:333337.
5 Higginbottom J: Estimation of fetal weight.
Ultrasound Med Biol 1977;3:59.
6 Merz E, Lieser H, Schicketanz KH, Harle J:
[Intrauterine fetal weight assessment using
ultrasound. A comparison of several weight
assessment methods and development of a
new formula for the determination of fetal
weight]. Ultraschall Med 1988;9:1524.
7 Ott WJ, Doyle S, Flamm S, Wittman J: Accurate ultrasonic estimation of fetal weight. Prospective analysis of new ultrasonic formulas.
Am J Perinatol 1986;3:307310.
8 Rose BI, McCallum WD: A simplified method
for estimating fetal weight using ultrasound
measurements. Obstet Gynecol 1987;69:671
675.
9 Shepard MJ, Richards VA, Berkowitz RL,
Warsof SL, Hobbins JC: An evaluation of two
equations for predicting fetal weight by ultrasound. Am J Obstet Gynecol 1982;142:4754.
10 Thurnau GR, Tamura RK, Sabbagha R, Depp
OR 3rd, Dyer A, Larkin R, Lee T, Laughlin C:
A simple estimated fetal weight equation
based on real-time ultrasound measurements
of fetuses less than thirty-four weeks gestation. Am J Obstet Gynecol 1983;145:557561.
11 Vintzileos AM, Campbell WA, Rodis JF,
Bors-Koefoed R, Nochimson DJ: Fetal weight
estimation formulas with head, abdominal,