**5.2. Fetal weight estimation in twin pregnancies**

Twin pregnancies are more likely to show deviations in fetal growth curve. Conditions such as prematurity, intrauterine growth restriction and fetal malformations are common in multiple gestations, raising the risk of mortality and perinatal morbidity to 3–7 times when compared to single pregnancies [21]. Prematurity is present in approximately 55% of twin pregnancies, with adverse consequences even in short and long term [22].

When comparing the weights of fetuses from twin pregnancies to those of single pregnancies, it is observed that fetuses of twin pregnancies have a lower weight than fetuses of single pregnancies, especially from the end of the second trimester. It is known that this variation between the weights starts at around 28 weeks and at 38 weeks the 50th percentile for a twin pregnancy corresponds to the 10th percentile for a single pregnancy [23], but this difference does not seem to increase neonatal mortality. Therefore, it is argued that the lower weight of twin fetuses, when compared to that of single pregnancies, may be physiological of this condition.

Accuracy in the estimation of fetal weight is of paramount importance for the proper followup of prenatal care and ultrasonography study has been the main tool for this evaluation.

Currently fetal weight estimation by ultrasonography is most commonly performed by the formula of Hadlock et al. [24], which uses two-dimensional measures of cephalic pole, abdominal circumference and femur length. However, studies have shown that formulas using two-dimensional parameters can generate variations of up to 15% in relation to the real weight of the fetus [25].

New methods have been sought to improve the accuracy of fetal weight estimation such as three-dimensional ultrasonography. In the early 2000s, Lee et al. [26] introduced a new sonographic parameter, the fraction limb volume. This parameter is based on evaluation of 50% of bone diaphysis length (arm and thigh).

This method has the advantage of reducing the time spent to perform the test, maintaining a good accuracy for the estimation fetal weight.

In general, the accuracy of estimation weight in twin pregnancies is worse than single pregnancies. Biometric measurement of these fetuses in the third trimester is greatly impaired due to the technical difficulty of examination. When using 33 formulas to assess the accuracy of estimation weight by two-dimensional ultrasonography, 25 of these formulas present a weight variation of less than 10% for single pregnancies, but only 3 of these formulas present the same result for twin pregnancies [27].

An ongoing study that has been developed in multiple pregnancy unit of Federal University of São Paulo has shown that the use of fraction limb volume in twin pregnancies can improve the accuracy of estimation weight in these pregnancies, as well as reduce the time of the examination.

Although evaluation of fetal body volume through the use of magnetic resonance imaging is still considered an expensive method, there is good accuracy in fetal weight estimation, besides being a good predictor in the diagnoses of small fetuses for gestational age when compared to two-dimensional ultrasonography [28].

Estimating weight in twin pregnancies remains a challenge. New research needs to be conducted in search for new methods in order to improve accuracy.

Fetal weight should not be considered when both fetuses are cephalic. In those cases, regardless the fetal weight, a vaginal delivery can be attempted. However, in cephalic/non-cephalic twin pregnancies, the influence of weight on mode of delivery is controversial. Most studies showed worst perinatal outcomes for vaginal deliveries when the second twin was noncephalic and under 1500 g [29, 30].

Weight difference is related to worst neonatal outcomes, regardless the delivery mode [31], and also to unsuccessful attempt of labor [32]. Furthermore, a weight difference above 40% has been associated with higher neonatal mortality rates in vaginal deliveries, regardless fetal presentation, in a retrospective study in 2005 [33].
