- Email: [email protected]
The Accuracy of Estimating Fetal Weight and Inter-Twin Weight Discordance by Ultrasound in Twin Pregnancies in Women With Increased Body Mass Index
OBSTETRICS
The Accuracy of Estimating Fetal Weight and Inter-Twin Weight Discordance by Ultrasound in Twin Pregnancies in Women With Increased Body Mass Index Sawsan Al-Obaidly, MBBS,1 Jacqueline Parrish,1 Kellie E. Murphy, MD,1 Phyllis Glanc, MD,2 Cynthia Maxwell, MD1 Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, Mount Sinai Hospital, University of Toronto, Toronto ON
1
Department of Medical Imaging, Sunnybrook Health Sciences Centre, University of Toronto, Toronto ON
2
Abstract Objectives: The purpose of this study was to determine whether increased maternal pre-pregnancy BMI decreases the ultrasound accuracy of fetal weight estimation and inter-twin weight discordance in twin pregnancies compared with women with normal BMI. Methods: We conducted a retrospective cohort study of women with a known pre-pregnancy or early pregnancy BMI who delivered a viable twin pregnancy after 28 weeks’ gestation between 2008 and 2011, and who had an ultrasound examination for fetal weight estimation within two weeks of delivery. The sonographically estimated fetal weight (EFW) was compared with the actual weight for each twin, and inter-twin weight discordance (defined as a weight difference between twins of more than 25%) was stratified for the patient’s BMI. We sought to determine if EFW and inter-twin weight discordance were affected if delivery occurred at eight to 14 days after ultrasound compared to within seven days of ultrasound. Results: A total of 300 twin pregnancies with known pre-pregnancy maternal BMI were identified. Of these, 179 were underweight or of normal weight (BMI < 25 kg/m2), 67 were overweight (BMI 25 to 29.9 kg/m2), and 54 were obese (BMI ≥ 30 kg/m2). Ultrasound accuracy among all BMI groups were compared when done between 8 and 14 days and within seven days from delivery. There was a significant increasing trend in mean absolute percent errors for both twins in the obese compared to normal weight (P < 0.05) if delivery happened between eight and 14 days from ultrasound. This difference was diminished if the ultrasound was performed within seven days of delivery. The ultrasound detection of inter-twin weight discordance was similar among the three BMI groups. Key Words: Body mass index, BMI, obesity, estimated fetal weight, EFW, mean absolute percentage error, MAPE, inter-twin weight discrepancy, mean absolute difference, MAD Competing Interests: None declared. Received on June 21, 2014 Accepted on November 26, 2014
696 l AUGUST JOGC AOÛT 2015
Conclusion: Estimation of fetal weight using ultrasound in obese women with twin pregnancies appears to be more reliable when performed close to delivery.
Résumé Objectifs : Cette étude avait pour objectif de déterminer si la présence d’un IMC maternel prégrossesse accru entraînait une baisse de la précision de l’échographie pour ce qui est de l’estimation du poids fœtal et de la discordance intergémellaire en matière de poids dans le cadre de grossesses gémellaires, par comparaison avec des femmes présentant un IMC normal. Méthodes : Nous avons mené une étude de cohorte rétrospective portant sur des femmes qui présentaient un IMC prégrossesse (ou aux débuts de la grossesse) connu, qui ont accouché après 28 semaines de gestation à la suite d’une grossesse gémellaire viable entre 2008 et 2011, et qui ont subi un examen échographique visant l’estimation du poids fœtal dans les deux semaines ayant précédé l’accouchement. Le poids fœtal estimé (PFE) par échographie a été comparé au poids réel de chacun des jumeaux, puis la discordance intergémellaire en matière de poids (définie comme une différence de poids entre les jumeaux de plus de 25 %) a été stratifiée en fonction de l’IMC de la patiente. Nous avons cherché à déterminer si le PFE et la discordance intergémellaire en matière de poids avaient été affectés lorsque l’accouchement était survenu de 8 à 14 jours à la suite de l’échographie, par comparaison avec un accouchement étant survenu dans les sept jours de la tenue de l’échographie. Résultats : Nous avons pu identifier, au total, 300 grossesses gémellaires pour lesquelles l’IMC maternel prégrossesse était connu : 179 femmes présentaient une insuffisance pondérale ou un poids normal (IMC < 25 kg/m2), 67 présentaient une surcharge pondérale (IMC = de 25 à 29,9 kg/m2) et 54 étaient obèses (IMC ≥ 30 kg/m2). Dans tous les groupes d’IMC, la précision de l’échographie menée entre 8 et 14 jours avant l’accouchement a été comparée à celle de l’échographie menée dans les sept jours de l’accouchement. Une tendance à la hausse considérable en matière d’erreur absolue moyenne en pourcentage pour les deux jumeaux a été constatée chez les femmes obèses,
Estimating Fetal Weight and Inter-Twin Weight Discordance by Ultrasound in Twin Pregnancies in Women With Increased BMI
par comparaison avec les femmes de poids normal (P < 0,05), lorsque l’accouchement avait eu lieu de 8 à 14 jours à la suite de l’échographie. Cette différence était moindre lorsque l’échographie avait été menée dans les sept jours de l’accouchement. La détection par échographie d’une discordance intergémellaire en matière de poids était semblable dans les trois groupes d’IMC. Conclusion : Chez les femmes obèses qui connaissent une grossesse gémellaire, l’estimation du poids fœtal par échographie semble être plus fiable lorsqu’elle est menée peu avant l’accouchement.
J Obstet Gynaecol Can 2015;37(8):696–701
INTRODUCTION
D
uring the last two decades, the rates of multiple pregnancies across the world and in Canada have increased, mainly because of the expanded use of assisted reproduction.1,2 There is an increased incidence of intrauterine growth restriction in one or both fetuses in twin pregnancies.3 Having accurate and timely identification of discordant growth is important because of its relationship to placental disorders, risk of IUGR, and perinatal morbidity and mortality.4 A significant weight discordance is assessed by taking the estimated fetal weight difference at 20% to 25%.5–7 Previous reports have classified a weight discordance of 25% as severe, with an associated overall risk of fetal death that is five-fold greater than in concordant twins.8,9 Theoretical reasons for the increased risk include a possible increased risk of head entrapment during delivery and birth trauma. Discordant twins may have a higher baseline neonatal mortality rate even after controlling for fetal growth and gestational age.10 On the other hand, there is another potential risk factor in twin pregnancies related mainly to maternal obesity. The WHO categorized the BMI into four different categories (underweight < BMI 18.5 kg/m2, normal weight BMI 18.5 kg/m2 to 24.9 kg/m2, overweight 25 kg/m2 to 29.9 kg/m2, and obese BMI ≥ 30 kg/m2). The number of patients with a BMI in the overweight and obese ranges has increased in the past two decades.11 Previous studies have noted the adverse effects maternal obesity can have
ABBREVIATIONS EFW
estimated fetal weight
MAD
mean absolute difference
MAPE
mean absolute percentage error
on the accuracy of sonographic fetal anatomy as well as on estimation of fetal weight.12,13 The purpose of this study was to determine whether increased maternal pre-pregnancy BMI in women with a twin pregnancy decreases the accuracy of fetal weight estimation by ultrasound and the estimation of inter-twin weight discordance when compared with women with a normal BMI and a twin pregnancy. METHODS
We performed a retrospective cohort study of women who delivered a viable twin pregnancy at > 28 weeks’ gestation at our institution during the period from 2008 to 2011 and who had a known pre-pregnancy or first trimester BMI. We included in the study women who had fetal weight estimation by ultrasound within 14 days of delivery. For purposes of comparison, study subjects were divided into two groups: group 1 had estimation of fetal weight by ultrasound within 8 to 14 days of delivery, and group 2 had the estimation of fetal weight within seven days of delivery. None of the neonates had any congenital anomaly, and monochorionic twins affected by twin–twin transfusion syndrome either resolved or were treated before delivery. Ultrasound biometric measurements were performed by registered sonographers, and images were reviewed either by radiologists who specialized in pregnancy imaging or by maternal–fetal medicine specialists. The standardized fetal weight estimation method used in our unit is the Hadlock regression formula, which incorporates the biparietal diameter, head circumference, abdominal circumference, and femur length.14 The accuracy of the estimations was assessed using several measures. The Pearson correlation coefficient was used to correlate EFW with the actual birth weight. The MAD was defined as the mean of actual birth weight minus the EFW (in grams) across the three different BMI groups. Absolute percentage errors were calculated by dividing the difference between sonographically estimated fetal weight and actual birth weight by the actual birth weight, and expressing the result as a percentage. Inter-twin weight discordance was defined as a weight disparity of more than 25%, using the larger twin as the index. All analyses were completed using Stata/IC release 12.1 for Windows (StataCorp LLP, College Station, TX). Linear regression and analysis of variance (ANOVA) was used to determine whether sonographic prediction of the actual birth weight varied according to BMI. Student t test was AUGUST JOGC AOÛT 2015 l 697
Obstetrics
Table 1. Demographic data of twin pregnancies in the three categories of pre-pregnancy BMI Normal (BMI < 25 kg/m2) n = 179
Overweight (BMI 25 to 29.9 kg/m2) n = 67
Obese (BMI ≥ 30 kg/m2) n = 54
P
Maternal age ≥ 35 years, n (%)
101 (56)
35 (52)
22 (40)
NS
Nulliparity, n (%)
109 (60)
40 (59)
27 (50)
NS
Maternal age, years, mean (SD)
34.4 (4.7)
34.5 (5.3)
33.2 (5.8)
NS
Gestational age at delivery, weeks, mean (SD)
35.8 (2.1)
35.4 (2.4)
35 (2.5)
NS
Total n = 300
Time from US to birth, days, mean (SD)
8 (4.6)
9 (5)
11 (3)
NS
Pre-pregnancy BMI, kg/m2, mean (SD)
21.4 (1.9)
27.2 (1.5)
37.6 (8)
< 0.005
BMI at delivery, mean (SD)
28.6 (3)
34.2 (2.8)
41.54 (8.3)
< 0.005
Dichorionic, diamnionic twins, n (%)
143 (79)
48 (71)
44 (81)
NS
Male sex, n (%)
160 (44)
74 (55)
50 (46)
NS
Vertex presentation of twin A, n (%)
126 (70)
36 (53)
26 (48)
NS
Vertex presentation of twin B, n (%)
78 (44)
26 (39)
15 (28)
NS
Intrauterine growth restriction, n (%)
14 (8)
5 (7)
1 (2)
NS
Twin-twin transfusion syndrome, n (%)
2 (1)
0
1 (2)
NS
used to compare continuous variables between groups, and chi-square was used for categorical variables. For all analyses, a P value of < 0.05 was considered significant.
Ultrasonographic EFW in group 1 was more accurate when the BMI was < 30 kg/m2 than when BMI was 30 kg/m2 or greater (Table 2).
The study was approved by the Mount Sinai Hospital Research Ethics Board.
There were statistically significant differences in both MAPE and MAD for both twins between the overweight and normal weight groups and between the obese and normal weight groups (Table 2). However these differences were absent if delivery occurred within seven days of the ultrasound (Table 3).
RESULTS
Electronic records of 1228 women who delivered viable twins were reviewed, and of these 300 women had available pre-pregnancy BMI or earliest BMI recorded in the first trimester and had an ultrasound for fetal weight estimation performed in our institution within 14 days of delivery (8 to 14 days group and 7 days group). There were 150 patients in each group. The demographic data collected are presented in Table 1. There were no significant differences between the groups except for the expected differences in mean pre-pregnancy BMI and mean BMI at delivery. To compare the difference between actual birth weight and last EFW by ultrasound, we measured the mean absolute percentage error and compared this between groups. The MAPE for twin A in group 1 (ultrasound performed 8 to 14 days before delivery) in the normal, overweight, and obese groups was 6%, 4%, and 20%, respectively, but in group 2 (ultrasound performed within 7 days of delivery) it was 4%, 2%, and 3%, respectively. The MAPE for twin B in group 1 in the normal, overweight, and obese groups was 4%, 5%, and 13%, respectively, and in group 2 it was < 1%, 3%, and 6%, respectively. 698 l AUGUST JOGC AOÛT 2015
The accuracy of ultrasonography in predicting inter-twin weight discordance is shown in Table 4. Overall, across the three BMI groups there were no significant differences in the inter-twin weight discordance assessed by ultrasound and identified by birth weights. DISCUSSION
Several reports have examined the accuracy of ultrasound in estimating fetal weight.15–17 Some studies have compared its accuracy in twin and singleton pregnancies and have found greater accuracy of ultrasound in singletons; for example, Danon et al. demonstrated that the accuracy of the ultrasonographic estimation of fetal weight seems to be less in twin pregnancies than in singletons, and that the accuracy is less for the second twin than for the first twin.18 Heer et al. showed in singleton pregnancies that if the time interval between weight estimation and delivery was more than seven days, the accuracy of the weight estimation will be reduced.19
Estimating Fetal Weight and Inter-Twin Weight Discordance by Ultrasound in Twin Pregnancies in Women With Increased BMI
Table 2. Sonographic prediction of fetal weight by increasing BMI category for twins A and B when delivery occurred 8 to 14 days after ultrasound
N
Estimated fetal weight, g (SD)
Birth weight, g (SD)
Mean absolute difference, g (SD)
Mean absolute error (SD)
Mean absolute percentage error (SD)
P
Pearson correlation coefficient*
Twin A BMI category
< 25 kg/m
89
2384 (536)
2531 (453)
147 (83)
0.06 (0.2)
5.8 (18)
0.05†
0.83
25 to 29.9 kg/m2
36
2348 (519)
2457 (517)
109 (2)
0.04 (0)
4.4 (0.4)
0.2‡
0.9
≥ 30 kg/m
25
1984 (576)
2475 (575)
491 (1)
0.2 (0)
19.8 (0.2)
0.01§
0.9
89
2424 (454)
2514 (509)
90 (55)
0.04 (0.1)
3.6 (10)
0.004†
0.85
36
2407 (529)
2525 (568)
118 (39)
0.05 (0.07)
4.7 (6.8)
0.08‡
0.9
25
2013 (638)
2321 (728)
308 (90)
0.1 (0.1)
13 (12)
0.001§
0.82
2
2
Twin B BMI category
< 25 kg/m2 25 to 29.9 kg/m
2
≥ 30 kg/m2
*Correlation between estimated fetal weight and actual birth weight. †Overall ANOVA of MAPE comparing overweight and obese with normal BMI group. ‡Comparing overweight with normal BMI. §Comparing obese with normal BMI. SD: standard deviation
Table 3. Sonographic prediction of fetal weight by increasing BMI category for twins A and B when delivery occurred within 7 days of ultrasound Estimated fetal weight, g (SD)
Birth weight, g (SD)
Mean absolute difference, g (SD)
Mean absolute error (SD)
Mean absolute percentage error (SD)
P
Pearson correlation coefficient*
90
2335 (586)
2442 (542)
107 (44)
0.04 (0.08)
4.4 (8)
0.5†
0.76
31
2387 (646)
2423 (668)
36 (22)
0.015 (0.03)
1.5 (3)
0.7‡
0.88
29
2266 (665)
2335 (635)
69 (30)
0.03 (0.04)
3.0 (4)
0.5§
0.9
< 25 kg/m2
90
2363 (585)
2370 (509)
7 (76)
0.003 (0.14)
0.3 (14)
0.4†
0.12
25 to 29.9 kg/m2
31
2394 (576)
2330 (608)
64 (32)
0.027 (0.05)
2.74 (5)
0.6‡
0.91
≥ 30 kg/m
29
2498 (625)
2343 (624)
155 (1)
0.066 (0.001)
6.6 (0.1)
0.7§
0.9
N Twin A BMI category
< 25 kg/m2 25 to 29.9 kg/m
2
≥ 30 kg/m2 Twin B BMI category
2
*Correlation between estimated fetal weight and actual birth weight. †Overall ANOVA of MAPE comparing overweight and obese with normal BMI group. ‡Comparing overweight with normal BMI. §Comparing obese with normal BMI. SD: standard deviation
AUGUST JOGC AOÛT 2015 l 699
Obstetrics
Table 4. Ultrasound estimated fetal weight discordance and actual birth weight discordance in different BMI categories
Total
Estimated fetal weight discordance > 25%
%
Actual birth weight discordance > 25%
%
P
BMI
300
< 25
179
20
6.7
20
6.7
NS
9
5.0
9
5.0
NS
≥ 25 to 29.9
67
≥ 30
54
4
5.9
2
3.0
NS
7
13.0
9
16.7
NS
No previous studies have reported an effect of chorionicity in twin pregnancies on ultrasound-based weight estimation.20,21 In our study there was no difference in chorionicity between the different BMI groups. The effect of BMI on the accuracy of estimating fetal weight remains uncertain. Gandhi et al. reported that increasing maternal obesity decreases the accuracy of sonographically determined fetal weight in twin pregnancies, particularly for twin B20; on the other hand, Heer et al. found no effect of BMI on mean percentage error of fetal weight estimations.19 However, in our study we found that the mean absolute percentage error of fetal weight estimation was greater for both twins in obese women, and this error was reduced when the ultrasound was performed close to the time of delivery in these women. Gernt et al. concluded that ultrasound prediction of a discrepancy of 25% or more in twins’ birth weights is reliable if done within 16 days of delivery.22 Previous reports of the association between maternal BMI and the accuracy of weight estimation by ultrasound in twin pregnancies have categorized BMI using measurements at or close to the time of delivery,20,23 and none have examined pre-pregnancy or early pregnancy BMI. We believe that prepregnancy BMI is more indicative of actual maternal body fat composition than BMI recorded at the time of delivery, as weight gain during pregnancy includes circulation and gestational components, both of which are greater in twin pregnancies. The populations in previous studies have also been restricted to women who had ultrasound estimation of fetal weight within three days of delivery.18 In clinical practice not all twin pregnancies will have estimation of fetal weight so close to delivery, and a common practice is to schedule scans at two-week intervals.3 In view of this common clinical practice, we chose a two-week interval to allow correlation with actual birth weight.
700 l AUGUST JOGC AOÛT 2015
A limitation of our study is its retrospective nature. We did not stratify the outcome according to the classification of obesity; to do so would require a larger number of subjects. However, the number of subjects in our study is greater than in previous published reports, and the use of pre-pregnancy BMI arguably provided a more reliable classification of maternal BMI status. CONCLUSION
Our findings suggest that the accuracy of estimating fetal weight by ultrasound in obese women with a twin pregnancy is less than in women with a twin pregnancy and a normal BMI. Providers of obstetrical care must be aware that estimation of fetal weight by ultrasound in twin pregnancies is less reliable when performed more than seven days before delivery. It is appropriate to consider scheduling a repeat ultrasound if delivery has not occurred within seven days in order to predict birth weight more accurately. REFERENCES 1. McDonald S, Murphy K, Beyene J, Ohlsson A. Perinatal outcomes of in vitro fertilization twins: a systematic review and meta-analyses. Am J Obstet Gynecol 2005;193:141–52. 2. Callahan TL, Hall JE, Ettner SL, Christiansen CL, Greene MF, Crowley WF. The economic impact of multiple-gestation pregnancies and the contribution of assisted-reproduction techniques to their incidence. N Engl J Med 1994;331:244–9. 3. Millar WJ, Wadhera S, Nimord C. Multiple births: trends and patterns in Canada, 1974–1990. Health Rep 1992;4(3):223–50 4. Barrett J, Bocking A. Management of twin pregnancies. SOGC consensus statement: part 1. no. 91, July 2000. J Soc Obstet Gynaecol Can 2000;22:519–29. 5. Caravello JW, Chauhan SP, Morrison JC, Magann EF, Martin JN Jr, Devoe LD. Sonographic examination does not predict twin growth discordance accurately. Obstet Gynecol 1997;89(4):529–33. 6. Nyberg DA, McGahan JP, Pretorius DH. Diagnostic imaging of fetal anomalies. New York (NY): Lippincott Williams & Wilkins; 2003. ISBN:0781732115. 7. Schröeder H, Sonntag J, Waltz S, Schollmeyer T, Schüppler U, Weisner D. Morbidity and mortality of discordant twins up to 34 weeks of gestational age. Eur J Pediatr 1996;155(3):224–9.
Estimating Fetal Weight and Inter-Twin Weight Discordance by Ultrasound in Twin Pregnancies in Women With Increased BMI
8. Rodis JF, Vintzileos AM, Campbell WA, Nochimson DJ. Intrauterine fetal growth in discordant twin gestations. J Ultrasound Med 1990;9(8):443–8.
16. Shamley KT, Landon MB. Accuracy and modifying factors for ultrasonographic determination of fetal weight at term. Obstet Gynecol 1994;84:926–30.
9. Erkkola R, Ala-Mello S, Piiroinen O, Kero P, Sillanpaa M. Growth discordancy in twin pregnancies: a risk factor not detected by measurements of biparietal diameter. Obstet Gynecol 1985;66(2):203–6.
17. Benacerraf BR, Gelman R, Frigoletto FD Jr. Sonographically estimated fetal weights: accuracy and limitation. Am J Obstet Gynecol 1988;159:1118–21.
10. Branum AM, Schoendorf KC. The effect of birth weight discordance on twin neonatal mortality. Obstet Gynecol 2003;101:570–4. 11. Statistics Canada. Adult obesity in Canada: measured height and weight. Ottawa (ON): Statistics Canada; 2005. Available at: http://www.statcan.gc.ca/pub/82–620-m/2005001/article/ adults adultes/8060-eng.htm. Accessed January 6, 2010. 12. Wolfe HM, Sokol RJ, Matier SM, Zador IF. Maternal obesity: a potential source of error in sonographic prenatal diagnosis. Obstet Gynecol 1990;76:339–42. 13. Catanzarite V, Quirk JG. Second-trimester ultrasonography: determinants of visualization of fetal anatomic structures. Am J Obstet Gynecol 1990;163:1191–5. 14. Nahum GG, Stanislaw H. Ultrasonographic prediction of term birth weight: how accurate is it? Am J Obstet Gynecol 2002;188:566–74. 15. Chien PFW, Owen P, Khan K. Validity of ultrasound estimation of fetal weight. Obstet Gynecol 2000;95:856–60.
18. Danon D, Melamed N, Bardin R, Meizner I. Accuracy of ultrasonographic fetal weight estimation in twin pregnancies. Obstet Gynecol 2008;112:759–64. 19. Heer IM, Kumper C, Vogtle N, Muller-Egloff S, Dugas M, Strauss A. Analysis of factors influencing the ultrasonic fetal weight estimation. Fetal Diagn Ther 2008;23:204–10. 20. Gandhi M, Ferrara L, Belogolovkin V, Moshier E, Rebaber A. Effect of increased body mass index on the accuracy of estimated fetal weight by sonography in twins. J Ultrasound Med 2009;28:301–8. 21. Melamed N, Yogev Y, Meizner I, Mashiach R, Bardin R, Ben-Haroush A. Sonographic fetal weight estimation: which model should be used? J Ultrasound Med 2009;28:617–29. 22. Gernt PR, Mauldin JG, Newman RB, Durkalski VL. Sonographic prediction of twin birth weight discordance. Obstet Gynecol 2001;97:53–6. 23. Ocer F, Ayden Y, Atis A, Kaleli S. Factors affecting the accuracy of ultrasonographical fetal weight estimation in twin pregnancies. J Matern Fetal Neonatal Med 2011;24(9):1168–72.
AUGUST JOGC AOÛT 2015 l 701
The Accuracy of Estimating Fetal Weight and Inter-Twin Weight Discordance by Ultrasound in Twin Pregnancies in Women With Increased Body Mass Index Sawsan Al-Obaidly, MBBS,1 Jacqueline Parrish,1 Kellie E. Murphy, MD,1 Phyllis Glanc, MD,2 Cynthia Maxwell, MD1 Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, Mount Sinai Hospital, University of Toronto, Toronto ON
1
Department of Medical Imaging, Sunnybrook Health Sciences Centre, University of Toronto, Toronto ON
2
Abstract Objectives: The purpose of this study was to determine whether increased maternal pre-pregnancy BMI decreases the ultrasound accuracy of fetal weight estimation and inter-twin weight discordance in twin pregnancies compared with women with normal BMI. Methods: We conducted a retrospective cohort study of women with a known pre-pregnancy or early pregnancy BMI who delivered a viable twin pregnancy after 28 weeks’ gestation between 2008 and 2011, and who had an ultrasound examination for fetal weight estimation within two weeks of delivery. The sonographically estimated fetal weight (EFW) was compared with the actual weight for each twin, and inter-twin weight discordance (defined as a weight difference between twins of more than 25%) was stratified for the patient’s BMI. We sought to determine if EFW and inter-twin weight discordance were affected if delivery occurred at eight to 14 days after ultrasound compared to within seven days of ultrasound. Results: A total of 300 twin pregnancies with known pre-pregnancy maternal BMI were identified. Of these, 179 were underweight or of normal weight (BMI < 25 kg/m2), 67 were overweight (BMI 25 to 29.9 kg/m2), and 54 were obese (BMI ≥ 30 kg/m2). Ultrasound accuracy among all BMI groups were compared when done between 8 and 14 days and within seven days from delivery. There was a significant increasing trend in mean absolute percent errors for both twins in the obese compared to normal weight (P < 0.05) if delivery happened between eight and 14 days from ultrasound. This difference was diminished if the ultrasound was performed within seven days of delivery. The ultrasound detection of inter-twin weight discordance was similar among the three BMI groups. Key Words: Body mass index, BMI, obesity, estimated fetal weight, EFW, mean absolute percentage error, MAPE, inter-twin weight discrepancy, mean absolute difference, MAD Competing Interests: None declared. Received on June 21, 2014 Accepted on November 26, 2014
696 l AUGUST JOGC AOÛT 2015
Conclusion: Estimation of fetal weight using ultrasound in obese women with twin pregnancies appears to be more reliable when performed close to delivery.
Résumé Objectifs : Cette étude avait pour objectif de déterminer si la présence d’un IMC maternel prégrossesse accru entraînait une baisse de la précision de l’échographie pour ce qui est de l’estimation du poids fœtal et de la discordance intergémellaire en matière de poids dans le cadre de grossesses gémellaires, par comparaison avec des femmes présentant un IMC normal. Méthodes : Nous avons mené une étude de cohorte rétrospective portant sur des femmes qui présentaient un IMC prégrossesse (ou aux débuts de la grossesse) connu, qui ont accouché après 28 semaines de gestation à la suite d’une grossesse gémellaire viable entre 2008 et 2011, et qui ont subi un examen échographique visant l’estimation du poids fœtal dans les deux semaines ayant précédé l’accouchement. Le poids fœtal estimé (PFE) par échographie a été comparé au poids réel de chacun des jumeaux, puis la discordance intergémellaire en matière de poids (définie comme une différence de poids entre les jumeaux de plus de 25 %) a été stratifiée en fonction de l’IMC de la patiente. Nous avons cherché à déterminer si le PFE et la discordance intergémellaire en matière de poids avaient été affectés lorsque l’accouchement était survenu de 8 à 14 jours à la suite de l’échographie, par comparaison avec un accouchement étant survenu dans les sept jours de la tenue de l’échographie. Résultats : Nous avons pu identifier, au total, 300 grossesses gémellaires pour lesquelles l’IMC maternel prégrossesse était connu : 179 femmes présentaient une insuffisance pondérale ou un poids normal (IMC < 25 kg/m2), 67 présentaient une surcharge pondérale (IMC = de 25 à 29,9 kg/m2) et 54 étaient obèses (IMC ≥ 30 kg/m2). Dans tous les groupes d’IMC, la précision de l’échographie menée entre 8 et 14 jours avant l’accouchement a été comparée à celle de l’échographie menée dans les sept jours de l’accouchement. Une tendance à la hausse considérable en matière d’erreur absolue moyenne en pourcentage pour les deux jumeaux a été constatée chez les femmes obèses,
Estimating Fetal Weight and Inter-Twin Weight Discordance by Ultrasound in Twin Pregnancies in Women With Increased BMI
par comparaison avec les femmes de poids normal (P < 0,05), lorsque l’accouchement avait eu lieu de 8 à 14 jours à la suite de l’échographie. Cette différence était moindre lorsque l’échographie avait été menée dans les sept jours de l’accouchement. La détection par échographie d’une discordance intergémellaire en matière de poids était semblable dans les trois groupes d’IMC. Conclusion : Chez les femmes obèses qui connaissent une grossesse gémellaire, l’estimation du poids fœtal par échographie semble être plus fiable lorsqu’elle est menée peu avant l’accouchement.
J Obstet Gynaecol Can 2015;37(8):696–701
INTRODUCTION
D
uring the last two decades, the rates of multiple pregnancies across the world and in Canada have increased, mainly because of the expanded use of assisted reproduction.1,2 There is an increased incidence of intrauterine growth restriction in one or both fetuses in twin pregnancies.3 Having accurate and timely identification of discordant growth is important because of its relationship to placental disorders, risk of IUGR, and perinatal morbidity and mortality.4 A significant weight discordance is assessed by taking the estimated fetal weight difference at 20% to 25%.5–7 Previous reports have classified a weight discordance of 25% as severe, with an associated overall risk of fetal death that is five-fold greater than in concordant twins.8,9 Theoretical reasons for the increased risk include a possible increased risk of head entrapment during delivery and birth trauma. Discordant twins may have a higher baseline neonatal mortality rate even after controlling for fetal growth and gestational age.10 On the other hand, there is another potential risk factor in twin pregnancies related mainly to maternal obesity. The WHO categorized the BMI into four different categories (underweight < BMI 18.5 kg/m2, normal weight BMI 18.5 kg/m2 to 24.9 kg/m2, overweight 25 kg/m2 to 29.9 kg/m2, and obese BMI ≥ 30 kg/m2). The number of patients with a BMI in the overweight and obese ranges has increased in the past two decades.11 Previous studies have noted the adverse effects maternal obesity can have
ABBREVIATIONS EFW
estimated fetal weight
MAD
mean absolute difference
MAPE
mean absolute percentage error
on the accuracy of sonographic fetal anatomy as well as on estimation of fetal weight.12,13 The purpose of this study was to determine whether increased maternal pre-pregnancy BMI in women with a twin pregnancy decreases the accuracy of fetal weight estimation by ultrasound and the estimation of inter-twin weight discordance when compared with women with a normal BMI and a twin pregnancy. METHODS
We performed a retrospective cohort study of women who delivered a viable twin pregnancy at > 28 weeks’ gestation at our institution during the period from 2008 to 2011 and who had a known pre-pregnancy or first trimester BMI. We included in the study women who had fetal weight estimation by ultrasound within 14 days of delivery. For purposes of comparison, study subjects were divided into two groups: group 1 had estimation of fetal weight by ultrasound within 8 to 14 days of delivery, and group 2 had the estimation of fetal weight within seven days of delivery. None of the neonates had any congenital anomaly, and monochorionic twins affected by twin–twin transfusion syndrome either resolved or were treated before delivery. Ultrasound biometric measurements were performed by registered sonographers, and images were reviewed either by radiologists who specialized in pregnancy imaging or by maternal–fetal medicine specialists. The standardized fetal weight estimation method used in our unit is the Hadlock regression formula, which incorporates the biparietal diameter, head circumference, abdominal circumference, and femur length.14 The accuracy of the estimations was assessed using several measures. The Pearson correlation coefficient was used to correlate EFW with the actual birth weight. The MAD was defined as the mean of actual birth weight minus the EFW (in grams) across the three different BMI groups. Absolute percentage errors were calculated by dividing the difference between sonographically estimated fetal weight and actual birth weight by the actual birth weight, and expressing the result as a percentage. Inter-twin weight discordance was defined as a weight disparity of more than 25%, using the larger twin as the index. All analyses were completed using Stata/IC release 12.1 for Windows (StataCorp LLP, College Station, TX). Linear regression and analysis of variance (ANOVA) was used to determine whether sonographic prediction of the actual birth weight varied according to BMI. Student t test was AUGUST JOGC AOÛT 2015 l 697
Obstetrics
Table 1. Demographic data of twin pregnancies in the three categories of pre-pregnancy BMI Normal (BMI < 25 kg/m2) n = 179
Overweight (BMI 25 to 29.9 kg/m2) n = 67
Obese (BMI ≥ 30 kg/m2) n = 54
P
Maternal age ≥ 35 years, n (%)
101 (56)
35 (52)
22 (40)
NS
Nulliparity, n (%)
109 (60)
40 (59)
27 (50)
NS
Maternal age, years, mean (SD)
34.4 (4.7)
34.5 (5.3)
33.2 (5.8)
NS
Gestational age at delivery, weeks, mean (SD)
35.8 (2.1)
35.4 (2.4)
35 (2.5)
NS
Total n = 300
Time from US to birth, days, mean (SD)
8 (4.6)
9 (5)
11 (3)
NS
Pre-pregnancy BMI, kg/m2, mean (SD)
21.4 (1.9)
27.2 (1.5)
37.6 (8)
< 0.005
BMI at delivery, mean (SD)
28.6 (3)
34.2 (2.8)
41.54 (8.3)
< 0.005
Dichorionic, diamnionic twins, n (%)
143 (79)
48 (71)
44 (81)
NS
Male sex, n (%)
160 (44)
74 (55)
50 (46)
NS
Vertex presentation of twin A, n (%)
126 (70)
36 (53)
26 (48)
NS
Vertex presentation of twin B, n (%)
78 (44)
26 (39)
15 (28)
NS
Intrauterine growth restriction, n (%)
14 (8)
5 (7)
1 (2)
NS
Twin-twin transfusion syndrome, n (%)
2 (1)
0
1 (2)
NS
used to compare continuous variables between groups, and chi-square was used for categorical variables. For all analyses, a P value of < 0.05 was considered significant.
Ultrasonographic EFW in group 1 was more accurate when the BMI was < 30 kg/m2 than when BMI was 30 kg/m2 or greater (Table 2).
The study was approved by the Mount Sinai Hospital Research Ethics Board.
There were statistically significant differences in both MAPE and MAD for both twins between the overweight and normal weight groups and between the obese and normal weight groups (Table 2). However these differences were absent if delivery occurred within seven days of the ultrasound (Table 3).
RESULTS
Electronic records of 1228 women who delivered viable twins were reviewed, and of these 300 women had available pre-pregnancy BMI or earliest BMI recorded in the first trimester and had an ultrasound for fetal weight estimation performed in our institution within 14 days of delivery (8 to 14 days group and 7 days group). There were 150 patients in each group. The demographic data collected are presented in Table 1. There were no significant differences between the groups except for the expected differences in mean pre-pregnancy BMI and mean BMI at delivery. To compare the difference between actual birth weight and last EFW by ultrasound, we measured the mean absolute percentage error and compared this between groups. The MAPE for twin A in group 1 (ultrasound performed 8 to 14 days before delivery) in the normal, overweight, and obese groups was 6%, 4%, and 20%, respectively, but in group 2 (ultrasound performed within 7 days of delivery) it was 4%, 2%, and 3%, respectively. The MAPE for twin B in group 1 in the normal, overweight, and obese groups was 4%, 5%, and 13%, respectively, and in group 2 it was < 1%, 3%, and 6%, respectively. 698 l AUGUST JOGC AOÛT 2015
The accuracy of ultrasonography in predicting inter-twin weight discordance is shown in Table 4. Overall, across the three BMI groups there were no significant differences in the inter-twin weight discordance assessed by ultrasound and identified by birth weights. DISCUSSION
Several reports have examined the accuracy of ultrasound in estimating fetal weight.15–17 Some studies have compared its accuracy in twin and singleton pregnancies and have found greater accuracy of ultrasound in singletons; for example, Danon et al. demonstrated that the accuracy of the ultrasonographic estimation of fetal weight seems to be less in twin pregnancies than in singletons, and that the accuracy is less for the second twin than for the first twin.18 Heer et al. showed in singleton pregnancies that if the time interval between weight estimation and delivery was more than seven days, the accuracy of the weight estimation will be reduced.19
Estimating Fetal Weight and Inter-Twin Weight Discordance by Ultrasound in Twin Pregnancies in Women With Increased BMI
Table 2. Sonographic prediction of fetal weight by increasing BMI category for twins A and B when delivery occurred 8 to 14 days after ultrasound
N
Estimated fetal weight, g (SD)
Birth weight, g (SD)
Mean absolute difference, g (SD)
Mean absolute error (SD)
Mean absolute percentage error (SD)
P
Pearson correlation coefficient*
Twin A BMI category
< 25 kg/m
89
2384 (536)
2531 (453)
147 (83)
0.06 (0.2)
5.8 (18)
0.05†
0.83
25 to 29.9 kg/m2
36
2348 (519)
2457 (517)
109 (2)
0.04 (0)
4.4 (0.4)
0.2‡
0.9
≥ 30 kg/m
25
1984 (576)
2475 (575)
491 (1)
0.2 (0)
19.8 (0.2)
0.01§
0.9
89
2424 (454)
2514 (509)
90 (55)
0.04 (0.1)
3.6 (10)
0.004†
0.85
36
2407 (529)
2525 (568)
118 (39)
0.05 (0.07)
4.7 (6.8)
0.08‡
0.9
25
2013 (638)
2321 (728)
308 (90)
0.1 (0.1)
13 (12)
0.001§
0.82
2
2
Twin B BMI category
< 25 kg/m2 25 to 29.9 kg/m
2
≥ 30 kg/m2
*Correlation between estimated fetal weight and actual birth weight. †Overall ANOVA of MAPE comparing overweight and obese with normal BMI group. ‡Comparing overweight with normal BMI. §Comparing obese with normal BMI. SD: standard deviation
Table 3. Sonographic prediction of fetal weight by increasing BMI category for twins A and B when delivery occurred within 7 days of ultrasound Estimated fetal weight, g (SD)
Birth weight, g (SD)
Mean absolute difference, g (SD)
Mean absolute error (SD)
Mean absolute percentage error (SD)
P
Pearson correlation coefficient*
90
2335 (586)
2442 (542)
107 (44)
0.04 (0.08)
4.4 (8)
0.5†
0.76
31
2387 (646)
2423 (668)
36 (22)
0.015 (0.03)
1.5 (3)
0.7‡
0.88
29
2266 (665)
2335 (635)
69 (30)
0.03 (0.04)
3.0 (4)
0.5§
0.9
< 25 kg/m2
90
2363 (585)
2370 (509)
7 (76)
0.003 (0.14)
0.3 (14)
0.4†
0.12
25 to 29.9 kg/m2
31
2394 (576)
2330 (608)
64 (32)
0.027 (0.05)
2.74 (5)
0.6‡
0.91
≥ 30 kg/m
29
2498 (625)
2343 (624)
155 (1)
0.066 (0.001)
6.6 (0.1)
0.7§
0.9
N Twin A BMI category
< 25 kg/m2 25 to 29.9 kg/m
2
≥ 30 kg/m2 Twin B BMI category
2
*Correlation between estimated fetal weight and actual birth weight. †Overall ANOVA of MAPE comparing overweight and obese with normal BMI group. ‡Comparing overweight with normal BMI. §Comparing obese with normal BMI. SD: standard deviation
AUGUST JOGC AOÛT 2015 l 699
Obstetrics
Table 4. Ultrasound estimated fetal weight discordance and actual birth weight discordance in different BMI categories
Total
Estimated fetal weight discordance > 25%
%
Actual birth weight discordance > 25%
%
P
BMI
300
< 25
179
20
6.7
20
6.7
NS
9
5.0
9
5.0
NS
≥ 25 to 29.9
67
≥ 30
54
4
5.9
2
3.0
NS
7
13.0
9
16.7
NS
No previous studies have reported an effect of chorionicity in twin pregnancies on ultrasound-based weight estimation.20,21 In our study there was no difference in chorionicity between the different BMI groups. The effect of BMI on the accuracy of estimating fetal weight remains uncertain. Gandhi et al. reported that increasing maternal obesity decreases the accuracy of sonographically determined fetal weight in twin pregnancies, particularly for twin B20; on the other hand, Heer et al. found no effect of BMI on mean percentage error of fetal weight estimations.19 However, in our study we found that the mean absolute percentage error of fetal weight estimation was greater for both twins in obese women, and this error was reduced when the ultrasound was performed close to the time of delivery in these women. Gernt et al. concluded that ultrasound prediction of a discrepancy of 25% or more in twins’ birth weights is reliable if done within 16 days of delivery.22 Previous reports of the association between maternal BMI and the accuracy of weight estimation by ultrasound in twin pregnancies have categorized BMI using measurements at or close to the time of delivery,20,23 and none have examined pre-pregnancy or early pregnancy BMI. We believe that prepregnancy BMI is more indicative of actual maternal body fat composition than BMI recorded at the time of delivery, as weight gain during pregnancy includes circulation and gestational components, both of which are greater in twin pregnancies. The populations in previous studies have also been restricted to women who had ultrasound estimation of fetal weight within three days of delivery.18 In clinical practice not all twin pregnancies will have estimation of fetal weight so close to delivery, and a common practice is to schedule scans at two-week intervals.3 In view of this common clinical practice, we chose a two-week interval to allow correlation with actual birth weight.
700 l AUGUST JOGC AOÛT 2015
A limitation of our study is its retrospective nature. We did not stratify the outcome according to the classification of obesity; to do so would require a larger number of subjects. However, the number of subjects in our study is greater than in previous published reports, and the use of pre-pregnancy BMI arguably provided a more reliable classification of maternal BMI status. CONCLUSION
Our findings suggest that the accuracy of estimating fetal weight by ultrasound in obese women with a twin pregnancy is less than in women with a twin pregnancy and a normal BMI. Providers of obstetrical care must be aware that estimation of fetal weight by ultrasound in twin pregnancies is less reliable when performed more than seven days before delivery. It is appropriate to consider scheduling a repeat ultrasound if delivery has not occurred within seven days in order to predict birth weight more accurately. REFERENCES 1. McDonald S, Murphy K, Beyene J, Ohlsson A. Perinatal outcomes of in vitro fertilization twins: a systematic review and meta-analyses. Am J Obstet Gynecol 2005;193:141–52. 2. Callahan TL, Hall JE, Ettner SL, Christiansen CL, Greene MF, Crowley WF. The economic impact of multiple-gestation pregnancies and the contribution of assisted-reproduction techniques to their incidence. N Engl J Med 1994;331:244–9. 3. Millar WJ, Wadhera S, Nimord C. Multiple births: trends and patterns in Canada, 1974–1990. Health Rep 1992;4(3):223–50 4. Barrett J, Bocking A. Management of twin pregnancies. SOGC consensus statement: part 1. no. 91, July 2000. J Soc Obstet Gynaecol Can 2000;22:519–29. 5. Caravello JW, Chauhan SP, Morrison JC, Magann EF, Martin JN Jr, Devoe LD. Sonographic examination does not predict twin growth discordance accurately. Obstet Gynecol 1997;89(4):529–33. 6. Nyberg DA, McGahan JP, Pretorius DH. Diagnostic imaging of fetal anomalies. New York (NY): Lippincott Williams & Wilkins; 2003. ISBN:0781732115. 7. Schröeder H, Sonntag J, Waltz S, Schollmeyer T, Schüppler U, Weisner D. Morbidity and mortality of discordant twins up to 34 weeks of gestational age. Eur J Pediatr 1996;155(3):224–9.
Estimating Fetal Weight and Inter-Twin Weight Discordance by Ultrasound in Twin Pregnancies in Women With Increased BMI
8. Rodis JF, Vintzileos AM, Campbell WA, Nochimson DJ. Intrauterine fetal growth in discordant twin gestations. J Ultrasound Med 1990;9(8):443–8.
16. Shamley KT, Landon MB. Accuracy and modifying factors for ultrasonographic determination of fetal weight at term. Obstet Gynecol 1994;84:926–30.
9. Erkkola R, Ala-Mello S, Piiroinen O, Kero P, Sillanpaa M. Growth discordancy in twin pregnancies: a risk factor not detected by measurements of biparietal diameter. Obstet Gynecol 1985;66(2):203–6.
17. Benacerraf BR, Gelman R, Frigoletto FD Jr. Sonographically estimated fetal weights: accuracy and limitation. Am J Obstet Gynecol 1988;159:1118–21.
10. Branum AM, Schoendorf KC. The effect of birth weight discordance on twin neonatal mortality. Obstet Gynecol 2003;101:570–4. 11. Statistics Canada. Adult obesity in Canada: measured height and weight. Ottawa (ON): Statistics Canada; 2005. Available at: http://www.statcan.gc.ca/pub/82–620-m/2005001/article/ adults adultes/8060-eng.htm. Accessed January 6, 2010. 12. Wolfe HM, Sokol RJ, Matier SM, Zador IF. Maternal obesity: a potential source of error in sonographic prenatal diagnosis. Obstet Gynecol 1990;76:339–42. 13. Catanzarite V, Quirk JG. Second-trimester ultrasonography: determinants of visualization of fetal anatomic structures. Am J Obstet Gynecol 1990;163:1191–5. 14. Nahum GG, Stanislaw H. Ultrasonographic prediction of term birth weight: how accurate is it? Am J Obstet Gynecol 2002;188:566–74. 15. Chien PFW, Owen P, Khan K. Validity of ultrasound estimation of fetal weight. Obstet Gynecol 2000;95:856–60.
18. Danon D, Melamed N, Bardin R, Meizner I. Accuracy of ultrasonographic fetal weight estimation in twin pregnancies. Obstet Gynecol 2008;112:759–64. 19. Heer IM, Kumper C, Vogtle N, Muller-Egloff S, Dugas M, Strauss A. Analysis of factors influencing the ultrasonic fetal weight estimation. Fetal Diagn Ther 2008;23:204–10. 20. Gandhi M, Ferrara L, Belogolovkin V, Moshier E, Rebaber A. Effect of increased body mass index on the accuracy of estimated fetal weight by sonography in twins. J Ultrasound Med 2009;28:301–8. 21. Melamed N, Yogev Y, Meizner I, Mashiach R, Bardin R, Ben-Haroush A. Sonographic fetal weight estimation: which model should be used? J Ultrasound Med 2009;28:617–29. 22. Gernt PR, Mauldin JG, Newman RB, Durkalski VL. Sonographic prediction of twin birth weight discordance. Obstet Gynecol 2001;97:53–6. 23. Ocer F, Ayden Y, Atis A, Kaleli S. Factors affecting the accuracy of ultrasonographical fetal weight estimation in twin pregnancies. J Matern Fetal Neonatal Med 2011;24(9):1168–72.
AUGUST JOGC AOÛT 2015 l 701