- Email: [email protected]
Ultrasound Fetal Weight Estimation: How Accurate Are We Now Under Emergency Conditions?
Ultrasound in Med. & Biol., Vol. -, No. -, pp. 1–5, 2015 Copyright Ó 2015 World Federation for Ultrasound in Medicine & Biology Printed in the USA. All rights reserved 0301-5629/$ - see front matter
http://dx.doi.org/10.1016/j.ultrasmedbio.2015.05.020
d
Original Contribution ULTRASOUND FETAL WEIGHT ESTIMATION: HOW ACCURATE ARE WE NOW UNDER EMERGENCY CONDITIONS? KAOUTHER DIMASSI,*y FATMA DOUIK,* MARIEM AJROUDI,* AMEL TRIKI,* and MOHAMED FAOUZI GARA*y * Obstetrics and Gynecology Unit, Mongi Slim Hospital, La Marsa, Tunisia; and y Faculte de Medecine, Universite de Tunis El Manar, Tunis, Tunisia (Received 17 January 2015; revised 20 May 2015; in final form 25 May 2015)
Abstract—The primary aim of this study was to evaluate the accuracy of sonographic estimation of fetal weight when performed at due date by first-line sonographers. This was a prospective study including 500 singleton pregnancies. Ultrasound examinations were performed by residents on delivery day. Estimated fetal weights (EFWs) were calculated and compared with the corresponding birth weights. The median absolute difference between EFW and birth weight was 200 g (100–330). This difference was within ±10% in 75.2% of the cases. The median absolute percentage error was 5.53% (2.70%–10.03%). Linear regression analysis revealed a good correlation between EFW and birth weight (r 5 0.79, p , 0.0001). According to Bland–Altman analysis, bias was 285.06 g (95% limits of agreement: 2663.33 to 494.21). In conclusion, EFWs calculated by residents were as accurate as those calculated by experienced sonographers. Nevertheless, predictive performance remains limited, with a low sensitivity in the diagnosis of macrosomia. (E-mail: [email protected]) Ó 2015 World Federation for Ultrasound in Medicine & Biology. Key Words: Fetal weight, Birth weight, Ultrasound estimation, Neonatal.
The aims of this study were to estimate the accuracy of sonographic estimation of fetal weight when performed during labor by trainees and to evaluate the effects of different maternal and fetal factors on this prediction.
INTRODUCTION Ultrasound estimation of fetal weight is routinely performed in labor rooms at due date. It is thought to be helpful in predicting fetal survival and making management decisions in very low birth weight infants and in managing delivery of large babies, in whom complications may occur (Dudley 2005). This is not trivial and has its consequences. As an example, one study suggested that after adjustment for confounding factors, overestimation of fetal weight remained associated with a high rate of cesarean delivery (CD) (Blackwell et al. 2009). Moreover, ultrasound examination during labor could potentially be problematic owing to the low position of the head and an increased risk of abdominal circumference distortion or posterior position of the femurs (Peregrine et al. 2007). Finally, in labor rooms, and when gestational age is .37 wk, ultrasound examinations are routinely performed by residents, and senior sonographers are sought only in the case of anomalies.
METHODS This prospective single-center study was conducted among pregnant women attending the Obstetrics and Gynecology Unit of Mongi Slim Hospital, La Marsa, Tunisia, between April 1 and November 30, 2014. The research protocol was approved by the hospital’s ethics committee. All participants gave informed consent, and data were analyzed anonymously. Inclusion criteria were: a singleton pregnancy in labor, and gestational age greater 37 wk of amenorrhea. Exclusion criteria were: detection of an intrauterine fetal demise and fetal pathology. The ultrasound examination was performed in the labor ward by one of the five residents involved in the study, who had at least 1 y of practical experience in measuring fetal biometry. A Toshiba SSA-510 A (famio5, Osaka, Yokohama, Japan) was employed with a 5- to 2-MHz probe. Fetal measurements of biparietal diameter,
Address correspondence to: Kaouther Dimassi, Residence Les printemps 2, Bloc G App 18, Cite taieb El Mihiri, 2045, Tunisia. E-mail: [email protected] 1
2
Ultrasound in Medicine and Biology
abdominal circumference and femur length were obtained, and the estimated fetal weight (EFW) was calculated using the formula of Hadlock et al. (1985): log EPF 5 1.335 1 0.0316 BIP 1 0.0457 PA 1 0.1623 LF 2 0.0034 PA LF), where EPF 5 EFW, BIP 5 biparietal diameter, PA 5 abdominal circumference and LF 5 femur length. EFW was compared with actual birth weight (BW). Data were collected on a standard spreadsheet (Microsoft Excel). Descriptive parameters are expressed as median (first–third quartiles) values. Frequencies are expressed as percentages. The analysis was performed in several ways: percentage error was calculated by subtracting the actual BW from the EFW and then dividing the difference by the actual BW and multiplying by 100. Mean percentage error (MPE), expressing the systematic error, was calculated from the percentage error. Absolute percentage error and mean absolute percentage error (MAPE) were calculated the same way using the absolute value of the difference between the EFW and actual BW. The proportion of estimates within 10% of the actual BW was also calculated. Correlation between BW and ultrasound EFW was evaluated using the Pearson coefficient, and agreement between these two measurements was assessed using Bland–Altman plots (Bland and Altman 1986). Percentage errors were compared using Student’s t-test with respect to maternal body mass index (BMI)—BW $ 4500 g, BW , 2500 g—and amount of amniotic fluid. We calculated the sensitivity, specificity, negative predictive value and positive predictive value of each EFW to detect fetal macrosomia. Statistical analysis was performed using MS excel software XLSTAT (2014.4.09; Addinsoft, New York, NY, USA). p , 0.05 was considered to indicate statistical significance.
Volume -, Number -, 2015
Table 1. Patient demographic data
Maternal age (y) Body mass index (kg/m2) Gestational age (wk of amenorrhea)
Median
First quartile
Third quartile
29 29.73 40
26 27.53 39
33 31.88 40
was 5.53% (2.70%–10.03%); 75.2% of the measurements had an error ,10%. There was a significant positive correlation between EFW and BW (r 5 0.79, 95% confidence interval: 2111.029, 259.091 (p , 0.0001) (Fig. 1). Figure 2 is the Bland–Altman analysis of these variables. Bias was 285.06 g (95% limits of agreement: 2663.33 to 494.21). Fetal macrosomia was associated with the worst accuracy of EFW (MPE 5 11.3%, p , 0.0001). However, neither low BW, nor maternal BMI, nor oligohydramnios nor polyhydramnios had an impact on the accuracy of the EFW (p 5 0.19, p 5 0.46, p 5 0.62 and p 5 0.82, respectively) (Table 2). The sensitivity, specificity, and positive and negative predictive values of predicting a BW $4000 g and BW $4500 g are given in Table 3. DISCUSSION The use of prenatal ultrasound scanning has increased in developed countries, but also in Tunisia. Today, ultrasound has become a common examination used daily in labor wards. Moreover, sonographic estimation of fetal weight is generally entrusted to residents. As the EFW is of major interest when the route of delivery
RESULTS During the study period, we analyzed 500 singleton pregnancies. The mean maternal age was 29.6 6 4.6 y (range: 17–42), and the mean gestational age at delivery was 39.6 6 1.3 wk (range: 37–42). Thirty-six patients (7.2%) had gestational diabetes mellitus, and 51 (10.2%) had a hypertensive disorder. Fifty-three patients (10.6%) had a BMI $ 35 kg/m2. Patient clinical and demographic data are summarized in Table 1. The median BW was 3500 6 476.8 g (3200–3850). Ten (2%) BWs were ,2500 g and 73 (14.6%) were .4000 g. The median EFW was 3480 6 431 g (3150– 3700). The median absolute difference between EFW and BW was 200 6 259.4 g (100–330), and the MAPE
Fig. 1. Correlation between sonographically estimated fetal weight and birth weight. The linear correlation coefficient R is close to 11 (R 5 0.79), indicating the strength of the positive correlation between estimated fetal weight and birth weight.
US fetal weight estimation d K. DIMASSI et al.
3
Table 3. Sensitivity, specificity and positive and negative predictive values of sonographic estimates of fetal macrosomia ($4000 g) and fetal extreme macrosomia ($4500 g) Birth weight $4000 g $4500 g
Fig. 2. Evaluation of agreement between estimated fetal weight and BW using Bland–Altman plots. Bias 5 285.06 g with 95% limits of agreement of 2663.33 to 494.21 g. BW 5 birth weight.
has to be determined (breech presentations, diabetes, suspected macrosomia), our aim was to estimate its accuracy when performed under those emergency conditions. Generally, we found that EFWs performed by obstetrics and gynecology trainees are as accurate as when performed by experienced sonographers. Indeed, in our study, the median absolute difference between EFW and BW was 200 g, and the MAPE was 5.53%. In previous studies, the MAPE ranged from 6% to 11% (Peregrine et al. 2007). Barel et al. (2014), in a retrospective cohort study including 9064 ultrasound EFWs performed during the week before delivery, estimated the systematic errors for the total population at 5.8% using the formula of Hadlock et al. (1985). In the last 25 y, numerous formulas for intrauterine estimation of fetal weight have been published. Dudley (2005) published a systematic review of ultrasound EFW and evaluated 11 different methods. No preferred method for sonographic EFW emerged from this review. For normal fetuses, the formula of Hadlock et al. (1985) is preferred and results in the smallest systematic Table 2. Factors influencing the accuracy of sonographic fetal weight estimation Median percentage error (first–third quartiles) Body mass index ,35 kg/m2 Body mass index $35 kg/m2 Amniotic fluid index Normal Oligohydramnios Polyhydramnios Birth weight ,4000 g Birth weight $4000 g Birth weight ,2500 g Birth weight $2500 g
5.5 (2.72–9.65) 5.9 (2.63–12.45) 5.5 (2.77–10.10) 5.4 (2.82–10.34) 4.9 (0.89–6.19) 5.2 (2.56–8.75) 11.3 (5.37–13.01) 11.3 (6.16–17.65) 5.5 (2.70–9.63)
p value 0.46
0.62 0.82 ,0.0001 0.19
Positive Negative Sensitivity Specificity predictive value predictive value 38.9% 37.5%
97.9% 99.4%
75.7% 50%
90.5% 99%
mean error (Dudley 2005). For this reason, we chose to use the formula of Hadlock et al. (1985) in our study. Barel et al. (2013) used sonographic fetal measurements taken up to 1 wk before delivery and calculated the expected birth weight using 23 different formulas. In this retrospective cohort study including almost 13,000 singleton pregnancies that considered both systematic error and random error and the percentage of estimates within 10% and 15% of BW, formulas that used three or more fetal measurements were more accurate than those that used one or two measurements; formulas that used abdominal circumference only were next. More than 80% of sonographic EFWs calculated by most models were within 15% of actual birth weights, but only about 65% were within 10%. It seems that no single model consistently predicts BW accurately (Barel et al. 2013). The estimates calculated with most models predict within 10% of BW only 65% of the time. For most models, 80% of sonographic EFWs predicted within 15% of BWs, but only about 65% were within 10%. Some models had a tendency to overestimate fetal weight, whereas others tended to underestimate it; these results were statistically significant and allow a better understanding of the specific formulas used by different institutions. Previous studies evaluated the performance of ultrasound EFWs performed at due date. However, only a few studies have evaluated EFW when performed by residents. Predanic et al. (2002) reported that a significant improvement in EFWs occurred with advanced training among residents: Data were within 10% in just less than half of the cases for inexperienced residents but rose to nearly three-fourths for the most experienced, whereas accuracy within 5% of BW rose steadily from less than one-fifth to more than twice that with experience. BenAroya et al. (2002) reported that resident fatigue affected the accuracy of clinical, but not sonographic EFWs. Nevertheless, the accuracy of EFWs was not influenced by the resident’s training level. Brunader (1996) studied the accuracy of family physicians at different levels of training in performing common prenatal sonographic measurements. No significant difference in accuracy was found between any of the resident year groups and family medicine faculty. This study concluded that the
4
Ultrasound in Medicine and Biology
skills necessary to perform accurate prenatal sonographic biometric measurements are easily learned within the confines of a 3-y family practice residency. Noumi et al. (2005) reported that neither the postgraduate year of the examiner, nor many other possible related variables affected the accuracy of EFW and concluded that EFWs performed during labor by residents were as accurate as EFWs performed by experienced sonographers. Barel et al. (2014) evaluated the effect of sonographer experience on the accuracy of sonographic EFW and concluded that sonographers with at least 2 y of experience had MAPEs closer to the actual BW. Five residents with at least 1 y of practical experience in measuring fetal biometry were involved in the present study. EFWs performed by obstetrics and gynecology trainees were as accurate as those performed by experienced sonographers in terms of MAPE. Thus, we can conclude that EFWs may be performed by residents in the labor room for patients at gestational ages .37 wk. The significant limits of agreement observed in this study may, however, be explained by interoperator variability. Having said that, the relatively high MAPE characterizing our work and most of the studies undermines the accuracy of the sonographic EFW and possibly affects clinical decisions regarding pregnancy follow-up and delivery (Barel et al. 2014; Little et al. 2012). In addition to the inherent random errors, various clinical and technical factors may affect the accuracy of the sonographic EFW. These factors may or may not include maternal factors such as BMI and pregnancy factors such as fetal sex, multiple pregnancy and amniotic fluid volume. In the literature, the effect of maternal obesity on EFW is uncertain. In our study, BMI had no impact on the accuracy of EFWs. Some authors have confirmed this result. Field et al. (1995), in a year-long study including 998 singleton pregnancies, observed that approximately two-thirds of the predicted fetal weights were within 10% of BWs and did not decrease with increasing maternal obesity. Blann and Prien (2000) reported that maternal weight did not affect the accuracy of sonographic measurements; however, it did result in a technically more challenging examination. In the prospective study of Farrell et al. (2002) comprising 96 women, the results indicated that the accuracy of ultrasound EFW is not influenced significantly by maternal BMI. Barel et al. (2014) concluded that high maternal BMI was not found to significantly affect weight estimation using the method of Hadlock et al. (1985), which is similar to our conclusion. Several studies have evaluated the amniotic fluid index (AFI) as a possible cause of ultrasonographic inaccuracy at term. Some have found that the AFI does not affect ultrasonographic accuracy, as we did. A recent study (Ashwal et al. 2015), including 1746 singleton
Volume -, Number -, 2015
pregnancies, reported that AFI has limited impact on the MPE in sonographic estimations performed a week before delivery and is not associated with EFW inaccuracy. Perni et al. (2004) concluded that there is no correlation between AFI and EFW during the third trimester, although a positive correlation between AFI and EFW was noted late in gestation. On the contrary, others studies have found that MPE tends to increase with the volume of amniotic fluid. Edwards et al. (2001) reported that oligohydramnios did result in a trend toward underestimated sonographic fetal weights. Durbin et al. (2005) concluded that an increased AFI led to overestimation of BW in 80% of cases and a decreased AFI led to underestimation of BW in 63% of cases. Barel et al. (2014) reported that the formula of Hadlock et al. (1985) is associated with an overestimation rate of 5.9% in cases of oligohydramnios, an overestimation rate of 3.4% in cases with normal AFI and a lower overestimation rate of 2.9% in cases of polyhydramnios (p , 0.05). The most accurate EFWs are observed for BWs between 2500 and 3500 g; EFW accuracy then decreases in fetuses weighing more than 4000 g (Acker et al. 1987). In our study, fetal macrosomia was associated with the worst accuracy, and the sensitivity of predicting BWs $4000 g was only 38.9% with 97.9% specificity. These results are consistent with data from the literature. Noumi et al. (2005) calculated a sensitivity of 50%, specificity of 97%, PPV of 50% and NPV of 97%. Peregrine et al. (2007) estimated the sensitivity at 40%, specificity at 94%, PPV at 59% and NPV at 87%. This poor correlation between ultrasound EFW and BW at the extremes of BW can be explained by the fact that the equations used are insufficiently accurate because they were derived from populations of fetuses within a restricted range of BW (Mongelli 1997). Previous work (Little et al. 2012) has suggested that sonographic overestimation of BW may be associated with an elevated risk of CD and that women with a false diagnosis of macrosomia are at increased risk of CD compared with BW-matched control patients. Melamed et al. (2010) found that the CD rate was 2 to 2.5 times higher when the EFW was 4000 to 4499 g, regardless of the actual BW. Given the low sensitivity and high NPV obtained in this work, we think that the risk is misdiagnosis of macrosomia, which may result in neonatal morbidity. CONCLUSIONS Sonographic estimation of fetal weight is one of the most common ways to assess the growth of a fetus in utero to evaluate an ongoing pregnancy or prepare for delivery. This examination is routinely performed at term and during labor by trainees and is as accurate as when performed by experienced sonographers. Nevertheless, the
US fetal weight estimation d K. DIMASSI et al.
relatively high percentage error undermines the accuracy of this examination and possibly affects clinical decisions regarding delivery. In addition to the inherent errors of this method, various clinical and technical factors may affect the accuracy of the EFW and must be borne in mind when decisions regarding obstetric management are formulated. REFERENCES Acker DB, Sach BP, Ransil BJ, Friedman EA. Ultrasonography for fetal weight estimation: The Birnholz equation. Ultrason Imaging 1987; 9:195–202. Ashwal E, Hiersch L, Melamed N, Bardin R, Wiznitzer A, Yogev Y. Does the level of amniotic fluid have an effect on the accuracy of sonographic estimated fetal weight at term? J Matern Fetal Neonatal Med 2015;28:638–642. Barel O, Maymon R, Vaknin Z, Tovbin J, Smorgick N. Sonographic fetal weight estimation—Is there more to it than just fetal measurements? Prenat Diagn 2014;34:50–55. Barel O, Vaknin Z, Tovbin J, Herman A, Maymon R. Assessment of the accuracy of multiple sonographic fetal weight estimation formulas. J Ultrasound Med 2013;32:815–823. Ben-Aroya Z, Segal D, Hadar A, Hallak M, Friger M, Katz M, Mazor M. Effect of OB/GYN residents’ fatigue and training level on the accuracy of fetal weight estimation. Fetal Diagn Ther 2002;17:177–181. Blackwell SC, Refuerzo J, Chadha R, Carren CA. Overestimation of fetal weight by ultrasound: Does it influence the likelihood of cesarean delivery for labor arrest? Am J Obstet Gynecol 2009;200: 340.e1–340.e3. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1: 307–310. Blann DW, Prien SD. Estimation of fetal weight before and after amniotomy in the laboring gravid woman. Am J Obstet Gynecol 2000;182: 1117–1120.
5
Brunader R. Accuracy of prenatal sonography performed by family practice residents. Fam Med 1996;28:407–410. Dudley NJ. A systematic review of the ultrasound estimation of fetal weight. Ultrasound Obstet Gynecol 2005;25:80–89. Durbin SA, Lee CW, Parker VG. The effect of amniotic fluid index on the accuracy of sonographic estimated fetal weight. J Diagn Med Sonogr 2005;21:329–335. Edwards A, Goff J, Baker L. Accuracy and modifying factors of the sonographic estimation of fetal weight in a high-risk population. Aust N Z J Obstet Gynaecol 2001;41:187–190. Farrell T, Holmes R, Stone P. The effect of body mass index on three methods of fetal weight estimation. BJOG 2002;109:651–657. Field NT, Piper JM, Langer O. The effect of maternal obesity on the accuracy of fetal weight estimation. Obstet Gynecol 1995;86: 102–107. 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: 333–337. Little SE, Edlow AG, Thomas AM, Smith NA. Estimated fetal weight by ultrasound: A modifiable risk factor for cesarean delivery? Am J Obstet Gynecol 2012;207:309.e1–309.e6. Melamed N, Yogev Y, Meizner I, Mashiach R, Ben-Haroush A. Sonographic prediction of fetal macrosomia the consequences of false diagnosis. J Ultrasound Med 2010;29:225–230. Mongelli M. Fetal weight estimation by symphysis fundus height and gestational age. Gynecol Obstet Invest 1997;43:20–24. Noumi G, Collado-Khoury F, Bombard A, Julliard K, Weiner Z. Clinical and sonographic estimation of fetal weight performed during labor by residents. Am J Obstet Gynecol 2005;192: 1407–1409. Peregrine E, O’Brien P, Jauniaux E. Clinical and ultrasound estimation of birth weight prior to induction of labor at term. Ultrasound Obstet Gynecol 2007;29:304–309. Perni SC, Predanic M, Cho JE, Kalish RB, Chasen ST. Association of amniotic fluid index with estimated fetal weight. J Ultrasound Med 2004;23:1449–1452. Predanic M, Cho A, Ingrid F, Pellettieri J. Ultrasonographic estimation of fetal weight: acquiring accuracy in residency. J Ultrasound Med 2002;21:495–500.
http://dx.doi.org/10.1016/j.ultrasmedbio.2015.05.020
d
Original Contribution ULTRASOUND FETAL WEIGHT ESTIMATION: HOW ACCURATE ARE WE NOW UNDER EMERGENCY CONDITIONS? KAOUTHER DIMASSI,*y FATMA DOUIK,* MARIEM AJROUDI,* AMEL TRIKI,* and MOHAMED FAOUZI GARA*y * Obstetrics and Gynecology Unit, Mongi Slim Hospital, La Marsa, Tunisia; and y Faculte de Medecine, Universite de Tunis El Manar, Tunis, Tunisia (Received 17 January 2015; revised 20 May 2015; in final form 25 May 2015)
Abstract—The primary aim of this study was to evaluate the accuracy of sonographic estimation of fetal weight when performed at due date by first-line sonographers. This was a prospective study including 500 singleton pregnancies. Ultrasound examinations were performed by residents on delivery day. Estimated fetal weights (EFWs) were calculated and compared with the corresponding birth weights. The median absolute difference between EFW and birth weight was 200 g (100–330). This difference was within ±10% in 75.2% of the cases. The median absolute percentage error was 5.53% (2.70%–10.03%). Linear regression analysis revealed a good correlation between EFW and birth weight (r 5 0.79, p , 0.0001). According to Bland–Altman analysis, bias was 285.06 g (95% limits of agreement: 2663.33 to 494.21). In conclusion, EFWs calculated by residents were as accurate as those calculated by experienced sonographers. Nevertheless, predictive performance remains limited, with a low sensitivity in the diagnosis of macrosomia. (E-mail: [email protected]) Ó 2015 World Federation for Ultrasound in Medicine & Biology. Key Words: Fetal weight, Birth weight, Ultrasound estimation, Neonatal.
The aims of this study were to estimate the accuracy of sonographic estimation of fetal weight when performed during labor by trainees and to evaluate the effects of different maternal and fetal factors on this prediction.
INTRODUCTION Ultrasound estimation of fetal weight is routinely performed in labor rooms at due date. It is thought to be helpful in predicting fetal survival and making management decisions in very low birth weight infants and in managing delivery of large babies, in whom complications may occur (Dudley 2005). This is not trivial and has its consequences. As an example, one study suggested that after adjustment for confounding factors, overestimation of fetal weight remained associated with a high rate of cesarean delivery (CD) (Blackwell et al. 2009). Moreover, ultrasound examination during labor could potentially be problematic owing to the low position of the head and an increased risk of abdominal circumference distortion or posterior position of the femurs (Peregrine et al. 2007). Finally, in labor rooms, and when gestational age is .37 wk, ultrasound examinations are routinely performed by residents, and senior sonographers are sought only in the case of anomalies.
METHODS This prospective single-center study was conducted among pregnant women attending the Obstetrics and Gynecology Unit of Mongi Slim Hospital, La Marsa, Tunisia, between April 1 and November 30, 2014. The research protocol was approved by the hospital’s ethics committee. All participants gave informed consent, and data were analyzed anonymously. Inclusion criteria were: a singleton pregnancy in labor, and gestational age greater 37 wk of amenorrhea. Exclusion criteria were: detection of an intrauterine fetal demise and fetal pathology. The ultrasound examination was performed in the labor ward by one of the five residents involved in the study, who had at least 1 y of practical experience in measuring fetal biometry. A Toshiba SSA-510 A (famio5, Osaka, Yokohama, Japan) was employed with a 5- to 2-MHz probe. Fetal measurements of biparietal diameter,
Address correspondence to: Kaouther Dimassi, Residence Les printemps 2, Bloc G App 18, Cite taieb El Mihiri, 2045, Tunisia. E-mail: [email protected] 1
2
Ultrasound in Medicine and Biology
abdominal circumference and femur length were obtained, and the estimated fetal weight (EFW) was calculated using the formula of Hadlock et al. (1985): log EPF 5 1.335 1 0.0316 BIP 1 0.0457 PA 1 0.1623 LF 2 0.0034 PA LF), where EPF 5 EFW, BIP 5 biparietal diameter, PA 5 abdominal circumference and LF 5 femur length. EFW was compared with actual birth weight (BW). Data were collected on a standard spreadsheet (Microsoft Excel). Descriptive parameters are expressed as median (first–third quartiles) values. Frequencies are expressed as percentages. The analysis was performed in several ways: percentage error was calculated by subtracting the actual BW from the EFW and then dividing the difference by the actual BW and multiplying by 100. Mean percentage error (MPE), expressing the systematic error, was calculated from the percentage error. Absolute percentage error and mean absolute percentage error (MAPE) were calculated the same way using the absolute value of the difference between the EFW and actual BW. The proportion of estimates within 10% of the actual BW was also calculated. Correlation between BW and ultrasound EFW was evaluated using the Pearson coefficient, and agreement between these two measurements was assessed using Bland–Altman plots (Bland and Altman 1986). Percentage errors were compared using Student’s t-test with respect to maternal body mass index (BMI)—BW $ 4500 g, BW , 2500 g—and amount of amniotic fluid. We calculated the sensitivity, specificity, negative predictive value and positive predictive value of each EFW to detect fetal macrosomia. Statistical analysis was performed using MS excel software XLSTAT (2014.4.09; Addinsoft, New York, NY, USA). p , 0.05 was considered to indicate statistical significance.
Volume -, Number -, 2015
Table 1. Patient demographic data
Maternal age (y) Body mass index (kg/m2) Gestational age (wk of amenorrhea)
Median
First quartile
Third quartile
29 29.73 40
26 27.53 39
33 31.88 40
was 5.53% (2.70%–10.03%); 75.2% of the measurements had an error ,10%. There was a significant positive correlation between EFW and BW (r 5 0.79, 95% confidence interval: 2111.029, 259.091 (p , 0.0001) (Fig. 1). Figure 2 is the Bland–Altman analysis of these variables. Bias was 285.06 g (95% limits of agreement: 2663.33 to 494.21). Fetal macrosomia was associated with the worst accuracy of EFW (MPE 5 11.3%, p , 0.0001). However, neither low BW, nor maternal BMI, nor oligohydramnios nor polyhydramnios had an impact on the accuracy of the EFW (p 5 0.19, p 5 0.46, p 5 0.62 and p 5 0.82, respectively) (Table 2). The sensitivity, specificity, and positive and negative predictive values of predicting a BW $4000 g and BW $4500 g are given in Table 3. DISCUSSION The use of prenatal ultrasound scanning has increased in developed countries, but also in Tunisia. Today, ultrasound has become a common examination used daily in labor wards. Moreover, sonographic estimation of fetal weight is generally entrusted to residents. As the EFW is of major interest when the route of delivery
RESULTS During the study period, we analyzed 500 singleton pregnancies. The mean maternal age was 29.6 6 4.6 y (range: 17–42), and the mean gestational age at delivery was 39.6 6 1.3 wk (range: 37–42). Thirty-six patients (7.2%) had gestational diabetes mellitus, and 51 (10.2%) had a hypertensive disorder. Fifty-three patients (10.6%) had a BMI $ 35 kg/m2. Patient clinical and demographic data are summarized in Table 1. The median BW was 3500 6 476.8 g (3200–3850). Ten (2%) BWs were ,2500 g and 73 (14.6%) were .4000 g. The median EFW was 3480 6 431 g (3150– 3700). The median absolute difference between EFW and BW was 200 6 259.4 g (100–330), and the MAPE
Fig. 1. Correlation between sonographically estimated fetal weight and birth weight. The linear correlation coefficient R is close to 11 (R 5 0.79), indicating the strength of the positive correlation between estimated fetal weight and birth weight.
US fetal weight estimation d K. DIMASSI et al.
3
Table 3. Sensitivity, specificity and positive and negative predictive values of sonographic estimates of fetal macrosomia ($4000 g) and fetal extreme macrosomia ($4500 g) Birth weight $4000 g $4500 g
Fig. 2. Evaluation of agreement between estimated fetal weight and BW using Bland–Altman plots. Bias 5 285.06 g with 95% limits of agreement of 2663.33 to 494.21 g. BW 5 birth weight.
has to be determined (breech presentations, diabetes, suspected macrosomia), our aim was to estimate its accuracy when performed under those emergency conditions. Generally, we found that EFWs performed by obstetrics and gynecology trainees are as accurate as when performed by experienced sonographers. Indeed, in our study, the median absolute difference between EFW and BW was 200 g, and the MAPE was 5.53%. In previous studies, the MAPE ranged from 6% to 11% (Peregrine et al. 2007). Barel et al. (2014), in a retrospective cohort study including 9064 ultrasound EFWs performed during the week before delivery, estimated the systematic errors for the total population at 5.8% using the formula of Hadlock et al. (1985). In the last 25 y, numerous formulas for intrauterine estimation of fetal weight have been published. Dudley (2005) published a systematic review of ultrasound EFW and evaluated 11 different methods. No preferred method for sonographic EFW emerged from this review. For normal fetuses, the formula of Hadlock et al. (1985) is preferred and results in the smallest systematic Table 2. Factors influencing the accuracy of sonographic fetal weight estimation Median percentage error (first–third quartiles) Body mass index ,35 kg/m2 Body mass index $35 kg/m2 Amniotic fluid index Normal Oligohydramnios Polyhydramnios Birth weight ,4000 g Birth weight $4000 g Birth weight ,2500 g Birth weight $2500 g
5.5 (2.72–9.65) 5.9 (2.63–12.45) 5.5 (2.77–10.10) 5.4 (2.82–10.34) 4.9 (0.89–6.19) 5.2 (2.56–8.75) 11.3 (5.37–13.01) 11.3 (6.16–17.65) 5.5 (2.70–9.63)
p value 0.46
0.62 0.82 ,0.0001 0.19
Positive Negative Sensitivity Specificity predictive value predictive value 38.9% 37.5%
97.9% 99.4%
75.7% 50%
90.5% 99%
mean error (Dudley 2005). For this reason, we chose to use the formula of Hadlock et al. (1985) in our study. Barel et al. (2013) used sonographic fetal measurements taken up to 1 wk before delivery and calculated the expected birth weight using 23 different formulas. In this retrospective cohort study including almost 13,000 singleton pregnancies that considered both systematic error and random error and the percentage of estimates within 10% and 15% of BW, formulas that used three or more fetal measurements were more accurate than those that used one or two measurements; formulas that used abdominal circumference only were next. More than 80% of sonographic EFWs calculated by most models were within 15% of actual birth weights, but only about 65% were within 10%. It seems that no single model consistently predicts BW accurately (Barel et al. 2013). The estimates calculated with most models predict within 10% of BW only 65% of the time. For most models, 80% of sonographic EFWs predicted within 15% of BWs, but only about 65% were within 10%. Some models had a tendency to overestimate fetal weight, whereas others tended to underestimate it; these results were statistically significant and allow a better understanding of the specific formulas used by different institutions. Previous studies evaluated the performance of ultrasound EFWs performed at due date. However, only a few studies have evaluated EFW when performed by residents. Predanic et al. (2002) reported that a significant improvement in EFWs occurred with advanced training among residents: Data were within 10% in just less than half of the cases for inexperienced residents but rose to nearly three-fourths for the most experienced, whereas accuracy within 5% of BW rose steadily from less than one-fifth to more than twice that with experience. BenAroya et al. (2002) reported that resident fatigue affected the accuracy of clinical, but not sonographic EFWs. Nevertheless, the accuracy of EFWs was not influenced by the resident’s training level. Brunader (1996) studied the accuracy of family physicians at different levels of training in performing common prenatal sonographic measurements. No significant difference in accuracy was found between any of the resident year groups and family medicine faculty. This study concluded that the
4
Ultrasound in Medicine and Biology
skills necessary to perform accurate prenatal sonographic biometric measurements are easily learned within the confines of a 3-y family practice residency. Noumi et al. (2005) reported that neither the postgraduate year of the examiner, nor many other possible related variables affected the accuracy of EFW and concluded that EFWs performed during labor by residents were as accurate as EFWs performed by experienced sonographers. Barel et al. (2014) evaluated the effect of sonographer experience on the accuracy of sonographic EFW and concluded that sonographers with at least 2 y of experience had MAPEs closer to the actual BW. Five residents with at least 1 y of practical experience in measuring fetal biometry were involved in the present study. EFWs performed by obstetrics and gynecology trainees were as accurate as those performed by experienced sonographers in terms of MAPE. Thus, we can conclude that EFWs may be performed by residents in the labor room for patients at gestational ages .37 wk. The significant limits of agreement observed in this study may, however, be explained by interoperator variability. Having said that, the relatively high MAPE characterizing our work and most of the studies undermines the accuracy of the sonographic EFW and possibly affects clinical decisions regarding pregnancy follow-up and delivery (Barel et al. 2014; Little et al. 2012). In addition to the inherent random errors, various clinical and technical factors may affect the accuracy of the sonographic EFW. These factors may or may not include maternal factors such as BMI and pregnancy factors such as fetal sex, multiple pregnancy and amniotic fluid volume. In the literature, the effect of maternal obesity on EFW is uncertain. In our study, BMI had no impact on the accuracy of EFWs. Some authors have confirmed this result. Field et al. (1995), in a year-long study including 998 singleton pregnancies, observed that approximately two-thirds of the predicted fetal weights were within 10% of BWs and did not decrease with increasing maternal obesity. Blann and Prien (2000) reported that maternal weight did not affect the accuracy of sonographic measurements; however, it did result in a technically more challenging examination. In the prospective study of Farrell et al. (2002) comprising 96 women, the results indicated that the accuracy of ultrasound EFW is not influenced significantly by maternal BMI. Barel et al. (2014) concluded that high maternal BMI was not found to significantly affect weight estimation using the method of Hadlock et al. (1985), which is similar to our conclusion. Several studies have evaluated the amniotic fluid index (AFI) as a possible cause of ultrasonographic inaccuracy at term. Some have found that the AFI does not affect ultrasonographic accuracy, as we did. A recent study (Ashwal et al. 2015), including 1746 singleton
Volume -, Number -, 2015
pregnancies, reported that AFI has limited impact on the MPE in sonographic estimations performed a week before delivery and is not associated with EFW inaccuracy. Perni et al. (2004) concluded that there is no correlation between AFI and EFW during the third trimester, although a positive correlation between AFI and EFW was noted late in gestation. On the contrary, others studies have found that MPE tends to increase with the volume of amniotic fluid. Edwards et al. (2001) reported that oligohydramnios did result in a trend toward underestimated sonographic fetal weights. Durbin et al. (2005) concluded that an increased AFI led to overestimation of BW in 80% of cases and a decreased AFI led to underestimation of BW in 63% of cases. Barel et al. (2014) reported that the formula of Hadlock et al. (1985) is associated with an overestimation rate of 5.9% in cases of oligohydramnios, an overestimation rate of 3.4% in cases with normal AFI and a lower overestimation rate of 2.9% in cases of polyhydramnios (p , 0.05). The most accurate EFWs are observed for BWs between 2500 and 3500 g; EFW accuracy then decreases in fetuses weighing more than 4000 g (Acker et al. 1987). In our study, fetal macrosomia was associated with the worst accuracy, and the sensitivity of predicting BWs $4000 g was only 38.9% with 97.9% specificity. These results are consistent with data from the literature. Noumi et al. (2005) calculated a sensitivity of 50%, specificity of 97%, PPV of 50% and NPV of 97%. Peregrine et al. (2007) estimated the sensitivity at 40%, specificity at 94%, PPV at 59% and NPV at 87%. This poor correlation between ultrasound EFW and BW at the extremes of BW can be explained by the fact that the equations used are insufficiently accurate because they were derived from populations of fetuses within a restricted range of BW (Mongelli 1997). Previous work (Little et al. 2012) has suggested that sonographic overestimation of BW may be associated with an elevated risk of CD and that women with a false diagnosis of macrosomia are at increased risk of CD compared with BW-matched control patients. Melamed et al. (2010) found that the CD rate was 2 to 2.5 times higher when the EFW was 4000 to 4499 g, regardless of the actual BW. Given the low sensitivity and high NPV obtained in this work, we think that the risk is misdiagnosis of macrosomia, which may result in neonatal morbidity. CONCLUSIONS Sonographic estimation of fetal weight is one of the most common ways to assess the growth of a fetus in utero to evaluate an ongoing pregnancy or prepare for delivery. This examination is routinely performed at term and during labor by trainees and is as accurate as when performed by experienced sonographers. Nevertheless, the
US fetal weight estimation d K. DIMASSI et al.
relatively high percentage error undermines the accuracy of this examination and possibly affects clinical decisions regarding delivery. In addition to the inherent errors of this method, various clinical and technical factors may affect the accuracy of the EFW and must be borne in mind when decisions regarding obstetric management are formulated. REFERENCES Acker DB, Sach BP, Ransil BJ, Friedman EA. Ultrasonography for fetal weight estimation: The Birnholz equation. Ultrason Imaging 1987; 9:195–202. Ashwal E, Hiersch L, Melamed N, Bardin R, Wiznitzer A, Yogev Y. Does the level of amniotic fluid have an effect on the accuracy of sonographic estimated fetal weight at term? J Matern Fetal Neonatal Med 2015;28:638–642. Barel O, Maymon R, Vaknin Z, Tovbin J, Smorgick N. Sonographic fetal weight estimation—Is there more to it than just fetal measurements? Prenat Diagn 2014;34:50–55. Barel O, Vaknin Z, Tovbin J, Herman A, Maymon R. Assessment of the accuracy of multiple sonographic fetal weight estimation formulas. J Ultrasound Med 2013;32:815–823. Ben-Aroya Z, Segal D, Hadar A, Hallak M, Friger M, Katz M, Mazor M. Effect of OB/GYN residents’ fatigue and training level on the accuracy of fetal weight estimation. Fetal Diagn Ther 2002;17:177–181. Blackwell SC, Refuerzo J, Chadha R, Carren CA. Overestimation of fetal weight by ultrasound: Does it influence the likelihood of cesarean delivery for labor arrest? Am J Obstet Gynecol 2009;200: 340.e1–340.e3. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1: 307–310. Blann DW, Prien SD. Estimation of fetal weight before and after amniotomy in the laboring gravid woman. Am J Obstet Gynecol 2000;182: 1117–1120.
5
Brunader R. Accuracy of prenatal sonography performed by family practice residents. Fam Med 1996;28:407–410. Dudley NJ. A systematic review of the ultrasound estimation of fetal weight. Ultrasound Obstet Gynecol 2005;25:80–89. Durbin SA, Lee CW, Parker VG. The effect of amniotic fluid index on the accuracy of sonographic estimated fetal weight. J Diagn Med Sonogr 2005;21:329–335. Edwards A, Goff J, Baker L. Accuracy and modifying factors of the sonographic estimation of fetal weight in a high-risk population. Aust N Z J Obstet Gynaecol 2001;41:187–190. Farrell T, Holmes R, Stone P. The effect of body mass index on three methods of fetal weight estimation. BJOG 2002;109:651–657. Field NT, Piper JM, Langer O. The effect of maternal obesity on the accuracy of fetal weight estimation. Obstet Gynecol 1995;86: 102–107. 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: 333–337. Little SE, Edlow AG, Thomas AM, Smith NA. Estimated fetal weight by ultrasound: A modifiable risk factor for cesarean delivery? Am J Obstet Gynecol 2012;207:309.e1–309.e6. Melamed N, Yogev Y, Meizner I, Mashiach R, Ben-Haroush A. Sonographic prediction of fetal macrosomia the consequences of false diagnosis. J Ultrasound Med 2010;29:225–230. Mongelli M. Fetal weight estimation by symphysis fundus height and gestational age. Gynecol Obstet Invest 1997;43:20–24. Noumi G, Collado-Khoury F, Bombard A, Julliard K, Weiner Z. Clinical and sonographic estimation of fetal weight performed during labor by residents. Am J Obstet Gynecol 2005;192: 1407–1409. Peregrine E, O’Brien P, Jauniaux E. Clinical and ultrasound estimation of birth weight prior to induction of labor at term. Ultrasound Obstet Gynecol 2007;29:304–309. Perni SC, Predanic M, Cho JE, Kalish RB, Chasen ST. Association of amniotic fluid index with estimated fetal weight. J Ultrasound Med 2004;23:1449–1452. Predanic M, Cho A, Ingrid F, Pellettieri J. Ultrasonographic estimation of fetal weight: acquiring accuracy in residency. J Ultrasound Med 2002;21:495–500.