The definition of estimated fetal weight below the 10th percentile on obstetric ultrasonography includes fetuses classified as small for gestational age (SGA), as well as those meeting diagnostic criteria for intrauterine growth restriction (IUGR). Conversely, large for gestational age (LGA) refers to fetuses with an estimated fetal weight above the 90th percentile [1]. Historically, a cutoff of 2.500 grams has been used to define low birth weight in newborns [2].

Both groups are associated with an increased risk of adverse neonatal outcomes. This may occur due to birth-related injuries in fetuses classified as LGA or increased perinatal morbidity and mortality in those classified as SGA [1][3][4], in both cases, there is also a higher likelihood of increased morbidity later in life [5].

The combination of maternal factors and ultrasonographic parameters at the appropriate gestational age has the potential to predict up to 90% of SGA neonates, including those at risk for adverse outcomes such as prematurity, perinatal hypoxia, hypothermia, and hypo-glycemia. This provides essential data for the development of strategies aimed at improving prenatal care [1].

On the other hand, fetuses classified as LGA are also associated with unfavorable outcomes [10], including shoulder dystocia, operative delivery, neonatal asphyxia, emergency cesarean section, severe perineal lacerations, postpartum hemorrhage, and, later in life, metabolic syndrome [6][7].

Third-trimester ultrasonographic screening can double the detection rate of fetuses at risk of being SGA or LGA and is best performed at approximately 36 weeks of gestation [1][5][8].

Finally, accurate estimation of fetal weight by ultrasonography requires adherence to well-established criteria. For this purpose, the Hadlock-4 formula was used, as described in the ISUOG Practice Guidelines [9]. Both SGA and LGA neonates present an increased risk of morbidity and mortality, making fetal weight estimation essential during prenatal care and delivery planning. Early identification and appropriate monitoring of these conditions may help detect potential complications and enable healthcare professionals to minimize associated risks. Therefore, the aim of this study was to evaluate the accuracy of ultrasonographic estimated fetal weight compared to actual birth weight in fetuses with estimated fetal weight (EFW) < 2500 g and ≥4000 g.

The general characteristics of the sample are presented in Table 1. Among the 98 included cases, 96 had EFW < 2500 g and 2 had EFW ≥ 4000 g. Because of the very small number of cases in the ≥4000 g subgroup, subgroup-specific estimates were interpreted cautiously. Most mothers self-reported white skin color and did not have a completed higher education degree. Maternal age ranged from 14 to 43 years, and gestational age at the time of the measurements ranged from 26 to 40 weeks. The prevalence of alcohol consumption and/or smoking was approximately 20%. Two mothers reported illicit drug use. More than half of the mothers were hypertensive, and slightly more than 15% had diabetes.

Table 2 presents the indicators of the relationship between ultrasound-estimated fetal weight (EFW) and neonatal birth weight (BW). EFW values obtained by ultrasonography were compared with neonatal birth weight to evaluate the accuracy of fetal weight estimation. A strong and statistically significant correlation was observed between the variables. Ultrasound-estimated fetal weight explained approximately 85% of the variation in birth weight. Adjustment variables contributed little to the model, reinforcing that birth weight prediction can be considered accurate regardless of maternal age and gestational age. The estimates obtained during the first three months (r = 0.903; adjusted β = 0.697; R² = 0.827) did not differ from those obtained during the remaining months (r = 0.960; adjusted β = 0.912; R² = 0.914); however, better estimates were observed after the initial months.

Table 1. General characteristics of the sample (n = 98).

Note: Missing or incomplete data were reported when applicable and excluded from variable-specific analyses. Education level: n = 90 due to missing data. Missing: 8; 8.2%.

Table 2. Correlation and explanatory capacity of ultrasound-estimated fetal weight in relation to birth weight.

Note: CI = confidence interval; R² = coefficient of determination; adjusted β = adjusted for maternal age and gestational age.

The Bland-Altman plot (Figure 2) demonstrates the agreement between ultrasound-estimated fetal weight and birth weight.

Figure 2. Agreement between ultrasound-estimated fetal weight and birth weight in grams (Bland-Altman plot). Note: SD = standard deviation.

A mean difference of 60.5 g was observed between the two methods, with limits of agreement (margin of error) ranging from −488.3 g to 609.3 g. Regarding categorical agreement (<2500 g or ≥4000 g), both methods classified 90% of the cases within the same category. No differences were observed between the estimates obtained during the first three months of residency training (bias = 60.2 [−536.4; 656.8]) and those obtained during the subsequent months (bias = 61.1 [−366.9; 489.1]) in the agreement analysis.

Based on the correlation analysis, it was possible to verify that fetal weight estimated by ultrasonography performed within 14 days prior to delivery showed a strong and significant correlation with birth weight, demonstrated good agreement with neonatal birth weight in fetuses with extreme estimated fetal weight. These results are consistent with the study conducted by Ben-Haroush et al. [12], which demonstrated that ultrasonographic estimated fetal weight (EFW) is highly correlated with birth weight, further reinforcing that obstetric ultrasonography has become an increasingly established imaging method for this purpose, with the advantage of being a low-cost and widely accessible examination in high-risk ser-vices at tertiary hospitals, such as HUSM.

However, the authors emphasize that physicians should be aware of the risk of overestimation in pregnancies with suspected large-for-gestational-age (LGA) fetuses and underestimation in pregnancies complicated by premature rupture of membranes and suspected intrauterine fetal growth restriction.

Correlation analysis was used to assess the strength of association between ultrasound-estimated fetal weight and birth weight [13]. A higher correlation coefficient indicates a stronger linear relationship between estimated and observed values. However, correlation alone does not assess agreement or clinical accuracy, since highly correlated measurements may still present clinically relevant individual differences. For this reason, agreement and clinical agreement were additionally evaluated using Bland-Altman analysis and error-based measures, including mean absolute error (MAE), mean percentage error (MPE), and the proportion of estimates within ±10% of birth weight. These complementary approaches provide a more clinically meaningful interpretation of ultrasonographic performance.

The present study is consistent with much of the literature in supporting that the equation proposed by Hadlock et al. remains one of the best options for predicting birth weight, with strong potential for clinical practice [14][15] in screening for fetal growth abnormalities [10]. Ricci et al. [16] highlighted that this formula, when compared with other formulas such as those proposed by Warsof et al. and Shepard et al., demonstrated greater predictive accuracy [16]-[18]. In the study conducted by Ricci et al. [16], adequate accuracy in fetal weight estimation using the Hadlock et al. formula, based on four parameters, was observed, and factors interfering with estimation error included the distance between the maternal skin and uterus, as well as fetal weight itself.

Regarding fetal growth, an expected weight gain of 30 grams per day from the third trimester onward was considered, regardless of the presence of maternal pathology. This may represent a limitation of the present study, since growth-restricted fetuses are expected to gain less intrauterine weight compared to fetuses considered “healthy.” On the other hand, fetuses exposed to hyperglycemic conditions tend to present greater daily weight gain compared to those from non-diabetic mothers.

Furthermore, it was observed that the ability of estimated fetal weight to predict actual birth weight was dependent on gestational age, with a higher rate of accuracy observed during the third trimester of pregnancy. Conversely, there was a tendency to overestimate estimated fetal weight (EFW) in fetuses weighing more than 4000 g, with a variation exceeding 10% between EFW and actual birth weight. However, the small number of cases in this category does not allow definitive conclusions regarding this finding, which may also be considered a limitation of the present study. The lower accuracy of the Hadlock formula in fetuses with suspected macrosomia may be related to the limitations of conventional two-dimensional biometric measurements, which may not fully reflect soft tissue deposition and overall fetal body composition. In larger fetuses, adipose tissue accumulation and body volume may not be adequately represented by standard parameters such as abdominal circumference, femur length, biparietal diameter, and head circumference, potentially reducing precision in fetal weight estimation. Similar limitations have been reported in previous studies evaluating ultrasonographic fetal weight estimation in macrosomic fetuses. The Bland–Altman limits of agreement also deserve clinical consideration, particularly in fetuses at the extremes of weight. In suspected macrosomic fetuses, an estimation error approaching 600 g may influence obstetric planning and risk assessment. For example, a fetus estimated near the threshold for macrosomia may in practice have a substantially lower or higher birth weight, potentially affecting delivery planning, anticipation of shoulder dystocia, or closer intrapartum monitoring. Therefore, estimated fetal weight should be interpreted together with clinical findings and not used in isolation for obstetric decision-making. Another limitation related to the study design was the absence of blinding of ultrasound measurements displayed on the monitor for the examiner, which may have influenced the obtained values.

Nevertheless, this study may be considered an evaluation of the agreement and accuracy of EFW in clinical practice, since EFW measurements were obtained by several examiners at different stages of training. It was hypothesized that the first three months of fetal medicine residency training could result in discrepancies in the measurements obtained due to the possible inexperience of new residents. However, no differences were observed between the estimates obtained during the first three months and those obtained during the remaining months of the year. This finding suggests that adequate super-vision and teaching were provided by the HUSM fetal medicine service throughout the entire training period of future specialists, considering the interval from March to February of the following year.

Furthermore, studies [15][16] have indicated that there is interobserver variability in estimated fetal weight (EFW) measurements, which may compromise its performance for clinical use. However, this was not observed in the present study, providing evidence that variability in obtaining EFW measurements may be reduced through the standardization of measurement techniques. According to Predanic et al., regarding the learning curve in fetal weight estimation, a significant improvement in accuracy is observed among residents in training after 24 months; however, even among experienced operators, interobserver differences may still occur [19].

According to Ricci et al., an adequate interpretation of fetal growth requires understanding the natural variability of ultrasonographic measurements [16]. The authors also emphasized that operator experience plays an important role in obtaining more accurate fetal weight estimates. The importance of proper training and qualification of obstetric ultrasonography operators should be highlighted in order to ensure the quality and accuracy of fetal weight measurements and estimations [16]. This approach may contribute significantly to successful obstetric follow-up and more informed clinical decision-making. Krispin et al. emphasized that ultrasound accuracy mainly depends on fetal presentation, spinal position, and placental location [15].

This study has some limitations that should be considered when interpreting the results. First, the retrospective design may introduce selection bias and limit the ability to establish causal relationships. In addition, although strict inclusion and exclusion criteria improved measurement standardization and comparability, they may have reduced external validity and contributed to potential selection bias, particularly considering the reduction from 693 eligible cases to 98 included patients. Second, the relatively small sample size, especially the limited number of fetuses with estimated fetal weight ≥ 4000 g, restricted subgroup analyses and reduced the stability of accuracy estimates for this category. In larger fetuses, the lower accuracy of the Hadlock formula in fetuses with suspected macrosomia may be related to the limitations of conventional two-dimensional biometric measurements, which may not fully reflect soft tissue deposition and overall fetal body composition. In larger fetuses, adipose tissue accumulation and body volume may not be adequately represented by standard parameters such as abdominal circumference, femur length, biparietal diameter, and head circumference, which may contribute to reduced precision in fetal weight estimation.

Ultrasound-estimated fetal weight showed good agreement with neonatal birth weight in fetuses with estimated fetal weight (EFW) < 2500 g and ≥4000 g. Most ultrasound estimations presented errors within clinically acceptable limits, and more than half of the examinations showed an estimation error within ±10% of actual birth weight. Ultrasound estimated fetal weight also demonstrated good explanatory capacity, explaining approximately 85% of the variation in birth weight. No relevant differences were observed between measurements obtained by examiners at different stages of training. However, findings regarding fetuses with EFW ≥ 4000 g should be interpreted cautiously due to the limited number of cases in this subgroup.