Sonographic estimation of fetal head circumference: how accurate are we?

Authors

  • N. Melamed,

    Corresponding author
    1. Department of Obstetrics and Gynecology, Helen Schneider Hospital for Women, Rabin Medical Center, Petach Tikva and Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel
    • Department of Obstetrics and Gynecology, Helen Schneider Hospital for Women, Rabin Medical Center, Petah Tiqwa 49100, Israel
    Search for more papers by this author
  • Y. Yogev,

    1. Department of Obstetrics and Gynecology, Helen Schneider Hospital for Women, Rabin Medical Center, Petach Tikva and Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel
    Search for more papers by this author
  • D. Danon,

    1. Department of Obstetrics and Gynecology, Helen Schneider Hospital for Women, Rabin Medical Center, Petach Tikva and Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel
    Search for more papers by this author
  • R. Mashiach,

    1. Department of Obstetrics and Gynecology, Helen Schneider Hospital for Women, Rabin Medical Center, Petach Tikva and Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel
    Search for more papers by this author
  • I. Meizner,

    1. Department of Obstetrics and Gynecology, Helen Schneider Hospital for Women, Rabin Medical Center, Petach Tikva and Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel
    Search for more papers by this author
  • A. Ben-Haroush

    1. Department of Obstetrics and Gynecology, Helen Schneider Hospital for Women, Rabin Medical Center, Petach Tikva and Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel
    Search for more papers by this author

Abstract

Objectives

To assess the accuracy of sonographic estimation of fetal head circumference (HC).

Methods

We compared sonographic estimations of fetal HC with actual measurements performed immediately after delivery using 3008 sonographic examinations performed within 3 days prior to delivery. The following measures of accuracy were calculated: correlation with actual HC, systematic error, random error, simple error, mean absolute percentage error and fraction of estimates within 5% of actual HC. Multivariate logistic regression analysis was used to identify factors affecting the accuracy of sonographic HC estimation.

Results

There was a high correlation between sonographic and postnatal measurements of HC (r = 0.845, P < 0.001). Overall, sonographic HC measurements consistently underestimated actual HC measured postnatally (mean simple error, − 13.6 mm; 95% CI, − 13.2 to − 13.9), and the difference increased with gestational age. A high cephalic index (> 0.81) (odds ratio (OR), 0.3; 95% CI, 0.2–0.4), HC > 90th centile (OR, 0.5; 95% CI, 0.3–0.6), delivery by vacuum extraction (OR, 0.6; 95% CI, 0.4–0.8), gestational week (OR, 0.7; 95% CI, 0.6–0.9) and male fetal gender (OR, 0.8; 95% CI, 0.6–0.9) were associated with decreased sonographic accuracy. At term, breech presentation at the time of sonographic examination was associated with a higher sonographic accuracy compared with vertex presentation (−12.0; 95% CI, − 10.5 to − 13.5 vs. − 13.9 mm; 95% CI, − 13.6 to − 14.3; P = 0.02). The random error was relatively constant, and was unaffected by any of the obstetric factors studied.

Conclusion

Sonographic estimation of HC is associated with significant underestimation compared with the actual postnatal HC. This measurement error may have important clinical implications and should be taken into account in the interpretation of sonographically measured HC. Copyright © 2011 ISUOG. Published by John Wiley & Sons, Ltd.

Introduction

Accurate sonographic estimation of fetal head circumference (HC) is important for the purpose of fetal weight estimation1, 2, as well as in cases in which abnormal fetal head growth is suspected. Surprisingly, there are few data regarding the accuracy of sonographic estimation of HC compared with actual postnatal HC3–5, and while some have found that sonographic measurements underestimate actual HC3, 5 others have found the differences between sonographic and actual HC to be statistically insignificant4. In addition to providing inconsistent results, these studies are further limited by a small sample size4, 5, a wide range of time interval between sonographic examination and delivery4, and differences in the timing of postpartum HC measurement. Furthermore, there are no data available regarding factors that may affect the accuracy of sonographic estimation of HC, such as fetal gender, fetal presentation at time of sonographic examination, mode of delivery and head shape.

The aims of our study were to assess the accuracy of sonographic estimation of fetal HC in an unselected cohort of women undergoing sonographic examination up to 3 days prior to delivery, and to identify factors that affect the accuracy of sonographic HC estimation.

Subjects and Methods

Data collection

This was a retrospective cohort study. Data were collected from a comprehensive database of sonographic examinations in a single center. Routine sonographic evaluations in the Rabin Medical Center include standard fetal biometry (abdominal circumference (AC), femur length (FL), biparietal diameter (BPD) and HC) and the findings are saved directly to the database. Antenatal data, gestational age at delivery, mode of delivery, and actual postnatal HC measurements were obtained from the hospital's perinatal database. The study was approved by the local institutional review board.

Study population

The database was searched for all sonographic fetal weight estimations performed within 3 days prior to delivery between the years 2002 and 2008. Inclusion criteria for the study were live birth, singleton pregnancy, birth weight > 500 g and gestational age at delivery > 24 weeks. Cases complicated by fetal malformations or hydrops, and cases in which there was uncertainty regarding gestational age, were excluded from the study.

Definitions

In our center, gestational age at the time of examination is recorded in the database along with the details of the sonographic examination and is calculated according to last menstrual period (LMP). When first-trimester ultrasound was available, the LMP was corrected based on the crown–rump length (CRL) when the discrepancy between the calculated LMP (based on Hadlock's CRL reference tables6) and the reported LMP exceeded 7 days, according to the recommendations of the American College of Obstetricians and Gynecologists7. The gestational age at the time of examination was further verified by comparing the interval (in days) between the ultrasound examination date and the delivery date with the interval between the gestational age at the time of examination and that at delivery (the latter being obtained from the perinatal database). Since these intervals are expected to be identical (considering that gestational age in both cases should have been calculated using the same LMP), cases in which the difference between these intervals was greater than 1 day were excluded.

All sonographic fetal weight estimations are performed in the ultrasound unit of the obstetrics and gynecology department. Weight estimations are performed by senior physicians who are ultrasound specialists or by experienced ultrasound technicians. In the latter case, the examination is reviewed and signed by a senior physician.

The BPD was measured from the proximal edge of the fetal skull to the proximal edge of the deep border (outer–inner) at the level of the cavum septi pellucidi. The HC was measured as an ellipse around the perimeter of the fetal skull8. The cephalic index was calculated as the ratio between the BPD and the occipitofrontal diameter (OFD)9. The cephalic index was defined as low or high when it was 1 SD below or above, respectively, the mean cephalic index of the study population.

Postnatal measurements of HC are routinely performed by a trained pediatrician within 2–6 h after delivery using a flexible measuring tape. The head measurement is made along the maximal horizontal plane along the occipital prominence at the back, above the ears, and just above the eyebrows at the front.

Measures of accuracy

Sonographic estimations of HC were compared with the actual HC measured postnatally, and the following measures of accuracy were calculated: correlation with actual HC (using Pearson's correlation coefficient); systematic error (mean of (sonographic HC − actual HC)/actual HC × 100), which reflects the systematic deviation from actual postnatal HC, expressed as a percentage of the actual HC; random error (SD of the systematic error × 100), a measure of precision (rather than accuracy) that reflects the random (or non-systematic) component of the prediction error; mean absolute percentage error (mean of absolute value of (sonographic HC − actual HC)/actual HC × 100), which reflects the unsigned deviation from the actual HC (in contrast to the systematic error) and is expressed as a percentage of the actual HC; simple error (sonographic HC − actual HC), which reflects the signed deviation, in millimeters, from the actual HC; and fraction of estimates within 5% of the actual HC.

Statistical analysis

Data analysis was performed with SPSS v15.0 software (SPSS Inc., Chicago, IL, USA). The one-sample t-test was used to assess whether the simple and systematic errors were significantly different from zero, and the one-sample Kolmogorov–Smirnov test was used to assess the same for the mean absolute percentage error. When comparing measures of accuracy for different subgroups, one-way analysis of variance (ANOVA) was used for the simple and systematic errors, Levene's test (equality of variance) for the random error and the Kruskal–Wallis test for the mean absolute percentage error. Bonferroni corrections were used as necessary to maintain an overall type I error rate of 0.05 within each set of comparisons. Multivariate logistic regression analysis was used to identify factors affecting the accuracy of sonographic HC estimation. Differences were considered significant when P < 0.05.

Results

Characteristics of the study group

A total of 3008 pregnancies met the inclusion criteria. The demographic and obstetric characteristics of these women are presented in Table 1. Most of the sonographic examinations were performed at term (88.5%) and within 48 h prior to delivery (67%).

Table 1. Demographic and obstetric characteristics of the study population
CharacteristicValue
  1. Data are presented as mean ± SD, n (%) or median (interquartile range). BPD, biparietal diameter; OFD, occipitofrontal diameter.

n3008
Maternal age (years)29.3 ± 5.1
Parity2 (1–3)
Nulliparous1227 (40.8)
Gestational age at delivery (weeks)38.9 ± 2.1
 Delivery at < 37 weeks346 (11.5)
 Delivery at < 34 weeks50 (1.7)
Time from fetal weight estimation to delivery (days)1 (0–2)
Cases with fetal weight estimation performed: 
 On day of delivery835 (27.8)
 1 day prior to delivery1183 (39.3)
 2 days prior to delivery598 (19.9)
 3 days prior to delivery392 (13.0)
Diabetes233 (7.7)
Male fetal gender1587 (52.8)
Birth weight (g)3248 ± 644
Head circumference 
 Sonographic (mm)328.9 ± 17.1
 Cephalic index (BPD/OFD)0.78 ± 0.03
 Actual (postnatal) (mm)342.5 ± 18.0

Relationship between sonographic and postnatal HC

Overall, the mean sonographic HC was lower than the postnatal HC (Table 1). This tendency for sonographic underestimation of HC persisted throughout gestation (Figure 1), and became more pronounced as gestational age increased, reaching a mean difference of − 14.4 mm (95% CI, − 12.6 to − 16.2 mm) or − 4.0% (95% CI, − 3.5 to − 4.5%) at 42 weeks (Figure 1). The simple and systematic errors (reflecting the tendency towards underestimation of HC) as well as the mean absolute percentage error (reflecting the unsigned deviation from actual HC) were all significantly different from zero (Table 2, first row). There was a high correlation between the sonographic and postnatal measurements of HC (r = 0.845, P < 0.001, Table 2).

Figure 1.

Simple error of the sonographic estimation of fetal head circumference (HC) according to gestational age at delivery. Values represent the difference between sonographic HC and actual HC (measured postnatally, within 3 days following sonographic evaluation), expressed in millimeters. The solid line represents the mean simple error of the sonographic estimation of HC at each gestational week. Negative and positive values reflect under- and overestimation of actual HC, respectively.

Table 2. Demographic and obstetric factors affecting the accuracy of sonographic estimation of fetal head circumference (HC)
FactornCorrelation with actual HCMean simple error (mm (95% CI))Mean absolute percentage error (% (95% CI))Sonographic HC within ± 5% of actual HC (% (95% CI))Systematic error (% (95% CI))Random error (% (95% CI))
  • *

    Significantly different from 0 (P < 0.001, using the one-sample Student's t-test for simple error and systematic error, and the one-sample Kolmogorov–Smirnov test for mean-absolute percentage error). Measures of accuracy for the different subgroups within each factor were compared using one-way analysis of variance (ANOVA) for the simple and systematic errors, Levene's test (equality of variance) for the random error, and the Kruskal–Wallis test for the mean absolute percentage error. For factors having more than two subgroups, Bonferroni corrections for multiple comparisons were used as necessary to maintain an overall type I error rate of 0.05. In these cases, the subgroup which was significantly different is indicated (†).

  • One, two and three missing values in the parity, gender and mode of delivery groups, respectively.

Overall30080.845− 13.6 (−13.2 to − 13.9)*4.2 (4.1–4.2)*66.1− 3.9 (−3.8 to − 4.0)*2.8 (2.7–2.9)
Parity
 Parous17800.851− 13.6 (−13.2 to − 14.1)4.1 (4.0–4.3)66.9− 3.9 (−3.8 to − 4.1)2.8 (2.7–2.9)
 Nulliparous12270.836− 13.6 (−13.0 to − 14.1)4.2 (4.0–4.3)65.1− 3.9 (−3.8 to − 4.1)2.8 (2.7–2.9)
 P  0.80.70.30.90.01
Gestational age       
 < 34 weeks500.964− 9.1 (−6.7 to − 11.4)†3.6 (3.0–4.2)76.0− 3.1 (−2.3 to − 3.9)†2.9 (2.4–3.4)
 34 to 36 + 6 weeks2960.806− 11.9 (−10.8 to − 13.0)4.0 (3.7–4.3)71.3− 3.7 (−3.3 to − 4.0)2.9 (2.7–3.1)
 37 to 39 + 6 weeks12760.784− 13.7 (−13.1 to − 14.2)4.1 (4.0–4.3)66.9− 4.0 (−3.8 to − 4.1)2.8 (2.2–2.9)
 ≥ 40 weeks13860.660− 14.1 (−13.5 to − 14.6)4.2 (4.1–4.3)64.0− 4.0 (−3.8 to − 4.1)2.8 (2.7–2.9)
 P  < 0.0010.0960.030.060.6
Diabetes
 No27750.845− 13.6 (−13.2 to − 13.9)4.1 (4.0–4.2)66.3− 3.9 (−3.8 to − 4.0)2.8 (2.7–2.9)
 Yes2330.849− 14.3 (−13.1 to − 15.5)4.3 (4.0–4.6)64.4− 4.1 (−3.8 to − 4.5)2.6 (2.4–2.8)
 P  0.30.30.560.30.9
Gender
 Male15870.846− 14.5 (−14.0 to − 14.9)4.3 (4.2–4.4)62.7− 4.1 (−4.0 to − 4.3)2.6 (2.5–2.7)
 Female14190.834− 12.6 (−12.1 to − 13.2)4.0 (3.9–4.1)70.0− 3.7 (−3.5 to − 3.9)3.0 (2.9–3.1)
 P  < 0.001< 0.001< 0.001< 0.0010.18
Mode of delivery
 Normal vaginal20420.807− 13.7 (−13.2 to − 14.2)4.2 (4.0–4.3)65.8− 4.0 (−3.8 to − 4.1)2.7 (2.6–2.8)
 Vacuum extraction2470.764− 15.4 (−14.1 to − 16.8)†4.5 (4.2–4.8)†57.5†− 4.4 (−4.0 to − 4.8)†2.8 (2.6–3.0)
 Cesarean section7160.894− 12.8 (−12.3 to − 13.4)4.0 (3.9–4.2)68.8− 3.7 (−3.5 to − 3.9)2.8 (2.7–2.9)
 P  0.0010.0130.0110.0020.6

Factors affecting accuracy of sonographic HC estimation

Tables 2 and 3 present the sonographic HC measurement error, stratified by different demographic, obstetric (Table 2) and sonographic (Table 3) characteristics that were thought to have the potential to affect measurement error. The measurement error and the degree of underestimation of HC were significantly higher in the case of male fetuses, fundal (compared with anterior or posterior) placenta, gestational age ≥ 34 weeks, high cephalic index (i.e. brachycephalic, compared with normal and dolichocephalic head shape) and postnatal HC > 90th centile. All these factors affected the systematic error (i.e. they resulted in a consistent deviation) but had no effect on the random measurement error (which is a measure of precision rather than accuracy) (Tables 2 and 3). Amniotic fluid abnormalities, parity, fetal presentation and maternal diabetes did not affect the sonographic measurement error. The difference between the sonographic and postnatal HC was also related to mode of delivery, being highest following vacuum extraction and lowest in cases of Cesarean delivery (Tables 2 and 3).

Table 3. Sonographic factors affecting the accuracy of sonographic estimation of fetal head circumference (HC)
FactornCorrelation with actual HCMean simple error (mm (95% CI))Mean absolute percentage error (% (95% CI))Sonographic HC within ± 5% of actual HC (% (95% CI))Systematic error (% (95% CI))Random error (% (95% CI))
  • *

    90th centile for postnatal head circumference in our study population = 363 mm.

  • Cephalic index cut-off values were selected based on the mean ± SD values of cephalic index in the study group (see Table 1).

  • Measures of accuracy for the different subgroups within each factor were compared using one-way analysis of variance (ANOVA) for the simple and systematic errors, Levene's test (equality of variance) for the random error, and the Kruskal–Wallis test for the mean absolute percentage error. For factors having more than two subgroups, Bonferroni corrections for multiple comparisons were used as necessary to maintain an overall type I error rate of 0.05. In these cases, the subgroup(s) which were significantly different are indicated (‡).

Placental location
 Anterior14690.810− 13.4 (−12.9 to − 13.9)4.1 (4.0–4.2)67.3− 3.9 (−3.7 to − 4.0)2.9 (2.8–3.0)
 Fundal2650.816− 14.8 (−13.6 to − 15.9)‡4.4 (4.1–4.7)63.4− 4.3 (−3.9 to − 4.6)‡2.8 (2.6–3.0)
 Posterior11430.882− 13.8 (−13.3 to − 14.3)4.2 (4.0–4.3)64.6− 4.0 (−3.9 to − 4.2)2.6 (2.5–2.7)
 P  0.0490.090.290.040.5
Amniotic fluid index
 50–250 mm21170.840− 13.5 (−13.1 to − 13.9)4.1 (4.0–4.2)67.0− 3.9 (−3.8 to − 4.2)2.8 (2.7–2.9)
 ≤ 50 mm4610.826− 13.7 (−12.7 to − 14.6)4.3 (4.0–4.5)65.7− 4.0 (−3.7 to − 4.3)3.0 (2.8–3.2)
 ≥ 250 mm1710.810− 14.1 (−12.8 to − 15.5)4.2 (3.8–4.5)65.5− 4.0 (−3.6 to − 4.3)2.5 (2.3–2.7)
 P  0.60.60.150.50.4
Presentation
 Vertex26850.837− 13.7 (−13.3 to − 14.1)4.2 (4.1–4.3)65.5− 4.0 (−3.8 to − 4.1)2.7 (2.6–2.8)
 Breech1980.827− 12.8 (−11.0 to − 14.6)4.0 (3.5–4.5)72.2− 3.8 (−3.3 to − 3.8)3.8 (3.5–4.1)
 P  0.30.10.10.70.6
HC > 90th centile*
 No27120.842− 12.9 (−12.5 to − 13.2)4.0 (3.9–4.1)69.0− 3.8 (−3.7 to − 3.9)2.7 (2.6–2.8)
 Yes2950.261− 20.0 (−18.9 to − 21.1)5.4 (5.1–5.7)44.0− 5.4 (−5.1 to − 5.7)2.6 (2.4–2.8)
 P  < 0.001< 0.001< 0.001< 0.0010.2
Cephalic index
 < 0.753290.915− 9.1 (−8.0 to − 10.2)‡3.2 (3.0–3.5)‡80.2‡− 2.7 (−2.3 to − 3.0)‡2.9 (2.7–3.1)
 0.75–0.8119560.854− 13.0 (−12.6 to − 13.4)‡4.0 (3.9–4.1)‡68.9‡− 3.8 (−3.6 to − 3.9)‡2.6 (2.5–2.7)
 > 0.816920.833− 17.3 (−16.6 to − 18.0)‡5.0 (4.9–5.2)‡51.7‡− 5.0 (−4.8 to − 5.2)‡2.6 (2.5–2.7)
 P  < 0.001< 0.001< 0.001< 0.0010.057

In order to adjust for possible confounding effects, we used multivariate logistic regression analysis, with sonographic HC within ± 5% of actual HC as the dependent variable. Brachycephaly (and, to a lesser degree, normal-shaped head) compared with dolichocephalic head, postnatal HC > 90th centile, delivery by vacuum extraction, gestational week and male fetal gender were significantly and independently associated with a decreased likelihood of the sonographic prediction being within 5% of the actual HC (Table 4). In cases with cephalic index > 0.81, HC > 90th centile, male fetus and gestational age > 37 weeks (n = 65) the simple error was − 24 mm (95% CI, − 22 to − 26 mm), and the systematic error was − 7.2% (95% CI, − 4.3 to − 10.1%).

Table 4. Factors affecting the accuracy of sonographic estimation of fetal head circumference (HC); multivariate analysis
FactorOdds ratio (95% CI)
  • Values reflect the results of multivariate logistic regression analysis, using sonographic HC within ± 5% of actual HC as the dependent variable.

  • *

    Compared with dolichocephalic head shape (cephalic index < 0.75).

  • 90th centile for postnatal head circumference in our study population = 363 mm.

  • Compared with normal vaginal delivery and Cesarean delivery.

Brachycephaly (cephalic index > 0.81)*0.29 (0.20–0.41)
Head circumference > 90th centile0.47 (0.34–0.65)
Vacuum extraction0.61 (0.43–0.85)
Normal-shaped head (cephalic index 0.75–0.81)*0.65 (0.47–0.90)
Gestational week0.74 (0.61–0.90)
Fetal male gender0.76 (0.63–0.91)

Effect of gestational age

To better understand the effect of the factors described above on the sonographic HC measurement error, we analyzed the effects of fetal presentation, fetal gender, cephalic index and postnatal HC with respect to gestational age at the time of the sonographic examination (Figure 2). Breech presentation was associated with an underestimation of postnatal HC that was relatively constant throughout gestation (Figure 2a), while, in contrast, vertex presentation was associated with a lower measurement error prior to 34 weeks, but the error increased with increasing gestational age, being higher than for breech presentation at term (Figure 2a, − 13.9 mm (95% CI, − 13.6 to − 14.3) vs. − 12.0 (95% CI, − 10.5 to − 13.5), P = 0.02). The higher measurement errors observed in cases of male (compared with female) fetal gender (Figure 2b), increasing cephalic index (Figure 2c) and postnatal HC > 90th centile (Figure 2d) persisted throughout gestation and were independent of gestational age.

Figure 2.

Simple error of the sonographic estimation of fetal head circumference (HC) according to gestational age at delivery, stratified by: (a) fetal presentation (●, vertex; ▴, breech); (b) fetal gender (●, male; ▴, female); (c) cephalic index (●, < 0.75; ▴, 0.75–0.81; □, > 0.81); and (d) postnatal HC (●, ≤ 90th centile; ▴, > 90th centile). Values represent the difference (mean and 95% CIs) between sonographic HC and actual HC (measured postnatally, within 3 days following sonographic evaluation), expressed in millimeters. Negative values reflect underestimation of actual HC.

Discussion

In this study we sought to assess the accuracy of sonographic estimation of fetal HC compared with actual HC measured postnatally, and to identify factors that affect the accuracy of HC estimation. Our study has three key findings: (1) sonographic measurements of HC consistently underestimate actual postnatal HC, by an average of ∼13.5 mm or 4%; (2) high cephalic index, postnatal HC > 90th centile, more advanced gestational age, male fetal gender, and vertex presentation (at term) are associated with an even greater tendency for sonographic underestimation of postnatal HC; and (3) the random error is relatively constant and is unaffected by any of the obstetric factors studied.

Despite the fact that sonographic measurement of HC is used routinely for the purpose of estimation of fetal weight or when abnormality of fetal head growth is suspected, there are very few data regarding the accuracy of this sonographic measure compared with the actual HC measured after delivery, and there are no data available regarding the obstetric and sonographic factors that affect the sonographic measurement error. Hadlock et al.3, in a study of 400 fetuses, found that the sonographic HC estimations were comparable to postnatal standards published previously by others (mean difference, − 0.94 ± 0.47 mm)10, but after 35 weeks of gestation the sonographic measurements were consistently smaller (by 4–8 mm) than the postnatal measurements. The same authors reported similar observations in another group of 100 fetuses at > 35 weeks of gestation for which HC was measured within 24 h of delivery (mean difference − 3.8 mm)3. Similarly, Fescina and Ucieda5 found that sonographic measurements of HC (n = 14) significantly underestimated postnatal HC (mean difference, − 11 mm). In contrast, Deter et al.4 evaluated sonographic HC measurements performed 1–10 days prior to delivery (n = 34) and reported the sonographic estimations to be similar to the postnatal measurements of HC (mean difference, 0.46%). These conflicting results may be related to the small sample sizes4, 5, differences in references values (i.e. published nomograms3 vs. self measurement of postnatal HC4, 5), differences in the time interval between the sonographic examination and delivery as well as in the timing of postnatal HC measurement (6 h3 vs. 2 h4 postpartum), and lack of information regarding factors that may affect the sonographic accuracy of HC estimation. In this study, we tried to overcome these limitations by analyzing a large, unselected cohort of women undergoing sonographic examination up to 3 days prior to delivery, for which data regarding obstetric and sonographic characteristics were available.

The reason for the consistent sonographic underestimation of postnatal HC is unclear. One possible cause, previously suggested by Hadlock et al.3, is that at term it can be difficult to distinguish fetal scalp from contiguous soft tissue of the uterus, so that sonographic measurements include only the bony calvaria of the fetal skull. Our finding that the differences between sonographic and actual HC were related to the mode of delivery (highest for vacuum extraction and lowest for Cesarean delivery) support the idea that the fetal scalp, which may not be fully represented in sonographic measurements at term, can significantly affect the overall HC. Another possible explanation may be the technical difficulty in obtaining the appropriate sonographic plane for HC measurement when the head is engaged. Our observations that the measurement error increased with gestational age and was higher in cases of vertex (compared with breech) presentation at term provide support to this assumption. The calculation of HC based on BPD and OFD (rather than direct measurement of HC) may also lead to underestimation of HC when the BPD is measured using the outer–inner technique. Finally, it has been suggested3 that at term the fetal head is often too large to be fitted totally into the screen, which may result in approximations being made for the HC ellipse. The extremely high measurement error in cases of HC > 90th centile that we observed (Table 3) supports this possibility.

Another issue that needs to be addressed is the reliability of the postnatal HC measurement, which may also affect the correlation between sonographic and postnatal HC measurements. Several studies have reported extremely high correlation coefficients for both intraobserver (0.999) and interobserver (0.979) measurements11, 12.

We found that the measurement error was consistently higher among male compared with female fetuses, even when controlling for gestational age and head size. The reason for this observation is unclear. It is possible that gender-related differences in the shape of the skull13 or in the thickness of soft tissues may have contributed. The finding that the measurement error was highest following vacuum extraction and lowest in cases of Cesarean delivery is probably related to the associated delivery-induced changes in head shape (i.e. molding and caput succedaneum) rather than a reflection of a genuine sonographic measurement error.

We also found that the random error was relatively constant (∼3%), and was unaffected by any of the obstetric factors evaluated in the current study. This finding is in concordance with previous studies, which reported the random error associated with HC measurement to be 3%4 and 3.8%14.

In conclusion, we found that the sonographic estimation of HC is associated with significant underestimation of the actual HC measured immediately postnatally, that the difference increases with increasing gestational age, and that the difference is even more pronounced in cases of a male fetus, vertex presentation, high cephalic index and large HC. This measurement error may have important implications and should be taken into account in the interpretation of sonographically measured HC in certain clinical scenarios. For example, in cases of borderline fetal weight estimation under the circumstances described above (i.e., advanced gestational age, high cephalic index, large HC, vertex presentation or male fetus), the use of alternative equations for fetal weight estimation that are based on AC and FL alone should be considered. Similarly, when sequential measurements of HC are performed during the course of pregnancy because of suspected abnormality of fetal head growth, the increasing underestimation of actual HC that occurs with increasing gestational age should be taken into consideration, especially when postnatal reference charts are used. Further studies are needed to determine whether other modalities, such as three-dimensional ultrasound or magnetic resonance imaging, can provide more accurate estimation of HC under these circumstances.

Ancillary