- Top of page
Ultrasound examination and measurement of fetal biometry has become an integral part of modern obstetric care. These measurements can serve for dating pregnancies or for assessment of fetal growth. However, to ensure accurate diagnosis, selection of the appropriate cross-sectional reference charts is of great importance. Some published reference charts are methodologically flawed. Common problems include repeated measurements on the same fetuses, formation of ‘super normal’ datasets by inappropriate exclusion of complicated pregnancies, failure to identify the statistical method of analysis, and the use of statistical methods which do not consider the variability of measurements with gestational age1. Appropriate methodology has been published1, 2, and fetal biometry charts and equations for various populations using the correct methodology are now available in the medical literature3–8.
It is well known that ethnicity has a significant influence on fetal biometry9, 10. Cross-sectional reference charts and equations from the Hong Kong ethnic Chinese population using appropriate methodology have not previously been published in the medical literature. The aim of this study, therefore, was to construct prospectively reference charts and equations for fetal biparietal diameter (BPD), head circumference (HC), abdominal circumference (AC) and femur length (FL) in the Hong Kong ethnic Chinese population using the recommended methodology1, and to compare these charts with published data from other populations3–5, 8, 11, 12.
- Top of page
This was a prospective observational study conducted in a university obstetric unit in Hong Kong over a 2-year period. This unit caters for both low- and high-risk obstetric populations, and ethnically over 98% of parturients were Chinese at the time of the study. Subjects were invited to participate in the study when they booked for a routine first-trimester dating ultrasound examination at our unit. Inclusion criteria were: (1) both parents ethnically Chinese; (2) no maternal medical disease; (3) singleton pregnancy; (4) sure menstrual dates and a regular 28–32-day cycle; (5) difference in gestational age according to last menstrual period and according to fetal crown–rump length (CRL) measurement in the first trimester13 of ≤ 4 days.
Subjects who fulfilled the inclusion criteria and who gave informed consent to participate in the study were assigned an appointment randomly at a gestational age of between 12 and 40 weeks, at which fetal biometric measurements were performed. This ultrasound examination was arranged specifically for the purposes of the study and each subject was examined sonographically only once for the study. Pregnancies that were found subsequently to be complicated by congenital abnormalities, were excluded from the construction of the nomogram. Cases with low birth weight, preterm delivery or other prenatal complications were not excluded.
All fetal biometric measurements were obtained by one of two sonographers. Realtime transabdominal ultrasound was performed using an Apogee 800PLUS (Apogee ATL, Bothell, WA, USA) ultrasound machine equipped with a 3.5-MHz curvilinear transabdominal probe. Fetal head measurements were made in the axial plane at the level where the continuous midline echo was broken by the cavum septi pellucidi in the anterior third and both thalami could be seen symmetrically14. Care was taken to ensure that the calvaria appeared smooth and symmetrical bilaterally. Fetal skull BPD was measured from the outer edge of the proximal calvarial wall both to the inner edge of the distal calvarial wall (outer–inner; BPDoi) and to the outer edge of the distal calvarial wall (outer–outer; BPDoo). Fetal HC was measured by placing the calipers over the landmarks for BPDoo and fitting a computer-generated ellipse to include the outer edges of the calvarial margins of the fetal skull. AC was measured on a transverse circular plane of the fetal abdomen at the level where the spine, descending aorta, anterior third of the umbilical vein and stomach bubble could be seen in the same plane15. FL was measured in a plane where the full femoral diaphysis was seen almost parallel to the transducer and the measurement was made from one end of the diaphysis to another5. In the third trimester, particular care was taken not to include the epiphysis5 in the measurement.
Statistical analysis was performed using the Statistical Package for Social Sciences, version 13.0 (SPSS Inc., Chicago, IL, USA), as described by Altman and Chitty1. Briefly, polynomial regression models were fitted to the measurements as a function of gestational age. The models were chosen based on the correlation coefficient, R2. Since the residuals were also dependent on gestational age, a polynomial regression analysis was performed between the absolute residuals and gestational age. The fitted values of this regression model were multiplied by √(π/2)( = 1.253), to give gestation-specific SDs. Centiles were calculated using the formula: centile = mean + K × SD, where K is ± 1.88 for 3rd and 97th centiles, ± 1.645 for 5th and 95th centiles and ± 1.28 for 10th and 90th centiles. To produce charts for dating of pregnancy, formulae were obtained by fitting the gestational age against the BPD, HC and FL16.
We compared our biometric measurements with those published from UK3–5, French8, Singaporean11 and central-south mainland Chinese12 populations, using the method described by Salomon et al.8: for each gestational age from 16 to 37 weeks, the 50th centiles of the published populations were calculated from the published equations3–5, 8, 11. For the central-south mainland Chinese population, the published median values were used, as such equations were not available12. The data were then expressed as Z-scores calculated with our reference equations using the formula: Z-score = (XGA − MGA)/SDGA, where XGA is data from other populations at a known gestational age, MGA is the mean value for our population calculated from the reference equations at this gestational age, and SDGA is the SD associated with the mean value at the same gestational age from our population. Results were presented graphically across the different gestational ages to allow visual comparison.
- Top of page
Included for analysis were fetal biometric measurements from 709 singleton pregnancies. Not all five fetal biometric measurements could be obtained in a few fetuses due to unfavorable fetal position or early gestational age. The numbers of valid observations for BPDoo, BPDoi, HC, AC and FL at each gestational week are shown in Table 1.
Table 1. Numbers of valid observations for biparietal diameter (BPD), head circumference (HC), abdominal circumference (AC) and femur length (FL) according to gestational age (GA) in completed weeks in Hong Kong Chinese singleton pregnancies
The raw data were fitted to the gestational age in weeks (GA) satisfactorily with a cubic polynomial model (Figure 1). The corresponding formulae for the regression models and their correlation coefficient (R2) were as follows (all biometric measurements in cm):
Figure 1. Raw data with 3rd, 50th and 97th fitted centiles in a Hong Kong Chinese population for: (a) biparietal diameter (outer–inner; BPDoi) (n = 705) (▴) and femur length (FL) (n = 708) (○), (b) biparietal diameter (outer–outer) (n = 705), (c) head circumference (n = 706) and (d) abdominal circumference (n = 679).
Download figure to PowerPoint
The absolute residuals of each biometric measurement across gestational age were fitted satisfactorily using a simple linear fit. The equations for the SDs were as follows (with SD in cm and GA in weeks):
Figure 2 gives the nomograms for each biometric measurement, including the 3rd, 5th, 10th, 50th, 90th, 95th and 97th centiles.
Figure 2. Fetal biometric charts with 3rd, 5th, 10th, 50th, 90th, 95th and 97th fitted centiles in a Hong Kong Chinese population for: (a) biparietal diameter (outer–inner; BPDoi) (above) and femur length (FL) (below), (b) biparietal diameter (outer–outer), (c) head circumference and (d) abdominal circumference.
Download figure to PowerPoint
For estimation of gestational age, a cubic polynomial model fitted all data satisfactorily (Figure 3). The formulae and SDs were as follows (with GA in weeks and all biometric measurements in cm):
Figure 3. Raw data with 3rd, 50th and 97th fitted centiles in a Hong Kong Chinese population for gestational age plotted against: (a) biparietal diameter (outer–inner; BPDoi) (n = 705) (▴) and femur length (FL) (n = 708) (○), (b) biparietal diameter (outer–outer) (n = 705) and (c) head circumference (n = 706).
Download figure to PowerPoint
Figure 4 shows the Z-scores for the 50th centiles of each of the biometric measurements in the four different previously published populations3–5, 8, 11, 12. Only BPDoo was used in the comparison of BPD measurements as most publications did not report measurements on BPDoi. The Z-score for the HC of the central-south mainland Chinese population was not calculated as data on HC were not available in the original publication12.
Figure 4. Comparison of our new equations for fetal biometry in a Hong Kong Chinese population with published data for French8 (●), UK3–5 (▴), Singaporean11 (▪) and central-south mainland Chinese12 (⧫) populations for: (a) biparietal diameter (outer–outer), (b) head circumference (data from central-south mainland Chinese population not available), (c) abdominal circumference and (d) femur length. The solid lines represent the expected Z-scores for the 3rd, 50th and the 97th centiles, calculated from our population.
Download figure to PowerPoint
- Top of page
This study was designed to provide reference equations for biometric measurements of a Hong Kong Chinese population between 12 and 40 weeks of gestation. The data were collected prospectively for the purposes of this study rather than being obtained from an available clinical dataset and we adhered strictly to the recommended methodology1, 2. Data for the study were obtained from only one designated ultrasound examination from each woman. After recruitment, we excluded only pregnancies complicated by congenital abnormalities. Our sample size of 709 was not as big as that in some other publications6, 8, 11, but was similar to that of Chitty et al.3–5. The larger the sample size, the greater will be the precision of resulting centiles; generally, for charts of fetal size, a sample of several hundred is sufficient to give a reasonable estimate of extreme centiles1. We chose a lower gestational age limit of 12 weeks because the nomograms for BPD and FL are useful at such early gestational ages, when an ideal fetal position for measurement of the crown–rump length cannot be obtained.
As well as providing fetal reference charts for use among the Hong Kong Chinese population, our study has established equations for sonographic dating of pregnancy. As previously indicated, size charts should not, in principle, be used for dating pregnancies, and accurate dating requires consideration of gestational age as a function of fetal size16. Based on our formulae, conversion tables for gestational age can be constructed based on fetal biometric measurements as well as the uncertainty in days.
In daily practice in our unit, we use BPDoi instead of BPDoo as the reference for BPD. We produced two BPD charts, one for BPDoo and another one for BPDoi, as in the publication by Chitty et al.3, because we are aware that different units might have their own preference in measuring BPD. The publication of both also allows comparison with other populations that provide only data on BPDoo6, 11, 12.
We used Z-scores to compare our fetal biometric measurements with those of the UK3–5, French8, Singaporean11 and central-south mainland Chinese12 populations across different gestational ages, with a Z-score of 1 representing a difference of 1 SD at that gestational age for our equation. There were some differences in the 50th centiles of the different populations (Figure 4).
The BPD measurements of all four populations were quite compatible with each other, although the UK population was slightly bigger (Figure 4a). This difference was more obvious in the HC measurements. It is perhaps surprising to find such a difference in HC between UK and French populations, although the studies differed in the way in which BPD was measured8. In the French population, it was measured by placing the calipers in the middle of the echo of each side of the fetal skull and the difference between the two methods ranged from 1.5 to 3 mm8. Compared with the SDGA of our equation for BPDoo, which ranged from 2.8 to 3.5 mm between 16 and 37 weeks of gestation, the difference between the two methods contributed to a Z-score of between 0.5 and 1. Comparing the HC between the French population and ours (Figure 4b), the Z-scores were progressively more negative in later gestation. This was not observed for BPD (Figure 4a). We believe that this difference in HC later in gestation has a methodological cause: we measured HC by fitting a computer-generated ellipse, while in the French population it was calculated/estimated by a formula using the BPD and occipital–frontal diameter, and there might be under-measurement of the occipital–frontal diameter later in gestation.
The AC of the four populations were also quite similar, although the UK and French populations were slightly larger, while the Singaporean and the central-south mainland Chinese populations were slightly smaller, compared with ours (Figure 4c).
A bigger difference was observed in the FL measurements in the four populations. Both the UK and French populations had longer femurs compared with our population, with a difference of around 1 SD across all gestations. The FL measurements of the Singaporean population were very similar to those in our population. The FL measurements from the central-south mainland Chinese population were unexpected: not only were their mean FL values scattered around our 97th centile, they were also larger than those of the UK and French populations. We do not believe that fetuses of the ethnically Chinese population had longer femurs compared with the Caucasian populations, and it is difficult to explain these findings; we suspect the difference might be due to the way in which FL was measured. The difference in FL measurements among the different populations illustrates the importance of selecting charts appropriate for one's own population.
In conclusion, we have constructed a new set of reference centiles for fetal biometric measurements and equations for dating of pregnancy for Hong Kong Chinese singleton pregnancies. Our charts were very similar to those of the Singaporean population. The main difference between our fetal biometric results and those of the UK and French populations was in the FL measurements. We believe that our charts are ready for clinical use and research among appropriate ethnic Chinese groups.