- Top of page
Objective To create reliable reference ranges and calculate Z scores for fetal head ultrasound biometry using a large sample size which is evenly distributed from 12 to 42 weeks of pregnancy.
Design A prospective, cross-sectional study.
Setting Obstetric clinics (outpatient and delivery units) at the University Hospital of Zurich.
Sample The study data were obtained from 6557 pregnant women.
Methods Only the first ultrasound examination between 12 and 42 weeks of each fetus with exactly established gestational age was used for analysis. No exclusions were made on the grounds of small-for-date birthweight, prematurity or other events several weeks after the examination. Separate regression models were fitted to estimate the mean and standard deviation at each gestational age for each parameter.
Results A total of 6217 fetal head biparietal diameters and 5510 occipito-frontal diameters were measured. Both head circumference and cephalic index were derived in 5462 cases where both biparietal diameter and occipito-frontal diameter could be measured on the same fetus. The centile charts, tables and regression formulae for biparietal and occipito-frontal diameters, head circumference and cephalic index are presented. An application to calculate 2 scores was developed using Excel (Microsoft Corporation, USA) and macros are presented in detail in the Figure 8 footnote. The comparison of our charts with those of the two most recent studies revealed almost no differences in biparietal diameter centiles. In one publication, occipito-frontal diameter charts, and in another, head circumference charts were different from the current study.
Conclusions We have presented centile charts, tables and regression formulae for fetal head ultrasound biometry derived from a large and minimally selected sample size in a carefully designed cross-sectional study. Complete tables and regression formulae to calculate reference ranges and Z scores are presented for use in computer-aided evaluation of fetal ultrasound biometry.
- Top of page
Measurements of the fetal biparietal diameter (BPD) and head circumference (HC) have become established methods of assessing fetal size1–3 and determining gestational age4,5. Reference ranges for fetal head ultrasound biometry have been reported by a number of investigator3,5–13. Royston and Wright14 and Altman15 have published methodology on how to design time-related reference ranges. According to them, reference curves should change smoothly with gestational age, fit well to the data at all ages and be mathematically as simple as possible. The common problems with published reference values and curves for fetal biometry include failure to identify the statistical method of analysis, centiles that do not change smoothly during gestation9, a ‘super normal’ data set, inadequate consideration of the changes in the variability of the measurements with gestation, and lack of scatter diagrams of the data with the fitted centiles superimposed16. The beginning and the end of the reference ranges contain significantly fewer observations compared with the middle part in most of the studies. In any case, many of the existing charts were produced using equipment now obsolete. Regression formulae have not been presented. The recent publications by Chitty et al.17 and Merz and Wellek18 correspond to the recommendations proposed by Royston and Wright14 and Altman15.
One main objective of our study was to create reliable reference ranges for fetal ultrasound biometry using a large sample size which is evenly distributed from 12 to 42 weeks of gestation. In contrast to both preceding studies, the measurements were done by many examiners and, as in the study of Chitty et al.17, only minimal exclusions were made. Another important objective was the calculation of Z scores, a very useful tool for evaluation in perinatal biology. Z scores are extensively used for data evaluation in paediatrics and now are increasingly used in perinatal medicine19–21. In this paper we are presenting all the necessary formulae required for computer-aided evaluation and documentation of fetal biometry data.
- Top of page
The study data were obtained from 6557 pregnant women routinely examined in obstetrics clinics (outpatient and delivery units) at the University Hospital of Zurich. The database, constructed by means of the dedicated Perisono programme22, comprised over 60,000 records with serial fetal ultrasound biometry data from the time period between January 1993 and July 1997. For each fetus only one routine ultrasound examination after the 12th and not later than the 42nd week of pregnancy was used for analysis. Informed consent was obtained from the women to use examination data in the study. Mean (SD) maternal age was 29.4 (5.7) years; 3270 were nulliparae and 3287 multiparae. Maternal booking weight was 62.0 (23.1) kg, height 161.9 (7.16) cm and body mass index (BMI) 23.5 (4.45). The country of origin of the women in the study is presented in Table 1.
Table 1. Country of origin of women in the study.
|Country of origin||Women n (%)|
|Former Yugoslavia||1309 (20.0)|
|Great Britain||22 (0.3)|
|Remaining Europe||287 (4.4)|
|Middle East||64 (10)|
Criteria for exclusion from the study were as follows: uncertain date of last menstrual period; missing the first trimester (dating) ultrasound examination; multiple pregnancies; pregnancies involving congenital malformations; pregnancies complicated by hypertension (arterial blood pressure 140/90 mmHg or greater), pre-eclampsia (arterial blood pressure 140/90 mmHg or greater with proteinuria of at least 0.3 g/24 hours) or severe (insulin dependent) maternal diabetes mellitus. No exclusions were made on the grounds of small-for- date birthweight, prematurity or other events several weeks after the examination.
Gestational age was determined from the date of the last menstrual period and confirmed by measurement of first trimester crown-rump length (CRL)23. Gestational age was corrected to the ultrasound value if the difference between the ultrasound estimated gestational age and the menstrual gestational age exceeded 5 days.
Examinations were carried out by many operators (senior house officers and registrars) using 3.5 MHz transducers (Acuson 128 XP4 and Acuson 128 XP10, Acuson Inc, USA; Aloka SSD-650, Aloka Ltd, Japan: Hitachi EUB-415, Hitachi Ltd, Japan).
Fetal head measurements were all made in the plane described by Campbell and Thoms1 as well as Hansmann2. This is an anatomic reference plane for the concurrent measurement of the BPD, occipito-frontal diameter (OFD), and head circumference (HC) at the level where the continuous midline echo is broken by the cavum septum pellucidum in the anterior third. Measurements of the BPD were made from fetal skull skin to fetal skull skin (outer-outer, skin-skin). This is accepted as the standard method in Germany and in Switzerland13. The OFD was measured in the same plane between the leading edge of the frontal bone and the outer border of the occiput. The head circumference was estimated from the measurement of the OFD and BPD using the formula for an ellipse:
The cephalic index (CI) was calculated as the ratio of the two diameters (BPD/OFD).
Z scores for all parameters were calculated using the following formula:
Where X is the measured value, M(GA), is the mean value for the appropriate gestational age and SD(GA) is the standard deviation for the appropriate gestational age.
The data were analysed according to the statistical methods described by Royston and Wright4 and Altman and Chitty16. The relation between the mean of each measurement and gestational age was modelled by a fractional polynomial. A full description of fractional polynomials and procedures for selecting the best fitting model are given in the appendix of Royston and Wright14. A standard deviation (SD) curve for each measurement was estimated by regressing the ‘scaled absolute residuals’ on gestational age, again using fractional polynomials. The scaled absolute residuals are the absolute differences (i.e. differences with the sign removed) between the measurements and the fitted mean curve, multiplied by the scaling factor 1.25315. For all the measurements, the SD increased with gestational age. The goodness of fit of each regression model was carefully assessed. The Shapiro-Francia W′ test24 was performed to check the normality of the Z scores. The reference limits at each gestational age were calculated as fitted mean ± 1.645 (fitted SD). Ninety-five percent intervals were calculated to indicate the precision of the 5th and 95th centiles. All statistical calculations and graphics were made using Stata (Stata Corporation, USA) and Microsoft Excel 4.0 (Microsoft Corporation, USA).
- Top of page
Fetal head measurements were obtained in 6557 pregnant women: due to unfavourable fetal position, the BPD could not be measured in 340, OFD in 1047, and calculated quantities like HC and CI in 1095 cases. Table 2 gives the number of observations per week of gestation.
Table 2. Number of examinations of biparietal diameter (BPD), occipito-frontal diameter (OFD), head circumference (HC) and cephalic index (CI) at different gestational weeks.
| ||Examinations (n)|
|Gestational age(weeks + days)||BPD||OFD||HC and CI|
|12 to 12 + 6||524||231||230|
|13 to 13 + 6||381||259||258|
|14 to 14 + 6||302||276||272|
|15 to 15 + 6||581||531||529|
|16 to 16 + 6||510||480||477|
|17 to 17 + 6||276||263||258|
|18 to 18 + 6||217||190||187|
|19 to 19 + 6||176||169||167|
|20 to 20 + 6||171||164||161|
|21 to 21 + 6||140||131||131|
|22 to 22 + 6||114||111||111|
|23 to 23 + 6||93||87||86|
|24 to 24 + 6||104||104||102|
|25 to 25 + 6||86||80||79|
|26 to 26 + 6||103||103||102|
|27 to 27 + 6||119||117||116|
|28 to 28 + 6||126||119||119|
|29 to 29 + 6||116||117||116|
|30 to 30 + 6||132||128||127|
|31 to 31 + 6||132||128||127|
|32 to 32 + 6||157||154||153|
|33 to 33 + 6||165||156||156|
|34 to 34 + 6||177||167||166|
|35 to 35 + 6||173||169||166|
|36 to 36 + 6||209||205||202|
|37 to 37 + 6||164||157||155|
|38 to 38 + 6||170||160||159|
|39 to 39 + 6||210||198||195|
|40 to 40 + 6||295||268||267|
|41 to 42||94||88||88|
In cases of the BPD, OFD and HC, linear-cubic regression models were fitted to the mean and the linear model to the SD. The 50th centile for the CI was estimated by a quadratic curve, and the SD was fitted by a fractional polynomial model with inverse and logarithmic powers of gestational age25. All the models were excellent fits to the data. The Shapiro-Francia W′ normality test for BPD was 0.99956, P= 0.9; for OFD, 0.99956, P= 0.9; for HC, 0.99966, P= 0.9; and for CI, 0.99938, with P= 0.7, thereby confirming the normal distribution of SD scores for all the variables. The regression formulae are shown in Table 3. The 5th and 95th centiles can be calculated by subtracting and adding, respectively, 1.645SD(GA), to the mean.
Table 3. Regression formulae used to generate ultrasound biometry charts and tables of biparietal diameter (BPD), occipito-frontal diameter (OFD), head circumference (HC) and cephalic index (CI). W = gestational age in exact weeks; D = gestational age in days; Ln = natural logarithm.
| ||Regression formulae|
| ||With gestational age in exact weeks||With gestational age in days|
| Mean||BPD =−28.04 + 4.18W −0.0006404W3||BPD =−28.04 + 0.597D −0.00000187D3|
| SD||SD = 1.648 + 0.0653W||SD = 1.648 + 0.00933D|
| Mean||OFD =−39.08 + 5.454W −0.001004W3||OFD =−39.08 + 0.779 1D −0.000002926D3|
| SD||SD = 1.266 +0.1216W||SD = 1.266 + 0.01738D|
| Mean||HC =−106.0 + 15.22W −0,002616W3||HC =−106.0 + 2.174D −0.000007626D3|
| SD||SD = 4.857 + 0.2249W||SD = 4.857+ 0.03213D|
| Mean||CI = 0.9383 −0.0101W + 0.0002031W3||CI = 0.9383 −0.001443D + 0.000004146D3|
| SD||SD =−0.1058 + 1.895/W + 0.07598Ln (W/10)||SD = 0.07873 + 13.27/D + 0.07598Ln (D/100)|
Figures 1–4 show the raw data for each measurement superimposed on the 5th, 50th and 95th centiles. Tables 4 and 5 show estimated centiles and SDs at each week of gestation as well as the standard errors for the 5th and 95th centiles.
Table 4. Fitted centiles, SD and standard errors (SE) for 5th and 95th centile of biparietal diameter and occipito-frontal diameter.
| ||Biparietal diameter (mm)||Occipito-frontal diameter (mm)|
|Week of gestation||5th||50th||95th||SD||SE||5th||50th||95th||SD||SE|
Table 5. Fitted centiles, SD and standard errors (SE) for 5th and 95th centile of head circumference and cephalic index.
| ||Head circumference (mm)||Cephalic index|
|Week of gestation||5th||50th||95th||SD||SE||5th||50th||95th||SD||SE|
Fig. 5 A shows the comparison of our BPD reference chart with that of Chitty et al.17, who measured 663 fetuses (outer-outer measurement, 5th and 95th centiles) and Fig. 5 B with that of Merz18 who measured 2032 fetuses. Chitty's chart had BPD centiles which were somewhat lower. A small difference (only a slight decrease in all centiles at the end of pregnancy) was found with Merz's BPD chart. Bigger differences were found in OFD and HC charts at the end of pregnancy. OFD centiles were higher in Chitty's charts after 25–27 weeks of pregnancy (Fig. 6 A). In Merz's OFD charts, the 5th and 95th centiles were very close together (Fig. 6 B). Head circumferences were similar in our and Chitty's charts from early pregnancy up to 32–35 weeks (Fig. 7 A). After that, our HC centiles lie lower. Very big HC chart differences were found with those of Merz (Fig. 7 B), where his 5th centile is similar to our 50th centile at the end of pregnancy.
Figure 5. Comparison of our biparietal diameter (outer-outer) 5th, 50th and 95th centiles (solid lines) with (A) those of Chitty et al.17 (dashed lines) and (B) those of Merz and Wellek18 (dashed lines).
Download figure to PowerPoint
Figure 6. Comparison of our occipito-frontal diameter (outer-outer) 5th, 50th and 95th centiles (solid lines) with (A) those of Chitty et al.17 (dashed lines) and (B) those of Merz and Wellek18 (dashed line).
Download figure to PowerPoint
Figure 7. comparison of our head circumference (calculated) 5th, 50th and 95th centiles (solid lines) with (A) those of Chitty et al.17 (dashed lines) and (B) those of Merz and Wellek18 (dashed line).
Download figure to PowerPoint
A computer application to calculate Z scores (Fig. 8) was developed using Excel (Microsoft Corporation, USA). The macros are presented in detail in the Figure 8 footnote.
- Top of page
The standards of fetal ultrasound biometry, particularly head biometry, was an object of interest in the late 70s and early 80s, after Willocks et al.26 published their paper in 1964 on fetal ultrasound cephalometry. Many reference charts and tables have been published since then1,4,8. However, a number of these were produced using old ultrasound equipment with low spatial resolution and different ultrasound velocities compared with today's ultrasound machine27. In the last 10 years the quality of ultrasound imaging has increased enormously, and this has opened up new improved measurement techniques. In addition many of the earlier publications had methodological flaws, falling short of the ideal attributes of gestational age-related reference curve design, namely: non-identification of the statistical method of analysis, a supernormal data set, inadequate account of the change in variability of measurement with gestation and failure to present scatter diagrams of the data with fitted centiles superimposed. In the publications of Royston and Wright14 and Altman and Chitty16, methodological guidelines were created for the development of age-related reference ranges. A series of articles of Chitty and co-authors 17,28,29 as well as a publication of Merz and Wellek18 revised the reference ranges of fetal ultra-sound biometry. However, in the first one there were not many ultrasound examinations and the beginning of the reference curves were not well covered with measurements, and in the second one, a ‘supernormal’ study cohort was used.
We used over 6000 ultrasound examinations to generate the ultrasound biometry reference ranges for the fetal head. The data were evenly distributed from 12 to 42 weeks of pregnancy. Many women had routine ultrasound examinations with complete biometry at term before delivery, which is routine practice in our clinic. This enabled us to have a high number of measurements at the end of the gestation period (Table 2).
The data were collected by many observers in our study. It is logical that the variability of measurements for one examiner will be much less than for many examiners. However, in the construction of reference intervals, it is more relevant to clinical practice for the variability to be based on several examiners. Even the most experienced examiner not can be free of inherent errors in his measurements. We have chosen to compare our data with the most recent studies17,18 which mostly correspond to the recommendations proposed by Royston and Wright14 and A1tman16. We found only minimal differences in centiles, with the exception of Chitty's OFD and Merz's HC charts. OFD is very difficult to measure at the end of pregnancy and the small amount of measurements with high variability in Chitty's study might be the reason for the wider outer centiles, because no difference in any of the centiles in the first and second trimester of pregnancy was found. The ellipse formula:
used to calculate HC in Merz's publication overestimates HC by 5% in comparison with ellipse formulas used by other authors30–33 and the ellipse formula presented in the book of scientific tables34. In all of Merz's charts (BPD, OFD, HC), higher centiies in early and lower centiles in late pregnancy were found. We assume that this is due to different regression models used by Merz or because there were less data at the beginning and the end of the reference ranges. However, we can not confirm the latter assumption, because the number of measurements in each of the different weeks of pregnancy was not given.
The close outer centiles may be because Merz and co-authors used stricter selection criteria (included only fetuses with a normal course of pregnancy and term delivery) and only one examiner performed the examinations. Wider outer centiles in our charts can be due to the fact that many examiners took part in our study. Although only three examiners took part in the study of Chitty, the distances between the outer centiles of our charts and Chitty's were very similar. However, Chitty's selection criteria were more similar to ours than Merz's. The minimal exclusions can be another reason for the higher variability of the data.
An obvious way of finding the members with very high or very low values in a dimensional spectrum is to rank them and note their centile locations or their positions at opposite ends of the spectrum. This process will identify the extreme members of the spectrum but it will not indicate how greatly they deviate from the mean. Even if we subtract their values from the mean, the deviations will be absolute rather than standardised magnitudes. To indicate a standardised magnitude for unusual measurements, we can use the Z score technique. The standardised Z scores immediately tell us several things that were not necessarily obvious in the original data.
Under the assumption that the data have an approximate Gaussian distribution, a standardised Z score is calculated for each item by subtracting the gestation- specific mean, then dividing by the gestation-specific SD. This transforms the original data into a new set of Z values that are independent of gestational age and expressed in standard deviation units of the variable. The calculation of Z scores in fetal ultrasound biometry is complicated by the dependence of SD on gestational age. However, using Royston's and Altman's methodology to design the gestational age-related reference ranges, it is possible to calculate Z scores, since SD may be mathematically described as a function of gestational age.
We developed a simple application to calculate Z scores of fetal BPD, OFD and HC measurements. The introduction of Z scores in fetal ultrasound biometry may be very useful for the assessment of fetal size, since a measurement below the 5th centile will have a 2 score >−1.645, while those above the 95th centile will have a Z score >1.645. In most clinical studies, the small for gestational age (SGA) fetuses were defined as fetuses with HC, abdominal circumference and other biometry values lying under the 3rd, 5th or 10th centile. However, such a definition of SGA fetuses has short- comings. First of all, it does not show the degree of growth retardation. Second, it is impossible to compare different fetuses at different gestational ages. Third, only graphical presentation of biometry data can con- vincingly show the real situation.
Z scores let us assign a value to all fetal biometry values independent of gestational age. Dimension-free Z scores allow the parametric comparison of the values with different dimensions. For example, the resistance index in the umbilical cord artery and fetal abdominal circumference Z scores can be compared independently of gestational age.