Second trimester growth velocities: assessment of fetal growth potential in SGA singletons

Abstract

Objective

To evaluate the validity of 2^nd trimester growth velocities as measures of fetal growth potential in small-for-gestational-age (SGA) singletons.

Methods

Second-trimester growth velocities for biparietal diameter (BPD), head circumference (HC), abdominal circumference (AC), and femur diaphysis length (FDL) were determined by linear regression analysis or direct measurement in 53 SGA singletons with normal growth outcomes (SGA N Group) and 73 with growth restriction [SGA GR Group] based on a composite Fetal Growth Pathology Score (FGPS1). The latter were subdivided into six groups based on their growth restriction pattern (Patterns Group). Similar data were available for 118 singletons with normal neonatal growth outcomes (NNGO Group). Coefficients of determination (R²) and growth velocities for each anatomical parameter were compared between Patterns subgroups and the SGA N, SGA GR, and NNGO Groups.

Results

Median R² values in the six Patterns subgroups ranged from 98.2% (Pattern 2, FDL) to 99.9% (Pattern 5, AC). Within each anatomical parameter set, no significant differences were found (Kruskal-Wallis). Patterns subgroup data were pooled to form the SGA GR Group for each anatomical parameter. Mean values for the three main Groups ranged from 98.4% (SGA N, FDL) to 99.6% [SGA N, HC]. No significant differences between Groups (ANOVA) were found for any anatomical parameter (ANOVA). Only 1.7% to 3.8% had R² values < 95th%.

No significant differences in median 2^nd trimester growth velocities among different Patterns subgroups were found for any anatomical parameter. In the SGA N and SGA GR Groups, mean BPD and HC values did not differ but were significantly smaller than the NNGO Group values. No differences in mean FDL values were seen. With AC, all three means were significantly different, having the following order: NNGO > SGA N > SGA GR. Of all 504 2^nd trimester growth rates, 92.5% were within their respective 95% reference ranges.

Conclusions

Growth in the 2^nd trimester is linear in fetuses at risk for growth restriction. Except for FDL, growth velocities were lower than those for fetuses with NNGO. Only AC had mean velocities that differed between the SGA N and the SGA GR Groups. Since most velocities (92.5%) were within normal reference ranges, they are reasonable measures of growth potential in fetuses at risk for growth restriction.

INTRODUCTION

In the study of fetal growth and growth abnormalities, both size^1–7 and velocity (change in size over time)^8–16 have been utilized. Although the use of size is more common, growth velocity is a more logical definition of ‘growth’^14,17. Working with growth velocities, however, requires serial measurements, the definition of an appropriate interval between scans and a means for acquiring velocity data over a substantial range of fetal ages^11–16. As a result, only a limited number of studies have been carried out except for those based on Individualized Growth Assessment (IGA)^{8–10, 18–22}, which utilizes comparisons of expected and measured third-trimester average growth velocities⁸.

Although most studies of fetal growth velocity have focused on its use in detecting growth abnormalities, IGA utilizes this parameter primarily as a measure of growth potential (see Appendix⁹). Measurements of 2^nd trimester growth velocities are used to specify Rossavik size models which then generate expected 3^rd trimester size trajectories and provide predicted birth characteristics⁹. Good agreement between actual and expected measurement has been obtained in fetus/neonates with normal neonatal growth outcomes (NNGO), as measured by Percent Deviations (%Dev) prenatally⁹ and Growth Potential Realization Index (GPRI) values in the neonate¹⁸. Both %Dev’s and GPRI’s have been shown to be proportional to the difference between the actual and expected growth velocities in the third trimester⁸.

Because of easily met fetal nutritional requirements due to the small size and the minimal variability in individual fetal growth during the 2^nd trimester (see Appendix), growth velocity measurements during this trimester have been proposed to be a manifestation of genetic and other constituent growth controllers^9,23. These characteristics and others led us to consider the 2^nd trimester growth velocity as an observable variable which can give estimates of the latent variable, Fetal Growth Potential (Appendix). This concept, of course, implies that each anatomical parameter has its own growth potential and only some composite of growth velocities (currently undefined) could represent the growth potential of the fetus as a whole. The justification for considering 2^nd trimester growth velocities to be optimal estimates of fetal growth potential is more thoroughly described in the Appendix.

Second-trimester growth velocities have been studied extensively in 119 fetuses with NNGO⁹. In only one small group (25) of small-for-gestational-age (SGA) cases have these velocities been studied when growth pathology could have been present²¹. However, a recent investigation of 184 SGA singletons identified 53 Normal and 73 Growth-Restricted cases in which concordance between 3^rd trimester growth status and neonatal outcomes was observed¹⁰. Our investigation characterizes and compares 2^nd trimester growth velocities of these two groups to each other and in relationship to fetuses having NNGO. Our principal objective was to determine if these growth velocities could be considered appropriate estimates of growth potential in SGA singletons.

METHODS

The sample, methods, and analytical techniques used in this study have been described previously¹⁰. Only those details pertinent to the current investigation will be summarized here.

Sample

This investigation was carried out using a sample of SGA neonates (birth weight <10^th percentile) obtained retrospectively from three centers: the Perinatology Research Branch, Detroit, Michigan, USA (97 cases); University of Toronto, Toronto, Ontario, Canada (20 cases); and Baylor College of Medicine, Houston, Texas, USA (9 cases)¹⁰. Data were from high-risk pregnancies in patients seen at the Center for Advanced Obstetrical Care and Research (PRB), the Placenta Clinic (Toronto), and the Maternal-Fetal Medicine Ultrasound Clinic (Baylor) under approved protocols: WSU IRB Number 110605MP4F and NICHD Protocol Number OH97-CH-N067 (PRB); REB Number 13-0266-C (Toronto); and IRB Number H-35518 (Baylor). Scans were either part of a research protocol or clinically indicated.

Using IGA, 126 cases were divided into 53 having normal growth (SGA N) and 73 having third-trimester growth restriction (SGA GR), as identified with the Fetal Growth Pathology Score (FGPS1); head circumference (HC), abdominal circumference (AC), femur diaphysis length (FDL), and estimated weight (EWT))¹⁰. Neonatal growth outcomes, evaluated using the average negative, pathological Growth Potential Realization Index (av − pGPRI) (WT, HC, and crown-heel length (CHL)^{10, 21}, were all correspondingly normal or growth-restricted¹⁰. Abnormal growth patterns have been investigated previously in the growth-restricted group²². Five distinct patterns, designated Patterns 1–5, and Unclassifiable, were identified in the 3^rd trimester (SGA Patterns Group). For this study, the 73 SGA GR cases were initially subdivided into the six SGA Patterns subgroups.

For comparisons with SGA singletons, 2^nd trimester growth data from 118 singletons with NNGO, based on a multivariable, modified Neonatal Growth Assessment Score which corrected for differences in birth age and growth potential²⁴, were used in this investigation⁹. These data were obtained as part of a prospective, longitudinal study in a single institution where all ultrasound measurements were made by one research ultrasonographer or Maternal-Fetal Medicine specialist. Details concerning the selection of this sample can be found in a previous publication⁹.

Fetal Age Determination

Crown-rump length measurements (before 12 weeks) or a composite-age parameter (biparietal diameter [BPD], HC, AC, and FDL) measured before 16 weeks, MA were used to estimate fetal age as previously described¹⁰.

Assessment of Fetal Size Parameters

Ultrasound size measurements (BPD, HC, AC, FDL) were made between 14 and 28 weeks, MA, using methods previously described¹⁰. In the SGA N Group, the number of scans had a median of 3, with a 100% range of 2 to 5. For the SGA GR Group, the median number was 3 in all specific SGA Patterns subgroups, with a 100% range of 2 to 4 (exception [Pattern 2]: 2 to 5). The three unclassifiable cases had a median of 4 scans with a 100% range of 2 to 5.

Data Analysis

Linear Regression

Measurements of specific anatomical parameters in each fetus were used in linear regression analyses to obtain estimates of parameter growth velocities in the 2^nd trimester. The validity of the assumed linear model was evaluated using the coefficient of determination (R²). R² values were available in 43/53 (81.1%) of SGA N Group cases. In the 73 SGA GR Group cases, R² values were available in 23/27 (85.2%) of the Pattern 1 cases, 14/20 (70%) of the Pattern 2 cases, 8/9 (88.9%) of the Pattern 3 cases, 7/8 (87.5%) of the Pattern 4 cases, 6/6 (100%) of the Pattern 5 cases, and 2/3 (66.7%) of the unclassifiable cases. The missing R² values were those cases with only 2 scans in the 2^nd trimester. For the NNGO Group, R² values were available in all 118 cases.

Descriptive statistics were obtained for the four parameters in the six growth-restriction pattern subgroups. Because of the small sample sizes, Kruskal-Wallis one-way analysis was used to compare the R² values for each anatomical parameter in the six Patterns subgroups. The similarities between all parameters in all Patterns subgroups (Table 1) justified pooling the subgroups to give a single SGA GR Group. The R² values for each anatomical parameter in the SGA N Group (43 cases), the SGA GR Group (60 cases), and the NNGO Group (118 cases) (Table 2) were compared using ANOVA.

Table 1.

Evaluation of Linear Functions Used in the Second Trimester

		Coefficients of Determination (R2)
A. Patterns Groups1		BPD			HC			AC			FDL
		Q1	Median	Q3	Q1	Median	Q3	Q1	Median	Q3	Q1	Median	Q3
	n	%	%	%	%	%	%	%	%	%	%	%	%
Pattern 1	23	98.5	99.7	99.9	99.1	99.7	99.9	98.5	99.7	99.9	99.1	99.6	99.9
Pattern 2	14	98.1	99.0	99.8	99.3	99.7	99.9	98.4	99.5	100.0	97.3	98.2	99.1
Pattern 3	8	96.4	98.6	99.9	97.6	99.6	99.9	98.2	99.4	99.4	97.4	99.0	99.9
Pattern 4	7	99.4	99.5	99.9	99.7	99.9	100	93.6	98.5	99.7	98.1	99.5	100.0
Pattern 5	6	90.6	99.8	99.9	98.9	99.7	99.9	99.1	99.9	99.9	96.6	98.6	99.7
Pattern U	2	----	99.8	----	---	99.8	---	----	99.4	---	---	99.5	---
B. Main Groups			BPD			HC			AC			FDL
			Mean	SD		Mean	SD		Mean	SD		Mean	SD
	n		%	%		%	%		%	%		%	%
SGA N2	43		99.1	±1.3		99.6	±0.5		98.8	±1.3		98.4	±3.6
SGA GR3	60		98.2	±4.5		99.4	±0.9		98.7	±1.9		98.6	±1.6
NNGO4	118		99.0	±1.6		99.3	±1.1		99.1	±1.2		98.7	±1.8

¹Growth restriction patterns defined in reference 22

²Small-for-Gestational-Age Normal Group;

³Small-for-Gesiational-Age Growth Restricted Group

⁴Normal Neonatal Growth Outcome Group; BPD = biparietal diameter; HC = head circumference; AC = abdominal circumference, FDL = femur diaphysis length

Table 2.

Second Trimester Growth Velocities in Normal and SGA Groups

		Second Trimester Growth Velocity
		BPD			HC			AC			FDL
A. Patterns Groups1		Q1 cm/wk	Median cm/wk	Q3 cm/wk	Q1 cm/wk	Median cm/wk	Q3 cm/wk	Q1 cm/wk	Median cm/wk	Q3 cm/wk	Q1 cm/wk	Median cm/wk	Q3 cm/wk
Pattern 1	27	0.275	0.300	0.313	1.050	1.087	1.139	0.918	0.964	1.054	0.257	0.270	0.293
Pattern 2	20	0.276	0.299	0.300	1.027	1.084	1.157	0.908	1.006	1.100	0.221	0.245	0.285
Pattern 3	9	0.291	0.300	0.330	1.091	1.154	1.158	1.025	1.075	1.262	0.256	0.270	0.287
Pattern 4	8	0.250	0.284	0.310	1.020	1.131	1.192	0.959	1.024	1.071	0.249	0.265	0.292
Pattern 5	6	0.245	0.299	0.312	1.050	1.081	1.120	0.939	1.038	1.178	0.254	0.275	0.302
Pattern U	3	0.268	0.301	0.400	1.087	1.096	1.317	0.989	1.000	1.128	0.252	0.255	0.283
B. Main Groups			Mean cm/wk	SD cm/wk		Mean cm/wk	SD cm/wk		Mean cm/wk	SD cm/wk		Mean cm/wk	SD cm/wk
SGA N2	53		0.299	±0.033		1.124	±0.076		1.078	±0.108		0.262	±0.035
SGA GR3	73		0.295	±0.041		1.106	±0.099		1.017	±0.226		0.265	±0.030
NNGO4	118		0.319	±0.039		1.164	±0.101		1.142	±0.121		0.272	±0.035

¹Growth restriction patterns defined in reference 22

²Small-for-Gestational-Age Normal Group;

³Small-for-Gesiational-Age Growth Restricted Group

⁴Normal Neonatal Growth Outcome Group; BPD = biparietal diameter; HC = head circumference; AC = abdominal circumference, FDL = femur diaphysis length

A secondary analysis determined the number of R² values below 95% for each of the four anatomical parameters in the Pattern Groups, the SGA N Group, the SGA GR Group, and the NNGO Group.

2^nd Trimester Growth Velocity

The sequence of statistical analyses described for the R² values was repeated with 2^nd trimester growth velocity data obtained with the Individualized Growth Assessment Program (iGAP; https://igap.research.bcm.edu). However, the SGA N Group contained 53 cases and the SGA GR Group 73 cases due to inclusion of cases where the growth velocities were calculated directly from two measurements. If significant differences were found with ANOVA, pair-wise comparisons of the main three groups were carried out using the t-test (p <0.05) with Bonferroni adjustment of the p-value (p<0.017) in multiple comparisons.

A secondary analysis compared growth velocities to the lower limits of their respective reference ranges (BPD: 0.25 cm/wk; HC: 0.90 cm/wk; AC: 0.90 cm/wk; and FDL: 0.20 cm/wk)⁹. These comparisons were carried out using the Abnormal Growth Velocity Scores (AGVS) obtained with iGAP software. The AGVS is a statistic used to quantify growth pathology in the 2^nd trimester by comparison of measured growth velocities to their 95% reference ranges, obtained in fetuses with NNGO⁹. This statistic is calculated by comparing the measured 2^nd trimester growth velocity to the appropriate reference range and determining if it is within or outside this range. If the velocity is within the reference range, an AGVS of zero is assigned as no growth abnormality was detected. If below the lower boundary (or above the upper boundary), the difference between the boundary and the growth velocity value is defined as the quantitative measure of the 2^nd trimester growth abnormality.

RESULTS

Linear Regression

As seen in Table 1A, median R² values for all anatomical parameters in all Growth Pattern subgroups were around 98%–99% with very little dispersion. No significant differences between Patterns subgroups were found by the Kruskal-Wallis analysis of variance test. These results justified pooling the data from all subgroups by an anatomical parameter to form the SGA GR Group. No significant differences (ANOVA) in R² values were found between the SGA N Group, SGA GR Group, and NNGO Group, mean values being around 98%–99% for all anatomical parameters in these three main groups (Table 1B).

Very few R² values below 95% were found in the 221 R² values obtained in this study (BPD: 3.6%; HC: 0.9%; AC: 2.7%; and FDL: 3.2%]. In the Patterns subgroups (all anatomical parameters pooled), there were 0.0% to 8.3% less than 95%. (Approximate values due to small sample size.) In the three major groups (SGA N Group, SGA GR Group, and NNGO Group), 1.7%, 3.8%, and 2.3% of the R² values were less than 95% (all anatomical parameters pooled).

2^nd Trimester Growth Velocities

Growth Velocity Comparisons between Groups

As seen in Table 2A, no significant differences between Patterns subgroups for any of the 4 anatomical parameters were found using the Kruskal-Wallis test (likely due to small sample sizes). These data were pooled by an anatomical parameter to form the SGA GR Group. ANOVA of the main groups (SGA N Group, SGA GR Group, and NNGO Group) revealed significant differences for all anatomical parameters except FDL (Table 2B). Subsequent pair-wise testing indicated significant differences between both the SGA N Group and SGA GR Group and the NNGO Group for BPD, HC, and AC. Only the AC growth velocities differed significantly between the SGA N and SGA GR Groups.

Comparisons of Growth Velocities to Lower Reference Range Boundaries

Comparison of 2^nd trimester growth velocities to their respective reference range lower boundaries⁹ in the NNGO Group revealed only 6 of 478 (1.3%) values below these boundaries (BPD: 5; HC: 0; AC: 0; FDL: 1). The corresponding differences from the boundaries were all −0.02 cm/wk with one exception (−0.10).

In the SGA N Group, there were 8/212 (3.8%) values below the boundaries (BPD: 2; HC: 1; AC: 3; FDL: 2). The corresponding differences from the boundaries were −.06, −.05, −.01, −.05, −.06, −.08, −.01, and −.03 cm/wk.

In the SGA GR Group, there were 30/292 (10.3%) values below the boundaries (BPD: 6; HC: 7; AC: 15; FDL: 2). The corresponding differences from the boundaries were −0.01(9 values), −0.02 (4 values), −0.03 (3 values), −0.04 (2 values), −0.05 (2 values), −0.07 (4 values), −0.08 (2 values), −0.09 (2 values), and −0.15, −0.17 cm/wk.

In the SGA Patterns subgroups, Pattern 1 had 18/108 (16.7%) values below the boundaries, Pattern 2 had 6/80 (7.6%), Pattern 3 had 1/36 (2.8%), Pattern 4 had 4/32 (12.5%), Pattern 5 had 2/24 (8.3%), and Pattern Unspecifiable had 1/12 (8.3%).

DISCUSSION

Primary Findings

Constant growth velocity in the 2^nd trimester is an important part of the rationale for proposing the use of these velocities as measures of fetal growth potential (Appendix). Second-trimester growth velocities have been previously evaluated in SGA singletons but the sample was small: 10 SGA Normal and 15 SGA Growth-Restricted²¹ vs. 53 and 73 in the current study. Our investigation represents the first comprehensive evaluation of 2^nd trimester growth in a large SGA sample where concordant fetal and neonatal growth evaluations provide the basis for subdividing the sample into normal and growth-restricted subgroups. The latter was further subdivided based on the pattern of growth restriction.

As seen in Table 1, very strong evidence for linear growth (and hence constant growth velocities) in the 2^nd trimester was found in all groups and subgroups of the SGA sample. These results were very similar to those found previously under ideal research conditions⁹, although the SGA data were obtained in a more clinic-like setting. Average and median R² values were in the 98%–99% range and the proportions below 95% were less than 4% in all three major groups. Special research conditions do not appear to be necessary for obtaining reasonable estimates of 2^nd trimester growth velocities. However, as intra- and inter-observer measurement errors can significantly affect these biometric parameters²⁵, careful measurements are necessary to avoid generating inaccurate individualized standards. This effect can be particularly significant if only two measurements are used in growth velocity calculations. Use of regression analysis mitigates this problem, but at least three measurements are required.

A more detailed examination of the growth velocities themselves (Table 2) indicates that fetuses in the SGA Group, on average, are growing more slowly than those with NNGO (exception: FDL) even in the 2^nd trimester. This suggests a lower growth potential, which may or may not be due to pathology occurring in the first trimester. Within the SGA Group, only the AC had a smaller growth velocity, on average, in the growth-restricted subgroup. This anatomical parameter also had the largest number of growth velocity values just below the lower limit of the reference range. These data suggest that even in the 2^nd trimester there is some evidence of impending growth abnormalities later in pregnancy, at least in certain individuals. However, in the SGA Group, 466/504 (92.5% [Normal: 96%; Growth-Restricted: 90%]) of the second-trimester growth velocities were within their 95% reference ranges.

The issue of 2^nd trimester growth velocities just below the lower boundaries of their 95% reference ranges has been discussed but not investigated in detail¹⁰. Since by definition, some of these growth velocities could be normal, further study was warranted. Such growth velocities were found in all three major groups but only the SGA Growth-Restricted Group had more than 4% of such values (10.3%). With two exceptions (AC: 0.15, 0.17 cm/wk), all differences were 0.09 or less cm/wk and present in all anatomical parameter categories. Among the different growth-restriction patterns, only Pattern 1 (growth abnormality increasingly worse during pregnancy) had a somewhat high number (16.7%) of these growth velocities and they were mainly in AC. These results suggest that differences from the lower boundary of 0.08 cm/wk (rather than 0.09 cm/wk¹⁰) or less could be used to identify growth velocities that might still be found in normally growing fetuses.

Previous Studies

Second-trimester growth velocities (BPD, HC, AC, FDL) of individual fetuses in SGA cases have only been studied in one previous publication²¹. Fifteen of 25 cases had growth restriction, and in 10 cases, growth was normal. Only 1/100 (1%) of the growth velocity measurements (AC) was outside the reference ranges and the difference from the lower boundary was 0.05 cm/wk.

Only four publications provide information on the variability of 2^nd trimester growth velocities for groups of normal fetuses (cross-sectional data). Fascina et al¹¹ reported mean velocities for BPD, HC, and AC of 0.40 to 0.30 cm/wk, 1.30 to 1.10 cm/wk, and 1.20 to 1.00 cm/wk, respectively, during the 14–26 weeks, MA, time period in 30 fetuses. For BPD and FDL, Gudihard-Costa et al¹² gave mean velocities of 0.39 to 0.32 cm/wk and 0.33 to 0.25 cm/wk, respectively, during the 14–26 wk, MA, time period in a study of 3433 cases. Deter and Harrist¹³, using instantaneous velocities obtained from 20 fetuses, reported mean values for BPD, HC, AC, and FDL of 0.30 to 0.30, 1.40 to 1.10, 1.20 to 1.20, and 0.30 to 0.30 cm/wk, respectively, for this same time period. Finally, Bertino et al¹⁶ presented graphs of growth velocity expected values for BPD, HC, AC, and FDL derived from 238 singletons. From these graphs, the beginning-peak-final growth velocities for the 14–26-week interval can be extracted. For BPD, the values were 0.34-0.35-0.28 cm/wk. For HC, the values were 1.21-1.29-1.02 cm/wk. For AC, the values were 0.96-1.06-1.05 cm/wk. For FDL, the values were 0.26-0.29-0.27 cm/wk. The analytical methods used in these studies were quite different, the data were for groups but not individuals, and they still contained the effects of uncorrected differences in growth potential (except for the study of Deter and Harrist¹³). However, the consistency of 2^nd trimester growth velocities (except perhaps for HC) provides additional support for the concept of relatively constant 2^nd trimester growth velocities.

Strengths and Limitations

A major strength of this investigation is the use of rigorous IGA criteria to identify individuals in the three main groups studied (NNGO, SGA N, SGA GR). For the two SGA groups, concordance between fetal and neonatal growth assessments was required. This investigation also benefited from the availability of a ‘gold standard’ for 2^nd trimester growth assessment in the results provided by the large, prospective, longitudinal study of fetuses with NNGO⁹. A primary limitation was the availability of only two scans in 18.3% of the cases, and the small number (3–5) in the other 81.7%, for evaluation of 2^nd trimester growth. This study was also limited by the small sample sizes (6 to 27) in the subgroups of the SGA Patterns Group. Such small samples restrict the reliability of our results.

Clinical Significance

The use of IGA requires a period of normal growth to establish individualized standards. The results of this study indicate that even for fetuses with subsequent evidence of growth restriction, 2^nd trimester growth is essentially normal. This permits the use of IGA in at-risk pregnancies and may allow detection of growth restriction in its earliest stages when better therapeutic options could be available.

CONCLUSIONS

Second-trimester growth in individual, singleton fetuses suggests that the growth of the BPD, HC, AC, and FDL is quite linear, indicating a constant growth velocity during this period of pregnancy. Similar results were found in fetuses at risk for growth restriction, both those who later developed this condition and those who did not. However, fetuses who were SGA at birth grow more slowly than those who are not, suggesting a difference in growth potential, although 92.5% of these growth velocities were within their 95% reference ranges. Different patterns of growth restriction appear to have similar 2^nd trimester growth. In most fetuses at risk for growth restriction, 2^nd trimester growth velocities can be considered appropriate measures of growth potential for use in IGA.

APPENDIX

Can the Growth Potential of Individual Fetuses be Measured?

Fetal growth potential is a concept (latent variable²⁶) that does not have a generally accepted definition. To measure such a quantity, it is necessary to choose something to represent the concept (observable variable). This first requires a definition of the secondary latent variable, growth, which is usually defined as the change in size of an anatomical parameter with age¹⁴. This definition indicates that there are multiple ‘growth potentials’, one for each anatomical parameter and specific measure of size. Change of size with age is commonly called ‘growth velocity’¹⁴.

Previous definitions of fetal growth potential have been limited to birth weight³ or skeletal parameters⁶. Expected weights at birth were calculated by regression analysis after adjusting for up to 19 known birth weight determinants^3,27. This approach only accounted for, at most, 36% of the weight variability²⁷. The growth of skeletal parameters was studied longitudinally in a multi-national, proscriptive sample which minimized growth abnormality risk factors and optimally supported fetal growth⁶. The observed empirical size parameter variability in this sample was taken as measures of growth potential. This approach tacitly assumes that normally growing fetuses follow group percentile lines, an assumption not supported in recent longitudinal studies²⁰.

In IGA, a different approach has been used. The observable variables for fetal growth potential were the individual, second trimester growth velocities of nine size parameters studied in fetuses with normal neonatal growth outcomes⁹. They were chosen for the following reasons:

• Second-trimester growth velocities are measures of change in size with time. not size alone, so are the most appropriate growth measurements⁸.

• Second-trimester growth velocities are empirical measures that reflect the effects of both known and unknown determinants of growth²⁷.

• Second-trimester growth velocities can be measured at a time when fetal nutritional requirements are low (as evidenced by similar 2^nd trimester growth in singletons, twins and triplets²⁸), thus reflecting the effects of intrinsic growth determinants rather than the environment provided by the mother.

• Second-trimester growth velocities of 1D, 2D and 3D anatomical parameters do not change appreciably during this period of pregnancy, provided the 2D and 3D measurements are appropriately transformed⁸.

• Second-trimester growth velocities can specify Rossavik size models that accurately predict 3^rd trimester size trajectories and birth characteristics in fetuses with normal neonatal growth outcomes^9,18.

• Rossavik model coefficient c values, predicted from second-trimester growth velocities⁹, usually have normal values in fetuses with subsequent growth restriction or macrosomia^29,30.

Although no single attribute can definitively establish that the second trimester growth velocity represents fetal growth potential, the set of attributes given above are consistent with a logical definition of this important biological characteristic of the growing fetus.