Unified standard for fetal growth velocity: the Eunice Kennedy Shriver National Institute of Child Health and Human Development Fetal Growth Studies

OBJECTIVE:

Fetal growth velocity has the potential to improve the identification of fetuses who are failing or who are exceeding their expected rate of growth compared with a single measure of fetal size.^1,2 The Eunice Kennedy Shriver National Institute of Child Health and Human Development Fetal Growth Studies – Singletons developed a calculator to compute fetal growth velocity standard percentiles for any given set of gestational week intervals in a contemporary, low-risk, and racially or ethnically diverse multisite population in the United States.¹ The calculator has the ability to compute velocity percentiles by maternal self-reported race and ethnicity, including non-Hispanic White (NHW), non-Hispanic Black (NHB), Hispanic, and Asian or Pacific Islander (Asian) women, that can be used in concert with race- or ethnic-specific fetal growth standards.^3,4 Subsequently, we have provided a unified fetal growth standard to provide a complementary option to the race- or ethnic-specific fetal growth standard.^5-7 In the current analysis, we provided a fetal growth velocity standard and calculator to complement our previous work on unified fetal growth size standards.⁷

STUDY DESIGN:

The cohort included 1737 pregnant individuals without obesity with low-risk prepregnancy and antenatal profiles who delivered at ≥37 weeks of gestation at 12 US clinical sites (2009–2013).^3,7,8 The analysis was limited to 1732 eligible individuals (99.7%) with ultrasound measurements, of whom 27.7% identified as NHW, 24.4% identified as NHB, 28.1% identified as Hispanic, and 19.8% identified as Asian. To address the original oversampling by race and ethnicity, the sample was weighted back to the US population distribution of pregnant women using natality statistics, employing the same methods as the unified fetal growth standard for size.^7,9 Human subjects’ approval was obtained from all participating sites, and all women provided informed consent before any data collection.

Research ultrasounds were performed at up to 6 visits where fetal biparietal diameter, head circumference (HC), abdominal circumference (AC), humerus, and femur length (FL) were measured following a standardized protocol. Estimated fetal weight (EFW) was calculated from HC, AC, and FL.¹⁰ The individual biometrics and EFW were log-transformed to stabilize variances across gestational ages (GAs) and improve normal approximations for error structures. Using GAs 11 to 41 weeks, we fitted a weighted linear mixed model with cubic splines for the fixed effects and a cubic polynomial for the random effects for each biometric measurement and EFW. Weights were chosen in the linear mixed model estimation so that inferences reflected a US-based population concerning race and ethnicity. Of note, 3-knot points (25th, 50th, and 75th percentiles) were chosen at GAs that evenly split the distributions. Velocity was computed and presented in 2 ways: (1) the difference between 2 consecutive weekly measurements and (2) the difference between any 2 measurements divided by the difference between the 2 measurements in weeks. In both cases, percentiles were derived from the fitted linear mixed model. Estimated 50th percentile velocity curves were defined as the mean change in each biometric measurement and EFW per gestational week. For detailed equations for the velocity calculation, see the Appendix. All analyses were implemented using SAS (version 9.4; SAS Institute, Cary, NC) or R (version 4.1.2; http://www.R-project.org).

RESULTS:

After weighting, the racial or ethnic representation of women in the analytical sample was 55.0% for NHW, 12.4% for NHB, 24.5% for Hispanic, and 8.1% for Asian, which is a distribution similar to all 2011 US births. Additional demographic details for the weighted cohort are as previously published.⁷ Average EFW growth velocity increased from 29 g/wk at 16 weeks, to 59 g/wk at 20 weeks, to 175 g/wk at 30 weeks, reaching a peak of 215 g/wk at 34 to 35 weeks. The 50th percentile for the unified weekly EFW velocity curve was lower than that of the NHW group, similar to the Hispanic group, and higher than the Asian and NHB groups through the peak EFW velocity at 35 weeks (comparative statistical testing not performed) (Figure). Data for weekly growth velocity percentiles for EFW and individual biometrics from weeks 11 to 41 (5th, 10th, 50th, 90th, and 95th percentiles) are presented in the Table.

FIGURE. EFW velocity (g/wk) in the NICHD Fetal Growth Studies – Singletons.

Velocity for the EFW by a unified multiracial or ethnic group, individual maternal self-reported race and ethnicities, and GA. Estimated 10th, 50th, and 90th percentiles for velocity were calculated using parameters from the linear mixed model as discussed in the methods. The unified velocity curve for the entire cohort was weighted for race and ethnicity using the 2011 natality data.⁹

EFW, estimated fetal weight; GA, gestational age; NICHD, Eunice Kennedy Shriver National Institute of Child Health and Human Development.

TABLE.

Percentiles for fetal growth velocity of anthropometric measurements by GA, Eunice Kennedy Shriver National Institute of Child Health and Human Development Fetal Growth Studies – Singletons

GA (wk)	Percentile
	5th	10th	50th	90th	95th
Biparietal diameter (mm)
11	2.4	2.6	3.4	4.2	4.4
12	2.4	2.6	3.6	4.6	4.9
13	2.3	2.6	3.7	4.9	5.3
14	2.0	2.4	3.7	5.1	5.5
15	1.7	2.1	3.6	5.2	5.7
16	1.3	1.7	3.4	5.2	5.7
17	0.9	1.4	3.3	5.2	5.8
18	0.6	1.1	3.1	5.2	5.9
19	0.3	0.9	3.1	5.4	6.0
20	0.1	0.8	3.1	5.6	6.3
21	0.0	0.6	3.1	5.7	6.5
22	−0.2	0.5	3.1	5.9	6.7
23	−0.4	0.3	3.1	6.1	6.9
24	−0.7	0.1	3.1	6.2	7.1
25	−0.9	−0.1	3.1	6.3	7.3
26	−1.2	−0.3	3.0	6.4	7.4
27	−1.4	−0.5	2.9	6.5	7.6
28	−1.7	−0.7	2.9	6.6	7.7
29	−1.9	−0.9	2.8	6.6	7.8
30	−2.2	−1.2	2.7	6.7	7.9
31	−2.5	−1.4	2.5	6.7	7.9
32	−2.8	−1.7	2.4	6.6	7.9
33	−3.1	−2.0	2.2	6.6	7.9
34	−3.5	−2.3	2.0	6.5	7.8
35	−3.8	−2.6	1.7	6.3	7.7
36	−4.2	−2.9	1.5	6.2	7.5
37	−4.4	−3.2	1.3	6.0	7.4
38	−4.7	−3.4	1.1	5.9	7.3
39	−4.9	−3.6	1.0	5.9	7.3
40	−5.0	−3.7	0.9	5.8	7.3
41	−5.1	−3.8	0.9	5.8	7.3
Head circumference (mm)
11	8.8	9.5	11.9	14.5	15.3
12	8.9	9.8	12.8	15.9	16.8
13	8.7	9.7	13.2	17.0	18.0
14	8.1	9.2	13.3	17.6	18.9
15	7.1	8.4	13.1	18.0	19.4
16	6.1	7.5	12.7	18.1	19.7
17	4.9	6.5	12.3	18.2	20.0
18	3.9	5.7	11.9	18.4	20.3
19	3.2	5.0	11.8	18.8	20.8
20	2.6	4.6	11.8	19.4	21.6
21	2.0	4.1	11.8	19.9	22.2
22	1.3	3.6	11.8	20.3	22.8
23	0.5	2.9	11.6	20.7	23.4
24	−0.3	2.3	11.4	21.0	23.8
25	−1.1	1.5	11.2	21.2	24.2
26	−2.0	0.7	10.9	21.4	24.5
27	−3.0	−0.1	10.5	21.5	24.7
28	−3.9	−0.9	10.1	21.5	24.9
29	−4.9	−1.7	9.6	21.5	25.0
30	−5.9	−2.6	9.1	21.4	25.0
31	−6.9	−3.5	8.6	21.2	24.9
32	−7.9	−4.5	7.9	20.9	24.7
33	−9.0	−5.5	7.3	20.6	24.4
34	−10.1	−6.5	6.5	20.1	24.1
35	−11.2	−7.5	5.7	19.5	23.6
36	−12.2	−8.5	4.9	19.0	23.1
37	−13.1	−9.4	4.3	18.5	22.6
38	−13.9	−10.1	3.7	18.1	22.3
39	−14.6	−10.7	3.2	17.7	22.0
40	−15.1	−11.2	2.8	17.5	21.8
41	−15.5	−11.6	2.6	17.4	21.7
Abdominal circumference (mm)
11	6.0	6.7	9.4	12.2	13.1
12	6.2	7.1	10.4	13.9	14.9
13	6.1	7.2	11.1	15.3	16.6
14	5.7	7.0	11.6	16.6	18.1
15	5.1	6.6	11.9	17.7	19.3
16	4.3	6.0	12.0	18.5	20.4
17	3.4	5.2	12.0	19.2	21.4
18	2.4	4.4	11.9	19.9	22.3
19	1.4	3.6	11.9	20.6	23.2
20	0.5	2.9	11.8	21.3	24.1
21	−0.6	2.1	11.7	21.9	24.9
22	−1.6	1.2	11.5	22.4	25.6
23	−2.7	0.3	11.2	22.9	26.3
24	−3.8	−0.6	11.0	23.3	27.0
25	−4.8	−1.4	10.8	23.8	27.6
26	−5.7	−2.2	10.6	24.3	28.3
27	−6.6	−2.9	10.5	24.9	29.1
28	−7.4	−3.5	10.6	25.6	30.0
29	−8.0	−4.0	10.7	26.4	31.0
30	−8.7	−4.5	10.9	27.3	32.1
31	−9.5	−5.1	11.0	28.0	33.1
32	−10.3	−5.8	10.9	28.7	33.9
33	−11.3	−6.6	10.8	29.2	34.6
34	−12.4	−7.5	10.5	29.6	35.2
35	−13.6	−8.5	10.0	29.8	35.6
36	−14.8	−9.6	9.5	29.9	35.9
37	−16.0	−10.6	9.0	29.9	36.1
38	−17.1	−11.6	8.6	30.0	36.3
39	−18.1	−12.5	8.1	30.1	36.6
40	−19.1	−13.4	7.8	30.2	36.8
41	−20.1	−14.2	7.4	30.4	37.2
Femur length (mm)
11	1.2	1.3	1.7	2.1	2.3
12	1.5	1.7	2.3	3.0	3.2
13	1.7	1.9	2.9	3.9	4.2
14	1.7	2.0	3.3	4.7	5.1
15	1.4	1.9	3.5	5.2	5.8
16	1.0	1.5	3.5	5.6	6.3
17	0.4	1.0	3.3	5.8	6.6
18	−0.2	0.5	3.1	6.0	6.9
19	−0.7	0.1	3.0	6.2	7.1
20	−1.1	−0.3	2.9	6.5	7.5
21	−1.5	−0.6	2.9	6.7	7.9
22	−2.0	−1.0	2.8	7.0	8.2
23	−2.4	−1.3	2.7	7.2	8.5
24	−2.8	−1.7	2.6	7.4	8.8
25	−3.3	−2.1	2.5	7.5	9.0
26	−3.7	−2.4	2.4	7.7	9.3
27	−4.1	−2.7	2.3	7.8	9.5
28	−4.4	−3.0	2.2	8.0	9.8
29	−4.7	−3.3	2.1	8.2	10.0
30	−5.0	−3.5	2.1	8.4	10.3
31	−5.3	−3.8	2.1	8.6	10.6
32	−5.6	−4.0	2.0	8.8	10.8
33	−6.0	−4.3	2.0	8.9	11.0
34	−6.3	−4.6	1.9	9.0	11.2
35	−6.6	−4.8	1.8	9.1	11.4
36	−6.9	−5.1	1.7	9.2	11.5
37	−7.3	−5.4	1.5	9.2	11.5
38	−7.6	−5.8	1.3	9.2	11.5
39	−8.0	−6.1	1.1	9.0	11.5
40	−8.4	−6.5	0.8	8.9	11.3
41	−8.8	−6.8	0.5	8.6	11.1
Humerus length (mm)
11	1.3	1.4	1.8	2.2	2.4
12	1.6	1.8	2.4	3.1	3.4
13	1.8	2.1	3.0	4.0	4.3
14	1.8	2.1	3.3	4.7	5.1
15	1.5	1.9	3.4	5.2	5.7
16	0.9	1.4	3.3	5.4	6.1
17	0.3	0.9	3.1	5.5	6.2
18	−0.3	0.3	2.8	5.5	6.3
19	−0.8	−0.1	2.6	5.6	6.5
20	−1.2	−0.4	2.5	5.8	6.8
21	−1.6	−0.7	2.5	6.0	7.0
22	−1.9	−1.0	2.4	6.1	7.3
23	−2.3	−1.4	2.3	6.3	7.5
24	−2.7	−1.7	2.2	6.4	7.7
25	−3.0	−2.0	2.0	6.5	7.8
26	−3.4	−2.3	1.9	6.6	8.0
27	−3.7	−2.5	1.8	6.7	8.1
28	−4.0	−2.8	1.7	6.8	8.3
29	−4.3	−3.0	1.7	6.9	8.5
30	−4.5	−3.2	1.6	7.0	8.6
31	−4.8	−3.4	1.6	7.1	8.8
32	−5.0	−3.6	1.5	7.2	9.0
33	−5.2	−3.8	1.5	7.3	9.1
34	−5.5	−4.0	1.4	7.5	9.3
35	−5.7	−4.2	1.4	7.6	9.4
36	−5.9	−4.4	1.3	7.6	9.5
37	−6.2	−4.6	1.2	7.6	9.6
38	−6.5	−4.9	1.0	7.5	9.5
39	−6.8	−5.2	0.8	7.4	9.4
40	−7.2	−5.6	0.5	7.2	9.2
41	−7.5	−5.9	0.1	6.8	8.9
EFW (g)
11	4	5	9	13	14
12	5	6	11	16	18
13	6	8	14	21	23
14	8	10	18	27	29
15	10	13	23	34	37
16	13	16	29	42	47
17	16	20	35	53	58
18	18	23	43	64	71
19	20	27	51	77	85
20	22	30	59	91	101
21	23	32	68	107	118
22	23	34	77	124	138
23	23	36	87	142	159
24	22	38	97	162	181
25	21	39	108	183	206
26	19	40	120	207	233
27	16	41	132	232	263
28	13	41	145	260	295
29	10	41	160	291	330
30	4	40	175	323	368
31	−3	37	189	356	406
32	−14	31	201	388	444
33	−29	21	210	418	480
34	−49	7	215	444	513
35	−73	−12	215	466	541
36	−101	−35	212	484	566
37	−129	−58	208	500	588
38	−158	−82	202	514	608
39	−187	−107	195	527	627
40	−216	−131	189	540	645
41	−244	−154	182	552	664

Week corresponds to exact week (eg, 11 weeks = 11.0 weeks). EFW was calculated using the Hadlock formula.¹⁰

EFW, estimated fetal weight; GA, gestational age.

CONCLUSION:

We have provided a unified, multiethnic fetal growth velocity standard to supplement our previous work.^3,4,7 This unified EFW velocity curve, which is weighted by race and ethnicity to represent the distribution of US pregnant women in 2011, generally fell below that for the fetuses of NHW women and above those for the fetuses of Asian and NHB women with peak velocity approximately 34 to 35 weeks of gestation. Our findings were generally consistent with other studies of fetal growth velocity, although the magnitude of change and patterns of the timing of accelerated growth and peak velocities for individual biometrics varied by population.^11-13 Fetal growth velocity indicates how a fetus arrived at a given EFW (ie, fast or slow growth), which provides additional information beyond that provided by a cross-sectional assessment of EFW at a single time point.¹ The week-specific velocity (change per week) is useful for understanding the expected amount of fetal growth between any 2 points during gestation and comparing it across groups. From a clinical perspective, we would not recommend performing sonograms 1 week apart to identify abnormal fetal velocity given that this is not sufficient time to allow actual differences in growth to be disentangled from random variation in fetal growth. However, as in the original velocity article, we presented a calculator that allows for the estimation of relative growth velocity based on any clinically appropriate interval between sonograms.¹ The optimal interval for estimating fetal growth velocity and using it to inform clinical practice is currently unknown and is complicated as it reflects the time between measurements and error in the biometric measurement. The interval may also vary based on the outcome of interest. Evidence suggests that fetal growth velocity might better distinguish fetuses that are failing to meet their growth potential from constitutionally small-for-gestational-age fetuses below the 10th percentile for EFW.² In summary, we provided a multiethnic, unified fetal growth velocity standard and calculator for fetal growth velocity. Future work is needed to determine whether adding growth velocity to traditional fetal size measures improves perinatal outcomes in terms of morbidity and mortality and, if so, whether racial- or ethnic-specific or a unified velocity standard is superior in improving outcomes.

APPENDIX

For a single longitudinal ultrasound measurement, we built a mixed-effects model with cubic spline based on 3-knot points for fixed effects and cubic function for random effects as follow:

$$\begin{gathered} Y_{ij}^{\ast }={\theta }_0+{\theta }_1{t}_{ij}+{\theta }_2{t}_{ij}^2+{\theta }_3{t}_{ij}^3+{\theta }_4{\left(t_{ij}-{\delta }_1\right)}_{+}^3 \\ +{\theta }_5{\left(t_{ij}-{\delta }_2\right)}_{+}^3+{\theta }_6{\left(t_{ij}-{\delta }_3\right)}_{+}^3 \\ +b_{i0}+b_{i1}{t}_{ij}+b_{i2}{t}_{ij}^2+b_{i3}{t}_{ij}^3+{\epsilon }_{ij} \end{gathered}$$ (1)

where $Y_{ij}^{\ast }$ denotes a log-transformed ultrasound measurement, t_ij3 is the j th gestational age for i th women, ${\left(t_{ij}-{\delta }_k\right)}_{+}^3=\max (0,{\left(t_{ij}-{\delta }_k\right)}^3)$, δ = (δ₁, δ₂, δ₃)′, is the knot sequence with δ₁ < δ₂ < δ₃, (1, t_ij, $t_{ij}^2$, $t_{ij}^3$, ${\left(t_{ij}-{\delta }_1\right)}_{+}^3$, ${\left(t_{ij}-{\delta }_2\right)}_{+}^3$, ${\left(t_{ij}-{\delta }_3\right)}_{+}^3$) ′ is a truncated polynomial basis functions of degree 3, θ = (θ₀, θ₁, θ₂, θ₃, θ₄, θ₅, θ₆)′ is a corresponding vector of parameter, b_i = (b_io, b_i1, b_i2, b_i3)′ denotes the random effects, for j = 1, 2, …, n_i, i = 1, 2, …, I, n_i denotes the number of repeated time points for i th women, and I denotes the number of women. Furthermore, we assumed that the residual errors ε_ij are distributed with independent normal distributions with mean 0 and variance. The random effects are assumed to follow the multivariate normal distribution with mean zero and unstructured variance-covariance matrix Ω. The random effects are interpreted as individual departures in an individual’s growth curve relative to the average fetal growth curve in the population. From the model in Eq. (1), the ratio and velocity of 2 ultrasound measurements can be computed, respectively, as follows:

$$r_{i,j{j}^{\prime }}=\text{exp}\left(\widehat{Y_{i{j}^{\prime }}^{\ast }}-\widehat{Y_{ij}^{\ast }}\right)\text{and}{v}_{i,j{j}^{\prime }}=\frac{\left(r_{i,j{j}^{\prime }}-1\right)Y_{ij}}{t_{i{j}^{\prime }}-t_{ij}}$$

where Y_ij is an original scaled ultrasound measurement at time t_ij, $\widehat{Y_{ij}^{\ast }}$ is an estimated log-transformed ultrasound measurement at time t_ij, r_i,jj′ is a ratio of 2 ultrasound measurements, and v_i,jj′ is a velocity of 2 ultrasound measurements. We incorporated weights in the linear mixed model where each individual’s component of the likelihood is weighted by the proportion of the US population reference for that individual’s race or ethnic group divided by the corresponding proportion in the Eunice Kennedy Shriver National Institute of Child Health and Human Development fetal growth twin study population.