Maternal weight gain and associations with longitudinal fetal growth in dichorionic twin pregnancies: a prospective cohort study

Abstract

Background: Maternal metabolic demands are much greater with twin gestations; however, there are no accepted recommendations for maternal weight gain in twin pregnancies.

Objective: We assessed the association of maternal weight gain and fetal growth in dichorionic twins throughout pregnancy.

Design: This was a prospective US cohort study (n = 143, 2012–2013) of dichorionic twin pregnancies with known birth outcomes followed from enrollment (11–13 wk) and for ≤6 research visits throughout gestation. Maternal prepregnancy weight was self-reported, and current weight was measured at each research visit and abstracted from prenatal records. Fetal biometry was assessed by ultrasound at each research visit. Maternal weight and twin-pair fetal size trajectories across gestation were modeled. The adjusted associations between maternal weight gain from 0 to 13, 14 to 20, 21 to 27, and 28 to 34 wk and fetal growth at the subsequent week (i.e., 14, 21, 28, and 35 wk, respectively) were estimated with the use of linear regression.

Results: The mean ± SD maternal weight gain from 0 to 13, 14 to 20, 21 to 27, and 28 to 34 wk was 2.8 ± 3.0 kg, 3.9 ± 1.2 kg, 3.8 ± 1.4 kg, and 4.4 ± 2.2 kg, respectively, with a total gain of 17.7 ± 7.4 kg. Maternal weight gain from 0 to 13 wk (first trimester) was not associated with fetal size at 14 wk. Maternal weight gain from 14 to 20 and 21 to 27 wk (second trimester) was significantly associated with increased fetal weight at 21 wk [increase of 10.5 g/kg maternal weight gain (95% CI: 1.2, 19.8 g)] and 28 wk [increase of 21.3 g/kg maternal weight gain (95% CI: 0.6, 42.0 g)]. Maternal weight gain from 14 to 20 wk was associated with increased twin abdominal circumference (AC) and biparietal diameter at 21 wk. Maternal weight gain from 21 to 27 wk was associated with increased femur and humerus lengths at 28 wk.

Conclusion: Maternal weight gain was associated with dichorionic twin fetal growth in the second trimester only, driven by an association with AC earlier in second trimester and long bones later in the second trimester.

INTRODUCTION

Twin pregnancies represent a growing proportion of pregnancies in the United States and worldwide (1, 2). The metabolic demands on the mother are much greater with a twin gestation, and the pattern of fetal growth differs for singletons compared with twins (3). However, there are no accepted recommendations for maternal weight gain in twin pregnancies. The Institute of Medicine (IOM) gestational weight gain recommendations for twin pregnancies remain provisional (4), representing the interquartile ranges of weight gain for women participating in a large multisite study who delivered twins with a mean weight of ≥2500 g at 37–42 wk gestation (5). To develop recommendations, more evidence on maternal weight gain during twin pregnancies and its association with short- and long-term maternal and offspring outcomes is needed. One of the most proximal outcomes to maternal weight gain is fetal growth, which is also an important predictor of long-term health in the offspring (6).

Previous research on maternal weight gain in twin pregnancies has been limited in many areas (7). Much of the literature on gestational weight gain in twin pregnancies has studied birth weight and not intrauterine growth, limiting the ability to examine specific variables of fetal growth and better understand the etiology of how maternal weight gain influences twin growth. In addition, most studies on maternal weight gain in twin pregnancies have only examined total gestational weight gain and thus are limited in their inferences regarding the timing of weight gain. Understanding the timing of weight gain may ultimately be useful for better developing trimester-specific clinical recommendations that can be monitored and remediated if necessary. Only a few studies have examined the timing of weight gain in twin pregnancies and its association with fetal growth with the use of estimated fetal weight (5, 8) or birth weight (9). To our knowledge, no studies have examined the specific variables of fetal biometry, such as the abdominal circumference (AC) or the length of long bones. Distinguishing the associations across these variables may help to inform the etiology of the maternal weight gain association with fetal growth because the timing and magnitude of the velocities of these variables are differential (10).

With the use of data from a prospective cohort of women with dichorionic twin gestations (11) followed from the first trimester onward, we sought to address these critical data gaps related to maternal weight gain and fetal growth in twin pregnancies. With the use of repeated maternal weight gain and ultrasound data, our aims were 1) to evaluate the association between gestational weight gain and twin fetal growth across different time periods of gestation, and 2) to assess if such associations were specific to certain fetal biometry variables.

METHODS

Study sample

The Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) Fetal Growth Studies – Twins was a prospective cohort study of 171 women with dichorionic twin gestations recruited from 8 US clinical sites (2012–2013). The study procedures have been previously described (11). Institutional review board approval was obtained by all participating institutions, the data coordinating center, and the NICHD. Consent was obtained from all participants before enrollment.

The current analysis excluded pregnancies with a miscarriage occurring at <20 wk (n = 3; 1.8%), pregnancies that were voluntarily terminated or had a fetal reduction (n = 4; 2.3%), pregnancies with a congenital anomaly (n = 15; 9%), and pregnancies in which either twin had an unknown birth outcome (n = 6; 3.5%). The final sample size for this analysis consisted of 143 dichorionic twin pregnancies (Supplemental Figure 1).

Data collection

Women were enrolled between 8 wk 0 d and 13 wk 6 d. After an initial sonogram between 11 wk 0 d and 13 wk 6 d, women were randomly assigned to 1 of 2 follow-up schedules (11). Each follow-up schedule consisted of ≤6 sonograms scheduled at prespecified gestational weeks. The only difference between the 2 follow-up schedules was the targeted gestational age to ensure that the research visits were equally distributed across the gestation to address the main aims of assessing longitudinal fetal growth in twins in the NICHD Fetal Growth Studies – Twins study (11).

At enrollment, women self-reported their prepregnancy weight, had their height and weight measured, and completed detailed sociodemographic and health questionnaires. At each subsequent research visit, maternal weight was measured. Anthropometric measurements were taken by trained staff according to a standardized anthropometric protocol (12, 13). To obtain a greater number of maternal weight observations, maternal weight was also abstracted from prenatal care records. To ensure the quality of the abstracted prenatal care data, the 2 sources were compared when the research visit (measured) and the previous clinical visit (abstracted) were within 3 d, but not on the same day. High concordance (r = 0.998) was observed, indicating that the prenatal care record abstraction provided a high-quality additional source of maternal weight data. Including weights abstracted from prenatal care visits, a median of 18 weight measurements were available for each woman. Prepregnancy BMI (in kg/m²) was calculated based on the self-reported prepregnancy weight and measured height. If the prepregnancy weight was missing, the measured weight at enrollment was used (n = 2; 1.4%). Prepregnancy BMI was categorized as underweight (<18.5), normal (18.5–24.9), overweight (25.0–29.9), or obese (≥30.0).

Fetal measurements were obtained according to a standardized ultrasound protocol (11, 14). Fetal measurements relevant to the current analysis included biparietal diameter (BPD, outer-to-inner), head circumference (HC), AC, humerus length (HL), and femur length (FL), all measured in millimeters. At ≥15 wk, estimated fetal weight (EFW, measured in grams) was calculated with the use of the Hadlock formula, incorporating HC, AC, and FL (15). The HC:AC ratio was calculated as an indicator of symmetric or asymmetric growth (16, 17). Birth weight and sex were abstracted from the neonatal medical record.

Statistical analysis

The statistical analysis took place in 2 stages. The first stage modeled the individual maternal weight gain and fetal growth trajectories. The second stage used maternal weight estimates and fetal size estimates calculated from the trajectories developed in the first stage to assess the association between maternal weight gain and fetal size at prespecified time points.

Individual trajectories for maternal weight across the gestation were modeled with the use of linear mixed models with cubic splines for the gestational age fixed effects (18). This modeling approach estimated the overall trajectory and each individual’s weight trajectory across the pregnancy. From these trajectory models, we calculated each individual woman’s weight gain at prespecified time points (including delivery), described below. The advantage of this approach for weight gain is that it accounted for the variation in the number and timing of weight measurements between women and measurement error in the observed data. The model included random effects corresponding to intercept and cubic terms for gestational age (18). The cubic splines included 3 knot points (25th, 50th, and 75th percentiles) chosen at gestational ages that evenly split the distribution. These trajectory models were unadjusted; the adjustment for covariates occurred in the second stage of the analysis described below. Maternal weight was log-transformed to stabilize variances across gestational ages and improve normal approximations for the error structures. We estimated each woman’s total weight gain (in kilograms) at delivery and weight change (in kilograms) in the first trimester (0–13 wk), second trimester (14–20 and 21–27 wk), and third trimester (28–34 wk) by back-transforming the calculated weight at each specified gestational week based on the fixed and random effects from the mixed-effects model. These time periods were chosen a priori based on trimester cutoffs and previous literature, which has shown the potential importance of weight gain for twin fetal growth late in the second trimester (5). Therefore, the second trimester was split into 2 time periods of equal duration. The third trimester was examined only ≤34 wk, because the mean gestational age of delivery was 35 wk. The rate of weight gain (in kilograms per week) in each period was calculated as the total weight gain divided by the number of gestational weeks in each period.

The mean twin-pair fetal size was modeled for each growth variable separately (BPD, HC, AC, HL, FL, HC:AC ratio, and EFW) according to the same linear mixed-model approach as described above for maternal weight. From these trajectory models, the mean fetal size was estimated for each twin pair at 14 (15 wk for EFW), 21, 28, and 35 wk of gestation. EFW at 15 wk was used instead of 14 wk because EFW is not typically assessed before 15 wk (11).

In the second stage of the analysis, we first examined differences in the mean maternal rate of weight gain in each trimester using a general linear model. Next, we examined the association between maternal weight gain as the exposure variable and fetal size as the outcome variable. Specifically, we examined the association between maternal weight gain in an interval (0–13, 14–20, 21–27, and 28–34 wk) with twin fetal size at the end of the interval (i.e., 14, 21, 28, and 35 wk, respectively). Linear regression models were used to estimate the mean fetal size increase per each kilogram gain in maternal weight. All analyses were performed unadjusted and adjusted for covariates determined a priori, including maternal age (19–24, 25–59, 30–34, 35–39, and 40–45 y), race/ethnicity (non-Hispanic white, non-Hispanic black, Hispanic, and Asian), height (continuous), prepregnancy BMI (continuous), nulliparity (yes or no), smoking (nonsmoker, smoked before pregnancy, or smoked during pregnancy), employment and student status (employed or full-time student: yes or no), marital status (married or living with a partner: yes or no), insurance (other, private, or managed care), education (less than high school, high school diploma or general equivalency degree, some college or associate’s degree, Bachelor’s degree, or Master’s degree or higher), fertility intervention (yes or no), chronic hypertension (yes or no), and fetal sex of the twin pair (male/male, male/female, female/female, and unknown/unknown). To account for the whole pattern of a woman’s weight gain, the models were also adjusted for the maternal weight gain in the previous time period (e.g., for a model with weight gain from 14 to 20 wk as the exposure, weight gain from 0 to 13 wk was considered as a covariate). Analyses with birth weight as the outcome were adjusted for the gestational age at delivery. Although generally considered a potential confounder in the literature (19), prepregnancy diabetes was not controlled for in the current analysis because it did not vary within the population (n = 2; 1.4%). Analyses for weight gain before 28 wk did not include adjustment for gestational diabetes, gestational hypertension, or preeclampsia because they could be on the causal pathway between maternal weight gain and fetal size. However, additional analyses for weight gain from 28 to 34 wk were performed adjusting for gestational diabetes, because it is typically diagnosed by 28 wk and may be a confounding variable for this later analysis. Finally, we evaluated whether there was effect modification by prepregnancy BMI status (underweight or normal weight compared with overweight or obese) by fitting an interaction term and performing stratified analyses. All analyses were performed with the use of SAS version 9.4 (SAS Institute) or RStudio version 0.98.1091.

RESULTS

In this cohort of 143 women with dichorionic twin pregnancies, the mean ± SD maternal age was 31.5 ± 5.9 y. The baseline characteristics of women in the cohort are shown in Table 1. Eighty-one women were non-Hispanic white (56.6%), 31 women were non-Hispanic black (21.7%), 25 women were Hispanic (17.5%), and 6 women were Asian (4.2%). Based on their prepregnancy BMI, 43.4% of the cohort had a normal weight (BMI: 18.5–24.9), 23.1% were overweight (BMI: 25.0–29.9), and 28.7% were obese (BMI: ≥30.0). On average, women delivered at 35.2 ± 4.1 wk. The majority of the twin pairs were live births (n = 137; 95.8%); 3 women had a fetal death of one of the twins (2.1%), and 3 women had fetal death of both twins (2.1%). Details on the distribution of the fetal growth outcomes are shown in Supplemental Table 1.

TABLE 1.

Maternal characteristics of dichorionic twin pregnancies (n = 143), National Institute of Child Health and Human Development Fetal Growth Studies – Twins (2012–2013)

Characteristic	n (%)
Race/ethnicity1
White/non-Hispanic	81 (56.6)
Black/non-Hispanic	31 (21.7)
Hispanic	25 (17.5)
Asian	6 (4.2)
Age, y
19–24	20 (14.0)
25–29	28 (19.6)
30–34	55 (38.5)
35–39	27 (18.9)
40–45	13 (9.1)
Prepregnancy BMI status, kg/m2
<18.5	7 (4.9)
18.5–24.9	62 (43.4)
25.0–29.9	33 (23.1)
≥30.0	41 (28.7)
Height, cm
<162	44 (30.8)
162–167	50 (35.0)
≥168	49 (34.3)
Smoking
Nonsmoker	123 (86.0)
Smoked before pregnancy only	16 (11.2)
Smoked during pregnancy	4 (2.8)
Nulliparous	82 (57.3)
Education
Less than high school	9 (6.3)
High school diploma or GED2 or equivalent	18 (12.6)
Some college or associate degree	23 (16.1)
Bachelor’s degree	60 (42.0)
Master’s degree or advanced degree	33 (23.1)
Insurance
Other	35 (24.5)
Private or managed care	108 (75.5)
Employed or full-time student	113 (79.0)
Married or living with partner	111 (77.6)
Any type of fertility intervention	68 (47.6)
Pregestational diabetes	2 (1.4)
Gestational diabetes
Diet control	9 (6.3)
Medication	5 (3.5)
Unknown control	1 (0.7)
Chronic hypertension	10 (7.0)
Pregnancy hypertensive diseases
Gestational hypertension	9 (6.3)
Preeclampsia	24 (16.8)
Unspecified hypertension	7 (4.9)
Pregnancy outcome
Live birth/live birth	137 (95.8)
Live birth/fetal death	3 (2.1)
Fetal death/fetal death	3 (2.1)
Twin sex
Male/male	37 (25.9)
Female/female	38 (26.6)
Male/female	65 (45.5)
Unknown/unknown	3 (2.1)
Zygosity (same-sex twins only)
Monozygotic	12 (16.0)
Dizygotic	56 (74.7)
Missing	7 (9.3)
Gestational age at delivery, wk
20–27	11 (7.7)
28–34	33 (23.1)
≥35	99 (69.2)
Birth weight discordance >20%	16 (11.2)

¹One woman self-identified as both non-Hispanic white and American Indian/Alaskan Native, and was grouped with the non-Hispanic white category.

²GED, general equivalency diploma.

The mean ± SD total weight gain for the entirety of the pregnancy was 17.7 ± 7.4 kg. The mean ± rate of weight gain in each of the relevant time periods was as follows: 0–13 wk: 0.21 ± 0.23 kg/wk; 14–20 wk: 0.64 ± 0.19 kg/wk; 21–27 wk: 0.64 ± 0.23 kg/wk; and 28–34 wk: 0.73 ± 0.37 kg/wk. There was a significant difference between weeks 0–13 and 14–20 (P < 0.001) and weeks 21–27 and 28–34 (P < 0.001), however, the rate of weight gain was not different in the 2 periods that comprised the second trimester, weeks 14–20 and 21–27 (P = 0.39).

First-trimester maternal weight gain was not associated with any of the fetal size variables assessed (Table 2). Second-trimester maternal weight gain from 14 to 20 wk (P = 0.03) and from 21–27 wk (P = 0.04) was significantly associated with an increase in the mean EFW of the twin pair. For example, for each kilogram gain in maternal weight from 14 to 20 wk, the mean EFW of each twin was significantly heavier by 10.5 g. At 14–20 wk, the increase in EFW coincided with significant positive associations of maternal weight gain with the AC (P = 0.01) and BPD (P = 0.02), whereas there was suggestion of a negative association with the HC:AC ratio (P = 0.05). At 21–27 wk, the increase in EFW coincided with significant positive associations of maternal weight gain with FL (P = 0.03) and HL (P = 0.03). Maternal weight gain from 28 to 34 wk was not associated with twin biometry. The results for weeks 28–34 were essentially similar, with the additional adjustment for pregnancy complications.

TABLE 2.

Adjusted association between each kilogram gain in maternal weight and subsequent fetal size the following week among dichorionic twin pregnancies, National Institute of Child Health and Human Development Fetal Growth Studies – Twins (2012–2013)¹

	Fetal growth outcome the following week
Maternal weight gain exposure2	EFW, g (95% CI)	AC, mm (95% CI)	BPD, mm (95% CI)	HC, mm (95% CI)	FL, mm (95% CI)	HL, mm (95% CI)	HC:AC ratio (95% CI)
0–13 wk (n = 143)	0.1 (−0.4, 0.6)3	0.0 (−0.3, 0.3)	0.0 (−0.1, 0.1)	0.0 (−0.3, 0.3)	0.1 (0.0, 0.1)	0.0 (−0.1, 0.1)	−0.001 (−0.005, 0.003)
14–20 wk (n = 141)	10.5 (1.2, 19.8)*	2.3 (0.5, 4.1)*	0.6 (0.1, 1.1)*	1.5 (−0.1, 3.2)	0.2 (−0.2, 0.7)	0.4 (−0.1, 0.8)	−0.007 (−0.014, 0.000)
21–27 wk (n = 132)	21.3 (0.6, 42.0)*	1.7 (−0.1, 3.5)	0.2 (−0.3, 0.6)	0.7 (−0.6, 2.0)	0.3 (0.0, 0.6)*	0.3 (0.0, 0.5)*	−0.005 (−0.011, 0.001)
28–34 wk (n = 99)	19.0 (−20.8, 58.9)	0.7 (−1.5, 3.0)	−0.2 (−0.8, 0.3)	−0.5 (−1.9, 1.0)	0.2 (−0.1, 0.5)	0.2 (−0.1, 0.6)	−0.004 (−0.010, 0.002)

¹Models adjusted for maternal age (19–24, 25–29, 30–34, 35–39, and ≥40 y), race/ethnicity (non-Hispanic white, non-Hispanic black, Hispanic, or Asian), prepregnancy BMI, height, nulliparity, married status, attained education (less than high school, high school diploma, general equivalency diploma or equivalent, some college or associate degree, Bachelor’s degree, Master’s degree, or advanced degree), employed or full-time student status, smoking (no, before pregnancy, or during pregnancy), fertility intervention, chronic hypertension, fetal sex (male/male, female/female, male/female, or unknown/unknown), and, where applicable, gestational weight gain in the previous time period. Estimates are based on linear regression models. *Indicates estimates that are significant (P < 0.05). AC, abdominal circumference; BPD, biparietal diameter; EFW, estimated fetal weight; FL, femur length; HC, head circumference; HL, humerus length.

²Sample size decreases due to early deliveries. Sample size is consistent across fetal growth outcomes.

³EFW outcome estimated at 15 wk, not 14 wk; EFW is not calculated before 15 wk.

The interaction of maternal weight gain and prepregnancy BMI status (underweight or normal weight compared with overweight or obese) was significant only for the association with weight gain from 28 to 34 wk and fetal EFW (P = 0.007), AC (P = 0.005), and HC (P = 0.009) at 35 wk. Stratified results presenting the associations of maternal weight gain from 28 to 34 wk with the outcomes of EFW, AC, and HC are shown in Supplemental Table 2. Maternal weight gain from 28 to 34 wk was significantly associated with a 4.8-mm (95% CI: 1.0, 8.5 mm) increase in fetal AC size, which corresponded to a 70.7-g (95% CI: 10.4, 131.0 g) increase in fetal size at 35 wk for women who were underweight or had a normal prepregnancy weight only. The interaction of maternal weight gain and prepregnancy BMI status was significant for the outcome of HC, however, the association between weight gain and HC was not significant in either BMI group.

Finally, we examined the association between maternal weight gain throughout pregnancy and the birth weights of the twins (Table 3). Maternal weight gain between 14 and 20 wk and 21 and 27 wk was significantly associated with increases in twin birth weights. For example, for each kilogram a woman gained between 14 and 20 wk, there was a mean increase in the birth weight of each twin of 132.8 g (95% CI: 56.6, 208.9 g). No significant interactions were observed for maternal weight gain and prepregnancy BMI status and birth weight.

TABLE 3.

Adjusted association between kilogram gain in maternal weight and mean birth weight among dichorionic twin pregnancies, National Institute of Child Health and Human Development Fetal Growth Studies – Twins (2012–2013)¹

Maternal weight gain2	Birth weight, g (95% CI)
0–13 wk (n = 141)	15.4 (−1.7, 32.6)
14–20 wk (n = 139)	132.8 (56.6, 208.9)*
21–27 wk (n = 131)	57.0 (2.7, 111.3)*
28–34 wk (n = 99)	14.3 (−37.6, 66.2)

¹Models adjusted for maternal age (19–24, 25–29, 30–34, 35–39, and ≥40 y), race/ethnicity (non-Hispanic white, non-Hispanic black, Hispanic, or Asian), prepregnancy BMI, height, nulliparity, married status, education (less than high school, high school diploma, general equivalency diploma or equivalent, some college or associate degree, Bachelor’s degree, Master’s degree, or advanced degree), employed or full-time student status, smoking (no, before pregnancy, or during pregnancy), fertility intervention, chronic hypertension, fetal sex (male/male, female/female, male/female, or unknown/unknown), gestational age at delivery, and, where applicable, gestational weight gain in the previous time period. Estimates are based on linear regression models. ^*Indicates estimates that are significant (P < 0.05).

²Sample size decreases due to early deliveries.

DISCUSSION

In this prospective cohort of women with dichorionic twin pregnancies, we observed that second-trimester maternal weight gain was significantly associated with twin-pair fetal size. Specifically, maternal weight gain from 14 to 20 wk and 21 to 27 wk was associated with a larger twin pair EFW. Greater maternal weight gain from 14 to 20 wk was significantly associated with increases in the AC and BPD of the twins and a decrease in the HC:AC ratio, whereas weight gain from 21 to 27 wk was significantly associated with an increase in the length of fetal long bones (FL and HL). The larger intrauterine size persisted to delivery, which was demonstrated by the significant association between maternal weight gain from 14 to 20 and 21 to 27 wk with birth weight. Another pertinent finding of this study was that maternal weight gain in the first trimester was not associated with the size of the twins. Lastly, there was a suggestion that the observed associations may be stronger for women who were underweight or normal weight before pregnancy, however, the sample size was small and this should be interpreted with caution.

Currently, the IOM gestational weight gain recommendations for twin pregnancies are provisional due to limited research specific to twins. Twin pregnancies put additional demands on maternal nutritional status due to the increased maternal and fetal tissue mass. For example, increases in both the blood volume and the number of erythrocytes in the blood are ∼20% higher in twin than in singleton gestations by 20 wk (20). Consequently, women with twin pregnancies require additional macro- and micronutrients; however, the specific amounts or the timing of the additional nutritional needs remain unclear. Much of the previous literature related to maternal weight gain in twin pregnancies has been limited to total weight gain (21–23), which limits the ability to determine whether there are critical windows of gestation when maternal weight gain is more or less associated with fetal growth of the twins. Furthermore, trimester-specific rates, as opposed to total weight gain recommendations, are important for twin pregnancies because most mothers deliver before reaching term (24).

Our findings suggest that, overall, only increased weight gain in the second trimester was associated with increases in twin fetal growth. A few previous studies examined the timing of maternal weight gain in twin pregnancies and the association with fetal growth or birth weight (5, 8, 9). One previous study suggested that weight gain before 20 wk was the most critical for women with twin gestation (9), whereas another study suggested that weight gain before 20 wk and from 20 to 28 wk was significantly associated with increases in mean twin size (8). Interestingly, when stratified by prepregnancy BMI status, third-trimester weight gain remained relevant for fetal growth for women who were underweight or normal weight before pregnancy. Our study builds on these previous studies by examining specific fetal size variables. Based on the models with each of the separate biometric outcomes, our study suggests that the association of maternal weight gain on overall fetal size may be due to an association with the AC (representing mass and organ size) earlier in the second trimester and the long bones (representing linear growth) later in the second trimester.

A main focus of our study was to examine the timing of maternal weight gain. With early data collection from the prospective cohort, we were able to divide the early period of <20 wk reported in previous studies into weight gain from 0 to 13 wk and 14 to 20 wk. We demonstrate that first-trimester weight gain in women with dichorionic twin pregnancies was not related to EFS or birth weight, which, notably, is consistent with singleton pregnancies (25). This may be reassuring to women who had an adequate prepregnancy nutritional status, but have limited first-trimester weight gain because of the increased nausea and vomiting common in twin pregnancies (26). We found that only second-trimester maternal weight gain was associated with fetal growth. During this time, maternal weight gain still tends to be in the maternal compartments, including volume expansion and subcutaneous fat increases (4). Later in gestation, a substantial component of maternal weight gain is the products of conception, including fetal weight. By the third trimester, there are many competing factors that may restrict fetal growth in twin gestations, such as placental insufficiency or the intrauterine environment, so that the association of maternal weight gain with fetal growth is attenuated (27). Stratified results by prepregnancy BMI status suggest that weight gain in the third trimester may remain relevant for fetal growth among women who were underweight or normal weight, however, the sample size was small (n = 51), and therefore the results should be interpreted with caution.

The strengths and limitations of this study warrant discussion. First, this study is generalizable only to dichorionic twin pregnancies. It is unclear if the findings are generalizable to monochorionic twin pairs, who are at higher risk of complications (28). Although the external validity is unknown, this study was based on contemporary practices and included a diverse sample of women from across the United States at 8 clinical sites. The sample size of our study was modest, and once stratified by prepregnancy BMI status, the groups were relatively small, which may have precluded the detection of a significant interaction with prepregnancy BMI status. Stratified results are presented only in instances where the interaction was significant. Our findings provide preliminary evidence that weight gain early in the third trimester (28–34 wk of gestation) had a stronger association in women who were underweight or normal weight than in heavier women, but corroboration of these findings is needed. In addition, the study was not suited for assessing the relation of the current IOM recommendations for twin pregnancies and obstetric outcomes; however, this study fills a much-needed data gap related to the underlying etiology of an association of maternal weight gain with fetal growth in dichorionic twins. In addition, not just the amount of weight that is gained but the pattern of weight gain may be important. We adjusted for weight gain in the previous time period so that clinically relevant estimates of weight gain in each trimester could be obtained while accounting for the pattern of gain. Self-reported prepregnancy weight recalled at enrollment between weeks 8 and 13 of gestation was used in this work. Although self-reported prepregnancy weight is clinically relevant and in line with the type of information health care practitioners have when women seek prenatal care, error may still be present in this measure. In previous validation studies based on a short length of recall, summarized in a recent review (29), self-reported prepregnancy weight was highly correlated with measured weight, and the error ranged from −0.34 to −2.94 kg. Notably, in these previous studies, reporting error had a minimal impact on the reported associations and conclusions developed from the studies. However, most previous validation studies in this area were based on studies of women with singleton pregnancies, and it is unclear if they are generalizable to women with twin pregnancies who tend to have different weight gain patterns. Thus, the degree of error in self-reported prepregnancy weight in our study is unknown. Additional notable strengths include the high-quality data on both maternal prepregnancy weight and fetal growth, which were collected according to standardized protocols at research visits.

In conclusion, maternal weight gain in the second trimester, but not in other trimesters, was associated with estimated twin fetal weight. This finding should help to direct other intervention studies to determine whether modification of a woman’s weight gain trajectory can enhance fetal growth and pregnancy outcomes in women with dichorionic twin pregnancies.