点击显示 收起
1 From the National Institute of Public Health, Cuernavaca, Morelos, Mexico (LMN); the Division of Nutritional Sciences, Cornell University, Ithaca, NY (JDH); the Institute of Nutrition of Central America and Panama, Guatemala City (RG); and the Department of International Health, Rollins School of Public Health, Emory University, Atlanta (RM).
2 Supported by NIH grant no. HD 29927; the Division of Nutritional Sciences, Cornell University, Ithaca, NY; and the Department of International Health, Emory University, Atlanta. 3 Address reprint requests to LM Neufeld, Epidemiología de la Nutrición, Instituto Nacional de Salud Pública, Avenida Universidad 655, Cuernavaca, Morelos 62508, México. E-mail: lneufeld{at}correo.insp.mx.
ABSTRACT
Background: Despite our knowledge of the negative consequences of stunting during early childhood and the important role that maternal nutritional status plays in the development of intrauterine growth retardation, we do not know the extent to which maternal nutritional status influences the growth in length of the fetus or whether a sensitive period for fetal linear growth exists during gestation.
Objective: Our objective was to explore the relation between maternal weight gain during different stages of pregnancy and linear growth of the fetus.
Design: Ultrasound examinations were conducted at 15–24 (
Conclusions: Maternal weight change from the first to the second trimester of pregnancy is strongly associated with fetal growth. Mid-gestation may be a sensitive period for fetal linear growth.
Key Words: Fetal linear growth • maternal weight gain • pregnancy • ultrasound • femur • tibia
INTRODUCTION
Maternal nutritional status has been identified as an important factor that contributes to poor fetal growth in developing countries (1–5). Most of the field studies designed to assess the influence of maternal nutritional status on fetal growth have focused on weight alone or on weight-for-gestational age at birth as the outcome of interest. The justification for this focus comes from the increased adverse outcomes for infants born with low weight (6–11). It has become evident that in populations living in economically poor areas, the prevalence of stunting at birth (short crown-heel length-for-gestational age) may be as high as 50% (12, 13).
The negative postnatal consequences and potential catch-up growth of infants who are born stunted have been documented. Stunting at birth is associated with an increased risk of stunting in later childhood and stunted adult height (13, 14). Stunting during childhood is associated with poor school performance (15) and poor physical work capacity (16, 17); the negative consequences last at least into adolescence and early adulthood (6). The timing of stunting may also be important. In one study, infants were measured from birth to 2 y of age, and those who were identified as stunted sooner after birth tended to be more severely stunted and to suffer from more negative outcomes (18). Infants who were born stunted were more likely to benefit from nutritional supplementation than were those who were not stunted at birth (14). This suggests that, given sufficient resources and early intervention, infants who are born stunted may be able to catch up in growth in the postnatal period. Of those infants who are born stunted because of genetic factors, not fetal growth faltering, one would not expect to see the adverse outcomes mentioned above nor catch-up growth in the postnatal period.
In a study from rural Malawi, a low rate of maternal weight gain in late pregnancy was associated with lower crown-heel length at birth (19). Because length at birth provides a cumulative summary of linear growth in utero, we do not know how these findings relate to patterns of fetal growth or when during gestation the differences in fetal length emerge. A better understanding of the relation between maternal nutritional status at different periods of pregnancy and fetal growth in length may help us to better understand the process of linear growth faltering in utero.
We report here the results of a study of fetal growth as measured by ultrasound in 4 rural villages in Guatemala. Our objective was to explore the relation between maternal weight gain during different stages of pregnancy and linear growth of the fetus.
SUBJECTS AND METHODS
Sample, recruitment, and anthropometry
All pregnant women between 19 and 34 y of age residing in 4 rural villages in eastern Guatemala were invited to participate in a longitudinal study conducted between August 1996 and June 1999. To ensure early identification of pregnancies, all women in the communities were visited once every 3 mo and asked to recall the date of their last menstrual period (LMP). At this time, weight was recorded with the use of an electronic scale (Tanita model 1582; Tanita Corp, Arlington Heights, IL). The weight measure from the visit that was closest but before the date of the LMP for each pregnancy was considered the prepregnant weight. This study received approval from the University Committee for Human Subjects at Cornell University and the Research Subjects Committee at the Institute of Nutrition of Central America and Panama (INCAP).
Pregnant women who accepted participation and signed an informed-consent form received free prenatal care 4 times during pregnancy from field staff nurses of INCAP. Two ultrasound examinations were scheduled for each woman; additional ultrasound measurements were taken during examinations that were scheduled for obstetric follow-up reasons.
Neonates were weighed on the electronic scale, and crown-heel length was measured within 48 h of birth by using a portable, combined stadiometer-infantometer (Schorr Productions, Glen Burnie, MD). For women, height was measured at prepregnancy or at the first prenatal visit. The current analysis includes the subset of women who met the following inclusion criteria: 1) =" BORDER="0"> 2 ultrasound examinations, with the first conducted before 24 wk of gestation; 2) infant weight and length measurement at birth; 3) maternal weight on =" BORDER="0"> 2 occasions; 4) singleton, term pregnancy with no congenital anomalies. Gestational age was estimated on the basis of the maternal recalled date of LMP.
Ultrasound measurements
All fetal measurements were made by using real-time ultrasound on a portable machine (Medison Eureka 600; Medison Inc, Seoul, South Korea) equipped with a 3.5-MHz transducer at a sound velocity of 1540 m/s. Repeat measures were taken from separate scans until 3 measures within 2 mm were recorded for femur length and 2 measures within 2 mm for tibia length. Femur length was measured according to the method of O’Brien and Queenan (20), and tibia length was measured as described by Hansmann et al (21). Both measures included only the diaphysis of the bone. One of 2 trained obstetricians made all of the ultrasound measurements. The results of the ultrasound examination were also recorded on a report form, which was given to the participant for her information and medical follow-up. A description of fetal size for gestational age compared with reference values is presented elsewhere (22).
Statistical analysis
Anthropometric and ultrasound measurements were divided into trimesters by using the following criteria: first trimester, 13 wk of pregnancy; second trimester, 14–26 wk of pregnancy, and third trimester, 27–34 wk of pregnancy. The actual range of weeks of pregnancy during each trimester varied slightly for the anthropometric and ultrasound measurements because the 2 activities were not programmed for the same session. Ultrasound examinations were conducted during the second and third trimesters only.
The measurement of femur length is presented as the mean and SD of the z score, by trimester, with the use of the reference of Chitty et al (23). This was done to adjust for the age at which the measurements were taken and to allow for a comparison between measurements in the second and third trimesters; this procedure was not conducted for the tibia measurements because no reference mean and SD were found.
Growth rates for the femur and tibia were calculated as the earlier measure subtracted from the latter measure divided by the number of weeks between them. This approach controls for differences in measurement periods but not for nonlinearities in growth during the period. Results are presented separately for femur length and tibia length. Intrauterine growth retardation was defined as birth weight less than the 10th percentile of a sex- and gestational age–specific reference (24), and stunting at birth was defined as birth length for gestational age less than the 10th percentile of the same reference. The rates of maternal weight gain from the first to the second and from the second to the third trimesters were calculated by subtracting the earlier from the latter and dividing by the number of weeks between them. Again, we did not account for the possibility of nonlinear weight gain during each period.
The influence of nutritional status before and during pregnancy, as measured by maternal weight, on fetal linear growth was assessed by using ordinary least-squares regression. Two series of models were run, first to test the influence of weight change by trimester (first to second and second to third trimesters) on fetal linear growth and length at birth and then to test the influence of weight change over the entire measurement period (first to third trimester) on the outcomes. Maternal prepregnant weight was included to adjust the models for nutritional status at the beginning of pregnancy, and maternal height was included to adjust for genetic and environmental influences. In each model we tested, as an interaction between prepregnancy weight or height and weight gain, whether the relation between changes in maternal weight and fetal outcomes was dependent on initial weight or height. We also tested whether the rate of gain in one period was modified by the rate of gain in the previous period. Three-way interactions between prepregnant weight and weight gain from the first to second and from the second to third trimesters were also tested. P values < 0.05 were considered statistically significant for main effects and P values < 0.10 were considered statistically significant for interactions.
Sample size was calculated to ensure sufficient power to detect a difference in femur length at 17 and 30 wk between infants who were and were not born stunted on the basis of a two-tailed test with a probability of type I error of 0.05 and a power of 0.8. At the time we planned this study, there was a dearth of information on the extent to which femur growth is compromised in infants born stunted. Therefore, we assumed that at 17 wk, the femur length of infants who would eventually be born stunted would be, on average, 1 SD below the reference mean (20) and that the average femur length of infants not stunted at birth would be similar to the reference mean. A similar assumption was made for femur length at 30 wk, assuming however that the difference would be 2 SD. A sample size of 126 would be needed to detect a 1-SD difference in femur length at 17 wk, and a sample size of 60 would be needed to detect a 2-SD difference at 30 wk. Data were entered in duplicate and cleaned at the INCAP computing center with the use of EPI-INFO (version 6; USD Inc, Stone Mountain, GA) and were analyzed with the use of SAS for WINDOWS (version 8; SAS Institute Inc, Cary, NC).
RESULTS
A total of 200 women participated in the study. Sample sizes for the regression analyses varied from 112 to 154, depending on the availability of ultrasound examinations and maternal weight measurements during specific periods of gestation. The ultrasound examinations were conducted at a mean of 17.5 ± 2.4 (range: 15–24) and 29.9 ± 0.9 (range: 28–32) wk for the second and third trimesters, respectively. Prenatal care, including the measurement of maternal height at the first visit only and the measurement of maternal weight at all visits, was conducted at a mean of 11.0 ± 1.6 (first trimester), 19.4 ± 1.5 (second trimester), 29.9 ± 0.7 (third trimester), and 37.0 ± 0.6 wk gestation. Weight measurements taken after 34 wk were not used in the regression analysis to ensure that the measurement was taken as close as possible to the time that the ultrasound examination was performed. In all subjects, there was =" BORDER="0"> 4 wk between the initial ultrasound examinations and maternal weight measurements and subsequent measurements. The sample consisted of 98 (49%) male and 102 (51%) female infants. The average birth weight of the group was 0.5 SD below the reference mean (birth weight z score: -0.46), and the average birth length was 1 SD below the reference mean (birth length z score: -1.03). Thirty-seven infants (18.5%) were classified as having intrauterine growth retardation, and 16 (8.0%) were classified as low birth weight (< 2500 g). Stunting at birth was observed in 73 (36.5%) infants. Maternal weight gain over the entire measurement period, prepregnancy to 37 wk (8.4 ± 4.1 kg), may have been underestimated as true gestational weight gain because of the dates of the measurements (Table 1).
View this table:
TABLE 1. Descriptive statistics for maternal anthropometric measures before and during pregnancy and for infant anthropometric measures at birth
Infants who were born stunted had a lower mean birth weight than did their nonstunted counterparts, but the mean gestational age did not differ significantly between the 2 groups (Table 2). The mothers of infants who were born stunted had a significantly lower mean weight and height prepregnancy and lower rates of weight gain from the first to the second trimester than did their nonstunted counterparts, but no significant difference in the rate of weight gain from the second to the third trimester was observed between the 2 groups. Although there was a tendency toward shorter femur length in the second trimester in those infants who were born stunted, the difference was not statistically significant until the third trimester. There was no difference in the rate of growth of the tibia from the second to the third trimester between the stunted and nonstunted infants, despite a significantly faster rate of growth of the femur in the nonstunted infants during the same period.
View this table:
TABLE 2. Maternal anthropometric measures and ultrasound measures of fetal size in infants who were stunted or not stunted at birth1
The rate of maternal weight gain from the first to the second trimester of pregnancy was a significant predictor of femur (model 1) and tibia (model 2) length in the second trimester (Table 3). The rate of maternal weight gain from the first to the second trimester was also a significant predictor of femur (model 3) and tibia (model 4) length in the third trimester, but the rate of maternal weight gain from the second to the third trimester did not predict femur or tibia length in the third trimester. The relation between maternal weight gain and fetal femur and tibia length was not modified by the prepregnant weight of the mother, tested as an interaction between maternal prepregnant weight and the rate of maternal weight gain, for any of the analyses. Maternal height was not a significant predictor of femur or tibia length in either trimester, and height did not modify the relation between maternal weight gain and femur and tibia length. Infant sex was not a predictor of femur or tibia length in either trimester. The omission of infant sex or maternal height in the model did not alter the regression coefficients for the other predictors; for this reason, these variables were not included in the final models shown in Table 3.
View this table:
TABLE 3. Influence of maternal weight gain during selected periods of pregnancy on femur and tibia lengths in the second and third trimesters1
The rate of maternal weight gain from the first to the second trimester of pregnancy was a significant predictor of infant length at birth (model 1, Table 4), after adjustment for covariates. The lack of association between the rate of maternal weight gain from the second to the third trimester, and infant birth length (model 2) was not changed when the rate of weight gain from the first to the second trimester was removed from the model (model 3). For comparison, the influence of the rate of weight gain from the first to the third trimester on infant birth length is also shown (model 4).
View this table:
TABLE 4. Influence of maternal weight gain during selected periods of pregnancy on infant recumbent length at birth1
DISCUSSION
The relation between the rate of maternal weight gain from the first to the second trimester and femur length, tibia length, and infant length at birth—independent of maternal prepregnant weight and height—suggests that this may be a sensitive period for fetal linear growth. A tendency toward smaller femur length in those infants who were born stunted compared with the infants who were not born stunted was apparent at our earliest ultrasound measurement (17 wk gestation). Although not statistically significant until the third trimester, the absolute size of that deficit increased only slightly in later gestation (from 0.44 to 0.62 z scores), which supports the suggestion of a sensitive period for fetal linear growth during early gestation.
The association between maternal weight gain during the first 2 trimesters of pregnancy and infant length at birth was documented in an older generation of this population (25). The association between maternal weight gain during the first 2 trimesters of pregnancy and infant length at birth was documented in an older generation of this population, ie, the mothers of the current study population. A statistically significant relation was found between changes in maternal weight during midpregnancy (3–6 mo), but not in later pregnancy (6–9 mo), and length at birth. In the current study, we also examined relations between maternal weight gain and infant length at birth in the current generation of women and, in addition, incorporated ultrasound measurements to investigate when the process of stunting is most evident in intrauterine life. Stein et al (26) also found a relation between maternal weekly weight gain and birth outcomes, only when weight gains were < 0.5 kg/wk. Considering that average weekly weight gains in both periods reported in the current analysis were considerably < 0.5 kg/wk, it is not surprising that we found strong relations between the rate of maternal weight gain and measures of fetal linear growth. Interestingly, although maternal height (a reflection of genetic potential and environmental influences, including nutrition) predicted infant length at birth, it was not a significant predictor of femur or tibia length in either trimester or of the rate of growth of either bone. After careful and varied analyses, we concluded that maternal weight gain (a reflection of changes in maternal muscle and fat stores and of recent dietary intakes) exerts a greater influence on femur and tibia growth than does maternal height.
It has been postulated that fetal linear growth is affected when factors that cause fetal growth restriction occur early in gestation (27). The growth velocity curves published by Tanner (28) form the basis of this hypothesis. The fetal linear growth rate is shown by Tanner to peak in velocity at 16 wk, followed by a fairly rapid decline. Ultrasound estimations of the growth velocity for femur length measured on healthy non-growth-retarded fetuses (29–31) suggest that, indeed, peak growth velocity occurs during the early second trimester but that the decline in velocity is relatively gradual, with femur growth continuing into the late third trimester. The relation between maternal weight gain in early pregnancy and growth of the fetal femur and tibia appears consistent with this pattern.
The relation between maternal nutrient deficiency and fetal growth is complex. In undernourished mothers, a series of metabolic responses to insufficient energy and protein availability result in a less-than-adequate exchange of nutrients between the mother and the fetus (32). The role of micronutrient deficiency in fetal growth has been the topic of many recent reviews (1, 33, 34), the results of which implicate the deficiency of many micronutrients (eg, zinc, iron, iodine, and folate) in fetal growth retardation. We used measures of maternal weight to reflect the overall adequacy of nutrient availability to meet maternal, placental, and fetal needs. The inability of this analysis to identify specific limiting nutrients for fetal linear growth in this population is acknowledged. Inasmuch as inadequate weight gain during pregnancy reflects inadequate energy intake, we suggest that, on average, energy intake in this population is marginally inadequate.
Maternal weight gain in the first and early second trimesters is largely a reflection of the deposition and expansion of maternal tissues, and gains from late in the second trimester until the end of pregnancy reflect primarily fetal and placental growth and accumulation of amniotic fluid (35–37). Early maternal weight gain may also reflect the adequacy of the nutrient supply to the placenta, which in turn ensures its adequate growth, development, and function (38). The association between weight gain from the first to the second trimester and fetal linear growth may actually reflect better placental nutrient transport and a better supply of endocrine growth factors in women who gain more weight. The exact nature of this relation is difficult to assess in this observational study, and further research is needed to understand the relation between maternal weight gain and fetal linear growth.
Contrary to our expectations, femur length throughout gestation was not shorter than reference values, even in infants born stunted. At this time, we have no adequate explanation for these findings. We chose to use the reference of Chitty et al (23), specifically because the measurement techniques they used were similar to those used in the current study. Considering the training and standardization of our ultrasonographers, we do not believe that the difference was due to measurement error. Nonetheless, technique may influence the results if our ultrasonographers systematically measured longer femur length than the reference. Femur lengths similar to reference populations were reported in other populations with low average birth lengths (39, 40). This apparent inconsistency could be explained by shorter trunk length in infants who were stunted at birth. The extent to which fetal trunk growth is compromised in infants who are born stunted in total length is not known, and trunk length cannot be measured accurately in utero. We did not measure crown-rump length at birth and, thus, could not compare proportions between leg and trunk length. Growth of the lower leg (knee-heel length) is highly correlated with growth in total recumbent length in stunted children 4–7 mo of age (41). Considering the positive relation between maternal weight gain and femur and tibia growth and infant length at birth, it seems probable that femur and tibia growth are compromised in infants who are born stunted. Studies that measure femur and tibia length with the use of similar ultrasound techniques in populations with and without stunting at birth may help clarify this issue.
Errors in the estimation of gestational age could influence the results of the study reported here. For example, we would expect that the relation between maternal weight gain and femur and tibia length would be attenuated if small fetuses were considered to be younger than their true gestational age. We avoided this potential bias by using the maternal recalled date of LMP to estimate gestational age. In a separate analysis, we compared the estimate of gestational age from the maternal recalled date of LMP (a second trimester measure of symphyseal fundal height) and the Capurro et al (42) examination of neonatal maturity with the use of age estimated from an ultrasound measurement of biparietal diameter made early in the second trimester as the gold standard. We found that the recalled date of LMP provided an excellent estimate of gestational age in this sample (LM Neufeld, unpublished observations, 2000).
Our results suggest that early gestation may be a sensitive period for fetal linear growth. Further research is needed to fully understand the etiology of fetal linear growth faltering and to determine how the finding of trimester-specific relations between changes in maternal weight and fetal linear growth can be applied to interventions designed to influence fetal growth.
ACKNOWLEDGMENTS
We gratefully acknowledge the support of Humberto Mendez, the computing center staff, and the field staff at INCAP and thank Kathy Rasmussen and David Pelletier for their comments on an earlier version of this manuscript.
JDH and RM were involved in the planning and design of the study. RG assisted with the implementation of the study. JDH, RM, and RG were involved in the interpretation of the results and writing of the manuscript. LMN developed the idea for the study, designed the data collection methods, and was responsible for the data analysis, the interpretation, the writing of the first draft, and for integrating the comments from the coauthors. None of the authors had any relation, either personal or financial, with the sponsors of this research.
REFERENCES