Patterns and success of fetal programming among women with low and extremely low pre-pregnancy BMI
- First Online:
- Cite this article as:
- Belogolovkin, V., Alio, A.P., Mbah, A.K. et al. Arch Gynecol Obstet (2009) 280: 579. doi:10.1007/s00404-009-0965-8
- 80 Views
To estimate the frequency of fetal programming phenotypes among women with low BMI and the success of these programming patterns-to determine if small for gestational age (SGA) is a biologically adaptive mechanism to improve chances for infant survival.
We examined the frequency of fetal programming phenotypes: SGA, large for gestational age (LGA), and adequate for gestational age (AGA) among 1,063,888 singleton live births from 1978 to 1997. We also estimated the success of fetal programming phenotypes using neonatal death as the primary study outcome.
Underweight gravidas with AGA and LGA babies had elevated risk of neonatal mortality when compared to normal weight mothers, while the risk for neonatal mortality among mothers with SGA babies was reduced.
The variation in relative degrees of fetal programming patterns and success observed suggests that underweight mothers are more likely to succeed in programming SGA fetuses rather than any other phenotype.
KeywordsLow BMISGAFetal programmingNeonatal mortality
The impact of pre-pregnancy body mass index (BMI) has previously been evaluated on a wide spectrum of maternal and neonatal complications of pregnancy including preterm birth, incidence of small and large for gestational age (LGA) infants, cesarean delivery incidence, gestational hypertension, gestational diabetes, preeclampsia and abruption. Whereas, low pre-pregnancy BMI has been reported to reduce the incidence of some maternal complications such as preeclampsia it has been associated with an increased incidence of small for gestational age infants (SGA) [1–5]. The underlying mechanism remains unclear but may be secondary to nutritional deficiency.
We have recently demonstrated (using epidemiologic data) that the success of fetal programming in utero depends largely on the underlying adaptive mechanisms and not necessarily on the resultant fetal phenotype . A seemingly adverse birth outcome (e.g., SGA) may not be indicative of in utero fetal programming failure but may indeed, represent the best chance of survival for the fetus given the underlying intra-uterine environment. As earlier stated, women with low pre-pregnancy BMI are at an elevated risk for SGA babies [1–5] but it remains unknown whether this higher likelihood for SGA programming in women with low BMI is a biologic selective mechanism to improve survival chances of the infant or rather a failure to program other fetal phenotypes (e.g., appropriate-for-gestational age infants) with enhanced survival chances. To answer this important question, we undertook this study, the main objective of which is to estimate the frequency of fetal program phenotypes among women with low BMI and the success of these programming patterns.
Materials and methods
We utilized in this analysis the Missouri maternally linked cohort data files covering the years 1978 through 1997. Unique identifiers were used to link siblings to their biological mothers in this dataset, and the algorithm used in linking birth data into sibships and the process of validation have been described in detail previously . The Missouri vital record system is a reliable one that has been adopted as “gold standard” to validate US national datasets that involve matching and linking procedures .
Live singleton births within the gestational age range of 20–44 weeks were used for the analysis. Maternal pre-pregnancy weight groups were obtained using BMI [weight (in kg) divided by height squared (in m2)]. Height and pre-pregnancy weight as reported at the first prenatal visit were used to calculate pre-pregnancy BMI . From previously published results [10, 11], women were categorized into the following BMI-based categories: normal (18.5–24.9) and underweight (<18.5). Underweight was further classified as: mild thinness (17.0–18.5), moderate thinness (16.0–16.9) and severe thinness (≤15.9) [12, 13]. We considered as referent category mothers of normal weight.
We examined the following fetal phenotypes: small for gestational age (SGA, defined as <10th percentile of birth weight for gestational age using population based national reference curves for singletons), large for gestational age (LGA, defined as ≥90th percentile of birth weight for gestational age) and adequate for gestational age (AGA, defined as those of normal birth weight children). Gestational age was largely based on the interval between the last menstrual period and the date of delivery of the baby (95% of cases). When the menstrual estimate of gestational age was inconsistent with the birth weight (e.g., very low birth weight at term), a clinical estimate of gestational age on the vital records was used instead . The accuracy of using gestational age as reported on the US birth certificate has previously been validated . In that validation study, the authors evaluated the concordance between date of last menses (DLM) reported among cases (very low birth weight infants or neonatal deaths) and a randomly selected non-case population. There was a very good agreement (84.2%) with medical records. Clinical estimate of gestation in completed weeks was 79.0% concordant for cases, and 94.0% for controls. This has recently been found to be consistent with an anticipated ±2 weeks variation in DLM . Hence, gestational age as used in this study is reasonably valid.
We determined the success of a specific fetal programming phenotype (AGA, SGA or LGA) using neonatal death (death of the newborn within the first 28 days of life) as the primary outcome of interest. We further categorized the time of occurrence of neonatal mortality as early neonatal mortality (death of the newborn within the first 7 days of life) and late neonatal mortality (death of the infant from day 8 through 28). We selected neonatal death as the primary outcome because it is more closely related to pregnancy-associated events (e.g., fetal programming) than infant and post-neonatal death. Interest lies in determining whether a given fetal phenotype programmed by a woman with low BMI had the same expected risk of mortality as a similar infant phenotype programmed by a woman with normal BMI values.
For the following selected maternal socio-demographic characteristics, we compared the distribution between normal and underweight gravidas to assess differences in baseline characteristics: maternal age, parity, education, marital status, smoking habits and adequacy of prenatal care. Adequacy of prenatal care was assessed using the revised graduated index algorithm, which has been found to be more accurate than several others, especially in describing the level of prenatal care utilization among groups that are high-risk [17, 18]. This index assesses the adequacy of care based on when the trimester prenatal care began, number of visits, and the gestational age of the infant at birth. In this study, inadequate prenatal care utilization refers to women who either had missing prenatal care information, had prenatal care but the level was considered sub-optimal, or mothers who had no prenatal care at all. The prevalence of common obstetric complications, namely, anemia, insulin-dependent diabetes mellitus, other types of diabetes mellitus, chronic hypertension, preeclampsia, eclampsia, placental abruption and placenta previa was compared between the two groups using crude frequency analysis.
Chi-square test was used to assess differences in socio-demographic characteristics and maternal pregnancy complications between the two groups (normal weight/underweight). We used logistic regression models to generate adjusted odd ratios and their 95% confidence intervals. The covariates loaded onto our model include: maternal age, maternal education, marital status, maternal race, prenatal smoking, adequacy of prenatal care, gender of the infant and year of birth. The model selection was based on the consideration of known risk factors for feto-infant morbidity, biologic plausibility and methodological considerations (e.g., adjustment for year of birth). Adjusted estimates were derived in all cases by using normal weight gravidas (BMI = 18.5–24.9) as the referent category.
Because the dataset also contains consecutive siblings, we estimated regression parameters by taking into account the presence of intra-cluster correlation using the methodology of generalized estimating equations (GEE) . We constructed the regression models and assessed goodness-of-fit using the −2 log likelihood ratio test. We estimated the significance of main effects by means of the Wald test . The GENMOD procedure in SAS (SAS Institute, Inc., Cary, North Carolina, version 9.1) was used to conduct the analysis.
All tests of hypothesis were two-tailed with a type 1 error rate fixed at 5%, and SAS version 9.1 (SAS Institute, Cary, NC) was used to perform all analyses. This study was approved by the Office of the Institutional Review Board at the University of South Florida.
Overall, 1,527,613 mothers with a singleton live birth were included in the study population which covers the period from 1978 to 1997. We excluded from our study population records with gestational age outside of the interval of 20–44 weeks (N = 74,895; 4.9%) and cases where information on BMI was either missing or implausible (N = 28,332; 2.0%). The resulting study population was then categorized by pre-pregnancy BMI status: underweight (N = 149,137 or 14.0%) and normal weight (N = 914,751 or 86.0%) totaling 1,063,888 singleton live births.
Comparison of underweight and normal weight mothers by selected socio-demographic characteristics for singletons, Missouri, 1978–1997
Underweight N = 149,137 (14.0%) %
Normal weight N = 914,751 (86.0%) %
Maternal age (≥34 years)
Education (>12 years)
Adequate prenatal care
Prevalence of common obstetric complications among mothers giving birth to singletons, Missouri, 1989–1997
Underweight N = 58,041 (13.3%)
Normal N = 379,773 (86.7%)
Other forms of diabetes
Infants of underweight mothers were born slightly earlier than those of normal weight mothers [mean (±SD) = 39.03 weeks (±SD = 2.69 weeks) vs. 39.30 weeks (±SD = 2.42 weeks) respectively]. The mean birth weights of infants of the two study groups were significantly divergent. Infants of underweight mothers weighed on average 201 g less than those born to mothers of normal weight [mean (±SD) = 3,164.90 g (±SD = 570.75 g) vs. 3,365.80 g (±SD = 563.76 g) respectively; P < 0.01.
Adjusted odds ratio for neonatal death and its subtypes by gestational age among mothers giving birth to singletons by underweight subclass, Missouri, 1978–1997
SGA OR (95% CI)
AGA OR (95% CI)
LGA OR (95% CI)
Early neonatal death
Late neonatal death
Early neonatal death
Late neonatal death
Early neonatal death
Late neonatal death
The results of our analysis showed that in comparison to normal weight mothers those who were underweight tended to have an increased frequency of SGA infants, and this tendency increased with ascending severity of maternal thinness in a dose-effect pattern. On the other hand, a decreasing trend was noted for AGA or LGA programming in utero. These results are not unexpected since fetal programming mirrors level of maternal resources available to the developing fetus. The line of evidence derived from Barker’s hypothesis posits that as a survival mechanism in an unfavorable or nutrient-deficient uterine environment, the developing fetus acquires adaptation mechanisms of survival through lowering of growth rate which matches demand to limited supply [21, 22]. Our findings are consistent with this widely held theory as well as the results of studies conducted on survivors of famine in Europe . The additional observation of a dose-effect pattern in which ascending severity of maternal thinness correlates with increasing tendency to program SGA infants further strengthens the association observed in this study.
The success of fetal programming as measured by ability of the neonate to survive the first 28 days of life was observed to vary across the fetal phenotypes programmed by underweight mothers. When underweight mothers programmed SGA infants, they had in general a 15% greater likelihood of succeeding (survival in the neonatal period) than their normal weight counterparts. By contrast, programming of AGA or LGA infants was less successful among underweight as compared to normal weight mothers. Indeed, when underweight mothers programmed AGA infants, the likelihood of neonatal mortality among these neonates was 22% greater than for those AGA neonates programmed in uteri of normal weight mothers. The widest disparity in fetal programming success was noted with respect to LGA phenotype. LGA neonates programmed in uteri of underweight mothers had a 84% greater likelihood of neonatal demise as compared to same phenotype programmed by normal weight mothers.
The variation in relative degrees of fetal programming observed tends to suggest that underweight mothers are more likely to succeed in programming SGA neonates than any other phenotype. This may also be explained by maternal nutritional constraint imposed by the underweight status of the mother. Although we used BMI to determine underweight status, it has been suggested that BMI is a reflection of nutritional status of the mother . This implies that underweight mothers are prone to nutrient deficiencies including those needed for optimal fetal programming (e.g., folate and zinc). The initiation and progression of optimal growth and development of the fetus and the placenta is to some extent under the control of epigenetic events during pregnancy [25, 26]. According to Reik et al. paternally derived imprinted genes enhance fetal and placental growth, whereas maternally imprinted genes down-regulate this growth . Gene imprinting is an important mechanism that controls and regulates fetal as well as trophoblastic invasion, and is also driven by maternal nutritional factors . For example, maternal nutritional status may impact epigenetic mechanisms such as DNA methylation and histone acetylation (processes that influence gene imprinting). Hence, underweight mothers may not have the sufficient nutritional ingredients required for optimal realization of epigenetic pathways for optimal trophoblastic and fetal growth and development. This theory may to some extent explain the increased creation of SGA phenotypes among underweight mothers, as well as the relative failure in programming AGA and LGA phenotypes since the latter variants may involve a more elaborate epigenetic processes requiring more nutritional input.
The women in the Missouri cohort data file were followed for a relatively long period of time. This is an important limitation to note, since aggregating different cohorts of infants together for analyses may induce bias associated with cohort effect. During this long period of follow-up, obstetric practices likely changed as new knowledge and technologies were introduced. This might have had an impact on our findings. To minimize this potential source of bias, we controlled for year of birth in our adjusted models.
A major source of strength in our study is that the data is population-based, and thus the validity of our findings are less likely to be threatened by selection biases, and are more likely to be generalizable. Another merit of the study is the knowledge gap it will potentially fill with respect to fetal programming patterns and their successes among women that have sub-optimal pre-pregnancy weight. To our knowledge this is the first paper that has examined fetal programming patterns and successes among underweight mothers across the spectrum of severity of maternal thinness.
This work was supported by an obesity grant from the Flight Attendant Medical Research Institute (FAMRI) to Dr. Hamisu Salihu (Last author). The funding agency did not play any role in any aspect of the study. We thank the Missouri Department of Health and Senior Services for providing the data files used in this study.
Conflicts of interest statement
The authors declare that they have no conflicts of interest with this research.