Risk factors for macrosomia in infants born to Latina women

Abstract

Objective

To assess risk factors for macrosomic infant birth among Latina women.

Study Design

Prospective study of Latina women recruited during pregnancy from prenatal clinic at San Francisco General Hospital. Information was obtained through a structured interview and review of medical records.

Result

A total of 11% of women delivered macrosomic infants (birth weight >4000 g). In unadjusted analyses, significant risk factors for macrosomia included older maternal age, increasing gravidity, previous history of macrosomic birth and pre-pregnancy overweight. After adjusting for confounders using multivariate analyses, older mothers (10-year increments) had an elevated risk of macrosomia (odds ratio (OR) 2.59; 95% confidence interval (CI) 1.28 to 5.24).

Conclusion

Efforts to reduce macrosomia in Latina women should focus on older mothers.

Introduction

In 2004, 8.5% of all United States (US) births were high birth weight (≥4000 g).¹ Macrosomic infants (defined as birth weight >4000 g using a standard cut point from the American College of Obstetricians and Gynecology Practice Bulletin on Fetal Macrosomia) have a greater risk for delivery complications such as shoulder dystocia, brachial plexus injury, clavicular fracture, meconium aspiration and perinatal asphyxia² as compared to infants of average birth weight (defined as 2500 to 4000 g).

Macrosomic infants also have a greater risk for overweight, obesity and diabetes mellitus during childhood and adulthood. Data from the Centers for Disease Control and Prevention’s (CDC) Pediatric Nutrition Surveillance longitudinal study demonstrates that children who were macrosomic at birth have a high prevalence of overweight during childhood (defined as body mass index (BMI) ≥95th percentile); one-third of macrosomic infants are overweight at age 3 to 4 years.³ Domestic and international longitudinal studies have demonstrated that higher birth weight increases the risk for adulthood obesity by a factor of 1.5 to 2.^4–6 Overweight in childhood and obesity in adulthood are important because of associations with the metabolic syndrome⁷ and cardiovascular disease.⁸

Previous epidemiologic studies have identified a number of maternal risk factors for macrosomic infant birth. These include older age (>40 years), tall height (≥165 cm), high pre-pregnancy BMI (BMI≥30), nonsmoking, preexisting and gestational diabetes mellitus (GDM), high gravidity, and prolonged gestation (>41 weeks).^9–14 Small studies have also identified excessive gestational weight gain as a risk factor for macrosomic birth.^15,16

US population data suggest that ethnicity may interact with known maternal risk factors and influence the incidence of macrosomic birth.¹ The CDC’s National Natality Vital Statistics indicate that the incidence of macrosomic birth among Latinas (7.9%) is lower than the incidence among whites (10.0%).¹ Among Latino births, infants born to Mexican-origin mothers had a macrosomic rate of 8.5% and infants born to Central and South American-origin mothers had a lower rate (7.8%). These national data are consistent with a study of the Northern California Region of Kaiser Permanente’s Medical Care Program, where Latina women had a 30% lower risk of macrosomia than White women.¹² It is surprising that Latinas have a lower rate of macrosomia than White women because Latinas have a higher incidence of certain risk factors for macrosomia (gestational diabetes, high pre-pregnancy BMI). However, the lower incidence of macrosomia among Latinas suggests that there may be protective factors such as environmental, behavioral or biologic factors that are specific to Latinas.^17–19

The objective of the current study was to evaluate the relationship between maternal characteristics and the risk of macrosomia in Latina women in a general population of pregnant women.

Methods

Study design

From July 1997 to September 1999, 350 pregnant Latina women were recruited from the prenatal clinic at San Francisco General Hospital, the only public hospital in San Francisco County. Women were recruited at ≥20 weeks’ gestation and followed prospectively to delivery, as previously described.²⁰ In this study, we defined Latina women as women who self-identified as Latina. Data were collected during one prenatal, structured interview that included questions about health history, diet and sociodemographics. Questions were asked about previous history of diabetes mellitus, GDM in this pregnancy and other health conditions. Interviews were conducted by three bilingual, bicultural research assistants and were conducted at a mean gestational age of 31.2±5.6 weeks. Infant birth weight was abstracted from the nursery logbook and measured by a digital scale following delivery, clamping and cutting the umbilical cord. Maternal pre-pregnancy weight was obtained by self-report and BMI (kg/m²) was calculated using self-reported pre-pregnancy weight and measured height. BMI <25 kg/m² was categorized as normal/underweight, BMI ≥25 and <30 was categorized as overweight, and BMI ≥30 was categorized as obese, using standard definitions from the Centers for Disease Control and Prevention.²¹ The study was originally designed and powered to test for acculturation differences in immunization outcomes. The Committee on Human Research at the University of California, San Francisco, approved the study and written consent was obtained from each woman.

Statistical methods

The dependent variable was infant birth weight, dichotomized as macrosomic infant birth weight (>4000 g), yes or no. The American College of Obstetricians and Gynecology Practice Bulletin on Fetal Macrosomia uses two cut points or grades for macrosomia (>4000 and >4500 g).²² We chose the >4000 g cut point (grade 1 macrosomia) so that we could compare our results with some of the larger cohort studies that use this cut point.^9,10,14 We also evaluated risk for higher infant birth weight using birth weight as a continuous variable (in increments of 100 g).

The primary independent variable in these analyses was pre-pregnancy BMI. Other independent variables were maternal age, number of pregnancies (gravidity), history of previous macrosomic birth, self-reported history of diabetes and hypertension, and nutritional factors. The nutritional factors included consumption of fresh fruit, vegetables, ‘American food’ (defined as spaghetti, potatoes and fast food), use of prenatal vitamins and prenatal iron supplements, and advice about what to eat. Questions on food consumption during pregnancy were asked using food frequency methods adapted from a larger food screener questionnaire developed by Block et al.²³ and validated against dietary recall methods (frequency defined as during the last week and throughout the pregnancy) with the categories of never, 1 to 3 days per week, 4 to 6 days per week, everyday, and two times a day. Beverage consumption (soda and coffee) during the last week and throughout the pregnancy was also ascertained using the following categories: none-<1, 1, 2 to 3 and >3 cups per day. We asked women whether they received advice about what to eat during pregnancy from their family, friends or health professionals, and this information was coded as a dichotomous variable. In addition, we considered two infant factors known to be associated with macrosomia:¹⁰ sex and gestational age. Gestational age was categorized as <38, 38 to 40 and >40 weeks. We used the <38 weeks cut point instead of the more standard <37 weeks because of the small number of infants born at <37 weeks in our study. We also evaluated risk for macrosomia based on maternal national origin, dichotomized as Mexican versus Central/South American/United States, and acculturation to the United States, using years residing in the United States, language and self-identification (Latina versus American) as indicators of acculturation. The frequencies of known risk factors for macrosomia were evaluated in relationship to maternal subgroup. Means and standard deviations were calculated for the following continuous variables: maternal age, pre-pregnancy weight, height, pre-pregnancy BMI, infant birth weight and gestational age. Unadjusted and adjusted logistic regression analyses were used to assess risk factors for macrosomic infant birth weight. In adjusted analysis, we used logistic regression models to analyze all maternal variables that were significant in the bivariate models (P<0.05) as well as known risk factors for macrosomic birth (maternal history of diabetes, infant sex and gestational age). We did not include the variable ‘history of macrosomic infant’ in the multivariate models because inclusion of this variable would have required restricting the analyses to parous women. Only women with complete data on all study variables were included in the multivariable models (n = 305).

Results

Of the 350 women recruited for this study, 348 had data on infant birth weight. The mean age of the participants was 25.3±6.0 years, mean pre-pregnancy weight was 62.5±12.3 kg, mean infant birth weight was 3.4±0.5 kg and mean gestational age was 39.1±1.7 weeks (Table 1). Approximately 5% of women reported having diabetes before or during their pregnancy. Nearly all women (94.8%) were foreign-born, including 56.3% born in Mexico and 38.5% born in Central or South America.

Table 1.

Mean, standard deviation and frequencies for selected characteristics among 348 Latina women, San Francisco, CA, 1997–1999

	Mean	Standard deviation	% (n/total)
Maternal characteristics
Age (years)	25.3	6.0
Pre-pregnancy weight (kg)	62.5	12.3
Height (cm)	155.0	6.2
Gravidity	2.5	1.6
Prior history of diabetes mellitus or diabetes mellitus with this pregnancy			4.9 (17/348)
History of preexisting hypertension			6.6 (23/348)
Maternal birthplace
USA			5.2 (18/348)
Mexico			56.3 (196/348)
Central/South America			38.5 (134/348)
Maternal pre-pregnancy BMI (kg/m2)
Underweight/normal (<25.0)			47.1 (146/310)
Overweight (25–29.9)			33.6 (104/310)
Obese (≥30.0)			19.4 (60/310)
Infant characteristics
Birth weight (kg)	3.4	0.5
Gestational age (weeks)	39.1	1.7

Abbreviation: BMI, body mass index.

Nearly 11% of the Latina women in the cohort delivered macrosomic infants and 95% of the women were interviewed in Spanish. We did not find a significant difference between Mexican-origin mothers and Central/South American-origin mothers in terms of known risk factors for macrosomic birth such as pre-pregnancy obesity, preexisting diabetes mellitus, gestational diabetes, gestational age or previous history of macrosomic birth (data not shown). Looking at measures of acculturation including language use, cultural self-identification and years lived in the United States, we did not find any statistically significant association with macrosomia. However, our results trended in the direction toward indicating that acculturation was associated with a reduced risk of macrosomia. Women who considered themselves very American were less likely to have macrosomic infants (3.5%) than women who considered themselves Latina (11.6%, P = 0.21). We also found a lower rate of macrosomia (6.8%) among women who had some English speaking skills in comparison with those who predominately spoke Spanish (11.5%) but our results were not statistically significant (P = 0.36; Table 2).

Table 2.

Rates (%) and unadjusted logistic regression odds ratios (ORs) for infant macrosomia, by selected maternal and infant characteristics, among 348 Latina women, San Francisco, CA, 1997–1999

	Rate of macrosomia % (n/N)a	OR (95% CI)b
Maternal sociodemographic characteristics
Age (years)
10–19	1.6 (1/61)	3.09 (1.80–5.32)c
20–29	11.3 (23/204)
30–39	13.7 (10/73)
40–49	40.0 (4/10)
Education (years)
<6	15.7 (8/51)	1.67 (0.71–3.85)
≥6	10.1 (30/297)	1.00
Marital status
Single/divorced	8.6 (7/81)	0.72 (0.31–1.70)
Married/living with partner	11.6 (31/267)	1.00
Language use
Only Spanish/Spanish>English	11.5 (35/304)	1.78 (0.52–6.05)
Spanish = English/English>Spanish/only English	6.8 (3/44)	1.00
Self-identification
Very Latina/mostly Latina	11.6 (37/319)	3.67 (0.49–27.80)
Latina = American/mostly American/very American	3.5 (1/29)	1.00
Maternal origin
Mexico	13.8 (27/196)	2.05 (0.98–4.27)
Central/South America/United States	7.2 (11/152)	1.00
Maternal health and reproductive characteristics
Height (cm)	1.57 (0.91–2.73)c
139.0–148.9	5.7 (3/53)
149.0–158.9	11.5 (22/192)
159.0–168.9	11.8 (9/76)
169.0–178.9	25.0 (2/8)
Pre-pregnancy BMI (kg/m2)
Underweight or normal (<25)	6.9 (10/146)	1.00
Overweight (25–29.9)	13.5 (14/104)	2.12 (0.90–4.97)
Obese (≥30.0)	18.3 (11/60)	3.05 (1.22–7.63)
Gravidity (number of pregnancies)
0–2	7.2 (15/209)	1.00
>2	16.6 (23/139)	2.56 (1.29–5.11)
History of diabetes mellitus or gestational diabetes in this pregnancy
No	10.3 (34/329)	1.00
Yes	17.7 (3/17)	1.81 (0.50–6.62)
History of hypertension
No	10.9 (37/340)	1.00
Yes	14.3 (1/7)	1.37 (0.16–11.65)
History of macrosomic birth
No	7.2 (21/291)	1.00
Yes	29.8 (17/57)	5.46 (2.66–11.23)
Maternal nutrition characteristics
Soda consumption during pregnancyd
None/<1 cup per day	12.0 (28/233)	1.00
1+cups per day	8.0 (8/100)	0.51 (0.26–1.01)
Coffee consumption during pregnancyd
None/<1 cup per day	11.0 (28/255)	1.00
1+ cups per day	9.8 (8/82)	0.88 (0.38–2.01)
Fresh vegetable/salad consumption during pregnancyd
No	13.5 (7/52)	1.00
Yes	10.2 (30/295)	0.73 (0.30–1.76)
Fresh fruit consumption during pregnancyd
<1 per day	14.4 (13/90)	1.59 (0.77–3.28)
≥1 per day	9.6 (24/250)	1.00
Frijoles consumption during pregnancyd
No	14.3 (3/21)	1.00
Yes	10.8 (35/325)	0.72 (0.20–2.58)
‘American food’ consumption during pregnancyd
No	13.1 (24/183)	1.00
Yes	8.5 (14/165)	0.51 (0.24–1.10)
Prenatal vitamin use during pregnancy
No	17.4 (4/23)	1.00
Yes	10.5 (34/325)	0.56 (0.18–1.73)
Iron supplementation during pregnancy
No	12.4 (36/290)	1.00
Yes	3.6 (2/55)	0.27 (0.06–1.14)
Received eating advice during pregnancy
No	9.6 (16/167)	1.00
Yes	12.5 (22/181)	0.77 (0.39–1.51)
Infant characteristics
Gestational age (weeks)
<38	6.3 (3/48)	0.62 (0.18–2.17)
38–40	9.7 (22/227)	1.00
>40	17.8 (13/73)	2.02 (0.96–4.25)
Infant sex
Male	14.1 (25/177)	1.00
Female	7.7 (13/170)	0.50 (0.25–1.02)

Abbreviations: BMI, body mass index; CI, confidence interval; OR, odds ratio.

^aTotal n = 348 but categories do not always add up to 348 due to missing data or applicability of the question to the total population.

^bConfidence interval.

^cAge group is analyzed in ten year age groupings and maternal height is analyzed in 10 centimeter groupings.

^dBased on questions that asked about eating patterns in the last week.

In unadjusted analyses, one sociodemographic factor was significantly associated with macrosomic infants: maternal age (odds ratio, OR 3.09, 95% confidence interval, CI 1.80 to 5.32; Table 2). In addition, pre-pregnancy obesity (OR 3.05, 95% CI 1.22 to 7.63), high gravidity (>2 previous pregnancies, OR 2.56, 95% CI 1.29 to 5.11) and a previous history of macrosomic birth (OR 5.46, 95% CI 2.66 to 11.23), were also associated with an elevated risk of macrosomia. Infant gestational age >40 weeks neared statistical significance (OR 2.02, 95% CI 0.96 to 4.25) compared to infants who were 38 to 40 weeks at time of delivery. Although maternal history of pre-pregnancy diabetes mellitus, any diabetes history (including GDM), maternal height and male infant sex all conveyed an elevated risk of macrosomia, none were statistically significant. A few dietary factors approached statistical significance; iron supplements, soda and ‘American food’ consumption conferred a lower risk of macrosomia but the confidence intervals spanned unity.

In multivariate regression analyses, maternal age (measured in 10-year increments) was the only maternal characteristic that was statistically associated with increased odds of macrosomic birth (OR 2.59, 95% CI 1.28 to 5.24; Table 3). In addition, female infants were half as likely to be macrosomic as male infants. When we changed the outcome to birth weight measured as a continuous variable and used the same variables presented in Table 3 in a multivariate model, we found that higher infant birth weight was associated with male sex, higher maternal BMI groups and longer gestation (results not shown).

Table 3.

Adjusted logistic regression for risk of infant macrosomia among 348 Latina women, San Francisco, CA, 1997–1999^a

Maternal variables	Odds ratio (95% CI)b
Increasing maternal age (10-year increments)	2.59 (1.28–5.24)
History of diabetes mellitus or gestational diabetes mellitus with this pregnancy
Yes	1.21 (0.29–5.03)
No	1.00
Pre-pregnancy BMI (kg/m2)
Normal weight/underweight (<25)	1.00c
Overweight (25–29.9)	1.43 (0.58–3.56)
Obese (≥30.0)	1.88 (0.68–5.16)
Gravidity (number of pregnancies)
1–2	1.00c
>2	1.14 (0.47–2.76)
Infant variables
Gestational age (weeks)
<38	0.55 (0.15–2.06)
38–40	1.00c
>40	2.16 (0.95–4.89)
Infant sex
Male	1.00c
Female	0.42 (0.19–0.93)

Abbreviation: BMI, body mass index; CI, confidence interval.

^aFixed model including covariates that were significant at P<0.05 in bivariate analysis and known risk factors for macrosomic birth.

^bConfidence interval.

^cReference category.

Discussion

In this cohort study of pregnant Latina women in San Francisco recruited without reference to any specific health care need, the rate of macrosomic infant birth (10.9%) was approximately 27% higher than the national rate for Latinas (7.9%), although the national rate uses a slightly different cut point (≥4000 g),¹ in comparison with the >4000 cut point used by American College of Obstetricians and Gynecologists. The 1-g difference in cut point is not likely to explain the difference in incidence of macrosomia. The higher rate of macrosomia in our population of Latina women is surprising because nearly all the women in our study were foreign-born, and foreign-born Latina women have fewer risk factors for macrosomic birth, including lower intake of fast food,²⁴ ratio of fat to total energy intake²⁵ and total caloric intake.²⁶ Furthermore, the women in our study had a relatively low prevalence of known risk factors such as pre-pregnancy obesity, older age (≥40 years) and preexisting diabetes. Thus, there may be an undetermined biologic or environmental factor for macrosomia associated with our population of Latinas that accounts for the elevated rate of macrosomia in our study. Previous studies have analyzed Latinas in a single national origin group (Mexican origin women)^27,28 whereas our study evaluated the risk for macrosomic birth among Latinas from several national origin groups.

In this population of Latina women, increasing maternal age was strongly associated with macrosomia. Our results are consistent with previous multiethnic studies, although the relationship between maternal age and macrosomia was stronger in our study. A large, population-based study reported a 40% increase in the odds of macrosomia in women 35 to 39 years old in comparison with younger women and a 20% increase in risk for women over 40 years.⁹ Metabolic changes are known to occur with age, and it has been hypothesized that there are specific metabolic factors that stimulate higher fetal growth velocity among pregnant older women, resulting in a higher risk of macrosomic birth, although these factors have yet to be delineated.⁹ When we assessed infant birth weight as a continuous variable, we did not find an association between maternal age and infant birth weight. It is possible that advanced maternal age is only associated with increased risk for macrosomia when infant birth weights are above a certain threshold (≥4000 g). Further study is needed to understand the physiologic basis for the elevated risk of macrosomic infants among older Latina women, and to determine whether there are other interactions between race/ethnicity, age and risk for macrosomia.

In this sample of Latina women, pre-pregnancy BMI was associated with an increased odds of macrosomic infant birth but this relationship was not statistically significant in multivariable analysis. In a prior study of a multiethnic population, women who weighed more than 300 lb²⁹ had a fourfold increase in risk of macrosomia, and women in a Danish study who weighed more than 80 kg (176 lb) had a twofold increase in risk of macrosomia¹⁴ in comparison with normal weight women. Similarly, in a small, multiethnic sample of women with gestational diabetes mellitus, pre-pregnancy weight was correlated with infant weight.¹⁶ We may have had insufficient power to detect a statistical difference in risk for macrosomic birth as a function of pre-pregnancy weight as we had a relatively small number of obese women (n = 60). In addition, as pre-pregnancy weight was self-reported and not confirmed by medical chart review, it is possible that these results are subject to recall bias. However, it is also possible that pre-pregnancy BMI may be less important as a determinant of macrosomic birth among nondiabetic Latina women, although this hypothesis must be tested in subsequent studies.

We also found an increased risk for macrosomic birth among Latinas with self-reported preexisting or GDM, but the risk was not statistically significant. Unfortunately, blood glucose levels were not measured in our study and because our interview was conducted at a mean gestational age of 31.2±5.6 weeks, it is possible that our interview was conducted before some women were diagnosed with gestational diabetes. Other studies have found that women with gestational diabetes who maintained good metabolic control of glucose levels throughout pregnancy were at no greater risk for delivering a large for gestational age infant than women without diabetes.³⁰ It is possible that had we tested for blood glucose levels, we would have found excess risk among Latinas with poorly controlled diabetes. On the basis of the timing of our interview, it is also possible that our interview was conducted before some women were diagnosed with gestational hypertension and preeclampsia, a risk factor for lower birth weight infants that may have influenced our results.

A unique aspect of our study was the inclusion of nutritional variables in relation to odds of macrosomic birth. Few previous studies have assessed the relationship between food consumption during pregnancy and risk of macrosomia among women recruited without reference to a specific health care need. One French study of women with gestational diabetes undergoing intensive prenatal care management reported that higher carbohydrate intake was associated with a decreased incidence of macrosomia,³¹ but our understanding of favorable diets and macronutrient balance in nondiabetic women is very limited. Previous studies in multiethnic samples have found that diabetic, overweight and obese Latina women are more likely to deliver macrosomic infants than Whites or African Americans,^17,18 and we hypothesize that ethnic-specific environmental or behavioral risk factors, such as diet, in addition to metabolic factors, could explain the elevated risk for macrosomic birth among at-risk Latinas.^17,18 In addition, as the rate of macrosomic birth was higher in our cohort than the national rate among Latinas, it is possible that behavioral and/or environmental factors could also increase the risk of macrosomic birth among healthy Latina women.

In unadjusted analyses, we found higher rates of macrosomic infants among women who consumed less fresh fruit (14.4 versus 9.6%), which contrasts with studies showing an association between high maternal serum vitamin C levels and high birth weight.³² Our group of Latina women reported a high frequency of fresh fruit consumption (72% reported eating fresh fruit every day or two times a day during pregnancy), which is consistent with other studies on the diets of immigrant Latina women.³³ Acculturation in Latinos has been shown to be associated with lower fruit and vegetable intakes.³³ In our study, the rate of macrosomic birth was lower in women who were taking iron supplements during pregnancy (3.6 versus 12.4%). As higher iron levels in pregnancy are associated with heavier infant birth weight,³⁴ it is possible that not taking iron pills in pregnancy is a marker of adequate iron serum levels. In our study of Latina women, we did not find an association for soda consumption and macrosomia even though soda is a known risk factor for obesity,³⁵ nor among women who consumed more ‘American food’ (defined as spaghetti, potatoes and fast foods) a proxy for foods with a higher glycemic index. In fact, although not statistically significant, we found that macrosomic birth was associated with lower levels of consumption of ‘American food’ and sodas during pregnancy, which may be associated with the reduced risk of macrosomia we found with acculturation (as indicated by self-identification and language use). Further study is needed to identify advantageous and disadvantageous nutritional practices among pregnant Latina women.

Additional study, with larger samples, is needed to assess and identify risk factors for macrosomia in Latina women. The most important limitation of our study was lack of statistical power. As the study was designed to test alternative hypotheses, we were unable to fully evaluate the relationships between all study variables and risk of macrosomia. We could not adequately test the effect of pre-pregnancy BMI, but with a high prevalence of overweight and obesity among Latinas, it is critical to understand the effect of maternal body mass on perinatal outcomes. Furthermore, if specific dietary components are identified as risks or protective factors for macrosomic births in Latinas, this information can be used to develop interventions to optimize maternal and infant health outcomes. In addition, if subsequent studies confirm the excess risk of macrosomia associated with older Latina women, then health care providers can alert Latinas of their increased risk for macrosomia associated with advanced maternal age. Given the high prevalence of pediatric overweight in Latinos, further studies are needed to ascertain risk factors for macrosomia, as macrosomia is an important precursor to pediatric overweight.