The risk of small-for-gestational-age⁶ (SGA) infants and preterm deliveries among pregnant adolescents has recently been shown to be associated with low gynecological age among primiparous adolescents, low body mass index (BMI) and prior poor outcomes among multiparous adolescents (Scholl et al. 1989 and 1992). Other risk factors may include changes in lifestyle-associated factors, such as adequacy of prenatal care, cigarette smoking, substance abuse, nutritional status and dietary intake (Kramer 1987, Scholl et al. 1989, Zuckerman et al. 1984). Dietary intake among low income adolescents has generally been found to be high in sugar, fat, and sodium, and low in other micronutrients (Portnoy and Christenson 1989, Rees 1992, Story and Alton 1987).

There is increasing evidence from animal studies that dietary sugar supplements elevate insulin and increase insulin resistance and blood pressure (BP) (Ahokas et al. 1986, Hulman et al. 1991, Hwang et al. 1987). Thus, as has been observed in animal studies (Ahokas et al. 1987), dietary sugar intake during pregnancy may affect pregnancy outcome. Because puberty is associated with hyperinsulinemia and insulin resistance (Amiel et al. 1991), as well as increasing BP (Szklo 1979), consuming a diet high in sugar may be especially detrimental during adolescent pregnancy.

In a previous study of 337 pregnant adolescents, drawn from the Camden County Adolescent Family Life demonstration project and using a single 24-h dietary recall, we showed that a high sugar diet was associated with an increased risk of SGA infants (Lenders et al. 1994). The objective of this study was to replicate the association between sugar intake and intrauterine growth restriction using a larger, independent sample of pregnant adolescents, particularly including more minority adolescents and a more precise assessment of dietary intake, i.e., more frequent dietary recalls. With the larger overall sample size and larger sample of minority adolescents, we were able to examine the effects of high sugar diets on gestational age at delivery and determine if there were ethnic differences related to outcome and sugar intake.

SUBJECTS AND METHODS

Study groups included primigravidas, ages 12-15 y at first pregnancy, and multigravidas, aged 19 y or younger, who were also aged 12-15 y at first pregnancy (Scholl et al. 1992). Exclusion criteria for the study were chronic disease or metabolic disease verified by medical record that could affect maternal growth, nutritional status or fetal outcome; intravenous drug use or cocaine addiction; heavy drinking (>50 g/d alcohol) or smoking (>2 packs/d). For these analyses, the sample was restricted to those pregnant adolescents who delivered live, singleton newborns; for the adolescents who delivered more than once during the study period, only their first pregnancy was included. Then, adolescents with a history of diabetes mellitus or gestational diabetes mellitus in the current pregnancy were excluded.

The obstetrical variables included age at menarche recalled in completed years, prenatal care, gestational diabetes, pregnancy-induced hypertension (PIH), duration of gestation, infant birth weight and infant sex. Early menarche was defined as age at menarche <12 y (Frisancho and Flegel 1982). Gynecological age was calculated by subtracting the age at menarche from age at the last menstrual period (LMP); low gynecological age was defined as two or fewer completed years since menarche (Zlatnick and Burmeister 1977). PIH and gestational diabetes were diagnosed by obstetricians at the prenatal clinic following standard recommendations (Creasy and Resnik 1984).

For length of gestation, the obstetrical estimate of gestational age was used, based on the LMP and confirmed by a routine obstetric ultrasound for the majority of the adolescents (75%) or by serial measurements of uterine fundal height when entry to care was early (25%). When using ultrasonography to confirm or establish dates at or before 12 wk, crown-rump length (CRL) was the standard; measurements of biparietal diameter (BPD) formed the basis for dating in the second trimester (Daya 1993, Kurtz et al. 1980). If the CRL was within 7 d of menstrual age for the LMP or the BPD within 10 d of menstrual age, then the estimated date of delivery was based on the LMP. Otherwise, the estimated date of delivery was based on the CRL or BPD measurements.

The outcome variables included both birth weight and gestational age at delivery. Several outcome categories were defined as follows: low birth weight (LBW) at <2,500 g, preterm delivery at <37 completed weeks, and SGA as <10th percentile of birth weight for gestational age (Brenner et al. 1976).

The anthropometric variables examined included pregravid weight, weight gain and height at entry to care. Pregravid weight was determined by patient recall at entry to prenatal care. An excellent correlation (r = 0.98) has been found between actual and recalled pregravid weights among adolescents (Stevens-Simon et al. 1986). Weight gain was computed by subtracting the pregravid weight from the last measured weight within 2 wk of delivery. Adequacy of gain was assessed using a standard chart for weight gain by gestational age (Butman 1983, Scholl et al. 1991). Using this schedule, total gestational weight gains <7.4 kg for a delivery at 32 wk, 9.0 kg at 36 wk, 9.7 kg at 38 wk, and 9.9 kg at 40 wk would be classified as inadequate. Height was measured at entry to prenatal care using standard methods, and body mass index (BMI) was calculated as pregravid weight-for-height² (kg/m²) (Institute of Medicine 1990).

Registered dietitians interviewed the adolescents up to three times during prenatal care (at entry, 28 and 36 wk gestation) and obtained 24-h dietary recalls. The 24-h dietary recalls were analyzed for total energy intake, macronutrients (total sugar, carbohydrate, protein and fat) and micronutrients (sodium, potassium, magnesium, calcium, phosphorus, manganese, zinc and iron).

A high sugar diet was defined as a daily intake of total sugar at or above the 90th percentile (206 g) for the sample. For purposes of analysis, the reference sugar group included all adolescents with intake of total sugar <206 g. Nutrient composition and the energy value of foods were calculated using the Campbell Master Nutrient Data Base. Nutrient data are based on the U.S. Department of Agriculture (USDA) Handbook 8 data Bases, literature information and brand product information from manufacturers.

The term "total sugar" refers to the simple carbohydrates, including monosaccharides, disaccharides and oligosaccharides. Total sugar does not include complex carbohydrate, such as starch or fiber. This definition is consistent with that provided by the Sugars Task Force of the Food and Drug Administration (FDA) (Glinsmann and Park 1995, Glinsmann et al. 1986).

RESULTS

Adolescents consuming high sugar diets were less likely than the other adolescents (P < 0.05) to be Medicaid recipients and have inadequate prenatal care (Table 1). The rates of SGA and LBW infants were nearly twice as high in the high sugar diet group, but both associations were of only borderline significance (P = 0.07).

Table 1. Adolescent characteristics by sugar intake group, comparing high sugar consumers with reference consumers [View Table]

After adjusting for energy intake, adolescents with a BMI  26.0 kg/m² were three times more likely than those with a BMI of 20.0-26.0 kg/m² to consume high sugar diets (AOR = 3.19, 95% CI 1.29-8.17). Those with BMI  20.0 kg/m² did not differ significantly from the adolescents with a BMI of 20.0-26.0 kg/m² (AOR = 1.83, 95% CI 0.90-3.72) in terms of sugar intake. After adjusting for energy intake, adolescents of low compared with higher gynecological age had a tendency to consume diets higher in sugar (AOR = 2.74, 95% CI 1.40-7.25), yet those adolescents with low gynecological age were of lower BMI (AOR = 2.18, 95% CI 1.13-4.19). There were no ethnic differences in terms of high sugar intake or high BMI, even after adjusting for energy intake (P  0.05).

Overall, the average energy intake was 9982 kJ/d, below the 10,465 kJ/d suggested for pregnant women by the Recommended Dietary Allowance (RDA). Unadjusted for energy intake, the mean sugar intake of the adolescents in the high sugar group was 267 ± 73 g and that of the reference group was 111 ± 46 g. The high sugar group consumed 44% of their total dietary energy intake as total sugar compared with 19% in the reference group. The high sugar group consumed significantly (P  0.05) more energy, total sugar, as well as all other macronutrients and micronutrients.

However, after adjusting for energy intake (Table 2), the high sugar compared with the reference group still showed higher dietary intake of total sugar and total carbohydrate, but relatively lower intake of protein and total fat. Again adjusting for energy intake, the high sugar group had lower consumption of all micronutrients except manganese, magnesium and iron (P  0.05).

Table 2. Dietary intake of pregnant adolescents comparing reference sugar consumers with the high sugar intake group1 [View Table]

The variables used in the models for birth outcomes included high sugar diet, ethnicity, a term combining high sugar diet and ethnicity, age, number of cigarettes smoked per day, inadequate weight gain, BMI (as a continuous variable), total energy intake, low gynecological age, parity and PIH.

The adjusted odds ratio for delivering an SGA infant was 2.01 (95% CI 1.05-7.53) for adolescents consuming high vs. reference-sugar diets, regardless of ethnicity (Table 3).

Table 3. Logistic regression analysis: small-for-gestational age infants born to mothers in high and reference sugar intake groups1 [View Table]

To determine whether there was an effect of high sugar intake on length of gestation, multiple linear regression analyses were used. These analyses (Table 4) showed that the association of high sugar diets with gestation duration was particularly strong for adolescents of Puerto Rican descent (1.69 ± 0.62 wk, P < 0.007) compared with all reference sugar consumers and white adolescents consuming high sugar diets. Black adolescents consuming high sugar diets did not exhibit a shortening of gestation. These findings of shortened gestation for Puerto Rican adolescents were confirmed using a two-way analysis of covariance, testing for differences in average gestation duration by ethnicity and sugar intake, adjusting again for age, number of cigarettes smoked per day, inadequate weight gain, BMI (as a continuous variable), total energy intake, low gynecological age, parity and PIH. For Puerto Rican adolescents with high sugar intake (n = 20), average gestation was 37.2 ± 0.6 wk (least square mean ± SEM), significantly shorter than that for Puerto Rican adolescents consuming reference sugar diets (38.6 ± 0.2 wk, P < 0.03). On the other hand, for both the black adolescents (39.3 ± 0.4 wk, high sugar vs. 38.9 ± 0.1 wk, reference sugar) and white adolescents (39.3 ± 1.5 wk, high sugar vs. 39.6 ± 0.4 wk, reference sugar), there were no differences in length of gestation by sugar intake group. Thus, in the model overall, there was a significant interaction (P = 0.05) between the main effects for ethnicity and sugar intake.

Table 4. Multiple regression analysis: factors associated with gestation duration in pregnant adolescents with varying levels of sugar intake [View Table]

The addition of variables such as Medicaid (an indicator of low socioeconomic status), infant gender, or a dummy for very low sugar diets (53 g, n = 60) to the model did not affect the outcome variables, magnitude or significance of the models.

DISCUSSION

The association of high sugar diets and birth outcome may be due in part to the overall poor quality of the adolescents' diets. No recommendation has been provided for sugar intake by the RDA (NRC 1989), but an intake below 10% of total energy intake is usually considered acceptable. In this study, the high sugar group consumed 44% of their total dietary energy intake as total sugar compared with 19% in the reference group, well above the suggested 10%. However, the overall percentage of dietary energy intake as sugar (22%) was remarkably similar to that of the 24% found for 15- to 18-y-old females from the 1977-1978 USDA Nationwide Food Consumption Survey (NFCS) and of the 25% for 15- to 18-y-old females from the more recent 1987-1988 NFCS (Gibney et al. 1995). In addition, our 90th percentile for the sample (206 g) is precisely double the average of 103 g/d sugar intake found in the NFCS (Gibney et al. 1995). Consistent with what we have found in Camden, carbonated beverages, fruit juices, ice cream, syrup added to pancakes and sweetened cereals are the most commonly listed products contributing to high sugar diets among adolescents (Lenders et al. 1994, Rees 1992).

In this study, the amounts of energy intake, macronutrients and micronutrients were all found to be higher absolutely in the high sugar group, but a relative deficiency or excess of some of the nutrients important for sugar and carbohydrate metabolism was observed. In keeping with what been noted repeatedly in studies of children and adults (Gibney et al. 1995, Hill and Prentice 1995, Prentice and Poppitt 1996), adolescents in our sample with high sugar consumption took in relatively less total fat, protein, calcium and zinc. Thus, the effects of pregnancy, puberty, ethnicity and high sugar or nutritionally unbalanced diets during adolescence on the risk for poor birth outcome might be compounded when several of these factors are present.

However, there are several mechanisms affecting glycemia, insulinemia, insulin resistance and cellular integrity that could account directly for an association between birth weight or gestational age and sugar intake. Although the association may not be confined to adolescents, this age group, because of their unique physiology, is probably particularly prone to the effect, and the effect among mature women is likely to be associated with frank pathology. That is, a relative hyperinsulinemia appears to be characteristic of adolescence, and hyperinsulinemia has been associated with changes of blood flow in peripheral flow in adolescents (Amiel et al. 1991). Hyperinsulinemia in general has also been associated with changes in blood flow to visceral tissues (Axelrod 1991) and increased systemic BP in both animal and human studies (Reaven 1991). Further, changes in sympathetic nervous activity (Landsberg and Young 1985), endothelin-1 production (Oliver et al. 1991, Svane et al. 1993), prostaglandin activity (Axelrod 1991), urinary sodium secretion (DeFronzo 1981) and changes in cellular ions (Resnick 1992) might then contribute to reduced placental blood flow and subsequent adverse pregnancy outcome. The period of adolescence is of special interest because of the hormonal changes occurring during this active period of growth and their association with hyperinsulinemia and insulin resistance (Amiel et al. 1991), particularly in girls with increased central body fat (Freedman et al. 1987) and increasing height velocity (Hindmarsh et al. 1988).

Finally, this study has several limitations. The underlying mechanisms for an association between high sugar diets and poor pregnancy outcomes have yet to be elucidated. The study sample was homogeneous for low economic status, limiting the generalizability of the findings to other economic groups. The assessment of dietary intake was based on several 24-h recall interviews rather than direct measurements of nutrients.

Nevertheless, the findings suggest that pregnant adolescents who consume high sugar diets are at increased risk for SGA infants and those of Puerto Rican ethnicity for shortened gestation. Further work on the association between sugar intake and birth outcome is needed, and, until more is known, current recommendations about sugar intake should be continued and adolescents who consume excessive amounts counseled to moderate their intake of sugar during pregnancy consistent with current recommendations.

ACKNOWLEDGMENTS