Table 1 presents descriptive statistics of the valid sample for the CDC’s 2017 natality data. The presentation follows the classification of selected variables in three groups, respectively related to the mother, the pregnancy, and the infant. All 3,864,754 children were included in the analyses.
Outcome variable: Overall, 319,427 children in the dataset present a low birth weight, making a relative frequency of 8.3%.
Maternal characteristics: data in table 1 indicate that almost 80% of women in the sample were between 19 and 35 years of age; 18% were older than 35 years and only 3% were 18 or below. The majority of mothers had a college degree or less, they were primarily white (74%) or Black Americans (15%). In terms of maternal risk factors, 38% of women were primipara and 55% were overweight, with a body-mass index of 25 or more. Nearly 40% of women in the sample were enrolled in a Women, Infants, and Children (WIC) nutrition supplement program.
Pregnancy history and prenatal care: in terms of pregnancy history and prenatal care, about 2% of women had no prenatal visit; 38% of them had between one and ten visits, and 60% more than 10 visits. Less than 1% had any congenital anomaly; 6% had pre-pregnancy diabetes and 6% had pre-pregnancy hypertension. More than 90% of the sample women gained 10 Kg or more during pregnancy
Infant's characteristics: Regarding the infant's characteristics, 51% are male, 97% were singleton births and 96% had a five-minute APGAR score of 7 or above. 88% of all births had a gestational age of fewer than 37 weeks and about 8% had a birth weight below 2500 grams. It is anticipated that the potential confounding factors may be gestational age and multiple births.
Table 2 presents bivariate analyses of birth weight (outcome variable) by the number of prenatal visits (independent variables), confounders (singleton births, and gestational age) as well as other mothers, pregnancy, and infant characteristics. The association between low birth weight and each of the characteristics is assessed using a chi-squared test with a 5% level of significance. The results indicated a statistically significant association between the number of prenatal visits and low birth weight with a p-value of 0.01. Moreover, it appears that the percent of children with low birth weight decreased from 19.5% for women with no visit to 12.1% as the number of prenatal visits increased and 5.3% respectively for women with 10 visits or less and 11 visits or more (see Figure 1).
Table 2 also indicates that all potential confounders and other selected covariates were significantly associated with birth weight, with p-values below 1%.
Regression analyses
Logistic regression analyses were conducted by first, including all explanatory variables one by one to estimate unadjusted odds ratios and, second including the independent, confounding, and all other covariates to estimate adjusted odds ratios. The results were presented in table 3 as both unadjusted and adjusted odds ratios with corresponding 95% confidence intervals. The data are presented using the same groupings as in tables 1 and 2. In the following lines, we highlight some of the main results of this multiple regression analysis.
Unadjusted odds ratios
Based on the unadjusted odds ratios, there is a statistically significant inverse relationship between the number of prenatal visits and low birth weight. Compared to women with more than 10 visits, women with 10 visits or less were almost 2.5 times more likely to give birth to an underweight child (OR=2.46; 95% CI: 2.44-2.48). The observed unadjusted odds ratios were statistically significant at the 5% level. As for women with no prenatal visit, they are 4.3 times more likely to give birth to a child with low birth weight compared with women who have more than 10 antenatal visits.
Regarding plurality, multiple births were 18.5 times more likely to be underweight compared with singletons (OR=18.48; 95%CI=18.27-18.69). Similarly, babies born before 37 weeks of gestational age were 25.6 times more likely to be underweight compared with those born at 37 weeks or later (OR=25.64; 95% CI=25.43-25.85).
Using age 18 or below as a reference, we found that the unadjusted odds ratio of low birth weight among those women aged 19 – 35 was 1.10. This means that there was a 10% increase in the probability of low birth weight for women aged 19-35 as compared with those aged 18 or less. In the same vein, the odds ratio for women aged more than 35 was 0.85, which means that there was a 15% decrease in the probability of low birth weight for that group compared with those aged 18 years or less. Regarding maternal education, women with a graduate degree were considered as a reference. While the unadjusted odds ratios of low birth weight among women with a college degree or less and those undergraduates compared with graduates were all statistically significant. Compared with White women, Black Americans were twice more likely to deliver low birth weight babies (OR=2.01; 95% CI=1.97-2.04) while American Indians and Alaskan natives have a 13% increased risk of delivering low birth babies (0R=1.13; 95% CI=1.06-1.21). The data also shows a 15% increase in the odds of low birth weight for Asian or Pacific Islander women, as compared with White women (OR=1.15; 95% CI=1.12-1.17).
On maternal risk factors, high BMI (overweight) and 10 or more kg weight gain present odds ratios that were below 1 (0.90 and 0.54 respectively) and statistically significant compared with their respective reference categories. This means that women with high BMI were 10% less likely to give birth to an underweight child compared with those of normal weight while those who gain 10 kg or more were 46% less likely to give birth to an underweight child compared with those who gain less than 10 kg.
The unadjusted data also show that participating in a WIC nutrition program resulted in a 20% increased risk of low birth weight. In terms of pregnancy risk factors, women with congenital anomalies were four times more likely to give birth to an underweight child compared with those with no congenital anomalies (OR=4.05; 95% CI=3.91-4.22). Women with pre-pregnancy diabetes present a 13% increase in their risk to give birth to an underweight child and those compared with women without pre-pregnancy diabetes. Also, women with pre-pregnancy hypertension are almost three times more likely to give birth to an underweight child compared with women with ho pre-pregnancy hypertension (OR=2.99; 95% CI=2.96-3.02).
Regarding the source of payment for the pregnancy and delivery care, we selected Medicaid as the reference group. The data, using both unadjusted results, show that women with private insurance are 28% less likely to give birth to an underweight child compared with those under Medicaid. The corresponding figure for women who paid by themselves is 31%.
In terms of the infant's characteristics, the data reveal that male infants are 16% less likely to be born underweight compared with females. In addition, children with an APGAR score above 7 are 6.7 times more likely to be underweight compared with those with an APGAR score of 7 or less.
Adjusted odds ratios
The data in the third column of Table 3 present adjusted odds ratios for each of the variables previously discussed. They confirm a statistically significant relationship between the number of prenatal visits and low birth weight. As with the unadjusted odds ratios, adjusted odds ratios are also statistically significant. The inverse relation found with unadjusted odds ratios remained. The corresponding adjusted odds ratios were smaller than the unadjusted values (1.74 and 2.40 respectively) but still statistically significant at the 5% level. This confirms our hypothesis that the number of prenatal visits impacts birth weight. However, the magnitude of changes in odds ratios from unadjusted to adjusted values indicated that other factors explain part of the relationship between the two variables.
Based on the data in table 3, it also appeared that multiple births have an adjusted odds ratio of 11.12 compared with singleton births. This is indicative that multiple births are 11 more likely to result in low birth weight. This is a sharp drop from the unadjusted value, indicating that other factors were at play in explaining the relationship between plurality and low birth weight. A similar conclusion could be drawn for premature children (resulting from pregnancies of less than 37 weeks gestational age), where the odds ratio dropped from 25.6 to less than 12.
On maternal characteristics, there is a significant and positive association between a mother's age and the probability of low birth weight. Using age 18 or below as a reference, we found that the adjusted odds ratio of low birth weight among those women aged 19 – 35 is 1.29. This means that there is a 29% increase in the probability of low birth weight for women aged 19-35 as compared with those aged 18 or less. In the same vein, the odds ratio for women aged more than 35 is 1.66, which means that there is a 66% increase in the probability of low birth weight for that group compared with those aged 18 years or less.
Regarding maternal education, women with a graduate degree were considered a reference. While the unadjusted odds ratios of low birth weight among women with a college degree or less and those undergraduates compared with graduates were all statistically significant, this was no longer the case for the adjusted odds ratios. This means that when other factors are accounted for, maternal education is no longer a predictor of low birth weight.
Compared with White women, Black Americans were about 57% more likely to deliver low birth weight babies while American Indians and Alaskan natives were 24% less likely. The data also showed that the odds ratio of low birth weight for Asian or Pacific Islander women was not statistically significant when compared with White women.
On maternal risk factors, high BMI (overweight) and 10 or more kg weight gain present odds ratios that were below 1 (0.66 and 0.57 respectively) and statistically significant compared with their respective reference categories. This means that women with high BMI were 34% less likely to give birth to an underweight child compared with those of normal weight while those who gain 10 kg or more were 43% less likely to give birth to an underweight child compared with those who gained less than 10 kg.
The data also show that participating in a WIC nutrition program results in a lower risk of low birth weight. In terms of pregnancy risk factors, women with congenital anomalies were three times more likely to give birth to an underweight child compared with those with no congenital anomalies. Women with pre-pregnancy diabetes were 22% less likely to give birth to an underweight child and those compared with women without pre-pregnancy diabetes. Also, women with pre-pregnancy hypertension were more than twice more likely to give birth to an underweight child compared with women with no pre-pregnancy hypertension.
Regarding the source of payment for the pregnancy and delivery care, we selected Medicaid as the reference group. The data, using both unadjusted and adjusted results, show that women with private insurance are 15% less likely to give birth to an underweight child compared with those under Medicaid. The corresponding figure for women who paid by themselves was 34%.
In terms of the infant's characteristics, the data reveal that male infants were 28% less likely to be born underweight compared with females. In addition, children with an APGAR score above 7 were 2.23 times more likely to be underweight compared with those with an APGAR score of 7 or less.