The mean BMI for the 880 children was 18.8 with a standard deviation (SD) of 4.5. Males had lower mean BMI (mean ± SD 18.6 ± 4.1; range of 11.9 to 36.0) than females (19.1 ± 4.9; range of 12.3 to 44.9) (P = 0.06). Children who were small for gestational age had a lower BMI (17.7 ± 3.5) than those born appropriate for gestational age (19.2 ± 4.7, P < 0.001) and those who were born pre-term had a lower BMI (18.5 ± 4.4) than those born full term (19.1 ± 4.5, P = 0.08). Cocaine exposure groups did not differ in BMI; with no exposure BMI was 18.7 ± 4.4, some exposure 18.8 ± 4.3 and high exposure 19.6 ± 5.4, respectively.
The frequencies of children in the three cocaine exposure levels is shown in . Overall 69% of children were born to women who used no cocaine in pregnancy, 21% with some use and 10% with high use. There were significant differences in maternal education, weight at delivery and alcohol and poly drug use between the three exposure groups. The frequencies of BMI categorized as normal, at risk for overweight and overweight are shown in . Overall, 16% of the children were at risk for overweight status at 9 years of age and 21% were overweight. There were significant differences among the BMI categories by birth weight, small for gestational age status, postnatal weight gain, race, and, mother’s weight at delivery.
Frequencies (and Percentages) of Children in Cocaine Exposure Levels by Demographic and Prenatal Factors
Frequencies (and Percentages) of Children in BMI Categories by Demographic and Prenatal Factors
The frequency and percentage of children in the sample classified as normal, pre-hypertensive, and hypertensive are presented in . Fifteen percent of the sample was classified as pre-hypertensive and 19 % were classified as having hypertension. The blood pressure categories varied significantly by birth weight.
Frequencies (and Percentages) of Children with Hypertension by Demographic and Prenatal Factors
The relationship between prenatal cocaine exposure and 9-year BMI, after adjusting for each of the control variables outlined above was explored. The interaction between term (vs. pre-term) birth and cocaine exposure was significant; F (2,677) = 5.36 (P = .005), suggesting that cocaine exposure has a differential impact on BMI for term and pre-term children. Therefore, separate models for these two groups were examined.
As shown in , high cocaine exposure was significantly associated with higher BMI among children born at term (P = 0.019). Higher BMI among children born at term was also associated with female gender, higher birth weight, higher postnatal weight gain for first four months of age in the child, higher maternal weight at delivery and watching 2 or more hours of TV on school days and consuming more calories at 9 years of age.
OLS Regression Model Predicting BMI at 9 Years: Children Born at Term (N= 522)
The corresponding regression model for pre-term children is shown in . Cocaine exposure was not significantly associated with BMI for either high (P = 0.69) or some (P = 0.90) exposure. Several of the other predictors of high BMI were similar to those for term children: female gender, higher birth weight, higher postnatal weight gain for child, and higher maternal weight at delivery. In addition, higher maternal education and not exercising regularly at 9 years of age were associated with higher BMI among children born at pre-term.
OLS Regression Model Predicting BMI at 9 Years: Children Born at Pre-Term (N= 358)
Regression analyses were then conducted to test whether cocaine exposure has a direct impact on blood pressure. For systolic and diastolic blood pressure, the term by cocaine interactions were not significant: Systolic: F (2,755) = 1.48 (P = .23); Diastolic: F(2,755) = 2.13 (P = .12). Hence, models that included both term and pre-term children were explored. As shown in and , cocaine exposure was not significant for either blood pressure measure. Higher systolic blood pressure was associated with greater postnatal weight gain and higher maternal weight at delivery while lower systolic blood pressure was associated with prenatal alcohol exposure. Similarly, lower diastolic blood pressure was associated with prenatal alcohol exposure.
OLS Regression Model Predicting Systolic Blood Pressure at 9 Years: All Children (N= 880)
OLS Regression Model Predicting Diastolic Blood Pressure at 9 Years: All Children (N= 880)
Based on the regression results, it appears that cocaine exposure does not have a direct impact on blood pressure after controlling for other possible predictors. Therefore, analyses exploring whether cocaine has an indirect effect on blood pressure through its impact on BMI was examined (). The other variables (i.e., demographic, pre/neonatal, and 9-year variables) are included as control variables and have direct paths to BMI. To control for possible variation in measurement error by site, clinical site has a direct path to both BMI and blood pressure. Path analyses were conducted to test this conceptual model for systolic and diastolic blood pressure. Because the regression results indicated no significant relationship between cocaine exposure and BMI for the pre-term children, the path analyses included only the term children. The path analysis model for systolic BP is shown in . This model fit very well (CFI = .97, TLI = .92, RMSEA = .03), supporting the hypothesis of an indirect cocaine effect on blood pressure. High cocaine exposure had a significantly positive relationship with BMI (path coefficient = 1.50, P = 0.02) while some cocaine exposure was not significant (path coefficient = 0.65, P = 0.23). There is a strong positive relationship between BMI and systolic blood pressure (path coefficient = 0.78; P < 0.001).
Relationship Between Cocaine Exposure, Body Mass Index, and Blood Pressure
Relationship of Cocaine Exposure and Systolic Blood Pressure at 9 years
Similar results were noted for diastolic blood pressure (). The model fit well (CFI = .99, TLI = .97, RMSEA = .02). High cocaine exposure was positively associated with BMI (path coefficient = 1.50, P = 0.02) and the relationship for some cocaine exposure was not significant (path coefficient = .65; p = 0.23). BMI and diastolic blood pressure were positively associated (path coefficient = 0.14; p = 0.02).
Path diagram for model of diastolic blood pressure at 9 years among children both at term (n=522). The analysis includes cocaine exposure and control variables.