Multilevel analysis of neighborhood factors and maternal behaviors in the prevalence of low birth weight and preterm births in a South Carolina PRAMS population

Background: Investigations of neighborhood contexts and maternal smoking behaviors on low birth weight or preterm births have yielded inconsistent results. We examined the independent effect of neighborhood contexts on maternal smoking behaviors and risks for low birth weight and preterm births among women participating in the South Carolina Pregnancy Risk Assessment and Monitoring System (PRAMS), 2000-2003.

Methods: We linked PRAMS data to Census 2000 for 8,064 mothers delivering live births. Smoking and other maternal-level covariates were obtained from PRAMS data, which were geocoded to the census tract level. Census data provided neighborhood-level factors. We used multilevel logistic regression analysis with SAS PROC GLIMMIX to determine odds ratios and corresponding 95% confidence intervals.

Results: In the fully adjusted models, only mothers resident in medium poverty neighborhoods were at increased odds of smoking (OR: 2.24, 95% CI=1.05-4.80. High poverty, predominantly African-American neighborhoods, upper quartiles of low education and second quartile of household crowding and maternal smoking were significantly associated with low birth weight. Mothers residing in predominantly African-American census tract areas were at a significantly increased risk of delivering preterm (OR: 2.2, 95% CI 1.29-3.78).

Conclusions: Maternal smoking has more consistent effects on low birth weight than preterm births and neighborhood contexts appear to be indirectly rather than directly related since medium poverty neighborhoods were associated with higher levels of smoking among pregnant women. Programs seeking to improve birth outcomes by reducing smoking during pregnancy need to take into account the ethnic and income make-up of neighborhoods where women reside.

Full Summary Background: Investigations of neighborhood contexts and maternal smoking behaviors on low birth weight or preterm births have yielded inconsistent results. Previous studies often used limited maternal-level covariates as control variables. We evaluated the independent effect of neighborhood contexts on maternal smoking behaviors and risks for low birth weight and preterm births among women participating in the South Carolina Pregnancy Risk Assessment and Monitoring System (PRAMS) survey, 2000-2003.

There is growing evidence to suggest that disparities in health are attributable to differences in neighborhood socioeconomic contexts in addition to individual-level determinants. Low birth weight (defined as birth weight less than 2,500 grams) and preterm births (defined as gestation of less than 37 weeks) are health indicators known to be sensitive to neighborhood socioeconomic contexts over and above individual-level socioeconomic status. Disentangling the relative importance of individual and contextual causal processes linked to disparities in birth outcomes is necessary in order to determine the targets of interventions.

Methods: We linked PRAMS data to Census 2000 for 8,064 mothers delivering live births. Smoking and other maternal-level covariates were obtained from PRAMS data, which were geocoded to the census tract level. Census data provided neighborhood-level factors. Selected variables from the 2000 Census and the geocoded PRAMS data sets were used to create a multilevel data set that included both individual maternal and neighborhood level variables as two-level data files. We restricted the analysis to singleton live births without congenital anomalies; respondents with records that lacked the code necessary to link with neighborhood level variables were also excluded. Analyses were further restricted to birth weights >500 grams, to gestation >20 weeks and to only African-American and White ethnic groups due to a small number of observations recorded for other racial groups. Dependent variables of birth weight and gestational age were dichotomized as <2500 g vs >2500 for birth weight and <37 vs >37 weeks for gestational age. Neighborhood level independent variables for the study were percent African-American population, proportion of households below 1.5 of the federal poverty line, proportion of residents with primary-level education, college education and household crowding (proportion of residents in a household with more than 1 person per room). The individual level variables were age, race, income, marital status, education and maternal smoking behaviors, based on self reported information from the PRAMS data. Information on cigarette smoking history was dichotomized as any smoking and non-smoking. The non-smoking category was used as the reference group for analysis. The data for analysis were weighted to adjust for differences in response rates and variations in the probability of PRAMS sample selection. SAS PROC GLIMMIX procedure was used to fit logistic regression models for binary responses (low birth weight, preterm delivery, and smoking) in multilevel models, assuming a binomial distribution and a logit link function. We used multilevel logistic regression analysis to determine odds ratios and corresponding 95% confidence intervals.

Results: The results showed that residence in a medium poverty area compared to a low poverty census tract was associated with more than a 3-fold increase in odds of smoking, while residence in predominantly African-American neighborhoods reduced the risk of smoking by 36%. However, in the adjusted estimates only percent poverty tract (census tracts with 10%-19.9% poverty) was significantly associated with smoking. In the fully adjusted models, only mothers resident in medium poverty neighborhoods were at increased odds of smoking (OR: 2.24, 95% CI=1.05-4.80). High poverty, predominantly African-American neighborhoods, upper quartiles of low education and second quartile of household crowding and maternal smoking were significantly associated with low birth weight. Mothers residing in predominantly African-American census tract areas were at a significantly increased risk of delivering preterm (OR: 2.2, 95% CI 1.29-3.78).

Discussion: This study found maternal smoking to be strongly related to increased risk of low birth weight or preterm birth. This association remained significant when maternal-level and neighborhood-level factors were included in the models, indicating that smoking had significant risks for birth outcomes independent of other maternal and neighborhood contexts. At the individual-level, African-American racial groups were at a greater risk for low birth weight and preterm births even when maternal smoking was controlled. Further, the association for non-smokers (compared with quitters) category for low birth weight and smoker (verses quitter) category for preterm disappeared after controlling for maternal-level and contextual factors, suggesting that these smoking categories are essentially determined by maternal socioeconomic and demographic status attainment.

Results of the multivariate analysis also showed that neighborhood contexts may be important in explaining observed disparities in low birth weight or preterm birth. Percent African-American population in a tract, high area poverty (census tract with >20 % poverty), neighborhood low education (census tracts with 28.84%-43.27%, and 43.27%-57.69% low education), and household crowding (census tracts with 3.8%-7.71% crowding) were significantly related to low birth weight, while percent African-American population in a neighborhood was most significantly related to preterm birth. These associations remained statistically significant when maternal smoking categories were included in neighborhood models, suggesting that neighborhood contexts may affect birth outcomes independent of maternal smoking behaviors. Higher percentage of the African-American population in a tract appears to result in greater risks for low birth weight and preterm birth outcomes. After adjustment for maternal-level covariates the association between neighborhood contexts and birth outcomes were still suggestive but statistically non-significant. Similar findings were made for the relationship among maternal smoking and contextual factors. Overall, the analysis showed that neighborhood contexts have greater effects on low birth weight than preterm births.

It is of public policy interest to note that although poverty across neighborhoods potentially affects ethnic groups equally, through racial residential segregation, the neighborhood environment of the average white mother is likely to be more affluent than that of the average black mother. Due to the nature of the environment, contexts may provide significant cues for increased cravings to use tobacco or for discouraging quitting during pregnancy. Further, because poor neighborhoods are associated with stress, this may elicit maternal neuro-biological responses to trigger cigarette smoking or encourage relapse among quitters. For these reasons, disadvantaged neighborhoods may be unhealthy due to increased exposure to incessant stress and other behavioral risks that may mediate or moderate the causal pathways to low birth weight or preterm births.

Limitations of this study include the use of birth certificate data and self reports to assess behavioral patterns of maternal smoking. If women were disinclined to accurately disclose details of levels of prenatal cigarette use then the true magnitude of effects of neighborhood factors on maternal cigarette smoking may be underestimated. The cross-sectional nature of the PRAMS survey and the Census data makes it difficult to infer causality.

The results of the study are generally consistent with previous investigations that have found an independent relationship between neighborhood contexts and birth outcomes. However, the results of this study differ from previous studies in a variety of ways. First, previous studies examining the relationship between neighborhood contexts and birth outcomes or smoking have often focused on neighborhood disadvantages without considering how different levels of contextual disadvantages are related to low birth weight and preterm or smoking behaviors. Results of this study suggest that it is not just living in a disadvantaged neighborhood that is associated with greater risk for low birth weight and preterm births or smoking. Risks for smoking and poor birth outcomes rather depend on whether the mother is resident in the most advantaged or disadvantaged census tract quartile. Second, previous studies have relied on the birth certificate data as a source for deriving maternal-level SES and smoking risks. Birth certificate data have limited information on maternal socioeconomic status and smoking. Therefore, it is most likely that previous studies may have overestimated the true neighborhood effects by minimizing maternal income and marital status as important determinants of neighborhood resources. Inability to include these factors in a model may have resulted in a spurious independent relationship between contexts and birth outcomes or smoking. Third, the results of this study provide a reasonable basis to suggest that contextual effects have greater impact on low birth weight than on preterm. Reasons for these effects are not clearly specified in the literature; but are most likely to be related to stress effects on fetal growth associated with limited resources in disadvantaged neighborhoods.

Future studies should consider using prospective longitudinal study designs, examine the women's perceptions of neighborhood quality as well as the contextual characteristics, and measure the length of stay in a particular neighborhood in a comprehensive theoretical framework. By this process, effects of neighborhood contexts on maternal smoking behaviors and birth outcomes could be more accurately estimated.

Conclusions: Maternal smoking has more consistent effects on low birth weight than preterm births and neighborhood contexts appear to be indirectly rather than directly related since medium poverty neighborhoods were associated with higher levels of smoking among pregnant women. Programs seeking to improve birth outcomes by reducing smoking during pregnancy need to take into account the ethnic and income make-up of neighborhoods where women reside.