Table 4.
Association of neighborhood racial composition and very preterm birth among Black and White
| mothers, Atlanta, GA | ||||||||||
| Main effect of segregation | Main effect of race | Segregation × Race Interaction | Model Fit (Deviance) | |||||||
| OR | 95% CI | OR | 95% CI | OR | 95% CI | |||||
| BASELINE | 2.84 | 2.70 | 2.99 | 62845 | ||||||
| Continuous | ||||||||||
| Tract % black | 1.75 | 1.35 | 2.24 | 2.45 | 2.28 | 2.64 | 0.69 | 0.53 | 0.90 | 62813 |
| 500-m % black | 1.59 | 1.30 | 1.95 | 2.43 | 2.26 | 2.62 | 0.75 | 0.60 | 0.94 | 62813 |
| 4000-m % black | 1.68 | 1.34 | 2.09 | 2.47 | 2.30 | 2.65 | 0.71 | 0.56 | 0.91 | 62814 |
| Tract-500-m† | 1.68 | 1.34 | 2.08 | 2.45 | 2.28 | 2.63 | 0.71 | 0.56 | 0.91 | 62811 |
| Trichotomous | ||||||||||
| Tract % black | ||||||||||
| < 0.3 | 1.00 | -- | -- | 2.75 | 2.55 | 2.96 | 1.00 | -- | -- | 62818 |
| 0.3-0.7 | 1.27 | 1.11 | 1.44 | 0.80 | 0.69 | 0.94 | ||||
| > 0.7 | 1.22 | 0.85 | 1.67 | 0.93 | 0.67 | 1.34 | ||||
| 500-m % black | ||||||||||
| < 0.3 | 1.00 | -- | -- | 2.75 | 2.53 | 2.99 | 1.00 | -- | -- | 62811 |
| 0.3-0.7 | 1.25 | 1.10 | 1.42 | 0.81 | 0.69 | 0.95 | ||||
| > 0.7 | 1.50 | 1.20 | 1.85 | 0.74 | 0.59 | 0.94 | ||||
| 4000-m % black | ||||||||||
| < 0.3 | 1.00 | -- | -- | 2.69 | 2.48 | 2.92 | 1.00 | -- | -- | 62814 |
| 0.3-0.7 | 1.23 | 1.10 | 1.38 | 0.87 | 0.76 | 1.01 | ||||
| > 0.7 | 1.43 | 1.10 | 1.82 | 0.81 | 0.62 | 1.06 | ||||
| Tract-500-m† | ||||||||||
| < 0.3 | 1.00 | -- | -- | 2.80 | 2.59 | 3.03 | 1.00 | -- | -- | 62811 |
| 0.3-0.7 | 1.31 | 1.16 | 1.47 | 0.76 | 0.66 | 0.88 | ||||
| > 0.7 | 1.27 | 0.96 | 1.64 | 0.88 | 0.67 | 1.17 | ||||
NOTE: Race is coded Black = 1, White = 0, and all continuous segregation indices are mean-centered so that the main effect of segregation is the effect for white women only, and the main effect of race is the excess relative risk for black as compared with white women at average segregation levels (e.g. mean-centered segregation equals zero for an neighborhood with average racial composition). The interaction term is then the excess relative risk for black women in the most compared with the least segregated neighborhoods.
†Tract-500-m are obtained by taking a population-weighted average of the racial composition in the 500-meter vicinity of each grid point within a census tract, thus approximating the average spatial isolation experienced by residents of each tract.
Data source: Georgia Department of Community Health, Office of Health Indicators for Planning; all singleton live births to non-Hispanic black and non-Hispanic white mothers in Atlanta MSA, 2000-2002