Multi-level analysis of factors related to low birth weight in a Midwestern County

Low birth weight (LBW) is associated with higher mortality rates, higher use of health care resources, and higher rates of chronic conditions. Multi-level determinants of health at the individual and environmental levels may influence LBW.

This study analyzed individual and environmental level factors associated with LBW in a county in a Midwestern state. Using county birth certificate data from 2007 to 2013 (n = 11,944), this study aggregated rates of lead poisoning, sexually transmitted diseases, and crime data to enable a multilevel analysis within Census Block Group (CBG) areas. All analyses were performed in R.

The Random intercept model with no predictors demonstrated that there was significant variation across the CBG areas (p < .001), but the size of this variation was modest (about 4% of residual variation).  When the model allowed random variation across the CBG areas with individual level predictors (e.g. age and race) in the random slope model, the amount of unexplained variance was even smaller and not significant.  In other words, the effect of the environmental factors was not significant in the overall model. In the final model, the odds of having a LBW infant were significantly higher for mothers: a) who were African American; 1.91 [CI: 1.55, 2.35], b) with fewer prenatal care visits; 3.78 [CI: 3.14, 4.56], and c) who smoked; 1.92 [CI: 1.58, 2.33]. This suggests county level interventions to reduce LBW should focus on individual factors and access to prenatal care.