Does the County Poverty Rate Influence Birth Weight and Infant Mortality in Kansas? A Multilevel Analysis of Vital Statistics and Census Data, 2006 – 2010

BACKGROUND: This study aimed to assess how birth weight and infant mortality varied in relation to the county poverty rate in Kansas from 2006 to 2010 and to evaluate if there was an independent effect separate and beyond that due to maternal characteristics and behaviors.

METHODS: This was an analytical cross-sectional multilevel analysis of linked birth and death certificate data for the years 2006 to 2010.  There were 105 counties and 191,309 infants included in the analysis. One-way analysis of variance and chi-squared tests were used to assess for differences in the distribution of birth weight and infant mortality among independent variables. Random effects linear and logistic regression was used to calculate the crude and adjusted effect of the county poverty rate on birth weight and infant mortality.

RESULTS: Birth weight and infant mortality were both associated with the county poverty rate. Birth weight decreased (β -3.11 grams, 95% CI -5.4, -0.8, p<0.05) and infant mortality increased (OR 1.02, 95% CI 1.01, 1.03, p<0.05) with each 1% increase in the county poverty rate in crude analysis. Both birth weight (β -0.2 grams, 95% CI -2.1, 1.7, p=0.85) and infant mortality (OR 1.00, 95% CI 0.98, 1.01, p=0.79) were attenuated to the null after adjusting for maternal characteristics and behaviors.

CONCLUSIONS: Neighborhood level poverty has been shown to effect human health in multiple studies. Macintyre, et. al. state that neighborhoods effect health through their physical features, built environment, service provisions, culture and reputation. The atomistic fallacy warns us of ignoring community context when describing causal relationships. These social factors can be considered risk regulators; not direct risk factors in the traditional sense, rather determinants of more proximal risk factors. Using this as a framework for interpretation, there is evidence that the county poverty rate does play a role in birth weight and infant mortality in Kansas. Increases in the county poverty rate were associated with worse outcomes for both birth weight and infant mortality, although a strong linear trend is not evident. This association remained after controlling for county urbanization. The effect of the county poverty rate did not remain after adjusting for maternal characteristics and behaviors. Rather than interpreting this as confounding, the risk regulator framework allows us to interpret the county poverty rate as a driver of maternal behaviors, which influence birth weight and infant mortality.