BACKGROUND: Preterm births (PTB), infants born <37 weeks gestation, increase the risk of perinatal morbidity/mortality and occur in about 7% of all live births in Multnomah County, Oregon. Concentrated Disadvantage (CD), an indicator developed by the Association of Maternal and Child Health Programs, may be an important predictor of preterm births; furthermore, the association between concentrated disadvantage and preterm births may vary spatially. Identifying locations where CD is most strongly associated with PTB may provide further guidance to policy-makers and program planners for prioritizing resource allocation for place-based interventions.
METHODS: Birth vital statistics (2011-2015) from Oregon Health Authority’s Center for Health Statistics were used to identify the number of preterm births for each census block within the urban growth boundary of Multnomah County. Concentrated disadvantage for each census tract was calculated by averaging the z-scores from 5 variables in the American Community Survey (2014 5-year estimates): poverty, public assistance, female-headed households, unemployment, and individuals < 18 years age. These data were joined with census block group boundary files from Oregon Metro’s Regional Land Information System. Ordinary least squares regression and geographically weighted regression were used to assess the association between concentrated disadvantage and preterm births per block group after adjusting for race.
RESULTS: During 2011-2015, there were 3,448 preterm births out of a total of 46,495 total live births. Ordinary Least Squares regression demonstrated a global association between CD and PTB (F-statistic = 63, p-value<.0001). On average, each 1-unit increase in CD predicted a 0.24 increase in number of preterm births per census block group (p-value<.001). Geographically weighted regression shows that the association between CD and PTB varies through space, with CD being a stronger predictor of preterm births per block group in the southern region of the study area compared to the northern. However, this spatial difference in association, though statistically significant, does not demonstrate meaningful variability in predictive ability, with each 1-unit increase in CD ranging from a 0.21 to a 0.26 increase in number of preterm births per block group depending on location.
CONCLUSIONS: Concentrated Disadvantage may be an important determinant for predicting preterm births per block group, but the strength of this association does not appear to vary widely through space. Further research adjusting for additional potential confounders is needed.