Association between Neighborhood-Level Smoking and Individual Smoking Risk: Maternal Smoking among Latina Women in Pennsylvania

Abstract

Objective. We examined whether or not high maternal smoking rates at the neighborhood level increase the likelihood of individual smoking by Latina women in the three months prior to and during pregnancy, independent of other individual and neighborhood factors.Methods. This study was observational in nature, using linked vital statistics records for 24,443 Latina women in Pennsylvania (2009Pennsylvania ( -2010 and U.S. Census data for 2,398 census tracts. We used multilevel logistic regression models to determine the individual odds of self-reported maternal smoking given different census tract-level rates of maternal smoking in the previous three years (2006)(2007)(2008), adjusting for maternal and census-tract characteristics, including ethnic density, population density, and poverty.Results. Higher levels of maternal smoking at the census-tract level were associated with increased individual odds of smoking among Latina mothers. In the fully adjusted model, a 10% increase in the neighborhood smoking rate was associated with a 1.28 (95% confidence interval 1.22, 1.34) increase in the individual odds of smoking.Conclusion. Latina women living in census tracts where more women have smoked during or immediately prior to pregnancy are themselves at higher risk of smoking during this period.

Spatial Autocorrelation Statistics of Areal Prevalence Rates under High Uncertainty in Denominator Data

“…Resorting to the EB approach by Marshall () does not completely solve the problem because it only considers variance heterogeneity based on population size, not on sampling error. Especially when reviewing spatial autocorrelation detection in public health studies, we find many uses of global Moran’s I and LISA that overlook uncertainty of ACS data (Dai et al ; Farber, Morang, and Widener ; Chesnokova et al ; Yang et al ; Ford et al ). As a first step toward improving statistical tools to account for uncertainty in ACS data, we focus on the most widely used spatial statistics, Moran’s I and LISA.…”

“…Thus, epidemiologists have easier access to small‐area estimates, such as population counts and birth counts, race and age, which are routinely used in denominator data when measuring prevalence or incidence rates by areas. Since the release of the first 5‐year ACS estimates in 2009, mapping and spatial autocorrelation detection of health‐related incidence with small‐area estimates have become prevalent in many place‐based public health studies (Dai et al ; Farber, Morang, and Widener ; Chesnokova et al ; Yang et al ; Ford et al ). Because of the sampling methodology, the numbers in small area are not actual counts, but estimates with associated confidence intervals.…”

Smoking Ban Law in Chile: Impact in Newborns’ Birth Weight by Women’s Age Groups and by City Population Density

“…Other studies have shown that the effectiveness of SBL in reducing tobacco behavior in the population could be modified by urban environments. With a higher number of public places for social interaction, more dense cities have experienced a greater reduction in second-hand smoking than less dense cities [ 16 , 17 ]. Some studies, the majority in high-income countries, have assessed the impact of SBLs on reducing LBW over the years.…”