Nancy L. Fleischer

Using directed acyclic graphs to guide analyses of neighbourhood health effects: an introduction

Background: Directed acyclic graphs, or DAGs, are a useful graphical tool in epidemiologic research that can help identify appropriate analytical strategies in addition to potential unintended consequences of commonly used methods such as conditioning on mediators. The use of DAGs can be particularly informative in the study of the causal effects of social factors on health. Methods: The authors consider four specific scenarios in which DAGs may be useful to neighbourhood health effects researchers: (1) identifying variables that need to be adjusted for in estimating neighbourhood health effects, (2) identifying the unintended consequences of estimating “direct” effects by conditioning on a mediator, (3) using DAGs to understand possible sources and consequences of selection bias in neighbourhood health effects research, and (4) using DAGs to identify the consequences of adjustment for variables affected by prior exposure. Conclusions: The authors present simplified sample DAGs for each scenario and discuss the insights that can be gleaned from the DAGs in each case and the implications these have for analytical approaches.

Built environment characteristics and perceived active park use among older adults: Results from a multilevel study in Bogota

OBJECTIVE Examine the associations between selected built environment (BE) attributes and perceived active park use among older adults in Bogota. METHODS A cross-sectional multilevel study was conducted. Participants included 1966 older adults in 50 neighborhoods. Socio-demographic covariates and BE attributes were measured. Multilevel logistic regression models were used for the analyses. RESULTS Residents from areas with higher park density and high land-use mix were more likely to report active park use while those from areas with high connectivity were less likely. CONCLUSIONS This study suggests that objective attributes of the residential BE are associated with perceived active park use. However, our study also points to the importance of surrounding environment, with the result of an inverse relationship between connectivity and physical activity, which highlights the potentially necessary interventions in the realm of traffic and pedestrian safety.

Socioeconomic Gradients in Chronic Disease Risk Factors in Middle-Income Countries: Evidence of Effect Modification by Urbanicity in Argentina

OBJECTIVES We investigated associations of socioeconomic position (SEP) with chronic disease risk factors, and heterogeneity in this patterning by provincial-level urbanicity in Argentina. METHODS We used generalized estimating equations to determine the relationship between SEP and body mass index, high blood pressure, diabetes, low physical activity, and eating fruit and vegetables, and examined heterogeneity by urbanicity with nationally representative, cross-sectional survey data from 2005. All estimates were age adjusted and gender stratified. RESULTS Among men living in less urban areas, higher education was either not associated with the risk factors or associated adversely. In more urban areas, higher education was associated with better risk factor profiles (P < .05 for 4 of 5 risk factors). Among women, higher education was associated with better risk factor profiles in all areas and more strongly in more urban than in less urban areas (P < 0.05 for 3 risk factors). Diet (in men) and physical activity (in men and women) were exceptions to this trend. CONCLUSIONS These results provide evidence for the increased burden of chronic disease risk among those of lower SEP, especially in urban areas.

Estimating the potential impacts of intervention from observational data: methods for estimating causal attributable risk in a cross-sectional analysis of depressive symptoms in Latin America

Background: The field of epidemiology struggles both with enhancing causal inference in observational studies and providing useful information for policy makers and public health workers focusing on interventions. Population intervention models, analogous to population attributable fractions, estimate the causal impact of interventions in a population, and are one option for understanding the relative importance of various risk factors. With population intervention parameters, risk factors are effectively standardised, allowing one to compare their values directly and determine which potential intervention may have the greatest impact on the outcome. Methods: The difference between total effects and population intervention parameters was examined using naïve, G-computation and inverse probability of treatment weighting approaches. The differences between these parameters and the intuitions they provide were explored using data from a 2003 cross-sectional study in rural Mexico. Results: The assumptions, specific analytic steps, limitations and interpretations of the total effects and population intervention parameters are discussed, and code is provided in Stata. Conclusion: Population intervention parameters are a valuable and straightforward approach in epidemiological studies for making causal inference from the data while also supplying information that is relevant for researchers, public health practitioners and policy makers.

Geographic Proximity of Waterpipe Smoking Establishments to Colleges in the U.S.

INTRODUCTION Waterpipe tobacco smoking is prevalent among college students in the U.S. and increasing in popularity. Waterpipe smoking establishments are almost completely unregulated, and limited information exists documenting the expansion of this industry. The objective of this study was to identify U.S.-based waterpipe establishments and measure their proximity to colleges/universities. METHODS Waterpipe establishments and their addresses were compiled using five Internet-based directories during 2014 and analyzed in 2015. Addresses were geocoded and overlaid on a U.S. map of accredited colleges/universities. Proximity of colleges/universities to the nearest waterpipe establishment was measured in 3-mile increments. Multinomial logistic regression was used to model the factors associated with proximity of waterpipe establishments to colleges/universities. RESULTS A total of 1,690 waterpipe establishments and 1,454 colleges/universities were included in the study. Overall, 554 colleges/universities (38.1%) were within 3 miles of a waterpipe establishment. Proximity of waterpipe establishments to colleges/universities was associated with higher full-time student enrollment. Public colleges/universities and those with a smoke-free campus policy were at lower odds of having waterpipe establishments within 3 miles of their campuses. CONCLUSIONS Waterpipe smoking establishments are more likely to be located near large colleges/universities. This study should inform initiatives aimed at reducing retail tobacco establishment exemptions.

Self-rated health and social inequalities, Buenos Aires, Argentina, 2005

Self-rated health is a quality-of-life indicator. This study investigates the impact of individual-level and neighborhood-level socioeconomic characteristics, considered simultaneously, on the state of self-rated health at the individual level in Buenos Aires, Argentina. The study employs a two-level (individual and neighborhood) multilevel analysis, and the data sources were the 2005 Argentina National Risk Factor Survey (multistage probabilistic sample) and the 2001 Population Census. Linear regression shows that higher schooling and income, as well as occupational category, are related to better self-rated health, and increasing age with worse health. In the multilevel analysis, an increase in the proportion (per census tract) of individuals with less schooling was associated with an increase in the proportion of individuals with worse self-rated health. Improving the general health of the population requires strategies and action that reduce the levels of social inequalities in their multiple dimensions, including the individual and neighborhood levels.

Neighbourhood deprivation and smoking and quit behaviour among smokers in Mexico: findings from the ITC Mexico Survey

Background In high-income countries (HICs), higher neighbourhood socioeconomic deprivation is associated with higher levels of smoking. Few studies in low-income and middle-income countries (LMICs) have investigated the role of the neighbourhood environment on smoking behaviour. Objective To determine whether neighbourhood socioeconomic deprivation is related to smoking intensity, quit attempts, quit success and smoking relapse among a cohort of smokers in Mexico from 2010 to 2012. Methods Data were analysed from adult smokers and recent ex-smokers who participated in waves 4–6 of the International Tobacco Control (ITC) Mexico Survey. Data were linked to the Mexican governments composite index of neighbourhood socioeconomic deprivation, which is based on 2010 Mexican Census data. We used generalised estimating equations to determine associations between neighbourhood deprivation and individual smoking behaviours. Findings Contrary to past findings in HICs, higher neighbourhood socioeconomic deprivation was associated with lower smoking intensity. Quit attempts showed a U-shaped pattern whereby smokers living in high/very high deprivation neighbourhoods and smokers living in very low deprivation neighbourhoods were more likely to make a quit attempt than smokers living in other neighbourhoods. We did not find significant differences in neighbourhood deprivation on relapse or successful quitting, with the possible exception of people living in medium-deprivation neighbourhoods having a higher likelihood of successful quitting than people living in very low deprivation neighbourhoods (p=0.06). Conclusions Neighbourhood socioeconomic environments in Mexico appear to operate in an opposing manner to those in HICs. Further research should investigate whether rapid implementation of strong tobacco control policies in LMICs, as occurred in Mexico during the follow-up period, avoids the concentration of tobacco-related disparities among socioeconomically disadvantaged groups.

Extending the income inequality hypothesis: ecological results from the 2005 and 2009 Argentine National Risk Factor Surveys.

Abstract A consensus on income inequality as a social determinant of health is yet to be reached. In particular, we know little about the cross-sectional versus lagged effect of inequality and the robustness of the relationship to indicators that are sensitive to varying parts of the income spectrum. We test these issues with data from Argentinas 2005 and 2009 National Risk Factor Surveys. Inequality was operationalised at the provincial level with the Gini coefficient and the Generalised Entropy (GE) index. Population health was defined as the age-standardised percentage of adults with poor/fair self-rated health by province. Our cross-sectional results indicate a significant relationship between inequality (Gini) and poor health (r=0.58, p<0.01) in 2005. Using the GE index, a gradient pattern emerges in the correlation, and the r values increase as the index becomes sensitive to the top of the distribution. The relationship between 2005 inequality and 2009 health displays a similar pattern, but with generally smaller correlations than the 2005 cross-sectional results. Further advances in the income inequality and health literature require new theoretical models to account for how inequalities in different parts of the income spectrum may influence population health in different ways.

Changes in health selection of obesity among Mexican immigrants: a binational examination.

Health selection is often measured by comparing the health of more recent immigrants to the native born of their new host country. However, this comparison fails to take into account two important factors: (1) that changes in the health profile of sending countries may impact the health of immigrants over time, and (2) that the best comparison group for health selection would be people who remain in the country of origin. Obesity represents an important health outcome that may be best understood by taking into account these two factors. Using nationally-representative datasets from Mexico and the US, we examined differences in obesity-related health selection, by gender, in 2000 and 2012. We calculated prevalence ratios from log-binomial models to compare the risk of obesity among recent immigrants to the US to Mexican nationals with varying likelihood of migration, in order to determine changes in health selection over time. Among men in 2000, we found little difference in obesity status between recent immigrants to the US and Mexican non-migrants. However, in 2012, Mexican men who were the least likely to migrate had higher obesity prevalence than recent immigrants, which may reflect emerging health selection. The trends for women, however, indicated differences in obesity status between recent Mexican immigrants and non-migrants at both time points. In both 2000 and 2012, Mexican national women had significantly higher obesity prevalence than recent immigrant women, with the biggest difference between recent immigrants and Mexican women who were least likely to migrate. There was also indication that selection increased with time for women, as the differences between Mexican nationals and recent immigrants to the US grew from 2000 to 2012. Our study is among the first to use a binational dataset to examine the impact of health selectivity, over time, on obesity.

Smoking Trends among U.S. Latinos, 1998–2013: The Impact of Immigrant Arrival Cohort

Few studies examine nativity disparities in smoking in the U.S., thus a major gap remains in understanding whether immigrant Latinos’ smoking prevalence is stable, converging, or diverging, compared with U.S.-born Latinos. This study aimed to disentangle the roles of period changes, duration of U.S. residence, and immigrant arrival cohort in explaining the gap in smoking prevalence between foreign-born and U.S.-born Latinos. Using repeated cross-sectional data spanning 1998–2013 (U.S. National Health Interview Survey), regressions predicted current smoking among foreign-born and U.S.-born Latino men and women (n = 12,492). We contrasted findings from conventional regression analyses that simply include period and duration of residence effects, to two methods of assessing arrival cohort effects: the first accounted for baseline differences in smoking among arrival cohorts, while the second examined smoking probabilities by tracking foreign-born arrival cohorts as they increase their duration of U.S. residence. Findings showed that Latino immigrants maintained lower prevalence of current smoking compared with U.S.-born Latinos over the period 1998–2013, and that longer duration of U.S. residence is associated with lower odds of smoking among men. Two findings are particularly novel: (1) accounting for immigrant arrival cohort dampens the overall protective effect of duration of residence among men; and (2) the earliest arrival cohort of Latino immigrant men experienced the steepest decline in smoking over duration of U.S. residence. Results have methodological and theoretical implications for smoking studies and the Latino mortality paradox.