Eric Bomberg HW 3/6/18

Eric Bomberg HW 3/6/18

by Eric -
Number of replies: 3

1.     What are 3 different ways to account for SES in an analytic model when investigating racial/ethnic health disparities? (describe a fourth for extra bonus points!). Briefly discuss the interpretations/implications of each approach as it relates to the interest in understand health disparities by race/ethnicity. 

3 ways to account for SES in an analytic model when investigating racial/ethnic health disparities are through mediation analysis, generalized estimating equations, and multilevel modeling.

a.     Mediation Analysis:  A mediation model is one that seeks to identify and explain the mechanism or process that underlies an observed relationship between an independent and dependent variable via the inclusion of a third hypothetical variable (mediator variable).  From a statistical standpoint, mediation analysis can be accomplished through sequential logistic regression models.  This technique was utilized in the Lorch article.  In this article, among Hispanic women, socioeconomic factors mediated 36% of the disparity in fetal deaths. 

b.    Generalized Estimating Equations (GEEs):  GEEs are used to estimate the parameters of a generalized linear model with a possible unknown correlation between outcomes.  This technique was utilized in the Headen article.  In this article, GEEs with log link functions were used to estimate risk ratios for race and gestational weight gain that accounted for clustering of pregnancies within women.  The authors concluded that normal weight Black and Hispanic women and underweight Black women all experienced an increased risk of inadequate gestational weight gain compared to White women. 

c.     Multilevel Modelling:  multilevel models are statistical models of parameters that vary at more than one level.  This technique was utilized in the Coley article.  In this article, multilevel models with cross-level interactions were used to identify variation in racial differences in low birth weight outcomes across neighborhood risk levels when controlling for maternal demographics and pregnancy behaviors.  The authors concluded that African American mothers were significantly more likely to have infants of low birth weight than White mothers and, even when controlling for confounding factors, racial disparities in low birth weight odds did not significantly vary by neighborhood risk level. 

2.     Describe a potential effect modifier, mediator, or contextual variable (for definition of contextual variable, see first page of option Merlo reading) for an association of interest to you and relevant to health disparities. For example, for investigating the association between SES and maternal mortality, I might be interested in the contextual variable of exposure to violence in the neighborhood. Describe how you would study whether this relationship exists.

Effect Modifier:  Effect modification occurs when the magnitude of a measure of association (between exposure and outcome) meaningfully differs according to a third variable.  One such variable that could be an effect modifier on the association between physical activity and obesity is neighborhood safety and walkability.  Sallis et al. (Prev Med, 2018) examined the association of neighborhood walkability to multiple activity-related outcomes and BMI among adolescents, and evaluated socioeconomic status as an effect modifier.  The authors discovered that walkability was positively related to physical activity and more frequent walking for transportation.  They also discovered that sedentary time and TV time were negatively related to walkability.  Further, time in vehicles was negatively related to walkability only among higher-income adolescents. Adolescents living in walkable neighborhoods reported less television time and less time in vehicles.


In reply to Eric

Re: Eric Bomberg HW 3/6/18

by Emilia Demarchis -

Hey Eric,

I really like your overview of the Sallis article--what a great way to demonstrate the importance of not only activity, but neighborhood level factors. Seems like this could also be evaluated in terms of contextual phenomenon--which I look forward to learning more about on Tuesday. 


In reply to Eric

Re: Eric Bomberg HW 3/6/18

by Hala Borno -

Thanks Eric, fascinating and important analysis. I wonder how they narrowed  what variables to include in their model. 

In reply to Eric

Re: Eric Bomberg HW 3/6/18

by Andrea Pedroza Tobias -

It is a fascinating study of walkability and physical activity. It is interesting and expected that the association between walkability and physical activity changes among SES. I think that it is expected because people with high SES are more likely to live in better neighborhoods, with better walkability score, but also people with high SES are more likely to have a car, and therefore they don't walk or use public transportation to go to work/school.  I am wondering if the authors did an analysis identifying the type of physical activity: maybe high SES have a higher prevalence of recreative physical activity (i.e. go to the gym or practice some sport), while low SES have a higher prevalence of physical activity due to active transportation.