Addition of Covariates

 



The addition of covariates to the basic univariate model will allow us to fully test the causal pathway model (link to birth weight page). We can determine which, if any, covariates affect infant mortality through birth weight and which affect infant mortality directly. An example is using maternal smoking. We analyzed data from the 2001 linked birth-infant death database for European Americans and African Americans. We found that in all populations maternal smoking impacts infant mortality directly. Additionally, there is some evidence that birth weight is also on the causal pathway to infant mortality in female populations but not male populations with respect to maternal smoking. This sex difference may be due to the canalization of females.


    img1

Primary population birth weight distributions and mortality curves (bold lines) with 95% confidence intervals (light lines) for African American Females (a) and African American Males (b). The mortality curves are plotted against standardized birth weight.

 

    img2

Secondary population birth weight distributions and mortality curves (bold lines) with 95% confidence intervals (light lines) for African American Females (a) and African American Males (b). The mortality curves are plotted against standardized birth weight.

 

    img3

Total population birth weight distributions and mortality curves (bold lines) with 95% confidence intervals (light lines) for African American Females (a) and African American Males (b). The mortality curves are plotted against standardized birth weight.

 

Further research on other exogenous variables will shed further light on the trends of birth weight causality and determine consistency of the causal hypothesis. Future projects will expand the current model to include the following covariates:

Plurality
Altitude
Maternal Age
Parity
Gestational Age
Education
Nativity
Prenatal Care