Manufacturing Plant Social Characteristics and Incident Hypertension, Diabetes and Ischemic Heart Disease

David H. Rehkopf, Stanford University
Mark Cullen, Stanford University

Characteristics of the workplace have long been associated with increased hypertension, diabetes and cardiovascular disease. However, this work suffers from a major limitation of self-report bias of work conditions. We will present results from regression models showing the extent to which manufacturing plant level characteristics predict incident hypertension, heart disease and diabetes. Our models use plant level (rather than individual level) characteristics in order to avoid bias inherent in most prior studies. Critical to our estimates are sensitivity analyses of ecological confounding based on the characteristics of the counties in which the manufacturing plants are located. Due to the large number of potential confounding variables available, we present results from machine learning methods for choosing appropriate confounders when there is a lack of clear theory of which measures would be most appropriate for avoiding bias, and compare these results to theory driven choice of confounding variables.

  See extended abstract

Presented in Session 30: Context, Health and Well-Being