Why Not Replace Quantitative Risk Assessment Models with Regression Models?

Chapter 2 suggested that dynamic simulation models, Bayesian networks, and causal analysis can add value to statistical regression modeling for understanding causal exposure concentration-response (C-R) relationships well enough to predict how changes in exposure would affect health risks—a task that typically requires causal insights that regression modeling alone cannot deliver (Pearl 2009). Chapters 3 , 4 , 5 , and 6 have discussed dynamic simulation models. This part of the book turns to Bayesian networks and causal analysis (Chaps. 9 , 10 , and 11 ). First, however, this chapter and Chap. 8 examine some ways in which regression has been misapplied in public health risk analysis, both to motivate the need for other methods and to explain why regression alone is not an adequate substitute for quantitative risk assessment (QRA), with its explicit emphasis on preventable causes of disease and the quantitative causal relationships between reductions in exposures and resulting reductions in health risks. Part 3 will apply these lessons specifically to air pollution and public health, with greatest emphasis on National Ambient Aie Quality Standards (NAAQS) for fine particulate matter (PM2.5). This chapter previews some of the issues developed in Part 3 by considering public health risks from a much more local form of air pollution: emissions from factory farms.