Hormesis is a nonlinear dose-response characterized by biological responses at low doses that are opposite to those observed at higher doses. Studies and review articles on hormesis are being published at an increasing rate by researchers from diverse disciplines and debate has emerged over the role hormesis in risk assessment. As a result, a survey was conducted to assess toxicologists and risk assessors knowledge and attitudes about the hormesis dose response. Study goals were to: (1) ascertain attitudes towards hormesis and other dose-response models, (2) identify whether acceptance or rejection of hormesis is based on knowledge of hormesis, predisposing values, or demographic characteristics, and (3) evaluate potential for response bias. The survey consisted of 44 questions pre-tested by 25 toxicologists and risk assessors. The survey was distributed via email to the membership of the Society of Toxicology and the Society for Risk Analysis, 9,500 potential respondents. The overall response rate was 17% (n= 1,463) with a completion rate over 87%. Major findings were that 50% of respondents indicated sufficient data exist to support the view hormesis occurs across a wide range of species and endpoints, 59% indicated evaluating potential benefits due to hormesis should be included in risk assessments, and 65% are in favor of modifying hazard assessment protocols to identify the presence of hormesis. Respondent characteristics such as: years of experience, society membership, education, residence, employment (excluding government and pharmaceutical companies), and political, economic or social views had little influence on opinion. One of the largest positive influences was experience with hormesis based on actual research; 79% of subjects who reported observing hormesis commonly in their studies agreed hormesis is broadly generalizable. The influence of non-response bias was evaluated through several internal and external measures. Despite a lower than hoped for response rate, but because of robust external validity measures, it is concluded that respondents’ opinions are likely a reasonable representation of the societies of which they are members. Because this is a baseline survey, a follow-up survey is in order. Future survey design should separately evaluate the science of dose-response from the regulatory approach to risk assessment.

This three-essay dissertation examines racial disparities in infant health outcomes and exposure to air pollution in Texas. It also asks whether the EPA's Risk-Screening Environmental Indicators Geographic Microdata (RSEI-GM) might be used to assess the effects of little-studied toxic air pollutants on infant health outcomes. Chapter 1 contributes to the “weathering” literature, which has shown that disparities in infant health outcomes between non-Hispanic black and non-Hispanic white women tend to widen with age. In this study, we ask whether the same patterns are observed in Texas and among Hispanic women, since other studies have focused on black and white women from other regions. We find that black and Hispanic women in Texas do “weather” earlier than white mothers with respect to rates of low birthweight and preterm birth. This differential weathering appears to be mediated by racial disparities in the distribution and response to socioeconomic risk factors, though a large gap between black and white mothers across all ages remains unexplained. Chapter 2 extends the statistical environmental justice literature by examining the distribution of toxic air pollution across infants in Texas. We find that, within Texas cities, being black or Hispanic is a significant predictor of how much pollution one is exposed to at birth. We further find that, among mothers who move between births, white mothers tend to move to significantly cleaner areas than black or Hispanic mothers. In Chapter 3, we use geocoded birth records matched to square-kilometer pollution concentration estimates from the RSEI-GM to ask whether the pollution-outcome relationships that emerge through regression analysis are similar to the effects found in previous research. If so, the RSEI-GM might be used to study the health effects of nearly 600 chemicals tracked in that dataset. We conclude, based on instability of results across various specifications and lack of correspondence to previous results, that the merged birth record-RSEI data are not appropriate for statistical epidemiology research.