Occupational Radiation, Neighborhoods and Circulatory Disease Incidence and Mortality

Abstract

Background: Exposure to ionizing radiation has been primarily linked with the development of cancer, but in recent decades it has also been recognized to be a risk factor for circulatory diseases (CDs). High doses of ionizing radiation above 500mGy are associated with damage of the circulatory system and increased mortality from circulatory diseases (CD). The biological mechanism is hypothesized to include inflammation, oxidative stress, changes in platelet activity, DNA damage, endothelial dysfunction and cell death. However, the risks of occupational exposure to low protracted doses of medical ionizing radiation on CD mortality are not well understood. The risk of CD is also influenced by an individual’s social environment, and this may modify the effect of occupational exposures like ionizing radiation. Most studies on the association between neighborhood context and coronary heart disease found increased risk of this disease among residents of neighborhoods of lower socioeconomic status. The socio-biological mechanism of this relationship is multifactorial, and it includes the physical environment, availability of services, neighborhood social interactions, experiential perceptions, spread of disease, stress levels, health behaviors and other shared factors. To understand the effect of cumulative ionizing radiation and the neighborhood context on CD, we used longitudinal data from the US Radiologic Technologists (USRT) study and the US Census Bureau. The USRT is a large national historical cohort of 146,022 radiation technologists that has followed participants since 1982 through four comprehensive study waves. It is known that the average participation rate across these four questionnaires was 68%, but the characteristics of participants and non-participants have not been compared to date. If non-participation and attrition are not a random process, this could introduce bias to any effect estimates calculated with this data, including estimates of CD incidence and mortality. Objective: The overall aim of the current thesis was to investigate the effect of protracted exposure to low doses of occupational radiation on CD incidence and mortality and assess if these associations are influenced by the residential neighborhood socioeconomic status (nSES) or by sequential participation bias. Manuscript 1: Aim: Examine the dose-response relationship between low chronic doses of occupational radiation exposure and CD mortality in a large cohort of radiologic technologists. Methods: 109,300 radiologic technologists in the US were followed for mortality from completion of baseline surveys during 1983-89 or 1994-98 through 2012. Using discrete time hazard models, HR and 95%CI were estimated for the association between cumulative ionizing radiation (as badge and organ doses) and mortality from any CD, ischemic heart disease (IHD) and cerebrovascular disease (CeVD). Calendar year was included as a time scale and adjusted for age as a time-varying covariate, sex, race, ethnicity, BMI and smoking. Results: We found evidence of increased risk of CD mortality with increased cumulative exposure to occupational radiation. For every 100mSv increase in cumulative badge dose, the risk of mortality from any CD increased in 3% (95% CI 2-4%) adjusted for age, sex, race, ethnicity, BMI and smoking. For every 100mGy increase in heart dose, the HR for IHD mortality was 1.07 (95% CI 1.02 – 1.13) and the hazard ratio per 100mGy increase in brain dose was 1.39 (95%CI 1.14 – 1.64) for CeVD. Categorical analyses supported this linear trend for all outcomes. There was evidence of effect modification by gender for CD and IHD mortality (p-value for interaction < 0.001 for both), but not by BMI or certification year. Manuscript 2: Aim: Determine whether the association between cumulative occupational radiation exposure on CD incidence and mortality is confounded or modified by residential nSES. Methods: Tertiles of nSES were created from an nSES index calculated for each block-group in the U.S. using 6 components from the 1990 Census and assigned it to participants by geocoding their 1990 mailing address. The confounding role of nSES tertiles was assessed by including it in discrete time hazard models along radiation exposure and adjusting for educational attainment. To determine the presence of effect modification by nSES tertiles, we tested a two-way multiplicative interaction between nSES tertiles and radiation, and assessed an additive interaction using the relative excess risk due to interaction (RERI). Results: Compared to residents from the top tertile of nSES, technologists from low-nSES areas had an elevated risk of overall CD mortality, IHD mortality and incidence, and CeVD incidence with (HRs ranging from 1.21 to 1.31), but not for CeVD mortality (HR: 0.97, 95%CI: 0.80-1.18). We found no evidence of confounding by nSES on the association between radiation and any CD outcomes. There was evidence of a multiplicative interaction between nSES tertiles and radiation exposure for all mortality outcomes (p-value for interaction: <0.001 for overall CD, 0.003 for IHD and 0.004 for CeVD), but not for incidence. People from high nSES had higher HRs for mortality than low-nSES residents for the same radiation dose. Manuscript 3: Aim: Compare the characteristics of participants and non-participants of each survey and estimate the impact of low chronic exposure to radiation on incidence and mortality from ischemic heart disease (IHD) and cerebrovascular disease (CeVD) adjusting for possible selection bias. Methods: Our main exposure was cumulative radiation exposure estimated for 110,374 technologists. After identifying predictors of participation from baseline and survey information, we created inverse probability weights (IPW) of participation to account for selection bias and included them in discrete time hazard models of the association of radiation and circulatory outcomes. Results: Being younger, female, white, married, non-smoker, having a normal BMI, reporting good health status, and living in a rural area was associated higher probability of participating in any survey. The difference between our estimates with or without IPWs ranged between 1-4%-points for IHD incidence, IHD mortality and CeVD mortality and supported evidence of increased risk of disease with higher chronic radiation exposure. For CeVD incidence, the weighted estimate was 15%-points lower than the unweighted estimate, but it still suggested an increased risk of CeVD incidence from chronic exposure to ionizing radiation. Conclusion: This study of medical radiation technologists who were chronically exposed to low-doses of radiation found consistent evidence of radiation-induced circulatory effects below the current 0.5Gy threshold. In addition, we found that nSES was independently associated with CD mortality and incidence in a group of US technologists occupationally exposed to low-doses of radiation, but there was no consistent evidence that it confounded the association between radiation and CD outcomes. We found a multiplicative two-way interaction between radiation and nSES, where residents of high nSES areas had higher risk of CD mortality than residents of low nSES exposed to the same radiation dose. In terms of our selection bias analysis, we found only a small impact of selection bias on our estimates of the association of cumulative radiation exposure and CD outcomes despite consistent differences in the characteristics of participants and non-participants.