Estimating Counterfactual Risk Under Hypothetical Interventions in the Presence of Competing Events: Crystalline Silica Exposure and Mortality From 2 Causes of Death.

Exposure to silica has been linked to excess risk of lung cancer and nonmalignant respiratory disease mortality. In this study we estimated risk for both these outcomes in relation to occupational silica exposure as well as the reduction in risk that would result from hypothetical interventions on exposure in a cohort of exposed workers. Analyses were carried out using data from an all-male study population consisting of 2,342 California diatomaceous earth workers regularly exposed to crystalline silica and followed between 1942 and 2011. We estimated subdistribution risk for each event under the natural course and interventions of interest using the parametric g-formula to adjust for healthy-worker survivor bias. The risk ratio for lung cancer mortality, comparing an intervention in which a theoretical maximum exposure limit was set at 0.05 mg/m3 (the current US regulatory limit) with the observed exposure concentrations, was 0.86 (95% confidence interval: 0.63, 1.22). The corresponding risk ratio for nonmalignant respiratory disease mortality was 0.69 (95% confidence interval: 0.52, 0.93). Our findings suggest that risks from both outcomes would have been considerably lower if historical silica exposures in this cohort had not exceeded current regulatory limits.

Exposure-Lag-Response in Longitudinal Studies: Application of Distributed-Lag Nonlinear Models in an Occupational Cohort.

Prolonged exposures can have complex relationships with health outcomes, as timing, duration, and intensity of exposure are all potentially relevant. Summary measures such as cumulative exposure or average intensity of exposure may not fully capture these relationships. We applied penalized and unpenalized distributed-lag nonlinear models (DLNMs) with flexible exposure-response and lag-response functions in order to examine the association between crystalline silica exposure and mortality from lung cancer and nonmalignant respiratory disease in a cohort study of 2,342 California diatomaceous earth workers followed during 1942-2011. We also assessed associations using simple measures of cumulative exposure assuming linear exposure-response and constant lag-response. Measures of association from DLNMs were generally higher than those from simpler models. Rate ratios from penalized DLNMs corresponding to average daily exposures of 0.4 mg/m3 during lag years 31-50 prior to the age of observed cases were 1.47 (95% confidence interval (CI): 0.92, 2.35) for lung cancer mortality and 1.80 (95% CI: 1.14, 2.85) for nonmalignant respiratory disease mortality. Rate ratios from the simpler models for the same exposure scenario were 1.15 (95% CI: 0.89, 1.48) and 1.23 (95% CI: 1.03, 1.46), respectively. Longitudinal cohort studies of prolonged exposures and chronic health outcomes should explore methods allowing for flexibility and nonlinearities in the exposure-lag-response.

Evaluation of the Webler-Brown model for estimating tetrachloroethylene exposure from vinyl-lined asbestos-cement pipes.

BACKGROUND: From May 1968 through March 1980, vinyl-lined asbestos-cement (VL/AC) water distribution pipes were installed in New England to avoid taste and odor problems associated with asbestos-cement pipes. The vinyl resin was applied to the inner pipe surface in a solution of tetrachloroethylene (perchloroethylene, PCE). Substantial amounts of PCE remained in the liner and subsequently leached into public drinking water supplies. METHODS: Once aware of the leaching problem and prior to remediation (April-November 1980), Massachusetts regulators collected drinking water samples from VL/AC pipes to determine the extent and severity of the PCE contamination. This study compares newly obtained historical records of PCE concentrations in water samples (n = 88) with concentrations estimated using an exposure model employed in epidemiologic studies on the cancer risk associated with PCE-contaminated drinking water. The exposure model was developed by Webler and Brown to estimate the mass of PCE delivered to subjects' residences. RESULTS: The mean and median measured PCE concentrations in the water samples were 66 and 0.5 microg/L, respectively, and the range extended from non-detectable to 2432 microg/L. The model-generated concentration estimates and water sample concentrations were moderately correlated (Spearman rank correlation coefficient = 0.48, p < 0.0001). Correlations were higher in samples taken at taps and spigots vs. hydrants (rho = 0.84 vs. 0.34), in areas with simple vs. complex geometry (rho = 0.51 vs. 0.38), and near pipes installed in 1973-1976 vs. other years (rho = 0.56 vs. 0.42 for 1968-1972 and 0.37 for 1977-1980). Overall, 24% of the variance in measured PCE concentrations was explained by the model-generated concentration estimates (p < 0.0001). Almost half of the water samples had undetectable concentrations of PCE. Undetectable levels were more common in areas with the earliest installed VL/AC pipes, at the beginning and middle of VL/AC pipes, at hydrants, and in complex pipe configurations. CONCLUSION: PCE concentration estimates generated using the Webler-Brown model were moderately correlated with measured water concentrations. The present analysis suggests that the exposure assessment process used in prior epidemiological studies could be improved with more accurate characterization of water flow. This study illustrates one method of validating an exposure model in an epidemiological study when historical measurements are not available.

Risk of learning and behavioral disorders following prenatal and early postnatal exposure to tetrachloroethylene (PCE)-contaminated drinking water.

This population-based retrospective cohort study examined the association between developmental disorders of learning, attention and behavior and prenatal and early postnatal drinking water exposure to tetrachloroethylene (PCE) on Cape Cod, Massachusetts. Subjects were identified through birth records from 1969 through 1983. Exposure was modeled using information from town water departments, a PCE leaching and transport algorithm, EPANet water flow modeling software, and a Geographic Information System (GIS). Mothers completed a questionnaire on disorders of attention, learning and behavior in their children and on potential confounding variables. The final cohort consisted of 2086 children. Results of crude and multivariate analyses showed no association between prenatal exposure and receiving tutoring for reading or math, being placed on an Individual Education Plan, or repeating a school grade (adjusted Odds Ratios (OR)=1.0-1.2). There was also no consistent pattern of increased risk for receiving a diagnosis of Attention Deficit Disorder (ADD) or Hyperactive Disorder (HD), special class placement for academic or behavioral problems, or lower educational attainment. Modest associations were observed for the latter outcomes only in the low exposure group (e.g., adjusted ORs for ADD were 1.4 and 1.0 for low and high exposure, respectively). (All ORs are based on an unexposed referent group.) Results for postnatal exposure through age five years were similar to those for prenatal exposure. We conclude that prenatal and early postnatal PCE exposure is not associated with disorders of attention, learning and behavior identified on the basis of questionnaire responses and at the exposure levels experienced by this population.