We observed no association between residence within 5 or 3 km of one or more dioxin-emitting facilities and NHL risk. However, we observed significantly elevated risk of NHL for individuals living within 3 km of cement kilns and an inverse association with proximity to municipal solid waste incinerators. The positive association for cement kilns did not appear to be due to selection bias based on our analyses of current residence for all eligible cases and controls; whereas, we found no association with proximity to municipal solid waste incinerators among this group. Our emission metric, which incorporated facility-specific dioxin emissions within 3 or 5 km of residences, was not associated with NHL risk overall or most NHL subtypes. Risk of marginal zone lymphoma was increased among those in the highest quartile, but there was no trend and the association was attenuated in analyses of all eligible cases and controls.
Our results do not support earlier observations that living near municipal solid waste incinerators is associated with increased risk of NHL ; however, our power to evaluate this association was limited due to the small numbers of study participants living within 5 km of these facilities. In a French study, NHL risk was significantly increased (OR = 2.3, 95% CI 1.4-3.8) for residence in the highest exposure zone (an area approximately 1 km by 4 km based on a Gaussian model) around a facility with high emissions (16.3 ng I-TEQ/m3 in 1998) that exceeded the European Union Standard of 0.1 ng I-TEQ/m3. The results from a subsequent case–control study  that estimated serum levels of the 1998 World Health Organization-TEQ among residents near this incinerator further support an association between environmental exposure to dioxins from municipal solid waste incinerators and risk of NHL. An ecological study of NHL risk and residence near 13 incinerators with lower emissions in France reported increased risk only among women (RR = 1.18, 95% CI 1.01-1.37) . Our emission metric was an index based on mass released per year (ng TEQ/yr) rather than concentration at a distance (ng TEQ/m3), which was the basis for French estimates. Therefore, it was not possible to directly compare exposure levels. A small area ecological study in Great Britain found no association between NHL incidence and distance from incinerators (within 3 km of 72 municipal solid waste incinerators compared to 3 to 7.5 km) . Differing results across studies may be due to different study designs, emission levels, and exposure misclassification.
We found no published studies that evaluated risk of NHL associated with environmental exposure to cement kiln emissions. The U.S. EPA average emission level for this facility type was lower than solid waste incinerators and some other facility types (Table 1). However, the TEQ is heavily weighted by PCDDs and 2,3,7,8-tetrachlorofuran dominates emissions from cement kilns (non-hazardous) . We conducted a pilot study to evaluate determinants of dioxins and furans in house dust in a subset of our study homes . Four facility types were located near the pilot study homes: non-hazardous waste cement kilns, coal-fired power plants, sewage sludge incinerators, and medical waste incinerators. We found 2- to 9-fold higher concentrations of 5 PCDDs and 5 PCDFs in homes within 5 km of these cement kilns compared with homes further away, which provides some evidence that our findings may be related to these exposures. With the exception of higher TCDD levels in homes within 5 km of sewage sludge incinerators, proximity to the other three facility types was not association with PCDD and PCDF concentrations in homes.
Dietary exposure to dioxins, mainly through consumption of meat, milk, eggs and fish, is generally believed to be the most important source of non-occupational dioxin exposure . Dioxin levels in blood and breast milk in non-occupationally exposed subjects residing near incinerators in Spain and Germany were not elevated in those who lived in close proximity to incinerators compared with those living farther away [31–34]. However, a recent Russian study found higher dioxin levels among women living within a few kilometers of a dioxin-emitting chemical plant and eating locally-produced foods . Similarly, in a French study, there was no association between serum dioxin levels and residential proximity to a municipal waste incinerator except among those who consumed animal products produced locally . Exposure routes were not determined in the previous positive studies [18–20], but it is possible that NHL associations with residential proximity were driven by local food consumption. Although dietary information was collected in our study, no information on local food consumption was available.
Our study had several strengths. We evaluated NHL risk in relation to a nationwide database of dioxin-emitting facilities, which allowed us to investigate several industrial sources separately and account for exposure from multiple sources simultaneously. Our residential histories allowed us to account for the mobility of the study population and to include information on changes in emissions over time. Previous studies were registry-based and were limited to the residence at diagnosis. Additionally, we had individual level information on potential NHL risk factors, while previous studies had limited data on potential confounders.
Our study had several limitations including small numbers for analyses of specific facility types and NHL subtypes. With the exception of municipal solid waste incinerators, facility-specific emissions were available only for 1995. We estimated changes in the facility’s emissions resulting from regulations in the late 1980s and early 1990s using an EPA national survey of facility-specific changes in air emissions . However, we did not have information on the years in which the facilities started and ceased to be operational, which may have led to misclassification of exposure. We estimated exposure 15 years before diagnosis; however, the latency of NHL in relation to environmental exposures is not known and earlier exposures may be important. Another limitation that may have resulted in exposure misclassification was the relatively simple proximity and distance-weighted emission metrics, which may not capture the complex and asymmetric exposure patterns that can result from local meteorological, facility, and terrain characteristics. Two of the prior studies included stack height, meteorological data, and surface topography, enabling exposure modeling using a Gaussian-type dispersion model [19, 20]; however, we did not have stack height for the facilities in our study. The amount of resulting exposure misclassification in our study depends on how much these factors influenced the true spatial distribution of dioxin emissions.
Our response rates were low especially among controls. However, analyses based on the current residence of all eligible cases and controls provided support for most of the associations we observed. Further, we found no notable differences in the associations across categories of gender, age, and education, factors associated with nonresponse in this population . Finally, the positive findings we observed should be interpreted with caution because we made many comparisons and some associations were based on small numbers.