A limited number of epidemiologic studies investigated the risk of congenital anomalies among populations directly exposed to the emissions of waste incinerators [5, 18–23]. Some studies did not find an increased prevalence of overall anomalies or of specific groups of birth defects, whilst others detected excess risks for nervous system anomalies , cardiovascular defects , facial cleft [19, 22], urinary defects , and overall infant deaths due to congenital malformations . However, methodological limitations considerably hamper the evaluation of results of most investigations: very few studies analyzed maternal residence during the first three months of gestation, or adjusted for maternal age and for socioeconomic status. Moreover, distance of maternal residence (at time of delivery or abortion) from the plant was generally considered in exposure assessment, without taking into consideration the characteristics of the plant (chimney heights and widths, type and amount of combusted waste), the amount of waste combusted and the meteorological factors, with the noticeable exception of the study by Cordier et al. who used an exposure index estimated from a Gaussian plume model.
Overall, results of the present study did not suggest the occurrence of an excess teratogenic risk in the vicinity of the incinerator plant, since prevalence increased only in the medium-exposure area and such increase was statistically very unstable, nor any evidence of reduction in risk during shut-down of the plant emerged. Moreover, a more specific analysis for single categories of anomalies did not generally lead us to identify excess risks for any disease group, though these results must be evaluated with caution since most of the computed estimates were statistically unstable. In particular, urogenital anomalies such hypospadias, an abnormality suspected to be associated with parental exposure to environmental endocrine disrupting chemicals [24, 25], did not increase in the exposed population. However, we found an excess prevalence of chromosomal anomalies in middle exposure area, which is difficult to interpret since there risk was not increased in the high exposure area, and no association was found in the two epidemiologic studies which specifically examined this category of birth defects [20, 22]. We therefore consider it useful to further monitor this finding in the study area or in other comparable contexts.
Some degree of exposure misclassification certainly occurred in the present study. First, since chlorinated compounds are contaminants characterized by persistency in the human body and in the environment, assessment of exposure based on residence during gestation and not on long-term residential history or the occupational environment might have biased to some extent the actual exposure burden experienced by the study subjects during the latest years. However, we checked in a random sample of case and referent women (n = 46, 10% of the study population) their residences three years before date of delivery or abortion, in order to ascertain the extent of long-term changes in exposure status. We found that 36 (78.3%) were then residing in the same exposure area we assigned in the present investigation according to residence at the beginning of pregnancy and 10 (21.7%) immigrated after that date into the municipality (8 towards the low-exposure area and 2 in the high-exposure area), thus suggesting a limited mobility of the study population across the different exposure areas.
Exposure misclassification might have also occurred due to additional sources of dioxins and of heavy metals in the study area, apart from the incineration plant, through different pathways of intake (inhalation, ingestion and dermal contact). However, concerning the first category of contaminants, in the study municipality no major industrial sources of dioxins were located such as electric arc furnaces, cement kilns, and copper and aluminium smelters, whilst the contribution of vehicle fuel combustion and of domestic wood burning to emissions is expected to be substantially lower that waste incineration and more evenly distributed [26, 27], thus suggesting that waste incineration was by far the major source of environmental contamination with dioxins in the study area, in line with other observations [28, 29]. Concerning exposure to heavy metals, findings from studies examining the specific burden of exposure attributable to waste incineration through measurement of biomarkers of exposure in nearby residents or occupationally-exposed workers are conflicting [30–32], and therefore the risk of exposure misclassification in our study area cannot be entirely ruled out.
Estimating if (and to which extent) exposure to the contaminants emitted by the incinerator actually changed during the study period as a consequence of the interruption in the activity of the incinerator is not easy. Few studies examined the kinetics of dioxins in humans, and half-life of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) was around 3 years in a Seveso population with a mean age of 24 years  and 7–9 years in other selected groups of adults . However, it is unclear if these indications may be extended to dioxins and related compounds other than TCDD, 'low-level' exposures might be much less efficient in the TCDD transfer to the foetus , and these kinetics appear also to be markedly influenced by the lipid status of the individuals , further complicating this issue. Overall, we estimate that the short-term interruption of waste incineration occurred in our study setting had limited effects on exposure status of the local population, also considering the prolonged half-life of dioxins in soils and in locally grown produces, and therefore we emphasize risk estimates based on the overall follow-up period for dioxin exposure. A separate analysis for each operation period might be more meaningful for heavy metals and for their potential teratogenic effects, since it seems likely that interruption of plant activity markedly decreased exposure to elements such as arsenic, lead and mercury having short half-lives in human blood, in the order of 3–40 days [10, 36, 37].
Results of the present study must be extended with caution to other contexts and particularly to older incinerators, to the relevant differences in amounts and types of contaminants which may be released into the environment by these plants, owing to the type of wastes combusted and the air pollution control technologies, as well as to potential differences in susceptibility of the exposed individuals. Moreover, we did not examine other reproductive issues apart from congenital anomaly risk such as altered male-to-female birth ratio, low birth weight or twinning, and therefore our investigation cannot be considered a comprehensive assessment of reproductive health in the exposed population.