Organic solvents are widely used in industry and exposure is common for workers in manufacturing and processing operations. Many organic solvents are known teratogens and embryotoxins for laboratory animals [1], and solvent abuse in pregnancy is associated with an increased risk of major foetal anomalies [2]. However, the epidemiological evidence concerning developmental effects in humans due to maternal exposure in occupational settings is still weak and ambiguous. Literature reviews have suggested an association between high exposure to some organic solvents and the risk of foetal loss [3, 4]. However, the findings concerning the impact of maternal occupational exposure to organic solvents on the risk of congenital anomalies have not been consistent [3, 5–10]. Studying this issue is complicated by the fact that often exposure involves a mixture of solvents and is therefore difficult to classify. Currently, ingestion of ethyl alcohol is the only single organic-solvent exposure with well-documented teratogenic effects in humans [11–13].

Due to the challenge of capturing a large number of outcomes, published studies rarely have a prospective cohort or controlled study design [14]. This inherent weakness and the above mentioned classification difficulty add to the likelihood of publication bias and null findings. Studies with a case-control design have reported associations between maternal organic-solvent exposure and being born with oral clefts, digestive anomalies, multiple anomalies, gastroschisis, and with neural tube defects [15–21].

In Russia, the proportion women working in industry has traditionally been high. In the borough of Mončegorsk, located in the far north-west of the Russian Federation (RF), the main employer for women has been the local nickel refinery and associated operations. Some of the jobs, as in other enterprises in the community, involved working with organic solvents on a daily basis. Further, all births in the borough since 1973 have been systematically registered retrospectively [22]. In the present study we made use of this established birth register.

The aim of this register-based prospective study was to assess the risk of non-chromosomal congenital anomalies for singleton newborns to women employed as painters and spoolers, which are jobs that have been related to a relatively high exposure to organic solvents on a frequent basis.

The prevalence of malformations appeared to be higher among newborns to women employed as painters in early pregnancy; mainly due to the higher proportions of malformations in the circulatory system and genital organs. However, the number of exposed was not large enough to detect a statistical difference in risk between the exposed and unexposed groups, and adjusting for the included risk factors did not change that finding. Nevertheless, the point estimate of the overall risk had a higher precision in the present study than that reported in previously published cohort studies [6, 8, 9].

During the embryonic and early foetal periods, susceptibility to induction of malformations of the circulatory, genital and musculoskeletal systems may be considered high. Animal studies with benzene, toluene and xylene, but not acetone and Stoddard solvent, suggest only the occurrence of skeletal malformations [28–30]. Defects of the circulatory and the genital systems were not reported, and perhaps were outside the scope of the experiments conducted. From the mechanism-of-action perspective, it is pertinent that metabolic activation of the three aromatic solvents is known to generate similar reactive species that are capable of causing oxidative damage and they thereby possess teratogenic potential [26].

Few studies of the association between organic solvent exposure and specific malformations or groups of malformations are available that focus on occupational exposures to aromatic hydrocarbons during pregnancy. A case-control study by McDonald et al. (1987) found an 8-fold increase in overall risk to offspring of women exposed to aromatic solvents, mainly toluene. The increased risk primarily concerned gastrointestinal and renal-urinary defects [31]. Exposure to toluene was also suggested as the explanation of the higher odds of oral cleft in newborns to women working in the leather and shoe industry [32, 33]. An increased risk of oral clefts has also been associated with several different classes of organic solvents [16, 18]. A post-hoc analysis of oral clefts in our study gave an adjusted OR of 1.9 (95% CI: 0.4-8.4), but there were only two such outcomes in the exposed group.

In the few prospective studies published, only the effect of exposure to a mixture of organic solvents has been examined. Both overall associations [5–7] and their absence [8, 10] have been reported. The finding of an overall association with non-chromosomal malformations, primarily reflecting the occurrence of oro-facial clefts and urinary, male genital [7] and musculoskeletal malformations [5], support our results (urinary malformations excluded). However, it was unexpected that malformations of the circulatory system appeared to have the largest difference in prevalence between the exposed and unexposed in our study. Most of the circulatory malformations in the exposed group were of the aortic and mitral valves (Q23). In case-control studies, ORs for cardiovascular malformations among the solvent-exposed subjects of 1.3 (95% CL: 0.8-2.2) [29] and 1.0 [17] have been reported. Furthermore, studies involving pregnant solvent abusers, who expose their foetuses to very high concentrations in each binge episode, have reported strong teratogenic outcomes. The latter had features of foetal alcohol syndrome (i.e., mainly microcephaly and craniofacial abnormalities) [2, 34].

Prevalences of malformations in different populations are not readily comparable [35, 36]. However, the overall rate in the study population was similar to that of Norway [37, 38], which suggests that the general perinatal detection of malformations at birth in Mončegorsk reflects comparable identification protocols. Some malformations, such as those of the genital organs and the circulatory systems, may be misclassified or under-ascertained before discharge from the birth clinic [35, 39, 40]. This would have influenced the estimated prevalences, but it is unlikely that the diagnostic procedures and specifications, or recording of cases in the delivery records, would have been differential between the exposed and unexposed. The malformations of the heart and other circulatory anomalies constituted 21 percent of all malformations in the exposed group, compared to 11 percent in the unexposed. Consequently, it is unlikely that the explanation of this difference was a differential likelihood of being diagnosed during the stay in the clinic. On the other hand if occupational organic-solvent exposure elevated the risk of malformations that inherently are difficult to detect in the early neonatal period, and/or of malformations that lead to termination of pregnancy after a prenatal diagnosis, the overall relative risk between the exposed and unexposed in our study would have been underestimated.

Several studies have reported an elevated risk of spontaneous abortions among women exposed to organic solvents [12]. If an increased incidence of anomalies promotes such an outcome, or terminations occur due to a pre-natal diagnosis of a malformation, the actual risk of malformations would only be revealed if spontaneous and induced abortions were included. No such association was observed with organic solvent exposure in our previous study of spontaneous abortion among nickel-exposed workers in Mončegorsk, but a weak risk could not be ruled out [41]. The incidence of induced abortions in Mončegorsk due to a pre-natal diagnosis of a severe malformation (including chromosomal) was 17/1000 births in the period 2000-2004, while the incidence of late spontaneous abortions (12-27 weeks of gestation) was 5/1000 births (based on clinical figures obtained from the hospital in Mončegorsk).

Due to the rareness of the specific malformations, studies tend to collapse them into one outcome to gain numbers [6, 8, 12]. This weakens the specificity of the statistical analysis because of the heterogeneity of the malformations. Mixed exposure (to more than one organic solvent) constitutes a similar limitation. Studies tend to treat them as a single group since they all cross the placenta easily and tend to be used as mixtures, or interchangeably in occupational settings. Since different malformations have different aetiologies, and perhaps solvent specificities, an overall increase in risk due to exposure to one specific organic solvent would be low. Thus, the likelihood of detecting a change in overall risk can be expected to be higher when exposure is to several solvents, and this pertains to our study.

Another possible limitation relates to confounding causative factors not adjusted for. However, diabetes mellitus and relevant infectious diseases were rather rare in the study population [25]. The same was observed for alcohol abuse, which was highly correlated with observed smoking habits. Post hoc inclusions of diabetes and alcohol abuse in the regression models did not change the parameter estimates for the exposure variable. The peri-conceptional folate status of the women in Mončegorsk, as in the RF, was not known, and prophylactic measures involving folate did not exist during most of the study period. However, there were no neural tube defects in the exposed group, which suggests that such participants did not have a poorer folate status than the non-exposed.

All confounders included were positively associated with the outcome. The year of birth (OR = 1.08; 95% C.L.: 1.05, 1.12), which was included in the model to adjust for structural and systematic changes over time, indicates that the prevalence of malformations increased slightly over time. Concominantly, the minimum duration of clinical observation before discharge was gradually shortened (from 7 to 3-4 days), and the number of women working as painters diminished from the late 1980s on. The regulations concerning work that may be hazardous during pregnancy remained the same throughout the study period. It is to be noted that the occupational regulations pertaining to hazardous substances during pregnancy remained more or less the same throughout the study period. On the other hand, the relative use of the different solvents used has changed over the years. For example, it is known to the authors that the use of benzene by the painters ceased after the dissolution of the Soviet Union, and that of glycol ethers appears to have increased. There have also been changes in the clinical practises over the years, with some improvements in detection and diagnosis of malformations during the perinatal period. These developments might have contributed to the increased prevalence over time.

The prospective nature of the exposure and outcome data was a strength of our study. The results were therefore not affected by selection bias based on exposure, nor by recall bias as in case-control studies. In addition, the size of the study cohort facilitated the inclusion of a homogenous group of workers from a well-defined population, permitting the univariate analyses of subgroups of congenital anomalies. An association between paternal exposure to organic solvents and an increased risk of malformations has only been been reported for neural tube defects [42]. Such an association was therefore unlikely at play in the present study.

In the exposed group, we included only women in jobs known to entail high daily exposure to a mixture of organic solvents. However, exposure misclassification remained a possibility. It constitutes a weakness in all studies of maternal occupational exposures that deduce exposure from maternal job title [43]. A mandatory job transfer for women in potentially hazardous jobs usually took place after the first antenatal visit, and the mean gestational age for this was 14 weeks (59% before 13 completed weeks). The critical window of suceptibility for impaired organ development varies between organs, but tends to be periconceptional and within a short period of organogenesis. Thus, some of the women in exposed jobs may have been absent from work during the criticial period, or were transferred to another job prior to the critical time for some types of malformations. Such a misclassification of exposure would have biased our findings towards the null, and impacted the diffent categories of malformations differently.

There appeared to be a higher risk of malformations of the circulatory system and genital organs at birth among newborns to women in occupations with solvent exposure during early pregnancy (predominantly employed as painters). However, the findings were not statistically conclusive. Considering that these two categories of malformations are not readily diagnosed perinatally with late manifestations a possibility, the difference in prevalence between the exposed and unexposed may have been underestimated.

^a Plasterers, hairdressers, manicurists, pedicurists, decorators, electricians, dry cleaners and cleaners; and petroleum, agriculture, shoe factory and laboratory workers.

^b Welders, nurses and industrial machine operators; gas, electronic, building, construction, textile, chemical and military workers.