Endocrine Disruptors in the Workplace, Hair Spray, Folate Supplementation, and Risk of Hypospadias: Case–Control Study

1Hypospadias is one of the most common urogenital congenital anomalies affecting baby boys. Prevalence estimates in Europe range from 4 to 24 per 10,000 births, depending on definition, with higher rates reported from the United States. Since the 1960s, prevalence appears first to have increased, then stabilized, along with increasing trends among other male fertility-related disorders. Little is known about the etiology of hypospadias. A role for endocrine-disrupting chemicals (EDCs) has been proposed, specifically antiandrogens. Animal studies suggest association of EDCs with various disorders related to male fertility, but data on humans are largely lacking. Studies of occupational risk factors have been inconclusive. Concerning diet and vitamin use, few data are available on effects of folic acid supplementation/antagonists, and two studies reported excess risk related to vegetarianism. We report here results of a large population-based case–control study of hypospadias in southeast England with primary aims of investigating risk of hypospadias and a) maternal occupational exposures, specifically to EDCs; b) folic acid supplementation during pregnancy; and c) vegetarianism.

2The study region included the health regions of North Thames, South Thames, and the Anglian part of Anglia and Oxford (Figure 1), comprising 120 London boroughs and local authority districts. Surgical centers in the study region and major surgical centers within 50 miles were visited. Hypospadias cases born in the study region over a 21-month period (1 January 1997–30 September 1998) were eligible if there was an abnormally positioned urethral orifice requiring surgery, with no major accompanying anomaly suggesting that it was part of a syndrome. Forty of 41 surgeons operating on hypospadias in the study region participated; 731 cases were identified from surgeons’ records and case notes. After initial contact by letter, up to two further invitations were sent. Of the 731 mothers, 610 replied, of whom 471 (77%; 64% of total eligible cases) agreed to participate.

3Controls born in the study region during the same period as the cases were randomly selected from the birth registry by the Office for National Statistics (ONS; London, UK). Ethical approval was given by the West Midlands Multi-centre Research Ethics Committee, Birmingham, United Kingdom. Following guidelines from the ethics committee, ONS asked the health authorities (now primary care trusts) to contact the general practitioners (GPs) of the control children. The GPs were then asked to pass on the invitation to the mother of the control child; in turn, the mothers were asked to contact the study team. In total, 1,568 control mothers were selected, but letters were not sent out to 81 mothers. From the possible 1,487 invitees, there were 758 replies, of whom 490 (65%; 33% of total eligible controls) agreed to take part. We attempted to contact a sample (n = 200) of the 729 mothers who had not replied and found that 144 (72%) had not received an invitation, mainly because the GP had not passed on the information (73%); the family had moved without a forwarding address (9.5%); or for other reasons. The cases and control mothers who agreed to participate were interviewed by telephone between September 2000 and March 2003.

4Case and control mothers were interviewed by telephone using a standard set of questions, with answers directly entered into a computer. The questionnaire included information on parental age, ethnicity, education, household income; family history of disease; pregnancy history; and maternal occupation, vegetarianism, folate supplements, smoking, alcohol use during pregnancy, and other questions related to diet history, vitamin use, demographics, and domestic and environmental exposures to chemicals.

5Approval for the study was obtained from the Multi-centre Research Ethics Committee and local research ethics committees in the study area, and participating mothers gave written consent before taking part in the study.

6To assess their occupational exposure to EDCs during the first 3 months of pregnancy, mothers were asked about their job title, department, company, their five main tasks, possible exposure to a list of 26 occupational substances (including hair spray, plastic fumes, cleaning agents such as disinfectants, solvents, paints and paint removers, printing ink, glue, heavy metal, welding and soldering fumes, anesthetics, cytostatics and antibiotics, and pesticides), and the hours per week that they were in contact with these exposures while at work.

7We used a job exposure matrix that included 348 possible job titles to classify job title, department, company, and main tasks into seven exposure categories assessed by a panel of occupational hygienists as to the likelihood of exposure to EDCs. We then dichotomized exposure into either exposed—including possible and probable exposure—or unexposed, blind to case or control status. The classes of EDCs were pesticides, polychlorinated organic compounds, phthalates, alkylphenolic compounds, biphenolic compounds, heavy metals, and other.

8We assessed correlations between variables using Spearman’s and kappa statistics. Exposure prevalence for cases and controls and unadjusted odds ratios (ORs) and 95% confidence intervals (CIs) were calculated. Chi-square tests were used to test for statistical significance.

9We performed multiple logistic regression in the statistical package R (version 2.2.0; R ). Variables were included in the multiple regression if they were statistically significant (p < 0.05) in the univariate model and improved fit of the multiple logistic regression model using Akaike Information Criteria. Positive family history of hypospadias and previous stillbirth were excluded from the multiple logistic regression models because among the control participants there were only one and two with the specific risk factor respectively. p-Values are uncorrected for multiple comparisons. Because of multicolinearity (kappa = 0.82), maternal occupational exposure to hair spray and phthalates were not entered together in the multiple logistic regression models. Income or level of education used to control for potential confounding by social class yielded similar findings; only models that included income are shown here. We also examined multiple logistic regression models that included maternal age.

10Table 1 shows prevalence of various risk factors and potential confounders for hypospadias. Gestational age and birth weight were correlated (Spearman r = 0.41), as were income and maternal age (r = 0.33) and maternal smoking and environmental tobacco smoke (r = 0.31); no other correlations exceeded ± 0.30. The hypospadias cases were more likely to be born preterm and to have low birth weight (< 2,500 g) (Table 1). Compared with mothers of control children, mothers of hypospadias cases tended to have lower income (p = 0.001), higher prevalence of smoking (p = 0.02), and lower prevalence of folate supplement use during the first 3 months of pregnancy (p = 0.01) (i.e., the critical period for development of hypospadias). For these variables, unadjusted ORs for the association with hypospadias ranged from 1.44 (95% CI, 1.05–1.97) for smokers compared with nonsmokers, to 2.74 (95% CI, 1.55–4.92) for household income < £10,000 compared with > £50,000, whereas folate supplementation during the first 3 months of pregnancy was associated with significantly reduced risk (OR = 0.64; 95% CI, 0.46–0.89) (Table 1). There was no association with vegetarianism (p = 0.36). Positive family history of hypospadias and previous stillbirth were also significantly associated with risk of hypospadias (not shown).

11Among occupational exposures (prevalence ≥ 5%), significantly increased risk was found for self-reported exposure to hair spray (exposed vs. nonexposed unadjusted OR = 2.30; 95% CI, 1.38–3.89) (Table 2). With the job exposure matrix, exposure prevalences were all < 5%; significant excess risk of hypospadias was found for boys of mothers exposed to phthalates (including hairdressers, beauty therapists/those working in beauty/hairdressing, research chemists, line operators, pharmaceutical operators, electrical assemblers, factory assistants) compared with those with no exposure to phthalates at work (unadjusted OR = 3.65; 95% CI, 1.19–11.20; 14 cases, 4 controls).

12After adjustment for multiple potential confounders, the significant positive association with maternal occupational exposure to hair spray was slightly strengthened (exposed vs. unexposed OR = 2.39; 95% CI, 1.40–4.17) (Table 3). Hairdressers as a group had a nonsignificant increased risk (unadjusted OR = 2.73; 95% CI, 0.72–10.38; adjusted OR = 2.59; 95% CI, 0.70–12.32). Folate supplementation during the first 3 months of pregnancy remained inversely associated with risk of hypospadias, with a virtually unchanged risk estimate (OR = 0.64; 95% CI, 0.44–0.93). With phthalates included in the model instead of occupational exposure to hair spray, the risk estimates for folate supplementation and other factors remained essentially unchanged; occupational phthalates exposure was associated with 3-fold elevated risk (OR = 3.12, 95% CI, 1.04–11.46) (Table 3). Inclusion of maternal age did not materially alter the risk estimates for the other variables in the model (data not shown).

13In this large case–control study, we found 2- to 3-fold increased risk of hypospadias among children of mothers exposed to hair spray and phthalates in the workplace during pregnancy and a 36% reduction in risk associated with folate supplementation during the first 3 months of pregnancy. These findings were robust to control for potential confounders. In contrast with two smaller studies, we did not find an association between vegetarianism/veganism and hypospadias.

14This is the first study to report a significant association between maternal occupational exposure to hair sprays, some of which may contain phthalates, and risk of hypospadias. A previous study reported that from 1980 to 1989, women hairdressers had slightly reduced risk of giving birth to a boy with hypospadias, whereas during 1992–1996, risk was significantly increased (OR = 1.50; 95% CI, 1.02–2.09). The risk was reduced after adjustment for parental social class (OR = 1.18; 95% CI, 0.80–1.64), suggesting possible confounding. In contrast, our study showed that adjustment for household income (or maternal educational level) as a proxy for social class did not materially affect the risk estimates.

15We are unaware of studies that have reported urinary phthalate metabolites in hairdressers or women applying hair sprays, so it is unclear to what extent their exposure may have been elevated at the time of the study. Phthalates, predominantly diethyl phthalate (DEP) and dibutyl phthalate (DBP), were present in many cosmetics including deodorants, fragrances, and nail and hair products. However, since 2005, certain phthalates including DBP have been prohibited for use in cosmetic products in Europe. The phthalates or their metabolites, for example, monoethyl phthalate and mono-n-butyl phthalate, are associated with androgen-lowering activities and abnormal Leydig cell function and have been linked to a decrease in anogenital distance in male infants; androgen lowering is associated with reproductive tract malformations including hypospadias. Inhalation contributes significantly to the uptake of these phthalates, which may explain some species differences for DEP between human and animal studies, where oral administration has mainly been used. A number of other substances included in hair sprays may have toxic effects if inhaled. They include polyvinyl alcohol, polyvinylpyrolidone, hydrofluoro-carbon, and propylene glycol, although for these substances the concern is acute effects on the cardiorespiratory system, skin, and eyes rather than on the reproductive system.

16This is also the first study to show a protective effect of folate supplementation on risk of hypospadias. A recent Dutch case–control study found no association of maternal folic acid supplements and hypospadias, nor was there an effect in a trial of folic acid/multivitamin supplementation based on small numbers of cases. However, in a Hungarian case–control study, use of dihydrofolate reductase inhibitors (folic acid antagonists) in pregnancy was associated with a (nonsignificant) 20% excess risk of hypospadias.

17Although we cannot exclude recall bias in our study, we believe it is unlikely to explain our findings, given that folate has not previously been associated with hypospadias, and, to explain the association reported here, the size of any such bias would need to be large. In the United Kingdom, 400 μg folate supplementation is recommended in the first trimester of pregnancy for the prevention of neural tube defects. Although we did not collect quantitative data on dietary intake of folate, the National Diet and Nutrition Survey estimated that mean intake of folate from foods ranged from 229 μg/day for women at 19–24 years of age to 255 μg/day at 35–49 years of age. Thus, folate supplementation is likely to have more than doubled the daily folate intake. Like the neural tube, the urethra is a midline structure. Biochemical, genetic, and epidemiologic observations suggest that folic acid may prevent neural tube defects by stimulating cellular methylation reactions, although this methylation hypothesis requires further exploration. Folic acid may also protect against other congenital anomalies such as orofacial clefts, cardiac, and urinary tract defects. In addition, recall bias for hair spray is unlikely, because only one previous study reported a possible association. Recall bias is not an issue for phthalate exposure, because it was assessed by job exposure matrix. However, some exposure misclassification is possible with the job exposure matrix because of uncertainty in expert assessment.

18Of other potential risk factors for hypospadias, some studies have suggested associations with occupational exposures of the father, including vehicle mechanics and exposure to solvents. In our study we did not collect information on occupation of the father. Family history and low birth weight have also been reported. Although possible associations with maternal smoking are inconsistent, recent reports suggest that if the father smoked, there was a higher risk of a boy being born with hypospadias than if the mother smoked, implying that environmental tobacco smoke (ETS) may play a part. In unadjusted analysis we found a borderline significant increased risk associated with ETS.

19Major strengths of our study include the large sample size, wide population-based coverage, and extensive interviewer-based questionnaire. Most previous studies of hypospadias have relied on routinely collected registry data with limited information on potential risk factors and confounders, and varying levels of quality control and completeness. We ascertained slightly more cases than the hospital registries and many more cases than the national congenital anomalies system. Our study also has limitations. One potential weakness is the low proportion of control women who replied to our invitation to participate. Because of constraints imposed by the ethics committee, we were not able to contact the women in the control group directly after they were randomly selected from the birth registry. Instead, we had to follow a convoluted procedure requiring both health authorities and the mothers’ GPs to forward our invitation pack, with the result that an estimated 72% of nonresponders never received the participant invitations. Although the controls appeared to be of slightly higher social class than the cases, adjusting for socioeconomic status (i.e., income or education) made no difference to our findings.

20Furthermore, we investigated more formally the potential for selection bias by socioeconomic status (SES) in selection of cases and controls using a weighting procedure analogous to poststratification for adjustment of item nonresponse in the survey literature. In brief, this involves reweighting the estimates of probabilities of exposure conditional on case/control status according to the distribution of SES in the target (i.e., unbiased population). The idea is that if, for example, individuals with low SES are underrepresented in the study, these estimates can be up-weighted using the SES distribution of the target population. We found no evidence of selection bias mediated by SES in the study.

21A particular difficulty affecting research on hypospadias is the wide variation in case definition, from mild displacement of the urethral orifice to severe anomalies requiring major corrective surgery. We relied on clinical judgment of surgeons who operate on cases, and thus excluded milder cases who were not referred for surgical correction. We also excluded cases where hypospadias was part of a wider syndrome. Thus our study is likely to have included a greater proportion of more severe, isolated cases than some previous studies.

22In conclusion, this is the first study to report increased risks of hypospadias associated with exposure to phthalates and hair sprays and a protective effect of folate supplementation. Previous association of vegetarianism with risk of hypospadias was not confirmed. Measurements of exposure to phthalates and/or biomonitoring may help to understand possible pathways of exposure and toxicology and provide quantitative estimates. Our findings with respect to folate use may have important implications for public health and prevention.