Fertility problems are an increasingly important public health issue in industrialised countries. The incidence of subfertility, defined as the inability to conceive within one year, is estimated to be 15%.1 2 In approximately half of the cases, there seems to be a male problem.1 It has also been suggested that health problems affecting the male genital tract, such as hypospadias, cryptorchidism, testicular cancer, and a decline in sperm count, have increased during the past 50 years.3 The hypothesis was put forward that the widespread exposure to environmental endocrine disruptors, such as some pesticides, might be an explanation for these adverse effects on male reproduction.3

Several investigators studied the effects of pesticide exposure in occupational settings on time-to-pregnancy (TTP) as a measure of reproductive capability. De Cock et al, who studied male fruit growers in The Netherlands, revealed an adverse effect of pesticides when these were used solely by the owner or with a low spraying velocity.4 Sallmen et al5 conducted a prospective study on greenhouse and garden workers and found that exposure to pyrethroids was related to decreased fecundability, while a study in Italy observed an increase in the risk of conception delay among greenhouse workers with high exposure to pesticides.6 On the other hand, some investigators did not observe an association between pesticides exposure and fertility. Thonneau et al found no increased TTP in male vineyard and greenhouse workers nor in farmers exposed to pesticides.7 Similar results were found by Larsen et al,8 who investigated time-to-pregnancy among traditional and organic farmers and by Curtis et al,9 who conducted a study on male fertility among farm occupants in Canada. In conclusion, the results of these studies differ considerably. Therefore, no uniform conclusions can be drawn about the effects of pesticides on male fertility.

One of the limitations of previously published studies is that the participants, mostly farmers, were only exposed to pesticides during the spraying season. In contrast, greenhouse workers, who are the subject of this study, are exposed throughout the whole year with a period of increased exposure during the summer. Greenhouse workers have relatively high exposure because they work in closed environments and, in the case of floriculture workers, they perform numerous tasks that require fine handwork for which gloves cannot be worn. Therefore, the aim of this study was to determine whether time-to-pregnancy is prolonged for male greenhouse workers in comparison with a non-exposed reference group.

METHODS

Population

We used three different approaches to obtain the study population (see fig 1). From the database of a major trade union, which registers information about a variety of Dutch workers, we identified all 1536 male greenhouse workers and a random sample of 2350 male cleaners of reproductive age (born in 1956–79) in 2002. In addition, we obtained addresses of 2414 greenhouse owners through the national database of an agricultural marketing company. These owners were born in 1950–79 or their date of birth was unknown. For the reference group, addresses of 4691 shopkeepers and market stallholders were obtained through the Chamber of Commerce. We only selected companies in which exposure to pesticides or other reproductive toxicants was unlikely (for example, builders’ merchants and shops for household goods). This reference group of cleaners, retail shopkeepers and market stallholders was chosen because of assumed comparability with greenhouse workers and owners with respect to educational level, socioeconomic status and working conditions, such as standing and carrying heavy loads.

Data collection

The 3886 men selected through the trade union database were sent an introductory letter, an information leaflet, and a self-administered questionnaire by mail. To the 7105 addresses obtained through the marketing company and the Chamber of Commerce, a mailing was sent which contained an introductory letter, an information leaflet and a self-administered questionnaire for the owner of the company, plus the same items for the partner of the owner if he or she worked in the same company. If not, we asked the owner to give the extra questionnaire plus information leaflet to an employee of the opposite sex. We also asked the owners for addresses of employees in the company so we could send them questionnaires as well. Alternatively, the owners could choose to personally distribute questionnaires to their employees. Only male owners and employees were included in this part of the study. To increase response rates, reminders were sent after two and six weeks. To avoid selective non-response and information bias, the study was presented to the potential participants as a general study on work, lifestyle and fertility not especially related to pesticides. We focused on the most recent pregnancy to avoid dependency issues of multiple pregnancies, assuming that the most recent pregnancy would be remembered best. This led to a mixture of first and subsequent pregnancies in the database.

Questionnaire

The design of the questionnaire was based on the literature and on previous reproductive studies performed by our research group.10^–12 In the questionnaire, men were asked whether their wife or female partner had ever been pregnant. If so, they were asked detailed questions about work (pesticide application, re-entry activities, and personal protective equipment), lifestyle factors (for example, smoking habits, coffee and alcohol consumption) and their partner. All questions related to a period of six months before the start of the most recent pregnancy to approximate the entire TTP-period, assuming that most men did not change jobs during this period. In addition, questions were asked about time-to-pregnancy and adverse pregnancy outcomes (for example, spontaneous abortion, preterm delivery and congenital malformations).

Non-response

Non-response was investigated at the end of data collection. We randomly selected 150 greenhouse workers and 150 cleaners from the trade union, 200 greenhouse owners from the marketing company, and 200 shopkeepers from the Chamber of Commence who did not return the questionnaire. These non-participants were called and asked for the reason why they did not respond and whether they had any children. When they did, we tried to ask questions about TTP and outcome of the most recent pregnancy.

Time-to-pregnancy

The primary outcome in this study was time-to-pregnancy, as a measure of fecundability. The question concerning TTP was phrased as follows: “For the most recent pregnancy, how many months did it take you and your partner to get pregnant (counted from the moment you and your partner stopped using contraceptive methods or started trying to get pregnant)?”.12 Only planned pregnancies, not caused by a failure of the birth control method used and regardless of pregnancy outcome, were included in the analysis. TTP was censured for couples with a waiting time longer than 14 months.

Exposure

To account for job changes and administrative errors in the original population files, all participants were classified as “exposed” (greenhouse and other horticulture workers with likely exposure to pesticides), “non-exposed” (cleaners, market stall and retail workers, and other workers not exposed to pesticides), or not working outside the home, based on reported occupation for the six-month period prior to the most recent pregnancy or pregnancy attempt. Work activities reported in the questionnaire, such as pesticide spraying and specific work with flowers, were also used as indicators for pesticide exposure.

Potential confounding factors

Potential confounding or effect modifying factors were recorded for both men and women. Smoking habits, alcohol and coffee consumption, and the use of oral contraceptives during one year preceding the most recent pregnancy were used dichotomously in the analyses (yes or no). Educational level was categorised as low (no education, primary school only, lower vocational education, or lower general secondary education) or high (intermediate vocational education, higher general secondary education, pre-university education, higher vocational education, or university) and the work status for women as working or not-working. Gravidity, divided into first (primi), second (duo), and third or subsequent pregnancies (multi), was considered as a potential confounder or effect modifier as well. Both male and female age were used as continuous variables in the analyses.

Analyses

The data from the questionnaires were put into a MS Access database and statistical analyses were performed using Statistical Analysis Systems (SAS) version 8.2. The time-to-pregnancy data were analysed with discrete proportional hazards regression analysis using the PHREG procedure and EXACT handling of ties.13 The resulting fecundability ratio (FR) represents the fecundability of the exposed group relative to the non-exposed group. Initially, crude FRs with 95% confidence intervals were calculated. Following, multivariable analyses were performed with confounder correction and checks for interaction between pesticide exposure and all relevant covariables. We used two criteria to select potential confounding factors: the distribution of the variable among the exposed and non-exposed groups and the relation of the potential confounder with TTP in the non-exposed reference group. The complete potential confounder set was then evaluated by reducing the full model with all potential confounders to gain precision without loosing validity. Only variables that did not change the FR of pesticide exposure by more than 0.1 upon removal were permanently deleted from the model.

Exposure model

The largest part of the study population (approximately 80%) consisted of floriculture workers which form a relatively homogeneous group concerning pesticide exposure, as they have to apply to the same rules for pesticide use which differ substantially from those for growers of edible crops. Therefore, an exploratory discrete proportional hazards regression analysis was carried out among the subgroup of floriculture workers (n = 541) to explain TTP from a number of work-related variables. These variables were asked for in the questionnaire and defined as: function in the company (employee/employer); number of working hours (continuous); average pesticide use per month (continuous); type of pesticides used: fumigants, fungicides, herbicides, and insecticides (yes/no); spraying pesticides (yes/no); kind of work activities: cultivating flowers, cutting flowers, preparing flowers for sale, and general activities without touching flowers (yes/no); type of flowers: alstroemeria, amaryllis, anthurium, carnation, chrysanthemum, freesia, gerbera, iris, lily, orchid, rose, tulip, summer flowers, plants and other flowers (yes/no). An entry probability of <0.10 and a removal probability of >0.15 were used in this forward stepwise model.

RESULTS

The composition of the study population and the pregnancy history of the final participants are depicted in figure 1. Of the 4127 employees and 7105 company owners invited for the study, at least 837 men proved to be ineligible because they were not a member of the trade union anymore, had another occupation, or were too old. Completed questionnaires were received from 1222 male greenhouse workers (31%) and 1235 men in the reference group (19%). Out of these 2457 participants, 619 men (25%) whose partner had never tried to become pregnant were excluded from the analysis. All other participants were categorised as greenhouse worker or belonging to the reference group based on the actual occupation held in the six-month period before the most recent pregnancy. This implied that some men from the original greenhouse worker group were shifted to the reference group and vice versa. A total of 96 men were excluded because they did not work outside the home in that period. Eventually, the exposed group consisted of 875 greenhouse workers: 851 men whose wife or girlfriend had been pregnant at least once and 24 men who did not conceive (table 1). The reference group was approximately the same size as the exposed group (n = 867), but included twice as many men who did not succeed in conceiving.

Table 2 shows the general characteristics for couples with one or more pregnancies who filled in the question about TTP of the most recent pregnancy. Couples who did not get pregnant (n = 72) or had an unexpected pregnancy (n = 198) and couples with a missing value on TTP (n = 165) were excluded. Men and their wives in the reference group were slightly younger then those in the greenhouse worker population, a larger number of them smoked, and fewer drank alcohol. Furthermore, the wives in the reference group were less educated and more often worked outside the home compared with the wives of the greenhouse workers. Twice as many pregnancies in the reference group were first pregnancies (22%) compared to those in the exposed group (10%), whereas the number of miscarriages was slightly lower in the reference group. Of all couples who had children, those in the exposed group on average had more children than those in the reference group (2.5 vs 2.0).

The crude fecundability ratio (FR), comparing greenhouse workers to the reference group, was 1.12 (95% CI 1.00 to 1.26) (table 3, model 1). The potential confounding factors in our data were age, educational level, and work status of the women, as well as smoking and coffee and alcohol consumption of the men. The female lifestyle factors were also different between exposed and non-exposed, but because of strong correlations between men and women for these factors, only the male lifestyle factors were included in the analyses. Possible interaction was also investigated for smoking and coffee and alcohol consumption of the men, but clear interaction effects were only found for gravidity (p = 0.02). Stratified models for primi-, duo- and multigravidity are shown in table 3. The fecundability for couples with a first pregnancy was decreased in the greenhouse worker group in comparison with the non-exposed reference group (crude FR = 0.72 (95% CI 0.52 to 0.99), model 2a). For couples who already had one (duogravidity) or more (multigravidity) children at the beginning of their most recent pregnancy, greenhouse workers seemed to be slightly more fecund than reference workers, which is reflected by crude FRs of 1.10 (95% CI 0.92 to 1.31) (model 2b) and 1.33 (95% CI 1.10 to 1.60) (model 2c), respectively. Correction for confounding decreased these FRs to 0.65 (95% CI 0.46 to 0.92) for primigravidous couples (model 3a), 1.09 (95% CI 0.90 to 1.33) for duogravidous couples (model 3b), and 1.17 (95% CI 0.96 to 1.43) for multigravidous couples (model 3c).

In addition, an exploratory analysis was done among the subgroup of floriculture workers to explain TTP from a number of work variables. Out of the 26 work variables, only function in the company (employee or employer), number of work hours, cultivating alstroemerias, and cultivating amaryllises remained in the model (table 4). Because the proportion of primigravidous couples was different between employees (22%) and employers (5%), however, we also entered gravidity in the model. After adjustment for primigravidity, only cultivating alstroemerias (FR = 0.69; 95% CI 0.47 to 1.02) and cultivating amaryllises (FR = 0.39; 95% CI 0.14 to 1.03) appeared to be associated with prolonged TTP in the final exploratory model.

DISCUSSION

The crude analyses did not show a decreased overall fertility among greenhouse workers, assumed to be exposed to pesticides, compared to the non-exposed reference group. However, when fecundability was assessed for primigravidous couples, duogravidous couples and multigravidous couples separately, greenhouse workers were found to be less fecund when trying to conceive their first pregnancy. Among couples who already experienced one or more pregnancies, no association was seen between pesticide exposure and TTP after adjustment for potential confounders. Before further evaluating these findings, however, the potential limitations of the study should be considered.

Possibilities for selection bias were present in several phases of the study. The response rate in general was low, but in particular among market stallholders and retail shopkeepers. This may be due to the selection of owners of small companies who were too busy to fill out the questionnaire. Alternatively, our questionnaires could have been labeled as junk mail in large companies. Moreover, the company owners were mostly men, who are known to have low response rates in reproductive studies.14 As selection for age was not possible for the reference group, a relatively large number of people could also have been too old to see the relevance of a questionnaire on reproductive issues. However, as the study was presented as general research on work and lifestyle habits, not putting any emphasis on pesticide exposure, greenhouse workers with adverse pregnancy outcomes had no other reasons to participate than cleaners, market stallholders, or retail shopkeepers with similar problems. This strongly reduces the chances of selection bias.

Because the response rate was low, a non-response study was done. We successfully contacted 243 male greenhouse workers and 200 male cleaners and shopkeepers. From these, 128 greenhouse workers and 79 cleaners and shopkeepers were willing to answer questions about the most recent pregnancy by telephone. The median TTP among both non-responder groups was 2.0 months, whereas the median TTP in both participant groups was 3.0. This means that slightly more men with a long TTP participated in our study, but this is seen in greenhouse workers as well as in the reference group. All in all, we have no reason to believe that non-response biased our results to a large extent.

Several different other types of bias, described by Weinberg et al, may occur in studies on time-to-pregnancy.15 One source of bias was excluded a priori in our study, as we prevented medical intervention bias by censoring TTP at 14 months. Moreover, time trend bias and behaviour modification bias seem unlikely, because it is rare for people in The Netherlands to change jobs, work habits, or behaviour during their waiting time to pregnancy. We have no reason to believe that pregnancy recognition bias, planning bias or wantedness bias were present in our data either, because the percentages of recognised spontaneous abortions (25% and 22%) and unintended pregnancies were comparable between greenhouse workers and the reference group, whereas the FR excluding the cycle one data (FR 1.06; 95% CI 0.93 to 1.20) is about the same as the FR including the cycle one data (FR 1.12; 95% CI 0.99 to 1.24). The number of unsuccessful pregnancy attempts was somewhat higher in the reference group (3.9%) than in the greenhouse worker group (2.2%), which could have led to sterility bias as couples who did not conceive were excluded from the analyses. However, the crude FR for pesticide exposure including couples that did not yet conceive was not different from the FR in the original analysis (FR 1.13 (95% CI 1.01 to 1.26) vs FR 1.12 (95% CI 1.00 to 1.26), respectively). This is in agreement with the hypothesis that pesticides would reduce fertility, but not cause complete sterility.

In this study, data were collected by means of questionnaires. This implies a probability of misclassification on the outcome measure, exposure and confounder variables. Not everyone in the study was able to produce an adequate TTP value, but in general TTP data are considered valid estimates of the waiting time to pregnancy.16 17 Therefore, any misclassification on the outcome measure could be considered non-differential leading to underestimation of the fecundability ratio. Concerning exposure, no distinctions could be made in amounts and kinds of pesticides used as many different pesticides were reported, some of which may not be reproductive toxicants. Moreover, not all greenhouse workers may actually have been exposed to relevant concentrations of pesticides, so the effects of pesticide exposure on TTP may have been underestimated. On the other hand, we assumed a link between working in greenhouses and exposure to pesticides, whereas greenhouse workers are also exposed to other risk factors, such as high temperatures, which we did not measure. Therefore, we cannot be absolutely sure that the relation between working in greenhouses and TTP reflects a causal link between pesticide exposure and TTP.

The major confounders for the relation between pesticide exposure and TTP mentioned in the literature were recognised and controlled for in our dataset. Apart from the potential confounders described before, we evaluated several female lifestyle and work-related factors, such as pesticide use, in sub-analyses, but none of these factors changed the results. However, residual confounding as a result of uncontrolled confounders and imperfect measurement of some confounders that we did control for may still play a role, as well as the fact that we did not correct for general health. Greenhouse workers in The Netherlands appear to have better general health, reflected in a better life expectancy in the regions with high greenhouse activity,18 which may also result in a higher reproductive capacity. If this were the case, the effects of pesticide exposure are underestimated, because the normal value of fecundability among greenhouse workers compared to the less healthy referent population would be above one.

The overall results found in this study are not in accordance with the hypothesis which suggests that pesticide exposure would reduce fertility, as we observed a prolonged TTP in primigravidous couples only. However, the study of Sallmen et al showed the same results.5 Male exposure to pesticides was associated with decreased fecundability among primiparous families (FR 0.3 (95% CI 0.1 to 1.0) and FR 0.4 (95% CI 0.2 to 0.8) for high and low exposure, respectively), and the FRs for male exposure to pesticides among parous families were leaning towards improved fecundability. Olsen stated that the main problem in non-experimental research of reproductive failure is that most pregnancies are planned and that past pregnancy experience is used in planning.19 Past reproductive failure might reduce risk behaviour when a new pregnancy is planned. To prevent this confounding by past experience, Olsen suggests restricting studies to first pregnancies only or using past reproductive experience to form groups that are comparable in terms of pre-exposure risk.20 Pregnancy history goes further than just parity and pregnancy order, as it will also take into account the desire for a given family size and the results of previous pregnancies. In our study, the average number of children was higher among greenhouse workers than among the reference group, but we did not collect any other data on pregnancy history. All this favours the analysis based on first pregnancies only, which gives a better representation of the pure fertility status in male greenhouse workers than the analyses of all pregnancies. Some other studies also based their analyses or a sub-analysis on the first pregnancy among pesticide exposed male workers and reported an association between pesticide exposure and prolonged TTP among primigravidous couples,6 7 which is in accordance with the findings in this study. In addition, a decreased FR for women working in flower production companies was found in a large study of first pregnancies in Colombia.21

In the first exploratory exposure model we observed that employers had a shorter TTP than greenhouse employees. However, primigravidity explained this difference in fecundability between employers and employees. After adjusting for gravidity, only cultivating amaryllises and/or cultivating alstroemerias were maintained in the final model, pointing towards longer TTPs for men who cultivate amaryllises and/or alstroemerias compared with men who cultivate other kinds of flowers, despite the small numbers of men involved. This may be because of the use of large amounts of granules with a high dose of active components against snails and large amounts of fungicides in these cultures at the same time.

In conclusion, we could not demonstrate an overall effect of pesticides on TTP among male greenhouse workers. However, prolonged TTPs were observed for male greenhouse workers who were trying to conceive their first pregnancy, which is the most valid analysis in which pregnancy planning issues are avoided. Indications were also found for reduced fertility associated with cultivating flowers which require the use of high doses of specific pesticides.