Time to pregnancy among female greenhouse workers

Objectives This study examined the possibility that work in greenhouses with potential exposure to pesticides entails a risk for reduced fecundity in terns of increased time to pregnancy. Methods Among 1767 female members of the Danish Gardeners Trade Union, telephone interview data were obtained on the 492 most recent pregnancies of women employed when they stopped contraception to get a child (the starting time). The pregnancies were classified according to job charactelistics at the starting time. The ratio between the likelihood of pregnancy during a month for the exposed persons versus the referents (the fecundability ratio) was estimated by discrete proportional hazards regression. Results The adjusted fecundability ratio for workers in flower greenhouses versus other union members was 1.1 1 [95% confidence interval (95% CI) 0.90-1.361. Among workers in flower greenhouses the handling of cultures many hours per week, the spraying of pesticides, and the nonuse of gloves was related to reduced fecundability [adjusted fecundability ratio 0.69 (95% CI 0.47-1.03), 0.78 (95% CI 0.59-1.06), and 0.67 (95% C10.46-0.98), respectively]. C ~ f l ~ l l l S i ~ f l S The findings suggest that female workers in flower greenhouses may have reduced fecundability and that exposure to pesticides may be part of the causal chain. Additional studies of fertility among women working in greenhouses are highly warranted.

During the past 2 decades it has become evident that occupational exposure to certain pesticides damages testicular function ( 1 -4 ) . With respect to women most interest has been directed towards possible developmental toxicity in the form of spontaneous abortion, birth defects, and still birth ( S ) , whereas only few studies have examined the reproductive toxicity of pesticides among women (6). Methodological limitations such as inaccurate assessments of the type, quantity, and timing of exposure are typical of earlier studies (6).
Time to pregnancy is the number of months it takes a couple to obtain a pregnancy after the discontinuation of contraception. This measure of fecundity has been used in several studies dealing with reproductive toxicity in the workplace (7)(8)(9). A study of fruit growers in The Netherlands indicated an increase in time to pregnancy after a seasonal increase in pesticide exposure of the male partner (10). In a multicenter study in France and Denmark no differences in time to pregnancy were found between pesticide-exposed male farmers and greenhouse workers and unexposed referents (1 1). A Canadian farm study included information on pesticide exposure for both husband and wife. No strong or consistent pattern of associations of pesticide exposure with time to pregnancy was found. However during exposure intervals in which women participated in pesticide activities, 6 of 13 pesticide exposure categories were associated with a decrease in fecundability (12). To our knowledge, the time-to-pregnancy values of the couples for which only the woman was exposed to pesticides have never been published.
In Denmark about 4000 women are occupationally exposed to pesticides in gardening. They are mainly exposed from the manual handling of cultures with pesticide residuals on the surface (called re-entry activity), but the spraying of pesticides may also contribute exposure. The use of pesticides is highest in greenhouses with flower production and lowest in greenhouses with the production of vegetables and outdoor gardening. The most used pesticides are insecticides, fungicides, and growth regulators, whereas herbicides are not used in greenhouses. Dermal exposure to pesticides is highly correlated with manual contact with pesticide-treated plants (13,14). Women typically perform the tasks involving the most plant contact (eg, nip cutting). Flower production in Denmark mainly uses pot cultures, which probably involve less exposure to pesticides than the production of cut flowers.
The purpose of this study was to examine whether work in greenhouses associated with dermal exposure to pesticides entails a risk for reduced fecundity among women.

Population
We studied female members of the gardeners' trade union in the 2 Danish regions with the highest density of greenhouses. Some 90-95% of Danish gardeners are members of the union. Our study population was identified in January 1995, and it included all female greenhouse or plant nursery workers aged 25-45 years with union membership during at least part of the previous 5 years. Altogether 1767 women were identified by name and address (figure 1). 1 Included in analysis, N-492

Questionnaire
A standardized questionnaire for a population-based survey of time to pregnancy was developed in the European Study on Infertility and Subfertility (15). The same questionnaire was used in the present investigation. A detailed description of the European study has been given elsewhere (15).

Data collection
The data were collected in a telephone interview during 1995 to 1996. The questionnaire data included reproductive history, time to pregnancy (for up to 4 pregnancies), as well as occupational and other exposures at the time when the couple started having unprotected intercourse to become pregnant (the starting time). For women not using contraception after a delivery, the starting time was defined by the return of menstruation. The key question on fecundity was phrased "How many months did it take you to become pregnant? That is, for how many months were you having sexual intercourse without doing anything to avoid a pregnancy?" As pregnancies we included all recognized pregnancies, but not induced abortions.

Exposure assessments
The potential exposure to pesticides was characterized by the following variables: Manual handling of cultures (hours per week): "How many hours per week did you have your hands in contact with plants or soil treated with pesticides?" The answers were dichotomized to at least 20 hours per week, yes;no. Use of gloves: "Did you use gloves when handling pesticide-treated cultures?" (a1ways;sometimes;never). Application of pesticides: "Did you spray pesticides?" (yes;no). The preceding variables were combined into an aggregated measure of working without gloves more than 20 hours per week or spraying (yes;no).

Study group
The most recent pregnancies were classified according to the job characteristic at the starting time as follows: (i) workers in flower greenhouses (N=253) (table 1) and (ii) a reference group, other members of the gardeners' union (N=239). The reference group included persons who were not gardeners at the starting time (N=198) or who worked in greenhouses not using pesticides (N=9) or who worked in greenhouses producing vegetables (N=13) or in outdoor gardening (N=8) or who worked in other gardener's jobs (N=l 1). The reference jobs were not associated with any significant occupational exposure to pesticides.
In addition we defined the indices of higher exposure for workers in flower greenhouses as follows: (i) manual handling of cultures at least 20 hours per week (N=220), (ii) not using gloves (N=213), (iii) applying pesticides (N=82), (iv) a combination of manual handling of cultures at least 20 hours a week, not using gloves and applying pesticides or handling cultures without using gloves more than 20 hours a week or applying pesticides (N=202).
Among those reached by telephone, 74.6% (909 of 1218) participated (figure 1). The age, number of children, and childlessness did not differ much between the respondents and the nonrespondents (table 2). At the time of the interview the workers in flower greenhouses were younger than the reference group of other union members, and a high percentage of workers in flower greenhouses had their most recent pregnancy after 1990. The workers in flower greenhouses differed in life-style from the reference group of union members; they smoked less, had a lower caffeine intake but used oral contraception more frequently. In the group of workers in flower greenhouses the high exposure group, defined by the aggregated information on exposure, had a high percentage of primipara pregnancies and a higher percentage of smokers than the low exposure group (table 3).

Statistical analyses
The time-to-pregnancy distributions were examined in the various groups. We compared exposed workers in flower greenhouses with other union members and with Table 3. Characteristics of the study p~pulation.~ subgroups of workers in flower greenhouses according to exposure. The analyses were performed for the first and the most recent pregnancy, only including a pregnancy if no contraception was used. Persons who were unemployed at the starting time were excluded.  members was slightly lower than in the analysis of the most recent pregnancy. In table 5 the unadjusted fecundability ratios are shown for the potential confounders included in the model.

Discussion
We found no difference in the fecundability ratio between exposed workers in flower greenhouses and other gardener union members. But for workers in flower greenhouses only the fecundability was decreased among the women doing tasks supposedly entailing a higher dermal exposure to pesticides. Pregnancies were excluded from the analysis due to birth control failure more often than among other gardener union members (14.3%) than among workers in flower greenhouses (7.9%). Highly fecund couples are more likely to have birth control failure, and as a consequence they were excluded from the study (17). Thus the fecundability in the reference group may have been biased towards values that were too low and therefore resulted in a bias in the risk estimate towards the null hypothesis. There were no big differences in the rates of induced abortions between the exposed workers in flower greenhouses and the other gardener union members.
We expected the other members of the gardeners' union to be a proper comparison group for the workers in flower greenhouses, but most of the other union members were not working in gardener jobs when they received their most recent child. They worked in factories, as cleaners, in public service taking care of children or old people or in offices.
The group of other union members included 41 women working as gardeners when they received their most recent child. They may have been exposed to pesticides although the exposure was probably low. Therefore the analysis comparing the exposed workers in flower greenhouses and the other gardener union members was also performed with the exclusion of those with other gardener jobs. This procedure did not change the fecundability ratio estimates.
Differential participation of pesticide-exposed women with reduced fertility would bias the results. In the introduction letter we did not specify explicitly that pesticides and time to pregnancy would be studied. The letter was kept in general terms about an investigation of occupation and reproductive health. In addition, the differences in fecundability were seen within the group of greenhouse workers, all exposed to pesticides, and with no knowledge of the exposure classification used in the study.
As we analyzed time to pregnancy conditional on pregnancy, women who had never achieved a pregnancy were not included. If pesticide exposure reduces fecundability, analyses of time to pregnancy may underestimate the effect.
Since low parity may be due to low fecundity, we analyzed data with and without controlling for parity. This procedure did not give rise to substantial differences in the fecundability ratios between the exposed workers in flower greenhouses and the other union members. For the greenhouse workers the fecundability ratio of the aggregated exposure variable declined to 0.59 (95% CI 0.40-0.81), but no substantial changes were seen in relation to the variables including only 1 determinant of exposure. By including parity in the model, we might have blurred an effect of pesticide exposure. But the analysis of the first pregnancy fecundability ratios were in the same range as those presented in table 4.
As an additional analysis we compared the time to pregnancy in our study (exposed workers in flower greenhouses and other union members) to a random sample of the Danish population. The population sample data were collected in 1991 by personal interview, using the same questionnaire as in this study. With the use of the same criteria for exclusion for the population sample, the exposed workers in flower greenhouses and the other union members compared with the population sample had a fecundability ratio of 0.84 (95% CI 0.73-0.97). This is an indication of lower fecundability among all gardener union members in comparison with a random population sample.
The main limitation of our study was the crude exposure assessment based on several indirect indicators of exposure and a lack of documentation of the exposure. As dermal exposure is by far the most important (13,14,18), it seems reasonable to assume that the number of hours in contact with plants reflects the level of exposure. At least 20 hours per week was chosen as an indicator of at least half-time dermal contact to cultures. The use of gloves when re-entry activities are performed can only be expected to lower exposure if used correctly, and for sprayers the incorrect use of gloves may in fact increase dermal exposure (19). The use of gloves when reentry activities are performed has become more widespread during recent years, and the awareness of risks associated with improper use may have increased. Thus it is reasonable to assume that the use of gloves during re-entry activities does decrease exposure. As the use of gloves has increased during recent years, the result achieved in the analysis of the use of gloves may have, in part, been a consequence of time trend bias (20). No exposure data exist to handle this potential bias.
Hoekstra et a1 (18) found that, compared with re-entry activities, spraying doubled the exposure per workhour. In this study the mean spraying time per week among sprayers was 5.5 (SD 8.0, median 2) hours. Thus we expected exposure from spraying activities to Scand J Work Environ Health 2000, "0126, no 2 contribute less to the total exposure than exposure froin re-entry activities. This assumption may be in accordance with the lower fecundability among the exposed persons classified according to the use of gloves than among the exposed persons classified according to the application of pesticides. The misclassification achieved by the use of proxy variables, and the fact that workers are exposed in several ways, would probably bias the results towards the null hypothesis. Part-time work may be related to family size, and part-time workers are also less likely to be highly exposed. Therefore the analysis of hours per week manually handling cultures and the aggregated exposure measure were restricted to full-time workers (75%). The fecundability ratio of manual handling was 0.55 (95% CI 0.33-0.92), and the fecundability ratio of aggregated exposure remained almost unchanged 0.63 (95% CI 0.40-0.99).
Most of the pregnancies among the exposed occurred after 1990. Thus the findings in the study can probably not be explained by earlier high exposure, unless pesticides accumulate in the body or have an irreversible effect. Some organochlorides such as lindane and endosnlfane have been used in greenhouses within the last few years, and they tend to interfere with the homeostasis of reproductive hormones. Disruption of endocrine regulation may interfere with many of the biological processes involved before a pregnancy is recognized (21,22).
The fungicide thiram and the insecticide amitraz in animal testing block ovulation. And the fungicides carbendazin and benomyl interfere with spindle formation, effects that, in humans, may decrease fecundity (23).
In conclusion our findings indicate that exposure to pesticides among women working in flower greenhouses may lead to reduced fecundity. Additional studies of fecundity among women working in greenhouses are highly warranted. Efforts to achieve measurements of exposure would add to the reliability of the results.