This is the first study to investigate the effect of any occupation on TTP in Africa and the first to do so for domestic workers in any setting. The results showed that domestic workers took significantly longer time to fall pregnant when compared to administrative workers, suggesting that domestic work may have negative effect on fecundity.
The finding is supported by a previous TTP study in the same population, which suggested that that fecundity may be significantly reduced for domestic workers. The study found increased TTP for employed women when compared to unemployed women [16]. Although the sample size for that study was too small to allow for occupation-specific analysis, a description of the occupation of the employed women showed that majority (40 %) were domestic workers.
It is well known that TTP data collected by means of a questionnaire provide valid estimates for how long it takes a woman to conceive [1, 18]. Occupation and other co-variates at the time of pregnancy attempt are unlikely to have been misclassified as women are well able to remember their reproductive events and the context around them [18]. Also, this study was based on the most recent pregnancy – 56 and 76 % of most pregnancies were in the last 5 and 10 years respectively. TTP recall of up to 20 years have been validated in Europe [1, 18, 19], and the high validity and reliability of the questionnaire used in this study has been previously reported [17]. However, the possibility of certain biases in TTP studies cannot be entirely excluded [1, 6, 20]. Therefore, we discuss the alternative explanations and plausible causal explanations for the finding. First alternative explanations, due to the potential role of confounding and bias, are discussed. Then plausible causal explanations for the results are discussed.
Confounding
The crude analysis showed a decreased fecundity for domestic workers in comparison to administrative workers (unadjusted FR = 0.46; 95 % CI 0.29–0.72). The results remained valid even after adjusting for six maternal confounders (adjusted FR = 0.53; 95 % CI 0.32–0.88). Furthermore, additional adjustment for partner age and smoking status did not change the effect (adjusted FR = 0.53; 95 % CI 0.31–0.92).
The number and types of confounders adjusted for in the study can be considered sufficient to get refined estimates of independent association between fecundity and occupational group, but the possibility of residual confounding cannot be entirely ruled out. An important potential confounder that was adjusted for in this study was coital frequency. This is because in South Africa, many domestic workers stay at the home of their employer during the week, only returning to their own homes on weekends, thereby reducing their coital frequency. The hypothesized effect of occupation in TTP studies is on fecundity. Therefore, any effect due to reduced coital frequency should be controlled for. However, the adjusted analysis showed that coital frequency was not a confounder in this study. This is because in this population, almost all domestic workers lived in the same district as their employers and went to work daily from home.
Bias
Planning bias
Further analysis was conducted to determine if planning bias played a role in this finding because the proportion of women who planned their pregnancy and reported a TTP value differed significantly by occupation. Domestic workers were less likely to plan their pregnancy than administrative workers and teachers. The effect of planning bias was assessed by assigning a TTP value of 0 to women who did not plan their pregnancy. The analysis did not show that planning bias had an effect on the results as the effect remained the same – adjusted FR = 0.53; 95 % CI 0.32–0.88. This is supported by the observation elsewhere that non-planners do not necessarily have higher fecundity than planners [21].
Time-trend bias
This population-based TTP study, as opposed to pregnancy based TTP studies, captured the entire TTP distribution from 1 month (high fecundity) to many years (low fecundity) for those who were still trying to fall pregnant at the time of the survey. Including women who were still trying to fall pregnant at the time of the survey has the benefit of eliminating fertility bias which can occur in pregnancy-based TTP studies. However, participants’ characteristics can change over time, for example they might change jobs and behaviours, as the waiting time increases. To evaluate the role of time-trend bias, we censored TTP at 14 months. The effect of domestic work remained similar (adjusted FR: 0.62, 95 % CI: 0.36–1.06) confirming that time-trend bias did not play a role in this finding.
Medical intervention (infertility treatment) bias
Medical intervention bias is unlikely to have played a role in this study. Respondents were asked if they sought medical intervention to become pregnant. Only 8 % sought medical intervention and the proportions were not significantly different by occupation (p = 0.690). Also, as described above, when TTP analysis was censored at 14 months [1], the results remained similar.
Occupational exposures in domestic work
In different parts of the world, domestic workers are exposed to a myriad of occupational exposures, including chemical, physical and psychosocial exposures, which can cause reproductive harm. This may even be more severe in less developed countries like South Africa where workplace regulation and employee education are minimal or non-existent. The most important of domestic work hazards is exposure to chemical agents. Domestic workers and cleaners are exposed to a number of volatile organic solvents, detergents, disinfectants and other chemical agents used for cleaning purposes in domestic work and other cleaning environments. Some cleaning solvents include halides, hydrocarbons, formaldehyde, alcohols, ketones, aldehyedes, esters, and ethers [22].
A number of chemicals used in domestic work have been implicated in reduced fecundity in other occupational groups [10, 11, 23, 24]. For example, chlorine, surfactants, formaldehyde, perchloroethylene (PERC) and ammonia commonly present in household cleaning products, soaps, air fresheners, scouring powders, etc., have been shown to be associated with increased time-to-pregnancy and pregnancy loss [10, 25]. A Finnish occupational study found that formaldehyde, which is commonly used in household cleaning substances, was associated with decreased fecundity: a fecundability ratio of 0.64 (95 % CI 0.43–0.92) was found in women with high levels of formaldehyde exposure in comparison to an unexposed group. A higher odds of endometriosis (OR = 4.5; 95 % CI 1–20) was also reported in the study [25].
For many of these substances, inhalation is the primary route of exposure, although dermal exposure is also important. Sadly, there are no regulations for the chemical contents in household products. While an average individual might use these products periodically, domestic workers use them routinely and in combinations. Such routine exposures can be frequent and chronic. Other chemical exposures could come from secondhand and thirdhand smoke. While most workplaces have banned smoking in South Africa, in private residences it is not banned. This can expose domestic workers to hazardous compounds from tobacco smoke. Acrolein, furan, acrylonitrile, and 1,3-butadiene are some of the most harmful volatile organic compounds to be identified in tobacco smoke residue [26].
In addition, domestic workers may be more exposed to physical and psychological stressors than administrative workers. Physical stressors such as awkward working postures, long working hours and physical exhaustion are common in domestic work, and these factors have been implicated in reduced fecundity. A study of the determinants of pregnancy outcome found that physical exertion in domestic work is associated with adverse outcomes [27]. Very early pregnancy loss, before its detection, can lead to prolonged time-to-pregnancy. Poor working conditions, high workload, poor remuneration and lack of developmental prospects can result in psychosocial stress for domestic workers. Domestic workers usually have limited skills for employment in formal sectors. This often means that they have to accept and work in any kind of environment domestic work presents, leading to psychosocial stress that can impact reproductive and overall health. Some studies have reported the role of psychosocial stress on increased TTP [28, 29].
External validity
Although the overall reproductive health study was representative of the study population, the women analysed in this TTP study have different characteristics from the overall sample. This is largely driven by the fact that pregnancy planning is low in South Africa and women who plan their pregnancy are different from those who do not, at least in terms of some basic socio-demographic characteristics. It is important to consider that the generalisability of these findings to all domestic workers in the population may be easily violated by the selective force of planning. It is not known to which extent those who participated in the study represent their occupational groups in the overall population. Therefore, external validity of the findings should be interpreted with caution.
However, the reduced sample size in the TTP analysis would be expected to be the case for TTP studies in many settings in South Africa as pregnancy planning is generally low in the population. In 2003, the South Africa Demographic and Health Survey which is a representative survey of the South African population reported a planned pregnancy rate of 50 % for the population [30]. Interestingly, low proportions of planned pregnancy have also been reported for some European countries: 37 % in Poland [31] and 41 % in East Germany [32]. Also, there is no reason to believe that the occupational groups in this study would have significantly different occupational exposure profiles from similar groups in other settings in South Africa.
Recommendations
Domestic workers may not know enough about the potential hazards of chemical exposure, and as a result they may not ask for or wear personal protection equipment when necessary. It is important for domestic workers to be educated about the potential hazards associated with chemical substances they use in their daily work, the need for safe handling and use of personal protective equipment. Employers should be key players in protecting the health of domestic workers. Also, the government should consider legislating that employers should provide domestic workers with training and protection equipment. Domestic workers’ unions and the government’s domestic workers Skills Development Project can be used as a platform for such training.
The rate at which new chemicals, designed for different purposes, are introduced into the market is high. Most of the chemicals are not tested for their toxicological or epidemiological effects. Simple epidemiological studies which compare exposed and unexposed groups, using data collected by short validated questionnaire, can be useful in identifying quantifiable risks to these agents. The advantage of detecting small increases in average TTP is that the reproductive hazard may be identified and controlled before irreversible damage occurs [33].
While focus on chemical exposure should be primary, physical and psychosocial hazards should also be curtailed. Domestic workers are some of the lowest paid employees in South Africa (median salary = ZAR 1,000) [33]. While salary increase negotiation can be knotty and not always possible, employers can ensure that they create good working conditions for their staff, which will go a long way in reducing the physical and psychosocial stress associated with domestic work. Recently, the South African government has taken actions to improve the working conditions and wage of domestic workers and to protect them. This includes setting rules and regulations for; their working conditions, minimum wage, housing conditions and so on. For example, according to recent regulations, domestic workers should not work more than 45 h a week, and should not work more than 15 h of overtime in a week. While these actions are welcomed, more work is still needed, especially in the area of health protection at work.