Spermatogenesis and spermiogenesis are the cellular processes that produce mature male sex cells. These processes take place within the seminiferous tubules of the testes of the sexually mature male, as shown in Figure 1. The human seminiferous tubules are 30 to 70 cm long and 150 to 300 mm in diameter (Zaneveld 1978). The spermatogonia (stem cells) are ppositioned along the basement membrane of the seminiferous tubules and are the basic cells for the production of sperm.

Figure 1. The male reproductive system

Sperm mature through a series of cellular divisions in which the spermatogonia proliferate and become primary spermatocytes. The resting primary spermatocytes migrate through tight junctions formed by the Sertoli cells to the luminal side of this testis barrier. By the time the spermatocytes reach the membrane barrier in the testis, the synthesis of DNA, the genetic material in the nucleus of the cell, is essentially complete. When the primary spermatocytes actually encounter the lumen of the seminiferous tubule, these undergo a special type of cell division which occurs only in germ cells and is known as meiosis. Meiotic cellular divison results in the splitting up of the chromosomes pairs in the nucleus, so that each resulting germ cell contains only a single copy of each chromosome strand rather than a matched pair.

During meiosis the chromosomes change shape by condensing and becoming filamentous. At a certain point, the nuclear membrane which surrounds them breaks down and microtubular spindles attach to the chromosomal pairs, causing them to separate. This completes the first meiotic division and two haploid secondary spermatocytes are formed. The secondary spermatocytes then undergo a second meiotic division to form equal numbers of X- and Y-chromosome bearing spermatids.

The morphological transformation of spermatids to spermatozoa is called spermiogenesis. When spermiogenesis is complete, each sperm cell is released by the Sertoli cell into the seminiferous tubule lumen by a process referred to as spermiation. The sperm migrate along the tubule to the rete testis and into the head of the epididymis. Sperm leaving the seminiferous tubules are immature: unable to fertilize an ovum and unable to swim. Spermatozoa released into the lumen of the seminiferous tubule are suspended in fluid pproduced primarily by the Sertoli cells. Concentrated sperm suspended within this fluid flow continuously from the seminiferous tubules, through slight changes in the ionic milieu within the rete testis, through the vasa efferentia, and into the epididymis. The epididymis is a single highly coiled tube (five to six metres long) in which sperm spend 12 to 21 days.

Within the epididymis, sperm progressively acquire motility and fertilizing capacity. This may be due to the changing nature of the suspension fluid in the epididymis. That is, as the cells mature the epididymis absorbs components from the fluid including secretions from the Sertoli cells (e.g., androgen binding protein), thereby increasing the concentration of spermatozoa. The epididymis also contributes its own secretions to the suspension fluid, including the chemicals glycerylphosphorylcholine (GPC) and carnitine.

Sperm morphology continues to transform in the epididymis. The cytoplasmic droplet is shed and the sperm nucleus condenses further. While the epididymis is the principal storage reservoir for sperm until ejaculation, about 30% of the sperm in an ejaculate have been stored in the vas deferens. Frequent ejaculation accelerates passage of sperm through the epididymis and may increase the number of immature (infertile) sperm in the ejaculate (Zaneveld 1978).

Ejaculation

Once within the vas deferens, the sperm are transported by the muscular contractions of ejaculation rather than by the flow of fluid. During ejaculation, fluids are forcibly expelled from the accessory sex glands giving rise to the seminal plasma. These glands do not expel their secretions at the same time. Rather, the bulbourethral (Cowper’s) gland first extrudes a clear fluid, followed by the prostatic secretions, the sperm-concentrated fluids from the epididymides and ampulla of the vas deferens, and finally the largest fraction primarily from the seminal vesicles. Thus, seminal plasma is not a homogeneous fluid.

Toxic Actions on Spermatogenesisand Spermiogenesis

Toxicants may disrupt spermatogenesis at several points. The most damaging, because of irreversibility, are toxicants that kill or genetically alter (beyond repair mechanisms) spermatogonia or Sertoli cells. Animal studies have been useful to determine the stage at which a toxicant attacks the spermatogenic process. These studies employ short term exposure to a toxicant before sampling to determine the effect. By knowing the duration for each spermatogenic stage, one can extrapolate to estimate the affected stage.

Biochemical analysis of seminal plasma pprovides insights into the function of the accessory sex glands. Chemicals that are secreted primarily by each of the accessory sex glands are typically selected to serve as a marker for each respective gland. For example, the epididymis is represented by GPC, the seminal vesicles by fructose, and the prostate gland by zinc. Note that this type of analysis pprovides only gross information on glandular function and little or no information on the other secretory constituents. Measuring semen pH and osmolality provide additional general information on the nature of seminal plasma.

Seminal plasma may be analysed for the presence of a toxicant or its metabolite. Heavy metals have been detected in seminal plasma using atomic absorption spectrophotometry, while halogenated hydrocarbons have been measured in seminal fluid by gas chromatography after extraction or protein-limiting filtration (Stachel et al. 1989; Zikarge 1986).

The viability and motility of spermatozoa in seminal plasma is typically a reflection of seminal plasma quality. Alterations in sperm viability, as measured by stain exclusion or by hypoosmotic swelling, or alterations in sperm motility parameters would suggest post-testicular toxicant effects.

Semen analyses also can indicate whether production of sperm cells has been affected by a toxicant. Sperm count and sperm morphology provide indices of the integrity of spermatogenesis and spermiogenesis. Thus, the number of sperm in the ejaculate is directly correlated with the number of germ cells per gram of testis (Zukerman et al. 1978), while abnormal morphology is probably a result of abnormal spermiogenesis. Dead sperm or immotile sperm often reflect the effects of post-testicular events. Thus, the type or timing of a toxic effect may indicate the target of the toxicant. For example, exposure of male rats to 2-methoxyethanol resulted in reduced fertility after four weeks (Chapin et al. 1985). This evidence, corroborated by histological examination, indicates that the target of toxicity is the spermatocyte (Chapin et al. 1984). While it is not ethical to intentionally expose humans to suspected reproductive toxicants, semen analyses of serial ejaculates of men inadvertently exposed for a short time to potential toxicants may provide similar useful information.

Occupational exposure to 1,2-dibromochloropropane (DBCP) reduced sperm concentration in ejaculates from a median of 79 million cells/ml in unexposed men to 46 million cells/ml in exposed workers (Whorton et al. 1979). Upon removing the workers from the exposure, those with reduced sperm counts experienced a partial recovery, while men who had been azoospermic remained sterile. Testicular biopsy revealed that the target of DBCP was the spermatogonia. This substantiates the severity of the effect when stem cells are the target of toxicants. There were no indications that DBCP exposure of men was associated with adverse pregnancy outcome (Potashnik and Abeliovich 1985). Another example of a toxicant targeting spermatogenesis/spermiogenesis was the study of workers exposed to ethylene dibromide (EDB). They had more sperm with tapered heads and fewer sperm per ejaculate than did controls (Ratcliffe et al. 1987).

Genetic damage is difficult to detect in human sperm. Several animal studies using the dominant lethal assay (Ehling et al. 1978) indicate that paternal exposure can produce an adverse pregnancy outcome. Epidemiological studies of large populations have demonstrated increased frequency of spontaneous abortions in women whose husbands were working as motor vehicle mechanics (McDonald et al. 1989). Such studies indicate a need for methods to detect genetic damage in human sperm. Such methods are being developed by several laboratories. These methods include DNA probes to discern genetic mutations (Hecht 1987), sperm chromosome karyotyping (Martin 1983), and DNA stability assessment by flow cytometry (Evenson 1986).

Figure 2. Exposures positively associated with adversely affecting semen quality

Figure 2 lists exposures known to affect sperm quality and table 1 provides a summary of the results of epidemiological studies of paternal effects on reproductive outcomes.

Table 1. Epidemiological studies of paternal effects on pregnancy outcome

¹ – no significant association; (+) marginally significant association; + significant association.
Source: Adapted from Taskinen 1993.

Neuroendocrine System

The overall functioning of the reproductive system is controlled by the nervous system and the hormones pproduced by the glands (the endocrine system). The reproductive neuroendocrine axis of the male involves principally the central nervous systems (CNS), the anterior pituitary gland and the testes. Inputs from the CNS and from the periphery are integrated by the hypothalamus, which directly regulates gonadotrophin secretion by the anterior pituitary gland. The gonadotrophins, in turn, act principally upon the Leydig cells within the interstitium and Sertoli and germ cells within the seminiferous tubules to regulate spermatogenesis and hormone production by the testes.

Hypothalamic–Pituitary Axis

The hypothalamus secretes the neurohormone gonadotrophin releasing hormone (GnRH) into the hypophysial portal vasculature for transport to the anterior pituitary gland. The pulsatile secretion of this decapeptide causes the concomitant release of luteinizing hormone (LH), and with lesser synchrony and one-fifth the potency, the release of follicle stimulating hormone (FSH) (Bardin 1986). Substantial evidence exists to support the presence of a separate FSH releasing hormone, although none has yet been isolated (Savy-Moore and Schwartz 1980; Culler and Negro-Vilar 1986). These hormones are secreted by the anterior pituitary gland. LH acts directly upon the Leydig cells to stimulate synthesis and release of testosterone, whereas FSH stimulates aromatization of testosterone to estradiol by the Sertoli cell. Gonadotropic stimulation causes the release of these steroid hormones into the spermatic vein.

Gonadotrophin secretion is, in turn, checked by testosterone and estradiol through negative feedback mechanisms. Testosterone acts principally upon the hypothalamus to regulate GnRH secretion and thereby reduces the pulse frequency, primarily, of LH release. Estradiol, on the other hand, acts upon the pituitary gland to reduce the magnitude of gonadotrophin release. Through these endocrine feedback loops, testicular function in general and testosterone secretion specifically are maintained at a relatively steady state.

Pituitary–Testicular Axis

LH and FSH are generally viewed as necessary for normal spermatogenesis. Presumably the effect of LH is secondary to inducing high intratesticular concentrations of testosterone. Therefore, FSH from the pituitary gland and testosterone from the Leydig cells act upon the Sertoli cells within the seminiferous tubule epithelium to initiate spermatogenesis. Sperm production persists, although quantitatively reduced, after removing either LH (and presumably the high intratesticular testosterone concentrations) or FSH. FSH is required for initiating spermatogenesis at puberty and, to a lesser extent, to reinitiate spermatogenesis that has been arrested (Matsumoto 1989; Sharpe 1989).

The hormonal synergism that serves to maintain spermatogenesis may entail recruitment by FSH of differentiated spermatogonia to enter meiosis, while testosterone may control specific, subsequent stages of spermatogenesis. FSH and testosterone may also act upon the Sertoli cell to stimulate production of one or more paracrine factors which may affect the number of Leydig cells and testosterone production by these cells (Sharpe 1989). FSH and testosterone stimulate protein synthesis by Sertoli cells including synthesis of androgen binding protein (ABP), while FSH alone stimulates synthesis of aromatase and inhibin. ABP is secreted primarily into the seminiferous tubular fluid and is transported to the proximal portion of the caput epididymis, possibly serving as a local carrier of androgens (Bardin 1986). Aromatase catalyses the conversion of testosterone to estradiol in the Sertoli cells and in other peripheral tissues.

Inhibin is a glycoprotein consisting of two dissimilar, disulphide-linked subunits, a and b. Although inhibin preferentially inhibits FSH release, it may also attenuate LH release in the presence of GnRH stimulation (Kotsugi et al. 1988). FSH and LH stimulate inhibin release with approximately equal potency (McLachlan et al. 1988). Interestingly, inhibin is secreted into the spermatic vein blood as pulses which are synchronous to those of testosterone (Winters 1990). This probably does not reflect direct actions of LH or testosterone on Sertoli cell activity, but rather the effects of other Leydig cell products secreted either into the interstitial spaces or the circulation.

Prolactin, which is also secreted by the anterior pituitary gland, acts synergistically with LH and testosterone to promote male reproductive function. Prolactin binds to specific receptors on the Leydig cell and increases the amount of androgen receptor complex within the nucleus of androgen responsive tissues (Baker et al. 1977). Hyperprolactinaemia is associated with reductions of testicular and prostate size, semen volume and circulating concentrations of LH and testosterone (Segal et al. 1979). Hyperprolactinaemia has also been associated with impotency, apparently independent of altering testosterone secretion (Thorner et al. 1977).

If measuring steroid hormone metabolites in urine, consideration must be given to the potential that the exposure being studied may alter the metabolism of excreted metabolites. This is especially pertinent since most metabolites are formed by the liver, a target of many toxicants. Lead, for example, reduced the amount of sulphated steroids that were excreted into the urine (Apostoli et al. 1989). Blood levels for both gonadotrophins become elevated during sleep as the male enters puberty, while testosterone levels maintain this diurnal pattern through adulthood in men (Plant 1988). Thus blood, urine or saliva samples should be collected at approximately the same time of day to avoid variations due to diurnal secretory patterns.

The overt effects of toxic exposure targeting the reproductive neuroendocrine system are most likely to be revealed through altered biological manifestations of the androgens. Manifestations significantly regulated by androgens in the adult man that may be detected during a basic physical examination include: (1) nitrogen retention and muscular development; (2) maintenance of the external genitalia and accessory sexual organs; (3) maintenance of the enlarged larynx and thickened vocal cords causing the male voice; (4) beard, axillary and pubic hair growth and temporal hair recession and balding; (5) libido and sexual performance; (6) organ specific proteins in tissues (e.g., liver, kidneys, salivary glands); and (7) aggressive behaviour (Bardin 1986). Modifications in any of these traits may indicate that androgen production has been affected.

Examples of Toxicant Effects

Lead is a classic example of a toxicant that directly affects the neuroendocrine system. Serum LH concentrations were elevated in men exposed to lead for less than one year. This effect did not progress in men exposed for more than five years. Serum FSH levels were not affected. On the other hand, serum levels of ABP were elevated and those of total testosterone were reduced in men exposed to lead for more than five years. Serum levels of free testosterone were significantly reduced after exposure to lead for three to five years (Rodamilans et al. 1988). In contrast, serum concentrations of LH, FSH, total testosterone, prolactin, and total neutral 17-ketosteroids were not altered in workers with lower circulating levels of lead, even though the distribution frequency of sperm count was altered (Assennato et al. 1986).

Exposure of shipyard painters to 2-ethoxyethanol also reduced sperm count without a concurrent change in serum LH, FSH, or testosterone concentrations (Welch et al. 1988). Thus toxicants may affect hormone production and sperm measures independently.

Male workers involved in the manufacture of the nematocide DBCP experienced elevated serum levels of LH and FSH and reduced sperm count and fertility. These effects are apparently sequelae to DBCP actions upon the Leydig cells to alter androgen production or action (Mattison et al. 1990).

Several compounds may exert toxicity by virtue of structural similarity to reproductive steroid hormones. Thus, by binding to the respective endocrine receptor, toxicants may act as agonists or antagonists to disrupt biological responses. Chlordecone (Kepone), an insecticide that binds to oestrogen receptors, reduced sperm count and motility, arrested sperm maturation and reduced libido. While it is tempting to suggest that these effects result from chlordecone interfering with oestrogen actions at the neuroendocrine or testicular level, serum levels of testosterone, LH and FSH were not shown to be altered in these studies in a manner similar to the effects of oestradiol therapy. DDT and its metabolites also exhibit steroidal properties and might be expected to alter male reproductive function by interfering with steroidal hormone functions. Xenobiotics such as polychlorinated biphenyls, polybrominated biphenyls, and organochlorine pesticides may also interfere with male reproductive functions by exerting oestrogenic agonist/antagonist activity (Mattison et al. 1990).

Sexual Function

Human sexual function refers to the integrated activities of the testes and secondary sex glands, the endocrine control systems, and the central nervous system-based behavioural and psychological components of reproduction (libido). Erection, ejaculation and orgasm are three distinct, independent, physiological and psychodynamic events which normally occur concurrently in men.

Little reliable data are available on occupational exposure effects on sexual function due to the problems described above. Drugs have been shown to affect each of the three stages xof male sexual function (Fabro 1985), indicating the potential for occupational exposures to exert similar effects. Antidepressants, testosterone antagonists and stimulants of prolactin release effectively reduce libido in men. Antihypertensive drugs which act on the sympathetic nervous system induce impotence in some men, but surprisingly, priapism in others. Phenoxybenzamine, an adrenoceptive antagonist, has been used clinically to block seminal emission but not orgasm (Shilon, Paz and Homonnai 1984). Anticholinergic antidepressant drugs permit seminal emission while blocking seminal ejection and orgasm which results in seminal plasma seeping from the urethra rather than being ejected.

Recreational drugs also affect sexual function (Fabro 1985). Ethanol may reduce impotence while enhancing libido. Cocaine, heroin and high doses of cannabinoids reduce libido. Opiates also delay or impair ejaculation.

The vast and varied array of pharmaceuticals that has been shown to affect the male reproductive system pprovides support for the notion that chemicals found in the workplace may also be reproductive toxicants. Research methods that are reliable and practical for field study conditions are needed to assess this important area of reproductive toxicology.

Back

Figure 1. The female reproductive system.

The female reproductive system is controlled by components of the central nervous system, including the hypothalamus and pituitary. It consists of the ovaries, the fallopian tubes, the uterus and the vagina (Figure 1). The ovaries, the female gonads, are the source of oocytes and also synthesize and secrete oestrogens and progestogens, the major female sex hormones. The fallopian tubes transport oocytes to and sperm from the uterus. The uterus is a pear-shaped muscular organ, the upper part of which communicates through the fallopian tubes to the abdominal cavity, while the lower part is contiguous through the narrow canal of the cervix with the vagina, which passes to the exterior. Table 1 summarizes compounds, clinical manifestations, site and mechanisms of action of potential reproductive toxicants.

Table 1. Potential female reproductive toxicants

Source: From Plowchalk, Meadows and Mattison 1992. These compounds are suggested to be direct-acting reproductive toxicants based primarily on toxicity testing in experimental animals.

The Hypothalamus and Pituitary

The hypothalamus is located in the diencephalon, which sits on top of the brainstem and is surrounded by the cerebral hemispheres. The hypothalamus is the principal intermediary between the nervous and the endocrine systems, the two major control systems of the body. The hypothalamus regulates the pituitary gland and hormone production.

The mechanisms by which a chemical might disrupt the reproductive function of the hypothalamus generally include any event that could modify the pulsatile release of gonadotrophin releasing hormone (GnRH). This may involve an alteration in either the frequency or the amplitude of GnRH pulses. The processes susceptible to chemical injury are those involved in the synthesis and secretion of GnRH—more specifically, transcription or translation, packaging or axonal transport, and secretory mechanisms. These processes represent sites where direct-acting chemically reactive compounds might interfere with hypothalmic synthesis or release of GnRH. An altered frequency or amplitude of GnRH pulses could result from disruptions in stimulatory or inhibitory pathways that regulate the release of GnRH. Investigations of the regulation of the GnRH pulse generator have shown that catecholamines, dopamine, serotonin, γ-aminobutyric acid, and endorphins all have some potential for altering the release of GnRH. Therefore, xenobiotics that are agonists or antagonists of these compounds could modify GnRH release, thus interfering with communication with the pituitary.

Prolactin, follicle-stimulating hormone (FSH) and luteinizing hormone (LH) are three protein hormones secreted by the anterior pituitary that are essential for reproduction. These play a critical role in maintaining the ovarian cycle, governing follicle recruitment and maturation, steroidogenesis, completion of ova maturation, ovulation and luteinization.

The precise, finely tuned control of the reproductive system is accomplished by the anterior pituitary in response to positive and negative feedback signals from the gonads. The appropriate release of FSH and LH during the ovarian cycle controls normal follicular development, and the absence of these hormones is followed by amenorrhoea and gonadal atrophy. The gonadotrophins play a critical role in initiating changes in the morphology of ovarian follicles and in their steroidal microenvironments through the stimulation of steroid production and the induction of receptor populations. Timely and adequate release of these gonadotrophins is also essential for ovulatory events and a functional luteal phase. Because gonadotrophins are essential for ovarian function, altered synthesis, storage or secretion may seriously disrupt reproductive capacity. Interference with gene expression—whether in transcription or translation, post-translational events or packaging, or secretory mechanisms—may modify the level of gonadotrophins reaching the gonads. Chemicals that act by means of structural similarity or altered endocrine homeostasis might produce effects by interference with normal feedback mechanisms. Steroid-receptor agonists and antagonists might initiate an inappropriate release of gonadotrophins from the pituitary, thereby inducing steroid-metabolizing enzymes, reducing steroid half-life and subsequently the circulating level of steroids reaching the pituitary.

The Ovary

The ovary in primates is responsible for the control of reproduction through its principal products, oocytes and steroid and protein hormones. Folliculogenesis, which involves both intraovarian and extraovarian regulatory mechanisms, is the process by which oocytes and hormones are produced. The ovary itself has three functional subunits: the follicle, the oocyte and the corpus luteum. During the normal menstrual cycle, these components, under the influence of FSH and LH, function in concert to produce a viable ovum for fertilization and a suitable environment for implantation and subsequent gestation.

During the preovulatory period of the menstrual cycle, follicle recruitment and development occur under the influence of FSH and LH. The latter stimulates the production of androgens by thecal cells, whereas the former stimulates the aromatization of androgens into oestrogens by the granulosa cells and the production of inhibin, a protein hormone. Inhibin acts at the anterior pituitary to decrease the release of FSH. This prevents excess stimulation of follicular development and allows continuing development of the dominant follicle—the follicle destined to ovulate. Oestrogen production increases, stimulating both the LH surge (resulting in ovulation) and the cellular and secretory changes in the vagina, cervix, uterus and oviduct that enhance spermatozoa viability and transport.

In the postovulatory phase, thecal and granulosa cells remaining in the follicular cavity of the ovulated ovum, form the corpus luteum and secrete progesterone. This hormone stimulates the uterus to provide a proper environment for implantation of the embryo if fertilization occurs. Unlike the male gonad, the female gonad has a finite number of germ cells at birth and is therefore uniquely sensitive to reproductive toxicants. Such exposure of the female can lead to decreased fecundity, increased pregnancy wastage, early menopause or infertility.

As the basic reproductive unit of the ovary, the follicle maintains the delicate hormonal environment necessary to support the growth and maturation of an oocyte. As previously noted, this complex process is known as folliculogenesis and involves both intraovarian and extraovarian regulation. Numerous morphological and biochemical changes occur as a primordial follicle progresses to a pre-ovulatory follicle (which contains a developing oocyte), and each stage of follicular growth exhibits unique patterns of gonadotrophin sensitivity, steroid production and feedback pathways. These characteristics suggest that a number of sites are available for xenobiotic interaction. Also, there are different follicle populations within the ovary, which further complicates the situation by allowing for differential follicle toxicity. This creates a situation in which the patterns of infertility induced by a chemical agent would depend on the follicle type affected. For example, toxicity to primordial follicles would not produce immediate signs of infertility but would ultimately shorten the reproductive lifespan. On the other hand, toxicity to antral or preovulatory follicles would result in an immediate loss of reproductive function. The follicle complex is composed of three basic components: granulosa cells, thecal cells and the oocyte. Each of these components has characteristics that may make it uniquely susceptible to chemical injury.

Several investigators have explored methodology for screening xenobiotics for granulosa cell toxicity by measuring the effects on progesterone production by granulosa cells in culture. Oestradiol suppression of progesterone production by granulosa cells has been utilized to verify granulosa cell responsiveness. The pesticide p,p’-DDT and its o,p’-DDT isomer produce supression of progesterone production apparently with potencies equal to that of oestradiol. By contrast, the pesticides malathion, arathion and dieldrin and the fungicide hexachlorobenzene are without effect. Further detailed analysis of isolated granulosa cell responses to xenobiotics is needed to define the utility of this assay system. The attractiveness of isolated systems such as this is economy and ease of use; however, it is important to remember that granulosa cells represent only one component of the reproductive system.

Thecal cells provide precursors for steroids synthesized by granulosa cells. Thecal cells are believed to be recruited from ovarian stroma cells during follicle formation and growth. Recruitment may involve stromal cellular proliferation as well as migration to regions around the follicle. Xenobiotics that impair cell proliferation, migration and communication will impact on thecal cell function. Xenobiotics that alter thecal androgen production may also impair follicle function. For example, the androgens metabolized to oestrogens by granulosa cells are provided by thecal cells. Alterations in thecal cell androgen production, either increases or decreases, are expected to have a significant effect on follicle function. For example, it is believed that excess production of androgens by thecal cells will lead to follicle atresia. In addition, impaired production of androgens by thecal cells may lead to decreased poestrogen production by granulosa cells. Either circumstance will clearly impact on reproductive performance. At resent, little is known about thecal cell vulnerability to xenobiotics.

Although there is a acuity of information defining the vulnerability of ovarian cells to xenobiotics, there are data clearly demonstrating that oocytes can be damaged or destroyed by such agents. Alkylating agents destroy oocytes in humans and experimental animals. Lead produces ovarian toxicity. Mercury and cadmium also produce ovarian damage that may be mediated through oocyte toxicity.

Fertilization to Implantation

Gametogenesis, release and union of male and female germ cells are all preliminary events leading to a zygote. Sperm cells deposited in the vagina must enter the cervix and move through the uterus and into the fallopian tube to meet the ovum. penetration of ovum by sperm and the merging of their respective DNA comprise the process of fertilization. After fertilization cell division is initiated and continues during the next three or four days, forming a solid mass of cells called a morula. The cells of the morula continue to divide, and by the time the developing embryo reaches the uterus it is a hollow ball called a blastocyst.

Following fertilization, the developing embryo migrates through the fallopian tube into the uterus. The blastocyst enters the uterus and implants in the endometrium approximately seven days after ovulation. At this time the endometrium is in the postovulatory phase. Implantation enables the blastocyst to absorb nutrients or toxicants from the glands and blood vessels of the endometrium.

Back

Paid employment among women is growing worldwide. For example, almost 70% of women in the United States are employed outside the home during their predominant childbearing years (ages 20 to 34). Furthermore, since the 1940s there has been an almost linear trend in synthetic organic chemical production, creating a more hazardous environment for the pregnant worker and her offspring.

Ultimately, a couple’s reproductive success depends on a delicate physiochemical balance within and between the father, the mother and the foetus. Metabolic changes occurring during a pregnancy can increase exposure to hazardous toxicants for both worker and concetus. These metabolic changes include increased pulmonary absorption, increased cardiac output, delayed gastric emptying, increased intestinal motility and increased body fat. As shown in figure 1, exposure of the concetus can produce varying effects depending on the phase of development—early or late embryogenesis or the foetal period.

Figure 1. Consequences of maternal exposure to toxicants on the offspring.

Transport time of a fertilized ovum before implantation is between two and six days. During this early stage the embryo may be exposed to chemical compounds that penetrate into the uterine fluids. Absorption of xenophobic compounds may be accompanied by degenerative changes, alteration in the blastocystic protein profile or failure to implant. Insult during this period is likely to lead to a spontaneous abortion. Based on experimental data, it is thought that the embryo is fairly resistant to teratogenic insult at this early stage because the cells have not initiated the complex sequence of chemical differentiation.

The period of later embryogenesis is characterized by differentiation, mobilization and organization of cells and tissue into organ rudiments. Early pathogenesis may induce cell death, failed cellular interaction, reduced biosynthesis, impaired morphogenic movement, mechanical disruption, adhesions or oedema (Paul 1993). The mediating factors that determine susceptibility include route and level of exposure, pattern of exposure and foetal and maternal genotype. Extrinsic factors such as nutritional deficiencies, or the additive, synergistic or antagonistic effects associated with multiple exposures may further impact the response. Untoward responses during late embryogenesis may culminate in spontaneous abortion, gross structural defects, foetal loss, growth retardation or developmental abnormalities.

The foetal period extends from embryogenesis to birth and is defined as beginning at 54 to 60 gestational days, with the concetus having a crown-rum length of 33 mm. The distinction between the embryonic and foetal period is somewhat arbitrary. The foetal period is characterized developmentally by growth, histogenesis and functional maturation. Toxicity may be manifested by a reduction in cell size and number. The brain is still sensitive to injury; myelination is incomplete until after birth. Growth retardation, functional defects, disruption in the pregnancy, behavioural effects, translacental carcinogenesis or death may result from toxicity during the foetal period. This article discusses the biological, sociological and epidemiological effects of maternal environmental/occupational exposures.

Embryonic/Foetal Loss

The developmental stages of the zygote, defined in days from ovulation (DOV), proceed from the blastocyst stage at days 15 to 20 (one to six DOV), with implantation occurring on day 20 or 21 (six or seven DOV), to the embryonic period from days 21 to 62 (seven to 48 DOV), and the foetal period from day 63 (49+ DOV) until the designated period of viability, ranging from 140 to 195 days. Estimates of the probability of pregnancy termination at one of these stages depend on both the definition of foetal loss and the method used to measure the event. Considerable variability in the definition of early versus late foetal loss exists, ranging from the end of week 20 to week 28. The definitions of foetal and infant death recommended by the World Health Organization (1977) are listed in table 1. In the United States the gestational age setting the lower limit for stillbirths is now widely accepted to be 20 weeks.

Table 1. Definition of foetal loss and infant death

Source: World Health Organization 1977.

Because the majority of early aborted foetuses have chromosomal anomalies, it has been suggested that for research purposes a finer distinction should be made—between early foetal loss, before 12 weeks’ gestation, and later foetal loss (Källén 1988). In examining late foetal losses it also may be appropriate to include early neonatal deaths, as the cause may be similar. WHO defines early neonatal death as the death of an infant aged seven days or younger and late neonatal death as occurring between seven and 29 days. In studies conducted in developing countries, it is important to distinguish between prepartum and intrapartum deaths. Because of problematic deliveries, intrapartum deaths account for a large portion of stillbirths in less developed countries.

In a review by Kline, Stein and Susser (1989) of nine retrospective or cross-sectional studies, the foetal loss rates before 20 weeks’ gestation ranged from 5.5 to 12.6%. When the definition was expanded to include losses u to 28 weeks’ gestation, the foetal loss rate varied between 6.2 and 19.6%. The rates of foetal loss among clinically recognized pregnancies in four prospective studies, however, had a relatively narrow range of 11.7 to 14.6% for the gestational period u to 28 weeks. This lower rate, seen in prospective versus retrospective or cross-sectional designs, may be attributable to differences in underlying definitions, misreporting of induced abortions as spontaneous or misclassification of delayed or heavy menses as foetal loss.

When occult abortions or early “chemical” losses identified by an elevated level of human chorionic gonadotrohins (hCG) are included, the total spontaneous abortion rate jumps dramatically. In a study using hCG methods, the incidence of post-implantation subclinical loss of fertilized ova was 22% (Wilcox et al. 1988). In these studies urinary hCG was measured with immunoradiometric assay using a detection antibody. The assay originally used by Wilcox employed a now extinct high affinity, polyclonal rabbit antibody. More recent studies have used an inexhaustible monoclonal antibody that requires less than 5 ml of urine for replicate samples. The limiting factor for use of these assays in occupational field studies is not only the cost and resources needed to coordinate collection, storage and analysis of urine samples but the large population needed. In a study of early pregnancy loss in women workers exposed to video display terminals (VDTs), approximately 7,000 women were screened in order to acquire a usable population of 700 women. This need for ten times the population size in order to achieve an adequate sample stems from reduction in the available number of women because of ineligibility due to age, sterility and the enrollment exclusively of women who are using either no contraceptives or relatively ineffective forms of contraception.

More conventional occupational studies have used recorded or questionnaire data to identify spontaneous abortions. Recorded data sources include vital statistics and hospital, private practitioner and outpatient clinic records. Use of record systems identifies only a subset of all foetal losses, principally those that occur after the start of prenatal care, typically after two to three missed periods. Questionnaire data are collected by mail or in personal or telephone interviews. By interviewing women to obtain reproductive histories, more complete documentation of all recognized losses is possible. Questions that are usually included in reproductive histories include all pregnancy outcomes; prenatal care; family history of adverse pregnancy outcomes; marital history; nutritional status; re-pregnancy weight; height; weight gain; use of cigarettes, alcohol and prescription and nonprescription drugs; health status of the mother during and prior to a pregnancy; and exposures at home and in the workplace to physical and chemical agents such as vibration, radiation, metals, solvents and pesticides. Interview data on spontaneous abortions can be a valid source of information, particularly if the analysis includes those of eight weeks’ gestation or later and those that occurred within the last 10 years.

The principal physical, genetic, social and environmental factors associated with spontaneous abortion are summarized in  table 2. To ensure that the observed exposure-effect relationship is not due to a confounding relationship with another risk factor, it is important to identify the risk factors that may be associated with the outcome of interest. Conditions associated with foetal loss include syphilis, rubella, genital Mycolasma infections, herpes simplex, uterine infections and general hyperpyrexia. One of the most important risk factors for clinically recognized spontaneous abortion is a history of pregnancy ending in foetal loss. Higher gravidity is associated with increased risk, but this may not be independent of a history of spontaneous abortion. There are conflicting interpretations of gravidity as a risk factor because of its association with maternal age, reproductive history and heterogeneity of women at different gravidity ranks. Rates of spontaneous abortion are higher for women younger than 16 and older than 36 years. After adjusting for gravidity and a history of pregnancy loss, women older than 40 were shown to have twice the risk of foetal loss of younger women. The increased risk for older women has been associated with an increase in chromosomal anomalies, particularly trisomy. possiblemale-mediated effects associated with foetal loss have been recently reviewed (Savitz, Sonnerfeld and Olshaw 1994). A stronger relationship was shown with paternal exposure to mercury and anaesthetic gases, as well as a suggestive but inconsistent relationship with exposure to lead, rubber manufacturing, selected solvents and some pesticides.

Table 2. Factors associated with small for gestational age and foetal loss

Employment status may be a risk factor regardless of a specific physical or chemical hazard and may act as a confounder in assessment of occupational exposure and spontaneous abortion. Some investigators suggest that women who stay in the workforce are more likely to have an adverse pregnancy history and as a result are able to continue working; others believe this group is an inherently more fit subpopulation due to higher incomes and better prenatal care.

Congenital Anomalies

During the first 60 days after conception, the developing infant may be more sensitive to xenobiotic toxicants than at any other stage in the life cycle. Historically, terata and congenital malformations referred to structural defects resent at birth that may be gross or microscopic, internal or external, hereditary or nonhereditary, single or multiple. Congenital anomaly, however, is more broadly defined as including abnormal behaviour, function and biochemistry. Malformations may be single or multiple; chromosomal defects generally produce multiple defects, whereas single-gene changes or exposure to environmental agents may cause either single defects or a syndrome.

The incidence of malformations depends on the status of the concetus—live birth, spontaneous abortus, stillbirth. Overall, the abnormality rate in spontaneous abortuses is approximately 19%, a tenfold increase in what is seen in the live born (Sheard, Fantel and Fitsimmons 1989). A 32% rate of anomalies was found among stillborn foetuses weighing more than 500 g. The incidence of major defects in live births is about 2.24% (Nelson and Holmes 1989). The prevalence of minor defects ranges between 3 and 15% (averaging about 10%). Birth anomalies are associated with genetic factors (10.1%), multifactorial inheritance (23%), uterine factors (2.5%), twinning (0.4%) or teratogens (3.2%). The causes of the remaining defects are unknown. Malformation rates are approximately 41% higher for boys than for girls and this is explained by the significantly higher rate of anomalies for male genital organs.

One challenge in studying malformations is deciding how to group defects for analysis. Anomalies can be classified by several parameters, including seriousness (major, minor), pathogenesis (deformation, disruption), associated versus isolated, anatomic by organ system, and aetiological (e.g., chromosomal, single gene defects or teratogen induced). Often, all malformations are combined or the combination is based either on major or minor categorization. A major malformation can be defined as one that results in death, requires surgery or medical treatment or constitutes a substantial physical or psychological handicap. The rationale for combining anomalies into large groups is that the majority arise, at approximately the same time period, during organogenesis. Thus, by maintaining larger sample sizes, the total number of cases is increased with a concomitant increase in the statistical power. If, however, the exposure effect is specific to a particular type of malformation (e.g., central nervous system), such grouping may mask the effect. Alternatively, malformations may be grouped by organ system. Though this method may be an improvement, certain defects may dominate the class, such as varus deformities of the feet in the musculoskeletal system. Given a sufficiently large sample, the optimal approach is to divide the defects into embryologically or pathogenetically homogenous groups (Källén 1988). Considerations should be given to the exclusion or inclusion of certain malformations, such as those that are likely caused by chromosomal defects, autosomal dominant conditions or malposition in utero. Ultimately, in analysing congenital anomalies, a balance has to be maintained between maintaining precision and compromising statistical power.

A number of environmental and occupational toxicants have been associated with congenital anomalies in offspring. One of the strongest associations is maternal consumption of food contaminated with methylmercury causing morphological, central nervous system and neurobehavioural abnormalities. In Japan, the cluster of cases was linked to consumption of fish and shellfish contaminated with mercury derived from the effluent of a chemical factory. The most severely affected offspring developed cerebral palsy. Maternal ingestion of polychlorinated biphenyl’s (CBs) from contaminated rice oil gave rise to babies with several disorders, including growth retardation, dark brown skin pigmentation, early eruption of teeth, gingival hyperplasia, wide sagittal suture, facial oedema and exophthalmoses. Occupations involving exposures to mixtures have been linked with a variety of adverse outcomes. The offspring of women working in the ul and aer industry, in either laboratory work or jobs involving “conversions” or aer refinement, also had increased risk of central nervous system, heart and oral cleft defects. Women working in industrial or construction work with unspecified exposures had a 50% increase in central nervous system defects, and women working in transportation and communication had two times the risk of having a child with an oral cleft. Veterinarians represent a unique group of health care personnel exposed to anaesthetic gases, radiation, trauma from animal kicks, insecticides and zoonotic diseases. Though no difference was found in the rate of spontaneous abortions or in birth weight of the offspring between female veterinarians and female lawyers, there was a significant excess of birth defects among veterinarians (Schenker et al. 1990). Lists of known, possible and unlikely teratogens are available as well as computer databases and risk lines for obtaining current information on potential teratogens (Paul 1993). Evaluating congenital anomalies in an occupational cohort is particularly difficult, however, because of the large sample size needed for statistical power and our limited ability to identify specific exposures occurring during a narrow window of time, primarily the first 55 days of gestation.

Small for Gestational Age

Among the many factors linked with infant survival, physical underdevelopment associated with low birth weight (LBW) resents one of the greatest risks. Significant weight gain of the foetus does not begin until the second trimester. The concetus weighs 1 g at eight weeks, 14 g at 12 weeks, and reaches 1.1 kg at 28 weeks. An additional 1.1 kg is gained every six weeks thereafter until term. The normal newborn weighs approximately 3,200 g at term. The newborn’s weight is dependent on its rate of growth and its gestational age at delivery. An infant that is growth retarded is said to be small for gestational age (SGA). If an infant is delivered prior to term it will have a reduced weight but will not necessarily be growth retarded. Factors associated with a preterm delivery are discussed elsewhere, and the focus of this discussion is on the growth-retarded newborn. The terms SGA and LBW will be used interchangeably. A low birth-weight infant is defined as an infant weighing less than 2,500 g, a very low birth weight is defined as less than 1,500 g, and extremely low birth weight is one that is less than 1,000 g (WHO 1969).

When examining causes of reduced growth, it is important to distinguish between asymmetrical and symmetrical growth retardation. Asymmetrical growth retardation, i.e., where the weight is affected more than the skeletal structure, is primarily associated with a risk factor operating during late pregnancy. On the other hand, symmetrical growth retardation may more likely be associated with an aetiology that operates over the entire period of gestation (Kline, Stein and Susser 1989). The difference in rates between asymmetrical and symmetrical growth retardation is especially apparent when comparing developing and developed countries. The rate of growth retardation in developing countries is 10 to 43%, and is primarily symmetrical, with the most important risk factor being poor nourishment. In developed countries foetal growth retardation is usually much lower, 3 to 8%, and is generally asymmetrical with a multifactorial aetiology. Hence, worldwide, the proportion of low birth-weight infants defined as intrauterine growth retarded rather than preterm varies dramatically. In Sweden and the United States, the proportion is approximately 45%, while in developing countries, such as India, the proportion varies between approximately 79 and 96% (Villar and Belizan 1982).

Studies of the Dutch famine showed that starvation confined to the third trimester depressed foetal growth in an asymmetric pattern, with birth weight being primarily affected and head circumference least affected (Stein, Susser and Saenger 1975). Asymmetry of growth also has been observed in studies of environmental exposures. In a study of 202 expectant mothers residing in neighbourhoods at high risk for lead exposures, prenatal maternal blood samples were collected between the sixth and the 28th week of gestation (Bornschein, Grote and Mitchell 1989). Blood lead levels were associated with both a decreased birth weight and length, but not head circumference, after adjustment for other relevant risk factors including length of gestation, socioeconomic status and use of alcohol or cigarettes. The finding of maternal blood lead as a factor in birth length was seen entirely in Caucasian infants. The birth length of Caucasian infants decreased approximately 2.5 cm per log unit increment in maternal blood lead. Care should be given to selection of the outcome variable. If only birth weight had been selected for study, the finding of the effects of lead on other growth parameters might have been missed. Also, if Caucasians and African Americans had been pooled in the above analysis, the differential effects on Caucasians, perhaps due to genetic differences in the storage and binding capacity of lead, may have been missed. A significant confounding effect also was observed between prenatal blood lead and maternal age and the birth weight of the offspring after adjustment for other covariables. The findings indicate that for a 30-year-old woman with an estimated blood lead level of about 20 mg/dl, the offspring weighed proximately 2,500 g compared with proximately 3,000 g for a 20-year-old with similar lead levels. The investigators speculated that this observed difference may indicate that older women are more sensitive to the additional insult of lead exposure or that older women may have had higher total lead burden from greater numbers of years of exposure or higher ambient lead levels when they were children. Another factor may be increased blood pressure. Nonetheless, the important lesson is that careful examination of high-risk subpopulations by age, race, economic status, daily living habits, sex of the offspring and other genetic differences may be necessary in order to discover the more subtle effects of exposures on foetal growth and development.

Risk factors associated with low birth weight are summarized in  Table 5. Social class as measured by income or education persists as a risk factor in situations in which there are no ethnic differences. Other factors that may be operating under social class or race may include cigarette smoking, physical work, prenatal care and nutrition. Women between the ages of 25 and 29 are least likely to deliver a growth-retarded offspring. Maternal smoking increases the risk of low birth-weight offspring  by about 200% for heavy smokers. Maternal medical conditions associated with LBW include placental abnormalities, heart disease, viral pneumonia, liver disease, re-eclamsia, eclamsia, chronic hypertension, weight gain and hyeremesis. An adverse pregnancy history of foetal loss, preterm delivery or prior LBW infant increases the risk of a current preterm low birth-weight infant two- to fourfold. An interval between births of less than a year triples the risk of having a low birth-weight offspring. Chromosomal anomalies associated with abnormal growth include Down’s syndrome, trisomy 18 and most malformation syndromes.

Smoking cigarettes is one of the primary behaviours most directly linked with lower weight offspring. Maternal smoking during pregnancy has been shown to increase the risk of a low birth-weight offspring two to three times and to cause an overall weight deficit of between 150 and 400 g. Nicotine and carbon monoxide are considered the most likely causative agents since both are rapidly and referentially transferred across the placenta. Nicotine is a powerful vasoconstrictor, and significant differences in the size of umbilical vessels of smoking mothers have been demonstrated. Carbon monoxide levels in cigarette smoke range from 20,000 to 60,000 m. Carbon monoxide has an affinity for haemoglobin 210 times that of oxygen, and because of lower arterial oxygen tension the foetus is especially compromised. Others have suggested that these effects are not due to smoking but are attributable to characteristics of smokers. Certainly occupations with potential carbon monoxide exposure, such as those associated with ul and aer, blast furnaces, acetylene, breweries, carbon black, coke ovens, garages, organic chemical synthesizers and petroleum refineries should be considered possible high risk occupations for pregnant employees.

Ethanol is also a widely used and researched agent associated with foetal growth retardation (as well as congenital anomalies). In a prospective study of 9,236 births, it was found that maternal alcohol consumption of more than 1.6 oz per day was associated with an increase in stillbirths and growth-retarded infants (Kaminski, Rumeau and Schwartz 1978). Smaller infant length and head circumference also are related to maternal alcohol ingestion.

In evaluating the possible effects of exposures on birth weight, some problematic issues must be considered. preterm delivery should be considered as a possible mediating outcome and the potential effects on gestational age considered. In addition, pregnancies having longer gestational length also have a longer opportunity for exposure. If enough women work late in pregnancy, the longest cumulative exposure may be associated with the oldest gestational ages and heaviest babies purely as an artifact. There are a number of procedures that can be used to overcome this problem including a variant of the Cox life-table regression model, which has the ability to handle time-dependent covariables.

Another problem centres on how to define lowered birth weight. Often studies define lower birth weight as a dichotomous variable, less than 2,500 g. The exposure, however, must have a very powerful effect in order to produce a drastic drop in the infant’s weight. Birth weight defined as a continuous variable and analysed in a multiple regression model is more sensitive for detecting subtle effects. The relative paucity of significant findings in the literature in relationship to occupational exposures and SGA infants may, in art, be caused by ignoring these design and analysis issues.

Conclusions

Studies of adverse pregnancy outcomes must characterize exposures during a fairly narrow window of time. If the woman has been transferred to another job or laid off work during a critical period of time such as organogenesis, the exposure-effect relationship can be severely altered. Therefore, the investigator is held to a high standard of identifying the woman’s exposure during a critical small time period as compared with other studies of chronic diseases, where errors of a few months or even years may have minimal impact.

Uterine growth retardation, congenital anomaly and spontaneous abortions are frequently evaluated in occupational exposure studies. There is more than one approach available to assess each outcome. These end-points are of public health importance due to both the psychological and the financial costs. Generally, nonsecificity in the exposure-outcome relationships has been observed, e.g., with exposure to lead, anaesthetic gases and solvents. Because of the potential for nonsecificity in the exposure-effect relationship, studies should be designed to assess several end-points associated with a range of possible mechanisms.

Back

The reconciliation of work and maternity is an important public health issue in industrialized countries, where more than 50% of women of child-bearing age work outside the home. Working women, unions, employers, politicians and clinicians are all searching for ways of preventing work-induced unfavourable reproductive outcomes. Women want to continue working while pregnant, and may even consider their physician’s advice about lifestyle modifications during pregnancy to be overprotective and unnecessarily restrictive.

physiological Consequences of pregnancy

At this point, it would be useful to review a few of the physiological consequences of pregnancy that may interfere with work.

A pregnant woman undergoes profound changes which allow her to adapt to the needs of the foetus. Most of these changes involve the modification of physiological functions that are sensitive to changes of posture or physical activity—the circulatory system, the respiratory system and water balance. As a result, physically active pregnant women may experience unique physiological and physiopathological reactions.

The main physiological, anatomical, and functional modifications undergone by pregnant women are (Mamelle et al. 1982):

1. An increase in peripheral oxygen demand, leading to modification of the respiratory and circulatory systems. Tidal volume begins to increase in the third month and may amount to 40% of re-pregnancy values by the end of the pregnancy. The resultant increase in gas exchange may increase the hazard of the inhalation of toxic volatiles, while hyperventilation related to increased tidal volume may cause shortness of breath on exertion.
2. Cardiac output increases from the very beginning of pregnancy, as a result of an increase in blood volume. This reduces the heart’s ability to adapt to exertion and also increases venous pressure in the lower limbs, rendering standing for long periods difficult.
3. Anatomical modifications during pregnancy, including exaggeration of dorsolumbar lordosis, enlargement of the polygon of support and increases in abdominal volume, affect static activities.
4. A variety of other functional modifications occur during pregnancy. Nausea and vomiting result in fatigue; daytime sleepiness results in inattention; mood changes and feelings of anxiety may lead to interpersonal conflicts.
5. Finally, it is interesting to note that the daily energy requirements during pregnancy are equivalent to the requirements of two to four hours of work.

Because of these profound changes, occupational exposures may have special consequences in pregnant women and may result in unfavourable pregnancy outcomes.

Epidemiological Studies of Working Conditions and preterm Delivery

Although there are many possible unfavourable pregnancy outcomes, we review here the data on preterm delivery, defined as the birth of a child before the 37th week of gestation. preterm birth is associated with low birth weight and with significant complications for the newborn. It remains a major public health concern and is an ongoing reoccupation among obstetricians.

When we began research in this field in the mid-1980s, there was relatively strong legislative protection of pregnant women’s health in France, with prenatal maternity leave mandated to start six weeks prior to the due date. Although the preterm delivery rate has fallen from 10 to 7% since then, it appeared to have levelled off. Because medical prevention had apparently reached the limit of its powers, we investigated risk factors likely to be amenable to social intervention. Our hypotheses were as follows:

• Is working per se a risk factor for preterm birth?
• Are certain occupations associated with an increased risk of preterm delivery?
• Do certain working conditions constitute a hazard to the pregnant woman and foetus?
• Are there social preventive measures which could help reduce the risk of preterm birth?

Our first study, conducted in 1977–78 in two hospital maternity wards, examined 3,400 women, of whom 1,900 worked during pregnancy and 1,500 remained at home (Mamelle, Laumon and Lazar 1984). The women were interviewed immediately after delivery and asked to describe their home and work lifestyle during pregnancy as accurately as possible.

We obtained the following results:

Work per se

The mere fact of working outside the home cannot be considered to be a risk factor for preterm delivery, since women remaining at home exhibited a higher prematurely rate than did women who worked outside the home (7.2 versus 5.8%).

Working conditions

An excessively long work week appears to be a risk factor, since there was a regular increase in preterm delivery rate with the number of work hours. Retail-sector workers, medical social workers, specialized workers and service personnel were at higher risk of preterm delivery than were office workers, teachers, management, skilled workers or supervisors. The prematurely rates in the two groups were 8.3 and 3.8% respectively.

Table 1. Identified sources of occupational fatigue

Source: Mamelle, Laumon and Lazar 1984.

Task analysis allowed identification of five sources of occupation fatigue: posture, work with industrial machines, physical workload, mental workload and the work environment. Each of these sources of occupational fatigue constitutes a risk factor for preterm delivery (see tables 1 and 2).

Table 2. Relative risks (RR) and fatigue indices for preterm delivery

Source: Mamelle, Laumon and Lazar 1984.

Exposure to multiple sources of fatigue may result in unfavourable pregnancy outcomes, as evidenced by the significant increase of the rate of preterm delivery with an increased number of sources of fatigue (table 3). Thus, 20% of women had concomitant exposure to at least three sources of fatigue, and experienced a preterm delivery rate twice as high as other women. Occupational fatigue and excessively long work weeks exert cumulative effects, such that women who experience intense fatigue during long work weeks exhibit an even higher prematurely rate. preterm delivery rates increase further if the woman also has a medical risk factor. The detection of occupational fatigue is therefore even more important than the detection of medical risk factors.

Table 3. Relative risk of prematurity according to number of occupational fatigue indices

Source: Mamelle, Laumon and Lazar 1984

European and North American studies have confirmed our results, and our fatigue scale has been shown to be reproducible in other surveys and countries.

In a case-control follow-u study conducted in France a few years later in the same maternity wards (Mamelle and Munoz 1987) , only two of the five previously defined indices of fatigue were significantly related to preterm delivery. It should however be noted that women had a greater opportunity to sit down and were withdrawn from physically demanding tasks as a result of preventive measures implemented in the workplaces during this period. The fatigue scale nevertheless remained a predictor of preterm delivery in this second study.

In a study in Montreal, Quebec (McDonald et al. 1988), 22,000 pregnant women were interviewed retrospectively about their working conditions. Long work weeks, alternating shift work and carrying heavy loads were all shown to exert significant effects. The other factors studied did not appear to be related to preterm delivery, although there appears to be a significant association between preterm delivery and a fatigue scale based on the total number of sources of fatigue.

With the exception of work with industrial machines, no significant association between working conditions and preterm delivery was found in a French retrospective study of a representative sample of 5,000 pregnant women (Saurel-Cubizolles and Kaminski 1987). However, a fatigue scale inspired by our own was found to be significantly associated with preterm delivery.

In the United States, Homer, Beredford and James (1990), in a historical cohort study, confirmed the association between physical workload and an increased risk of preterm delivery. Teitelman and co-workers (1990), in a prospective study of 1,200 pregnant women, whose work was classified as sedentary, active or standing, on the basis of job description, demonstrated an association between work in a standing position and preterm delivery.

Barbara Luke and co-workers (in press) conducted a retrospective study of US nurses who worked during pregnancy. Using our occupational risk scale, she obtained similar results to ours, that is, an association between preterm delivery and long work weeks, standing work, heavy workload and unfavourable work environment. In addition, the risk of preterm delivery was significantly higher among women with concomitant exposure to three or four sources of fatigue. It should be noted that this study included over half of all nurses in the United States.

Contradictory results have however been reported. These may be due to small sample sizes (Berkowitz 1981), different definitions of prematurely (Launer et al. 1990) and classification of working conditions on the basis of job description rather than actual workstation analysis (Klebanoff, Shiono and Carey 1990). In some cases, workstations have been characterized on a theoretical basis only—by the occupational physician, for example, rather than by the women themselves (peoples-Shes et al. 1991). We feel that it is important to take subjective fatigue—that is, fatigue as it is described and experienced by women—into account in the studies.

Finally, it is possible that the negative results are related to the implementation of preventive measures. This was the case in the prospective study of Ahlborg, Bodin and Hogstedt (1990), in which 3,900 active Swedish women completed a self-administered questionnaire at their first prenatal visit. The only reported risk factor for preterm delivery was carrying loads weighing more than 12 kg more often than 50 times per week, and even then the relative risk of 1.7 was not significant. Ahlborg himself points out that preventive measures in the form of aid maternity leave and the right to perform less tiring work during the two months receding their due date had been implemented for pregnant women engaged in tiring work. Maternity leaves were five times as frequent among women who described their work as tiring and involving the carrying of heavy loads. Ahlborg concludes that the risk of preterm delivery may have been minimized by these preventive measures.

preventive Interventions: French Examples

Are the results of aetiological studies convincing enough for preventive interventions to be applied and evaluated? The first question which must be answered is whether there is a public health justification for the application of social preventive measures designed to reduce the rate of preterm delivery.

Using data from our previous studies, we have estimated the proportion of preterm births caused by occupational factors. Assuming a rate of preterm delivery of 10% in populations exposed to intense fatigue and a rate of 4.5% in non-exposed populations, we estimate that 21% of premature births are caused by occupational factors. Reducing occupational fatigue could therefore result in the elimination of one-fifth of all preterm births in French working women. This is ample justification for the implementation of social preventive measures.

What preventive measures can be applied? The results of all the studies lead to the conclusion that working hours can be reduced, fatigue can be lessened through workstation modification, work breaks can be allowed and prenatal leave can be lengthened. Three cost-equivalent alternatives are available:

• reducing the work week to 30 hours starting from the 20th week of gestation
• prescribing a work break of one week each month starting in the 20th week of gestation
• beginning prenatal leave at the 28th week of gestation.

It is relevant to recall here that French legislation provides the following preventive measures for pregnant women:

• guaranteed employment after childbirth
• reduction of the workday by 30 to 60 minutes, applied through collective agreements
• workstation modification in cases of incompatibility with pregnancy
• work breaks during pregnancy, prescribed by attending physicians
• prenatal maternity leave six weeks prior to the due date, with a further two weeks available in case of complications
• postnatal maternity leave of ten weeks.

A one-year prospective observational study of 23,000 women employed in 50 companies in the Rhône-Ales region of France (Bertucat, Mamelle and Munoz 1987) examined the effect of tiring work conditions on preterm delivery. Over the period of the study, 1,150 babies were born to the study population. We analysed the modifications of working conditions to accommodate pregnancy and the relation of these modifications to preterm delivery (Mamelle, Bertucat and Munoz 1989), and observed that:

• Workstation modification was reformed for only 8% of women.
• 33% of women worked their normal shifts, with the others having their workday reduced by 30 to 60 minutes.
• 50% of women took at least one work break, apart from their prenatal maternity leave; fatigue was the cause in one-third of cases.
• 90% of women stopped working before their legal maternity leave began and obtained at least the two weeks leave allowed for in the case of complications of pregnancy; fatigue was the cause in half the cases.
• In all, given the legal prenatal leave period of six weeks prior to the due date (with an additional two weeks available in some cases), the real duration of prenatal maternity leave was 12 weeks in this population of women subjected to tiring work conditions.

Do these modifications of work have any effect on the outcome of pregnancy? Workstation modification and the slight reduction of the workday (30 to 60 min) were both associated with non-significant reductions of the risk of preterm delivery. We believe that further reductions of the work week would have a greater effect (table 4).

Table 4. Relative risks of prematurity associated with modifications in working conditions

¹ In a reduced sample of 778 women with no previous or present obstetric pathology.

Source: Mamelle, Bertucat and Munoz 1989.

To analyse the relation between prenatal leave, work breaks and preterm delivery, it is necessary to discriminate between preventive and curative work breaks. This requires restriction of the analysis to women with uncomplicated pregnancies. Our analysis of this subgroup revealed a reduction of the preterm delivery rate among women who took work breaks during their pregnancy, but not in those who took prolonged prenatal leave (Table 9).

This observational study demonstrated that women who work in tiring conditions take more work breaks during their pregnancies than do other women, and that these breaks, particularly when motivated by intense fatigue, are associated with reductions of the risk of preterm delivery (Mamelle, Bertucat and Munoz 1989).

Choice of preventive Strategies in France

As epidemiologists, we would like to see these observations verified by experimental preventive studies. We must however ask ourselves which is more reasonable: to wait for such studies or to recommend social measures aimed at preventing preterm delivery now?

The French Government recently decided to include a “work and pregnancy guide”, identical to our fatigue scale, in each pregnant woman’s medical record. Women can thus calculate their fatigue score for themselves. If work conditions are arduous, they may ask the occupational physician or the person responsible for occupational safety in their company to implement modifications aimed at alleviating their workload. Should this be refused, they can ask their attending physician to prescribe rest weeks during their pregnancy, and even to prolong their prenatal maternity leave.

The challenge is now to identify preventive strategies that are well adapted to legislation and social conditions in every country. This requires a health economics approach to the evaluation and comparison of preventive strategies. Before any preventive measure can be considered generally applicable, many factors have to be taken into consideration. These include effectiveness, of course, but also low cost to the social security system, resultant job creation, women’s references and the acceptability to employers and unions.

This type of problem can be resolved using multicriteria methods such as the Electra method. These methods allow both the classification of preventive strategies on the basis of each of a series of criteria, and the weighting of the criteria on the basis of political considerations. Special importance can thus be given to low cost to the social security system or to the ability of women to choose, for example (Mamelle et al. 1986). While the strategies recommended by these methods vary depending on the decision makers and political options, effectiveness is always maintained from the public health standpoint.

Back