Contraception and Congenital Malformations

Allegations that inadvertent pregnancies occurring in users of contraception are associated with congenital anomalies are common. Fortunately, in virtually all instances. there is little to no scientific basis for such claims. Evaluating these claims requires consideration of the two general mechanisms responsible for human malformation: teratogenesis and mutagenesis. A teratogen deleteriously affects the embryo that was otherwise destined to develop normally and thus can only exert its effect during a pregnancy. A mutagen causes a fresh mutation in the sperm or oocyte responsible for fertilization; thus, potential risk persists after the contraceptive agent has been discontinued. This chapter reviews potential teratogenicity and potential mutagenicity of the various contraceptive methods.

Principles of teratogenesis are detailed elsewhere; however, repeating salient principles might be useful here.

Whether a specific agent is teratogenic in humans depends on several factors: the specific agent and its dosage, the time point in gestation during which the fetus was exposed, and the genetic susceptibility of the mother and the fetus. The embryo is relatively resistant to teratogenic insults in the first few weeks of life (the first 2 embryonic weeks or the first 4 weeks following menstruation). During this time, a major insult would probably be embryotoxic; however, if the embryo survives, no organ-specific anomalies would result. By the third embryonic week, however, organ differentiation begins. During organogenesis (embryonic weeks 3 to 8 or gestational weeks 5 to 10), the embryo is susceptible to the effects of teratogens. Organogenesis is then followed by a period of organ growth. During this time, a teratogen can deleteriously affect growth of the fetus or a particular organ but usually not produce a specific malformation. Exceptions to this rule are brain and gonadal tissues, where insults later in gestation may result in abnormalities. Susceptibilities to specific agents vary with genetic predisposition.

Various types of studies have been conducted to investigate the potential teratogenicity of contraceptive methods, but all have pitfalls. Cohort (prospective) studies are theoretically preferable; however, such studies are expensive and laborious and suffer from few subjects providing new information (affected). Therefore, case-control (retrospective) studies are more often used. In such studies, women with a specified abnormal outcome (e.g., mother having an infant with a cardiac anomaly) are matched with control subjects who differ only with respect to the variable being tested (e.g., drug exposure). Unfortunately, major pitfalls frequently exist with the retrospective study design. Often years have passed from the occurrence to the study interview (memory bias). Women whose pregnancies result in an abnormal outcome may search much harder for factors potentially responsible (recall bias). Therefore, case-control studies frequently yield spurious positive associations. In an analysis of epidemiologic studies of teratogens, Khoury and colleagues¹ concluded that the potential for bias in such studies was so likely that weak to moderate associations should be interpreted with caution. This included values of relative risks (RRs) up to 3.

General Comments

Both progestogens (progesterone as well as synthetic progestins) and estrogens have been implicated as teratogens and mutagens. Claims made could apply to oral progestogens as well as injectable or implantable progestogens used for pregnancy diagnosis, pregnancy maintenance, or birth control. Certain progesterones, namely the 19-nor-testosterones, norethisterone and norgestrel, can virilize female fetuses if administered in high doses at the susceptible period of pregnancy;² however, this dose is considerably in excess of that present in oral or injectable/implantable contraceptives. More relevant clinically are allegations that oral contraceptives are associated with cardiac defects, limb reduction defects, neural tube defects, hypospadias, and other anomalies. We shall show that there appears to be no substantive evidence that either oral contraceptives or injectable/implantable progestins are teratogenic or mutagenic.

In evaluating claims of teratogenicity of hormones, it would be optimal to consider separately exposure to progestins and progestogens only, estrogens only, and combined oral contraceptives. Available data do not permit such an analysis. In fact, such data probably never will exist because the progestin exposure during pregnancy is becoming uncommon. In a cohort study of 729 United States women, only two received progestins.³ Investigations usually pool outcomes following exposure to a variety of sex hormones. Fortunately, the only estrogen implicated as a teratogen is diethylstilbestrol, an agent that is not a component of oral contraceptives. Therefore, pooling progestogen and oral contraceptive exposure still provides useful information in evaluating risks to the fetus.

Cardiac Anomalies

Levy and associates⁴ were the first to claim that progestins cause cardiac anomalies. In a 1973 case-control study, seven of 76 mothers delivered of infants with transposition of the great vessels received hormones in the first trimester; none (0) of the 76 controls had been exposed (p < 0.007). Similar findings were published by Nora and Nora,⁵ whose initial retrospective case-control study found that 20 of 224 (8.9%) mothers delivered of infants with cardiac defects recollected receiving an estrogen/progesterone compound; 4 of 262 controls (p < 0.001) were similarly affected. However, in a follow-up prospective study, Nora and coworkers⁶ reported no significant differences in hormone use between the 60 mothers and their controls. Despite these negative findings, a second prospective study was undertaken with two controls per subject. In this second study, 31 of 176 mothers with affected offspring were said to have received hormones, compared with 21 of 352 control mothers (p < 0.001).

Two other early case-control studies reported positive correlations. Janerich and partners⁷ identified 104 infants with cardiac defects through birth certificates in New York state. Of the 104, mothers of 18 received hormones; 16 cases involved hormones for pregnancy diagnosis and two involved inadvertent contraceptive use. Significantly fewer controls reported exposure. A relationship was also found between progestins and overall anomaly rates in a study by Greenberg and coworkers;⁸ cardiac anomalies were among the birth defects responsible for this finding.

Although troublesome, these studies were counterbalanced by other negative case-control studies. In a case-control study of 1370 children with congenital malformations and exposure to oral contraceptives “around the time of conception,” Bracken and colleagues⁹ found no significant increase in the rates of certain cardiac malformations (tetralogy of Fallot, ventral septal defect, atrial septal defect). Moreover, no association was found with any malformation when exposure occurred in the year before conception or during pregnancy, nor was there any relationship when exposure was stratified by specific estrogen or progestin. In fact, the issue of cardiac teratogenicity by progestogens may not have been raised into the Klieg lights were it not for a report from the US Collaborative Perinatal Project^10,11 of a positive association between sex hormone exposure during gestation and cardiac anomalies. Of the 1042 offspring said to be exposed, 19 had cardiac defects (1.82%); 385 of 49,240 (0.78%) unexposed offspring were affected (RR, 2.3, p < 0.05). Although analysis by specific hormones was not possible because of the relatively small number of cases, inadvertent oral contraceptive use was noted to carry the greatest RR (2.4). Exposure only to progesterone carried a lower but still significantly increased RR (1.8, p < 0.05); exposure to estrogens only carried a RR of 1.44.

The conclusions of Heinomen and associates^10,11 were questioned. Wiseman and Dodds-Smith¹² reevaluated original data from the US Collaborative Perinatal Project and found several shortcomings. First, a priori risk of anomalies were dissimilar between subjects and controls. In four of the 17 hormone-exposed cases, a previous pregnancy was characterized by a major malformation (ventral septal defect, Down syndrome with a cardiac defect, neonatal death with serious malformations, and a stillborn); only one case in a control group of 100 (selected from the 1023 who had a noncardiac malformation and were exposed to progestogens) had a previous history of a major anomaly. Thus, subject and control groups were not comparable. In addition, two infants with cardiac anomalies in the exposed group had Down syndrome; the cardiac anomaly was surely related not to drug exposure but rather to underlying trisomy. Timing of exposure was also implausible in many cases. Among the 19 progestogen-exposed infants with cardiac defects, four were exposed during the first lunar month, a period when anomalies are not ordinarily produced (all-or-none period); three other infants were exposed only in the fourth month, long after heart development is completed (42 embryonic or 56 gestational days). Wiseman and Dodd-Smith¹² concluded that no significant association between hormone use and cardiac anomalies existed in the US Collaborative Perinatal Project data.

In 1994, E.B. Hook¹³ added to the controversy by publishing an evaluation of the Wiseman and Dodd-Smith analysis. He began by noting the attributes of the original study by Heinomen and colleagues.¹¹ The drug exposure history was obtained before delivery; twins and infants exposed to rubella were excluded. He criticized Wiseman and Dodd-Smith for redefining the time of pertinent exposure for the affected cases, but not the unaffected cases, and for excluding infants with Down syndrome after the data were obtained, rather than excluding all syndromes associated with congenital heart lesions before analysis. He added that there is no proof that prenatal hormone exposure does not predispose offspring with Down syndrome to heart disease. Including the two cases of Down syndrome, Hook calculated an RR of 2.48; without these two cases, it is 1.87. Hook states that confounding variables may still account for the association and that his reanalysis does not establish a true causal connection between hormones and congenital heart malformations. Hook concludes by stating the hypothesis should be tested in other data sets and offering an interesting theory: hormones may be antiabortifacients and therefore increase survival to livebirth affected fetuses that would otherwise spontaneously abort.

In further support of the thesis that progestins are not teratogenic for cardiac defects, several other studies have failed to detect an association between cardiac anomalies and sex hormones. In a 1979 study, Harlap and partners¹⁴ found a positive association between progestins and overall abnormal outcome (Table 1). In the first study by Harlap and partners,¹⁴ exposure during the first lunar month was, as in the US Collaborative study, erroneously considered to be a susceptible interval. In a 1985 study, Harlap and colleagues¹⁵ did not find such an association but did find, albeit before corrections for multiple comparisons, an association with coarctation of the aorta and heart valve defects. This later study¹⁵ had shortcomings by vaguely defining exposure as oral contraceptive use after the last menstrual period. Representative of the frequent negative studies is that of Spira and coworkers,¹⁶ who observed on a follow-up basis more than 20,000 French women throughout their gestation; almost half (n = 9566) received hormones, usually for pregnancy diagnosis or pregnancy maintenance. The anomaly rate was no different in the exposed group than in the unexposed group. When the same population was reexamined by Goujard and Rumeau-Rouquette,^17,18 no differences in the rates of cardiac anomalies were observed in exposed (43%) versus unexposed mothers (41%). Cohort studies in Finland,¹⁹ Germany,²⁰ Sweden,²¹ Great Britain,^22,23,24 and the United States^25,26 similarly revealed no increased incidence of cardiac anomalies. In another type of study, Nishimura and colleagues²⁷ also failed to detect cardiac anomalies in 108 microdissected embryos exposed to hormones, despite several controls having cardiac defects.

TABLE 1. Major Prospective Studies Evaluating Effects of Progestin Exposure during Pregnancy*

*The consensus is that progestins in the doses received were not teratogenic.
(Modified from Simpson JL, Phillips OP: Spermicides, hormonal contraception and congenital malformations. Adv Contracept 6:141, 1990.)

Several studies have complemented the above by offering information on specific hormones or specific clinical situations. Katz and colleagues²⁸ prospectively followed mothers who were treated with hormones because of bleeding during the first trimester. One group (n = 1608) was treated with progestins (either medroxyprogesterone acetate or injectable 17α-hydroxyprogesterone caproate) beginning in the first trimester; their outcomes were compared with those of 1146 mothers who were not exposed. No significant difference was observed between the two groups with respect to any malformation. Another study of 988 Minnesota offspring exposed in utero to progestins, 17α-hydroxyprogesterone caproate and progesterone, failed to demonstrate an increased risk of cardiovascular anomalies or other anomalies.²⁹ Check and partners³⁰ reported outcomes of 382 women treated for recurrent abortions with either 17-hydroxyprogesterone (17-OHP) or progesterone vaginal suppositories. Five anomalies were identified, two of which were cardiac (ventral septal defects and transposition of great vessels) and in patients exposed to progesterone suppositories; the authors concluded that their data support those of others. There is no increased risk in the congenital anomaly rate as result of in utero exposure to natural progesterone or 17-OHP.

In a study of 8816 births in Thailand, Pardthaisong and associates³¹ compared the incidence of cardiovascular anomalies among several groups: users of oral contraceptives either before conception or during gestation (n = 3038), injectable contraceptive depo-medroxyprogeserone (DMPA) (Depo-Provera) (n = 1229), and noncontraceptors (n = 4023). No significant differences were found among the groups. Interestingly, the incidence of anomalies among oral contraceptive users was lower than the incidence in the general population. Yovich and associates³² examined the fetal effects of medroxyprogesterone in an especially well-designed cohort study of 449 Australian women having 508 pregnancies. Recurrent abortion was the indication for treatment in 199 cases; therapy was begun at 5 weeks gestation and continued until 16 weeks. Threatened abortion was the indication in 309 cases. Pregnancy outcome was compared with matched controls of women conceiving after infertility treatments. Only one cardiac defect was noted in the exposed group, and this in an individual with Noonan syndrome. In 1998, Martinez-Frias and colleagues³³ published a case-control study that included 20,388 liveborn malformed infants. This large Spanish study found no association between first-trimester exposure to sex hormones, including oral contraceptives, and children with cardiac malformations.

Wiseman³⁴ provided additional information by inquiring whether the rates of congenital malformations in the United Kingdom were altered as patterns of progestin usage for diagnosing pregnancy were altered. Over a 14-year period after its introduction, progestin use first increased sharply and then rapidly declined. During this interval, there was no correlation between incidence of congenital malformations and progestin use. The congenital malformation rate remained stable over years when use of hormone pregnancy test was declining. The observation of only two progestin exposures among 729 women observed by Simpson and associates³ is further consistent with these reassuring secular trends.

Overall, the consensus of many studies is that progestins are not cardiac teratogens. Even the US Food and Drug Administration has belatedly accepted this, recently deleting allusions to cardiac teratogenicity in its warning labels for progestins.

Limb Reduction Deformities

Limb reduction defects are defined as shortening or absence of a limb, finger, or a toe. Although genetic heterogenicity exists, most limb reduction defects can be considered polygenic/multifactorial in etiology.

Janerich and coworkers³⁵ were the first to report an association between progestogens and limb reduction defects. Of 108 women with an affected infant, 15 had received hormones (inadvertent oral contraceptive exposure, hormone pregnancy test, or hormones for pregnancy maintenance). Only four of 108 controls were exposed (p < .05). The association was not significant when oral contraceptives were considered separately. Greenberg and colleagues⁸ found an overall increase in anomalies following progestogen exposure; limb reduction defects contributed to this increase.

More recently, a retrospective case-control study in Australia by Kricker and associates³⁶ observed an association between oral contraceptive use and limb abnormalities. Of 155 limb-deficient children, 18 were reported as having been exposed, compared with one normal control (RR, 16.6; p < .05). No association was found when progestin use only was considered. The anomalies were varied, including dysmelic longitudinal defects (e.g., absence of radius or thumb) and transverse defects (e.g., amputation of limbs or digits). Unfortunately, serious methodologic shortcomings exist with this study. The potential for recall and memory bias was enormous. Interviews were conducted on the average 4.5 years after the birth of the child! Moreover, the crucial issue of when during gestation the fetus was exposed was not addressed.

Far better designed case-control studies have failed to confirm an association between limb abnormalities and maternal hormonal exposure. In a study of 1370 exposed women, Bracken and partners⁹ showed no statistically significant association. Oakley and coworkers³⁷ failed to observe a relationship in a well-constructed case-control study in which the control group consisted of women who had delivered children with chromosomal abnormalities. Use of this abnormal control group obviated potential for recall bias. Among recent case-control studies is that of Lammer and Cordero.³⁸ Among 1091 infants with major malformations, no significant association was found between hormone exposure and limb reduction defects. Pardthaisong and partners³¹ also failed to demonstrate an association between limb reduction defects and use of injectable contraceptive medroxyprogesterone or oral contraceptives, either prior to conception or during gestation.

Cohort studies have further failed to confirm an association, as shown in Table 1. Prospective studies investigating limb reduction defects have specific validity because missing digits or severe limb shortening should be obvious to even the casual observer, (By contrast, cardiac defects may pass undiagnosed until later in childhood.) Of most recent interest is the study of Yovich and colleagues,³² who found no limb reduction defects among 508 exposed pregnancies. Also of interest is the failure of Nishimura and associates²⁷ to observe limb reduction deformities in 108 microdissected embryos recovered from progestin-exposed mothers.

In summary, most retrospective studies and all cohort studies have failed to show an association between progestogen exposure and limb reduction defects.

Hypospadias

The first claim that antenatal exposure to progestins, medroxyprogesterone in particular, caused penile or perineoscrotal hypospadias was by Aarskog³⁹. This study was uncontrolled, but supportive case-control studies were published. A Latin American collaborative case-control study (ECLAMC)⁴⁰ found a 2.4 RR for progestin exposure. Of 314 cases of hypospadias, 24 (7.6%) reported exposure to progestins; 12 of 319 controls (3.8%) (p < .05) were exposed. However, details concerning the time of exposure were not provided, and incidence of hypospadias varied widely among the countries surveyed. Czeizel and colleagues⁴¹ found a similar association in Hungary; 28 of 294 (9.5%) mothers delivered of male children with hypospadias received sex hormones, compared with 12 of an unspecified number of controls. This difference was said to be significant, but there is little confidence that controls were properly matched. The high prevalence of hypospadias in the subjects' male relatives suggests unwitting selection bias.⁴²

More recent studies of better experimental design have failed to confirm an association between progestins and hypospadias.^9,27,42,43 In addition, neither the US Collaborative Perinatal Project^10,11 nor other cohort studies reviewed in Table 1 found an effect. Of interest, Katz and colleagues²⁸ found no cases of pseudohermaphroditism among infants exposed to progestins in early gestation following bleeding. Incidence of hypospadias and clitoromegaly was the same in the exposed (n = 1605) as in the control group (n = 1146). Yovich and associates³² found two cases of hypospadias among the 508 exposures with medroxyprogesterone, an incidence surely differing little from the incidence in western Australia (said to be 0.5%). In the cohort study of 988 offspring exposed in utero to exogenous progestins, Resseguie and associates²⁹ found no increased risk of hypospadias. No cases of hypospadias were observed in offspring of 382 women treated with vaginal progesterone suppositories or 17-hydroxyprogesterone caproate early in gestation.³⁰

Kallen and associates⁴⁴ published the largest study on hormone therapy during pregnancy and hypospadias to date. This international case-control study identified 846 cases of isolated hypospadias. Questionnaires were obtained from mothers of cases and controls. Hormone use was reported in 39 case-control pairs: in 27 the case was exposed; in 12, the control was exposed. The odds ratio (OR) was 2.3. Several confounding factors were identified. Early pregnancy bleeding appeared to have a strong positive effect. Furthermore, there was a lack of association between the prenatal timing of exposure to hormones and location of the urethral meatus or the severity of the disorder. Those authors contend that this was evidence against a causal relationship, and that recall or interviewer bias likely explains the association.

In summary, exposure in utero to progestins seems unlikely to result in abnormal development of the male genitalia. Despite our belief that the issue is well studied, it should be noted that the US Food and Drug Administration still considers the topic in need of additional studies.

Urinary Tract Anomalies

Embryonic development of urinary tract is closely linked to that of the genital tract. Therefore, it is plausible that prenatal hormone exposure may influence normal development of the kidneys and renal collecting system. Surprisingly few studies have specifically addressed the association between urinary tract anomalies and exposure to oral contraceptive. One study investigating this hypothesis, in our opinion, is characterized by several pitfalls that cloud the conclusions.

Using the Washington State Birth Defect Registry for 1990 to 1991, De-Kun and partners⁴⁵ scanned for codes designating congenital urinary tract anomalies (CUTA) in singleton liveborns. Included were all renal, ureter, bladder, and urethral anomalies except those associated with chromosome abnormalities. Infantile polycystic kidneys were excluded because of its well-known autosomal-recessive inheritance pattern. Of 187 eligible cases found, 117 mothers were ultimately interviewed. Questionnaires were administered through an in-person interview about contraceptive use at the time of conception and afterward. The time of conception was determined by the respondents. Risk of CUTA was evaluated for contraceptive use 4 or fewer weeks after conception and more than 4 weeks after conception.

Over 44% of the cases had anomalies of other organ systems. Nine cases (7.6%) reported use of oral contraception compared with eight of 368 controls (2.2%) (OR, 4.8). Both cases and controls reported about the same use of prenatal ultrasound (96.6% versus 93%) (If prenatal ultrasound had detected asymptomatic anomalies, this would have been included spuriously as a case.)

Recall bias is possible. More than a year may have passed between the birth and the interview. Although interviewers and respondents were “not aware of the study hypothesis,” the questionnaire appeared to focus almost entirely on the contraceptive use. Despite acknowledging potential recall bias, the authors concluded that their study suggests an association between oral contraceptive use and CUTA.

There are more serious problems with this study. At least 11 types of renal anomalies were included. Whereas renal agenesis is thought to have a polygenic/multifactorial etiology, multicystic kidney may have a vascular etiology. Congenital hydronephrosis is the result of a mechanical obstructive process. All were included in the analysis, along with anomalies of the kidney, ureter, bladder, and urethra in an “other” category. No exposure to any single agent would be expected to be responsible for such diverse anomalies. Furthermore, cases with multiple anomalies were included, with no apparent attempt to exclude syndromes whose etiology was established and known not to be associated with teratogens (e.g., Meckel-Gruber or Mohr type II).

The mother was asked to establish the day of conception and relate this to her contraceptive use. There is an innate potential for error in dating contraceptive use using this method. Those authors state that to allow for this potential error, risk was assessed for the two different durations of oral contraceptive use (4 weeks or less and more than 4 weeks from conception). However, 4 weeks or less after conception includes the time period when exposure would not have a teratogenic effect (the all-or-none period). If cases with exposure in this time frame were excluded, few cases would remain for statistical analysis. The number of cases analyzed in the study is already too small from which to draw conclusions. Overall, the study of De-Kun and coworkers⁴⁵ does not make a case for a teratogenic effect of oral contraceptives on the urinary tract.

Harlap and associates¹⁵ included renal anomalies in their case-control study of malformed infants and maternal contraceptive use and showed no association. Our conclusion is consistent with theirs.

Neural Tube Defects and Hydrocephalus

Neural tube defects (NTDs) and to a lesser extent hydrocephalus were at one time presumed to be associated with hormone exposure. This claim now has been discounted almost completely.

The original claim was that of Gal and associates,⁴⁶ a 1967 case-control study in which 19 of 100 women delivered of infants with myelomeningocele or hydrocephalus recollected receiving hormones (estradiol plus ethisterone or norethisterone) for pregnancy diagnosis. Only four of 100 controls recalled hormone exposure (p < .01). A study by Greenberg and associates⁸ demonstrated a possible relationship between NTDs and hormone exposure, 25 of 93 infants with NTDs having had a history of hormone exposure. An important variable ignored in the study by Gal was prior reproductive history, especially relevant in the United Kingdom because of its high incidence of NTDs. (In the United Kingdom, the recurrence risk for first degree relatives is 5%.) The stage of embryogenesis at which exposure occurred was also not considered. Because the neural tube closes at 28 embryogenic days, it is highly likely that hormones administered for pregnancy diagnosis were given after neural tube closure. Harlap and associates¹⁵ observed three infants with anencephaly or spina bifida in 850 women (0.35%) exposed to oral contraceptives in utero, compared with 27 infants in 32695 (0.08%) unexposed women (RR, 4.3). However, this study defined exposure as any contraceptive use after the last menstrual period. In some cases, it is likely that oral contraceptives may have been discontinued when the anticipated menstrual period did not occur. If true, exposure would have been restricted to the all-or-none period and not at all during the interval of organogenesis.

Later case-control studies failed to show an association between NTDs and progestins. No significant increase was seen in a UK study by Laurence and associates,⁴⁷ whose study population was larger than that of Gal and colleagues.⁴⁶ Other large negative case-control studies include those of Bracken and partners⁹ in Connecticut and Lammer and Cordero³⁸ in Georgia. Moreover, not a single prospective study has shown an association between progestins and NTDs and hydrocephalus (see Table 1). Finally, Nishimura and associates²⁷ failed to detect NTDs in 108 microdissected embryos exposed to progestins.

In summary, little credence should be given to the hypothesis that progestins cause neural tube defects or hydrocephalus.

Esophageal Atresia

That esophageal atresia is associated with progestin teratogenicity was raised in a case-control study by Lammer and Cordero.³⁸ They studied first trimester sex hormone exposure in 1091 infants having at least one of 11 major types of malformations. These included cleft lip and palate, small bowel atresia, esophageal atresia, anal atresia, anterior abdominal wall defects, and diaphragmatic hernia. A positive association was found between esophageal atresia and hormone exposure (OR, 2.84); no correlations were seen with other anomalies.

In the study by Bracken and associates,⁹ 1370 anomalies were considered, including tracheoesophageal fistulas, inguinal hernias, cleft lip and palates, pyloric stenoses, musculoskeletal anomalies, and polysyndactyly. No statistical significance was found for any individual malformation. In a cohort study of 1608 newborns exposed to progestogens, Katz and colleagues²⁸ studied various malformations, including polydactyly, talipes, genitourinary malformations, cleft lip and palate, and tracheoesophageal fistula. For none of these anomalies was a difference found between the study group and controls.

Martinez-Frias and associates,³³ in a Spanish case-control study of 1,264,962 liveborn infants, identified 23,740 malformed infants. Cases due to syndromes known to have dominant or recessive inheritance patterns or due to teratogenic effects were excluded, leaving 20,388 study cases. Mothers were questioned within the first 3 days after delivery; controls were selected from the same hospital, matched for sex. Any case with first-trimester exposure to sex hormones was studied; sex hormone exposure oral contraceptives, estrogen alone, progestogen alone, and estrogen plus progestogen. After analyzing for 600 different categories of defects, four significant associations were found. Prenatal exposure to oral contraceptives was associated with anomalies of the face, mouth, ear, and neck (OR, 2.05). Exposure to progestogens alone was associated with genital defects (OR, 1.56), cleft lip with or without cleft palate (OR, 5.11) and tracheoesophageal fistula (OR, 10.47). Whereas the first two mentioned associations were thought due to chance, the latter two, with their stronger risk values, received further analysis. It was found that in both categories, a higher proportion of cases of malformed infants had maternal vaginal bleeding, previous spontaneous abortions, other first-degree relatives with congenital anomalies, and maternal infertility than controls did. Those authors concluded that these confounders played a more important role in predisposing the infants to malformation than sex hormone exposure. Moreover, corrections for multiple comparison were not made. Table 1 summarizes the many cohort studies, none of which suggest an association.

Polydactyly and Syndactyly

In a case-control study of 8816 newborns in Thailand, Pardthaisong and partners³¹ obtained a contraceptive history on the mothers finding 1229 women who used injectable contraceptive DMPA, 3038 having used oral contraceptives before conception or during pregnancy and 4549 who used no contraceptive use before conception. The frequency of peripheral limb abnormalities, polydactyly and syndactyly was significantly increased in DMPA users (RR, 4.9/1000) compared to users of oral contraceptives (RR, 1.2/1000); no significant difference was seen in comparison to women receiving no contraception (1.7/1000). However, only three cases of polysyndactyly were exposed during the critical period of limb organogenesis (days 28 to 52). The rates of polysyndactyly observed in DMPA users were higher than those reported in other studies, for which reason, the authors concluded that the observed association may be the result of genetic predisposition in this population. We concur, noting as well that no other case-control study nor cohort study (see Table 1) reported an association between polydactyly and progestogens.

Multiple Anomalies (Nonspecific Increase in Anomalies)

A final teratogenic claim is that there exists a generalized (nonspecific) increase in anomalies following progestin or estrogen exposure. Actually, such a claim is biologically implausible. Every known human teratogen produces an anomaly or spectrum of anomalies characteristic for a given agent. It is not likely that a generalized increase in anomalies would result from a single teratogen. Investigations searching for an overall increase in anomalies associated with an agent should thus be considered as merely hypothesis-generating studies, the denouement awaiting studies dealing with specific anomalies.

These warnings notwithstanding, several case-control studies have claimed generalized increased anomaly rates associated with oral contraceptive exposure. One such study is that of Greenburg and colleagues;⁸ however, the validity of this study is in doubt because only a small and possibly unrepresentative proportion of eligible women participated. Pardthaisong and colleagues³¹ found a statistically significant increase in the overall rate of major malformations among users of DMPA when compared with oral contraception users; however, the rate among oral contraceptive users was unexpectedly low, suggesting self-selection bias. By contrast, no significant associations were found in case-control studies of Bracken⁹ and Oakley and respective partners.³⁷ The frequency of anomalies was not increased in 541 “pill-failure” pregnancies gathered by Harlap and Eldor.⁴⁸

Of 18 other cohort studies (see Table 1), only one claimed an association between progestins and overall anomaly rate. One of the two Harlap and associates'¹⁴ studies showed a small increase in RR. The other 17 prospective studies failed to show any generalized increase in anomalies. Most recently, Resseguie and colleagues²⁹ and Katz and associates²⁸ failed detect an increase in the major anomaly rate of offspring exposed in utero to progestins. Yovich and partners³² found no significant increase in congenital abnormalities in pregnancies exposed to medroxyprogesterone (15 of 366 or 4.1%) as compared to controls (15 of 428 or 3.5%).

A special aspect of the claim that progestins are responsible for an overall nonspecific increase in anomalies are claims that the VACTERL complex is associated with progestin exposure. Diagnosis of the VACTERL complex is made when three of the following seven organ systems are anomalous: vertebral, anal, cardiac, tracheal, esophageal, renal, and limb. In 1978, Nora and Nora⁶ found that 11 of 30 (36.6%) VACTERL probands reported progestin exposure, compared with 5 of 60 (8.3%) controls (p < .001). However, the sample size was small, and recall bias would probably be amplified in mothers whose infants had multiple malformations. Later studies^{2,49,50,51,52} failed to confirm such an association, nor has any cohort study found an association between hormone exposure and the VACTERL complex or any of its component parts (e.g., cardiac or limb abnormalities) (see Table 1).

In conclusion, little evidence supports the hypothesis that progestin exposure results in a nonspecific increase in malformations.

A deleterious agent may cause abnormalities not only by teratogenic mechanisms but also by inducing mutations in individual germ cells. A child subsequently conceived with that germ cell would be anomalous. Both chromosomal and gene mutations have been claimed to result from sex hormone exposure.

Numeric Chromosomal Abnormalities

That earlier hormone exposure could induce chromosomal abnormalities is not an unreasonable hypothesis. Oocytes resting in dictyotene of meiosis I until ovulation might complete meiosis sluggishly as result of prior hormone exposure, resulting in numeric abnormalities of the chromosomes.

The original concern was raised in 1970 by Carr,⁵³ who reported an increase in chromosomally abnormal abortuses among women previously using oral contraceptives. In such cases, 48% of abortuses were chromosomally abnormal, compared with 22% in controls; polyploidy was responsible for most of the excess. However, these observations were not confirmed by later studies. In the study of Boué and Boué,⁵⁴ 16% of abortuses from 243 prior contraceptive users had abnormal complements (16% polyploidy), compared with 63% (18% polyploidy) of 604 controls. Lauritsen⁵⁵ found a slight excess of abnormal complements in previous oral contraceptive users (61% versus 49%). The increase was due to monosomy X and structural abnormalities, not to polyploidy as Carr⁵³ had described. Alberman and colleagues⁵⁶ observed abnormalities in 32% of 524 prior users, compared with 26% in 428 controls. Similar differences by Dhardial and associates⁵⁷ were likewise statistically insignificant. Klinger and partners⁵⁸ found a higher frequency of abnormalities in induced abortions in prior contraceptive users (1%) than in controls (0.5%), but the difference was not statistically significant.

That no study could confirm Carr's initial findings⁵³ is especially noteworthy because any of several biases might be expected to increase anomaly rates spuriously in exposed groups. For example, surreptitious induced abortuses are more likely to exist in the control group than in past users of oral contraceptives, who are more likely to be attempting to achieve a pregnancy. In fact, this may prove to be the explanation for the unusually low (22%) frequency of chromosomally abnormal abortuses in Carr's controls, because that study was conducted prior to the legalization of abortion in North America.

Other evidence supports our thesis that oral contraceptives do not induce numeric chromosomal abnormalities. If prior contraceptive use results in chromosomal abnormalities, the frequency of spontaneous abortions in previous users should be increased substantially because 50% to 60% of first-trimester abortuses have cytogenetic abnormalities. An increased incidence of Down syndrome would also be expected because the same cytologic mechanism (nondisjunction in maternal meiosis I) is usually responsible for trisomy in abortuses and liveborns. Pardthaisong and partners³¹ found a statistically significant association between DMPA for contraception and chromosomal abnormalities (RR, 5.5; p < .05). Last use of DMPA occurred from 7 to 13 months before conception in three of four cases; the other mother of an infant with Down syndrome conceived while using DMPA. No association was noted among oral contraceptive users. Those authors stated that the positive association was more likely due to chance effects than causal association. In their case-control study, Bracken and associates⁹ found no increased incidence of Down syndrome among women who used oral contraceptives up to 12 months before conception. Moreover, follow-up of prior oral contraceptive users has failed to show an increase in Down syndrome.

In summary, initial concern generated by Carr's observation⁵³ of an excess of polyploid abortuses in prior oral contraceptive users has dissipated. The consensus is that neither progestins nor estrogens predispose to chromosomally abnormal abortuses.

Structural Chromosomal Abnormalities

Several in vitro studies claimed increased chromosomal breakage in lymphocytes of contraceptive users.² Often such studies are performed without blind analysis. However, there are several other positive lines of reassurance. No increase in structural chromosomal abnormalities was reported in the abortus studies already cited.^{53,54,55,56,57,58} Structural chromosomal abnormalities would also result in anomalous liveborns, and we have already concluded that no increase in the overall anomaly rate exists.

In conclusion, the possibility that children of women exposed to oral contraceptives have an increased frequency of structural chromosomal abnormalities seems remote.

Gene Mutations

Could progestin exposure result in mutations responsible for Mendelian or polygenic/multifactorial disorders? To evaluate this hypothesis, each locus should ideally be assessed for mutability. This is mathematically impossible, even if a large increase occurs over the baseline mutation rate of 10^-5 to 10^-6 locus/gamete/generation. As a result, studies can generally offer only a comparison between total anomaly rates in women exposed and not exposed to hormones. Anomalies are mendelian, polygenic/multifactorial, chromosomal, or environmental in etiology. This etiologic heterogeneity diminishes the power of available conclusions. Fortunately, available data are sufficiently reassuring that more refined investigators would not be expected to yield contrary results.^15,51 Not a single Mendelian disorder has increased in frequency since introduction of oral contraceptives.

An indirect method of researching gene mutation rates involves analysis of the gender ratio. Induction of lethal X-linked recessive mutations decreases the proportion of liveborn males; thus, lethal mutations at any X-linked loci contribute to a decreased male/female gender ratio. In fact, the larger cohort studies provide no evidence of an altered gender ratio, refuting several earlier studies of smaller sample size.⁴⁶

Finally, neither progestins nor estrogens are representative of classes of compounds plausibly implicated as mutagens. Progestins fail to yield mutations in the Ames test,⁵⁹ nor is there evidence that progestins induce mutagenic effects (e.g., cancer) later in life.

Although various maternal hazards have been associated with the use of an intrauterine device (IUD), consideration of adverse fetal effects was not raised until 1976. Barrie⁶⁰ reported two infants whose mothers were using IUDs at conception; the devices remained in situ throughout pregnancy (one Grafenberg ring, probably one Dalkon Shield). Both infants showed limb reduction defects in the upper and lower extremities. Leighton and associates⁶¹ reported a similar case associated with an IUD that contained copper.

Associating limb reduction defects with a coexisting IUD appears superficially to be an attractive hypothesis. Indeed, at least one animal study⁶² reported a teratogenic relationship. Fetuses born of rats with a Silastic device showed increased anomaly rates compared with animals without devices; deformed skull and brain, kinked ribs, cleft palate, and exomphalos were observed. Causal relationship between anomalies and an IUD presumably assumes that the device penetrates both chorion and amnion to exert a direct mechanical action on the developing embryo. However, even if the pregnancy envelops the IUD later in gestation, limb reduction defects and other skeletal anomalies would seem unlikely. The only plausible pathogenesis would involve ruptured amniotic bands, which in humans would produce a somewhat different spectrum of anomalies.

Following the cited case reports, uncontrolled observational studies of pregnancies conceived despite a coexisting IUD were published. Results were reassuring. The one study claiming an association is that of Guillebaud,⁶³ who reported five anomalies among 167 full-term infants whose mothers had an IUD in situ throughout pregnancy. However, the five anomalies were not related (bald spot, lipoid tumor, eyelid ptosis, spina bifida, congenital hip location). By contrast, Tatum and colleagues⁶⁴ evaluated 918 women who conceived while using a copper IUD device. After excluding 465 women who opted for an elective termination, abortion rates and anomaly rates were reviewed in the remaining 453. The spontaneous abortion rate was 20.3% (24 of 118) with the device removed or expelled and 54.1% (85 of 157) with the device left in situ. The stillborn rate was 0.9% with removal and 1.9% without removal. However, only one anomaly (vocal cord fibroma) was observed among the 166 embryos that developed to a stage at which anomalies could be assessed. Albert⁶⁵ also failed to detect an increase in anomalies in women conceiving with a copper device. In a UK study, Snowden⁶⁶ observed no anomalies in 317 pregnancies. At least four other series have failed to detect increased frequency of anomalies.^67,68,69,70

In addition to the cited descriptive studies, case-control studies have been undertaken. Layde and associates⁷¹ found no significant increased risk in a case-control study of 96 mothers delivered of infants with limb reduction defects. In the limb reduction defect group, 2.1% (2 of 96) were exposed. This compared with an IUD prevalence of 1.6% (15 of 915) in mothers of infants with all other major anomalies and 1.2% (2 of 169) in mothers of infants with chromosomal abnormalities. (Both these abnormal control groups would obviate potential for recall bias.) The Connecticut case-control study of Bracken and Vita⁷² also provides information. Comparisons were made between IUD users and users of other nonhormonal contraceptive methods. The only hint of association among 154 categories was that of multiple malformation. However, only 2 mothers with an IUD in situ at conception were in this category; thus, corrections for multiple comparisons rendered the association nonsignificant.

In conclusion, there is no evidence for an increased frequency of anomalies in women who inadvertently become pregnant with an IUD. The relatively increased frequencies of ectopic pregnancy and spontaneous abortions observed in association with IUD use is probably due to an infectious etiology, rather than teratogenesis.

Each year, 300,000 to 600,000 US women become pregnant while using vaginal spermicides.⁷³ Consequently, claims that spermicides are mutagenic or teratogenic are cause for potential concern. Actually, it does not appear particularly reasonable to hypothesize a teratogenic effect because women who become pregnant usually discontinue contraceptive use as soon as pregnancy is suspected. Continued use until the missed period would generally result only in exposures during the first few weeks following conception, a time not likely to result in anomalies (all-or-none period). However, maternal storage and long-term mutagenic or even teratogenic effects of nonoxynol (the most common active ingredient in vaginal spermicides), alcohol, and mercury cannot be dismissed completely.

Extensive controversy was generated by the 1981 report of Jick and associates.⁷⁴ In a study conducted in a large health maintenance organization in Seattle, a woman was considered exposed if a “prescription for spermicide was filled 600 days or less before delivery.” Of 4772 pregnant women, 790 (17%) were subjects. Among their 4665 liveborns were 84 anomalous infants. Of the 84, 18 infants were excluded because their defects were “familial, minor, or positional.” The frequency of spermicide prescription was 2.2% in the remaining 66 anomalous infants, compared with only 1% in the control group not receiving a prescription. An excess of cases existed in four groups: limb reduction defects, neoplasms, chromosomal abnormalities, and hypospadias. One other report claimed a relationship between spermicidal usage and teratologic effects; Smith and associates⁷⁵ observed an association between limb reduction defects and spermicides in a case-control study of 93 newborns with this abnormality.

Studies like that of Jick and associates⁷⁴ are intended more to generate hypotheses than to provide definitive conclusions. Pitfalls are obvious. Some women may never have used the prescribed contraceptive, whereas women in the control group may have obtained spermicides without prescription. Information concerning time of exposure was not available.

Conversely, far better designed studies have failed to find an association between anomalies and spermicide usage. The case-control study by Bracken and Vita⁷² found no association with any nonhormonal contraceptive use. In the US Collaborative Perinatal Project data,⁷⁶ 462 women used spermicides during the first 4 lunar months, with 438 of these also reporting use during the month preceding the last menstrual period. Twenty-three anomalous children were delivered (5%) compared to 2254 anomalies among 49,825 nonexposed controls (4.5%). The difference remained statistically insignificant when anomalies claimed significant by Jick and associates⁷⁴ were studied: Down syndrome, hypospadias, and limb reduction defects. In a case-control study by Polednak and partners,⁷⁷ RRs were 0.43 for limb reduction defects (n = 108) and 1.10 for hypospadias (n = 99). No significant differences were observed for neural tube defects, cardiovascular defects, or “multiple anomalies.”

Cordero and Layde⁷³ studied limb reduction defects in their case-control analysis in the Metropolitan Atlanta Congenital Defects Program (MACDP). They found no evidence that spermicide use during the first 4 lunar months of pregnancy increased the risk of limb reduction defects (RR, 1.0). The same results persisted when they took into account and addressed cumulative exposure. When the MACDP data were reanalyzed for use of a spermicide in the 600 days before delivery (the study design of Jick and colleagues⁷⁴), there was still no increased risk.

Using data from the Birth Defects Study of the Drug Epidemiology Unit (Boston), Warburton⁷⁸ studied the same anomalies as Jick and colleagues.⁷⁴ Again, no association was found. This study also tested the issue of cumulative exposure by stratifying the sample into three group according to duration of exposure: (1) 1 month before until 1 month after the last menstrual period, an interval that should capture exposure around the time of conception; (2) 6 months before conception through the fourth month of pregnancy; and (3) cumulative lifetime exposure to spermicides. Controls consisted of women who had never used spermicides. No increased risk of specific anomalies were observed in any group.

Louik and colleagues⁷⁹ evaluated 1138 newborns with specified birth defects (265 Down syndrome, 396 hypospadias, 146 limb reduction defects, 116 neoplasms, and 215 neural tube defects), comparing their spermicide exposure with 3442 controls having a wide variety of other malformations (i.e., abnormal control group to mitigate against recall bias). None of these defects was associated with exposure to spermicides, either in the first 4 months of pregnancy, at the time of conception, or before conception. RRs were all close to 1.0.

The most extensive investigation exonerating a teratogenic or mutagenic effect for spermicides is the case-control study of 34,660 women by Mills and associates.⁸⁰ A total of 3146 women reported spermicide usage only before their last menstrual period; 2282 used spermicides only after their last menstrual period. Other contraceptive usages were documented, confounding variables considered, and analysis by time of exposure conducted. No significant differences were observed between controls and spermicide users for any of 157 different types of malformations, for infants with three or more anomalies, or for infants with patterns of anomalies.

In summary, great publicity and several legal allegations have been raised concerning spermicides and birth defects. However, the clear scientific consensus is that vaginal spermicides are not associated with an increased rate of malformations in children of women who became pregnant while using either of these agents (teratogenic).

One category of anomalies that Jick and colleagues⁷⁴ claimed to be associated with use of vaginal spermicides was chromosomal abnormalities. Such an outcome could occur only if spermicides were mutagenic, not teratogenic (barring mosaicism due to mitotic nondisjunction). A mutagenic effect would presumably be exerted on the sperm responsible for fertilization, on the developing oocyte, or directly on the zygote. Rothman⁸¹ reported increased spermicidal exposure in Down syndrome infants with cardiac defects, compared with normal controls. However, spermicide use was not increased in mothers of infants with cardiac defects not due to Down syndrome.

Data relating to claims of a mutagenic effect have, in fact, already been raised. Five investigations already discussed (i.e., Shapiro,⁷⁶ Polednak,⁷⁷ Warburton,⁷⁸ Mills,⁸⁰ and Louik⁷⁹ and their respective coworkers) included Down syndrome as one of the conditions studied. For example, Louik and coworkers⁷⁹ ascertained 264 infants with Down syndrome, 39 of whose mothers reported periconceptual spermicide use. No association between spermicides and Down syndrome were found in any of these studies. In addition, Mills and associates⁸⁰ reported no change in the gender ratio following spermicide exposure, offering reassurance that lethal X-linked mutations and thus lethal mutations in general were not being induced.

Warburton⁷⁸ evaluated the potential mutagenic effect of spermicides by studying chromosomal results of amniocentesis in women seeking prenatal diagnosis. Women were interviewed before learning results of their amniocentesis, obviating recall bias. Stratification by duration and timing of spermicide exposure was also performed. No association with trisomy (most commonly trisomy 21) was seen in the 135 study women compared with controls.

Given that most chromosomally abnormal gestations result in spontaneous abortions (at least 50% to 60% of abortuses are chromosomally abnormal), the mutagenic effect of spermicides can also be assessed by studying abortus material. Aneuploidy would be expected to be increased after prior spermicide usage if the latter agent were mutagenic. Strobino and colleagues⁸² evaluated abortus material from 1556 women from whom contraceptive history was obtained. Exposure to spermicides was classified according to duration of use and timing of exposure. ORs were generated for euploidy, trisomy, triploidy, tetraploidy, and monosomy. For none was a statistically significant difference found in comparison with controls of 2766 women delivered of newborns after 28 weeks' gestation. However, there was a positive association between women over 30 who reported use of spermicides for more than a year and trisomy 16 in their abortus (OR, 2.2).

In summary, there is no substantive evidence to support the claim that vaginal spermicides are mutagenic. Studies on abortuses, second trimester amniotic fluid cells, and liveborns have all failed to show associations between spermicide exposure and chromosomal abnormalities.

Women using natural family planning (NFP) avoid pregnancy by abstaining from intercourse during the fertile time period. Conceptions that do occur involve aged gametes. There presumably could be fetal risks arising from fertilization involving aging gametes, and animal studies have provided experimental evidence for such deleterious effects. However, Simpson and partners have recently published human data that failed to find an association.⁸³ Reviewed below are selected animal studies that illustrate the consensus in the biologic community that led to the concern about aging gametes and the results of the recent studies in NFP users.

Animal Studies

SPERM RETENTION IN THE MALE TRACT.

Fertilization of normal ova may occur by sperm retained for prolonged periods in the male tract before ejaculation. Tesh and Glover⁸⁴ studied the effects of insemination by aged sperm in the rabbit. Many of the offspring sired with sperm aged 4 weeks in the epididymis showed structural abnormalities, principally involving the skeletal system or gallbladder. Martin-DeLeon and colleagues⁸⁵ found an increase in chromosomal abnormalities in the embryos of rabbits inseminated with sperm from males who had undergone bilateral epididymal ligation. Sperm were collected 7 to 35 days after ligation. Eight of 72 blastocysts (11%) showed chromosomal abnormalities, compared with only one of 125 (0.8%) controls. The primary abnormality was trisomy. It was subsequently verified that trisomies induced by sperm aged between 8 and 20 days were of paternal origin.⁸⁶

SPERM RETENTION IN THE FEMALE TRACT.

Deleterious effects may also occur following prolonged retention in the female tract before fertilization. Martin-DeLeon and Shaver⁸⁷ observed abnormalities in blastocysts recovered from rabbits inseminated by sperm aged in utero (18 to 32 hours). In the group fertilized by aged sperm, 13 of 134 blastocysts (10%) were cytogenetically abnormal.

AGING OVA.

Postovulatory aging of ova is termed delayed fertilization. The original studies concerning delayed fertilization were by Witschi⁸⁸, who found that in the frog Rana temporaria, prolonged retention of ova in the uterus leads to abnormal embryonic development after fertilization. Fertilization of ova retained 3 to 4 days at elevated temperatures results in (1) cleavage abnormalities, resulting in grossly abnormal embryos unlikely to survive, (2) defects in the neural fold, (3) abnormal division causing multiple births, (4) postaxial duplications resulting in polymelia and polydactyly, and (5) reduction in individual organ sizes. Using Rana pipiens, Witschi and Laguens⁸⁹ later demonstrated that most of these malformed embryos had chromosomal abnormalities.

In mice, Vickers⁹⁰ found that delayed fertilization significantly increased the incidence of polyploidy; aneuploidy occurred less frequently. In rabbits, Shaver and Carr⁹¹ found 15% to 20% of zygotes induced by delayed matings were polyploid. They studied 6-day embryos recovered after matings delayed up to 10 hours following intravenous injection of chorionic gonadotropin. The frequency of chromosomal abnormalities in matings occurring immediately after injection (a 6-hour delay being the injection-ovulation interval) was 7% (5 of 73 blastocysts), compared with only 2% (1 of 58) in control animals with no delay.

To summarize animal data, delayed fertilization clearly may cause chromosomal abnormalities; both sperm and ova may be involved. However, rates of abnormalities differ between species.

Aging Gametes in Humans: Circumstantial Evidence

Potential deleterious effects of aging gametes in humans can only be assessed indirectly, such as through studying outcomes in women with decreased or irregular coital frequency. In 1968, German⁹² suggested that trisomy 21 associated with advanced maternal age was caused by delayed fertilization resulting from decreased coital frequency. German's rationale was derived from observations of Kinsey and colleagues,⁹³ who concluded that coital frequency is inversely related to duration of marriage. After corrections for age, German found that mothers delivered of trisomic offspring had a significantly longer length of marriage and concluded that this correlated with the maternal age-related increase in aneuploidy.

Juberg and partners⁹⁴ reported that the interval preceding the birth of a child with trisomy 21 was significantly greater than the interval preceding a normal birth. Milstein-Moscati and Becak^95,96 attempted to document the frequency of intercourse in women with Down syndrome offspring. Using a case-control design in women under age 35, coitus less than once per week during the conception cycle was reported more commonly in mothers of Down syndrome offspring (75%) than controls (10%). Juberg and partners⁹⁴ also reported coital frequencies to be lower in parents of Down syndrome probands than in parents of offspring with abnormalities other than chromosomal defects.

Several investigators have compared Down syndrome rates in members of various religious groups, assuming indirectly that certain groups were practicing periodic abstinence and were at increased likelihood for fertilization involving aging gametes. Jongbloet and Paesrkote⁹⁷ and Mulcachy⁹⁸ believe that Roman Catholic populations have about a twofold increased risk for Down syndrome compared with findings in Protestants.

If aging gametes indeed predispose to chromosomal abnormalities in humans, one would expect an increased incidence of spontaneous abortions because at least 50% to 60% of abortuses are chromosomally abnormal. In studying the chromosomes of spontaneously aborted fetuses, Boué and partners⁹⁹ found the prevalence of polyploidy was 32% when intercourse occurred 15 days or more after the last menstrual period. When intercourse was said to occur less than 15 days after the last day of the last menstrual period, prevalence was 10% The mean delay between last menstrual period and conception was 15 days for cytogenetically normal abortuses and 17 days for polyploid abortuses. Using basal body temperature charts, ovulation was determined. Delays between ovulation and fertilization in excess of 2 days were associated with a nonsignificant higher prevalence of polyploidy (35%), compared with delays less than or equal to 2 days (23%); however, no differences were observed with respect to monosomy X or autosomal trisomy. These findings suggest that delayed fertilization could increase the risk of polyploidy.

Chromosomal Abnormalities in NFP Populations

Given potential concern for aging gametes, questions must be raised concerning fetal risks of NFP. Unfortunately, few studies in NFP populations have addressed the issue of chromosomal abnormalities or congenital anomalies. Those conducted have failed to distinguish pregnancies resulting from failure by NFP users from pregnancies resulting from failure of NFP method. Only the latter would be associated with aging gametes. Combining method-failure and user-failure pregnancies diminishes the power of a study to identify an effect. Irrespective, two studies have addressed directly the relationship between Down syndrome and NFP. Harlap and partners¹⁰⁰ studied 33,551 abortions and livebirths among northern Californian women between 1975 and 1977. At the first prenatal clinic visit, women completed a questionnaire seeking social and contraceptive history. A pregnancy was defined as unplanned if the woman used a contraceptive method the first day of the cycle and throughout the month that conception occurred. Among 338 “rhythm-unplanned” pregnancies were two cases of trisomy 21. Although the crude RR is 4.59, the absolute risk is obviously low, even without adjustment for potentially confounding variables. The case-control study of Bracken and Vita⁷² found a nonsignificant increase in the risk of Down syndrome in women using “periodic abstinence.”

There is also no indication that spontaneous abortion rates are increased in pregnancies occurring in practitioners of NFP. Crude loss rates of 15% or less were observed in NFP populations studied by Marshall (14.5%),¹⁰¹ World Health Organization (9.8%),¹⁰² Roetzer (8.3%),¹⁰³ France and coworkers (7%),¹⁰⁴ Guerrero and Rojas (7.8%),¹⁰⁵ and Simpson and associates (10.2%).⁸³ Consistent with the hypothesis that delayed fertilization causes deleterious effects are studies by Guerrero and Rojas.¹⁰⁵ Basal body temperature charts were used to determine probable time of conception in a series of couples undergoing artificial insemination. Among 965 pregnancies were 75 in which spontaneous abortion occurred. The fetal loss rate was 24% when insemination was performed 3 or more days after ovulation, but only 3.2% when performed within 2 days of the thermal shift. Gray and associates¹⁰⁶ and the World Health Organization¹⁰² similarly found higher spontaneous abortion rates in NFP users when conception occurred 3 days before or 3 days after ovulation.

One of us (Simpson)^83,107 recently reported the findings of the collaborative group, the LatinAmerican Collaborative Study of Malformations (ECLAMC), on the frequency of congenital anomalies in NFP users in South America. During 1992 to 1994, data were gathered by questionnaire during the immediate postpartum period on 5324 case-control pairs from 18 hospitals. Cases were identified by examination for anomalies in the nursery; the same gender nonmalformed liveborn born next in temporal sequence to the malformed baby was chosen as the control. Mothers of case infants and controls were questioned about fertility control. Of the 262 Down syndrome cases, 28 (10.69%) mothers reported use of NFP, compared with 16 (6.11%) of the matched controls. The OR was 1.84. However, the difference narrowed significantly when the OR was adjusted for maternal age and parity.

In a cohort study of the same population,¹⁰⁷ the risk of spontaneous abortion in NFP users was also assessed, given that at least 50% to 60% of such abortuses are chromosomally abnormal. No significant association between spontaneous abortions and NFP was found in women without a history of prior losses. However, there was an increased risk in NFP users who reported previous losses; this could be related to recurrent aneuploidy, although the biologic basis is unclear.

In conclusion, NFP methods are safe with respect to Down syndrome and do not increase the overall rate of spontaneous abortion. These findings should be reassuring to NFP users.

Other Anomalies In NFP Populations

Other studies have determined, de facto, the association between NFP and Mendelian or polygenic/multifactorial disorders by assessing overall major anomaly rates. Jongbloet and colleagues^108,109,110 offer circumstantial data that delayed fertilization has deleterious effects. In one report,¹⁰⁹ parents of 127 institutionalized mentally retarded children were interviewed regarding the circumstances surrounding conception of their abnormal child. After excluding couples in which the child had a known genetic or metabolic defect, there remained 49 couples who experienced unintended pregnancies while practicing NFP; 93 pregnancies were considered voluntary. Among the 49 method-failure pregnancies, 28 (57.1%) progeny were abnormal; only 11 (11.8%) of the 93 voluntary pregnancies were abnormal. In another study of 211 Roman Catholic couples,¹¹⁰ Jongbloet identified 127 pregnancies in users of NFP; of these, 30 were associated with NFP failure. Of these 30, seven (23%) progeny were “abnormal”; only three (7%) of 46 progeny of planned pregnancies were “abnormal.” Major pitfalls in the analyses of Jongblont include limited information on methodology, probable recall bias, memory bias (given years having passed between conception and interviews), and failure to exclude postnatal causes of abnormalities with respect to the later. In fact, nine of 39 (23%) children with abnormalities had retardation from a postnatal event, for example, encephalitis.

Jongbloet and Zwets¹⁰⁹ further claimed that regions in Europe with predominantly Roman Catholic populations have a higher incidence of both anencephaly and Down syndrome, indirectly implicating NFP on religious grounds. This claim fails to take into account genetic factors, the later reproductive age of among Roman Catholics who are not practicing contraception, and the low proportion of Roman Catholics who actually use NFP.

The case-control study of Bracken and Vita⁷² found significantly increased risks for both cleft lip and palate and congenital hydrocele. However, those authors concluded these findings were probably chance findings, reflecting multiple comparisons on a small number of NFP users. This study carries potential for two types of bias: misclassification bias due to inability to distinguish between method and user NFP failures, a phenomenon that would decrease the likelihood of finding an association; and recall bias due to less rigorous recall by normal controls, a phenomenon that would increase the likelihood of finding an association. The only other case-control study is that of Kuhr¹¹¹, who studied neural tube defects. The risk for this polygenic/multifactorially disorder in users of the rhythm method was actually lower than in controls (RR, 0.66).

Several observational studies have been reported. Marshall¹⁰¹ found no increased incidence of birth defects among pregnancies that resulted from NFP failures (those who measure basal body temperature) than controls. Of 53 children resulting from spermatozoa aged 1 to 8 days, three had congenital anomalies; of the 15 children who resulted from spermatozoa aged 9 to 16 days, two had congenital anomalies. In this small sample size, no significant difference was noted. Oescheli¹¹² reported 5.14% of birth defects among 779 users of NFP and 4.73% among 2718 women who used no contraception. Among 617 NFP pregnancies, Roetzer¹⁰³ found only five major malformations (0.8%). Moreover, these five pregnancies were the result of intercourse on or just before the day of ovulation. A prospective five-country study of the ovulation method of NFP, by The World Health Organization (1984)¹⁰² found no evidence of a relationship between the most likely day of conception and either congenital malformations or spontaneous abortion. Of 175 pregnancies with known outcome, there were 2 congenital malformations (1.2%) and 16 spontaneous abortions (9.8%). Shortcomings of this study include small sample size, failure to distinguish method from user pregnancies, and inability to obtain outcome on 12 pregnancies.

Recently, one of us (Simpson)¹⁰⁷ published the outcome of an international group the purpose of which was to clarify fetal risks of NFP through a US Agency for International Development-funded program for Natural Family Planning research. The collaborative effort included the University of Tennessee (Memphis), Johns Hopkins (Baltimore), Universidad de Chile (Santiago), Pontificia Universidad Católica de Chile (Santiago), Asociacion de Trabajo Laico Familiar (Lima), Javeriana Universidad (Bogota), Centro Ambrosiano Methodi Natural (Milan), and the Maryland and District of Columbia NFP Providers Association (Washington). Through a cohort study, all women using NFP who became pregnant were interviewed during pregnancy, thus avoiding recall bias. Fertile and infertile days will be determined using cervical mucus interpretation, basal body temperature, and symptoms of ovulation. NFP charts were used to assess the probable day of coitus that led to conception, the precise method of NFP used and dates of LMP. The pregnancies were classified as planned pregnancies or pregnancies resulting from method failure or user failure. A systematic evaluation was performed on all neonates for major and minor anomalies.

The part of the study designed to assess risk of fetal anomalies is still in progress. However, a case-control study on the population of NFP users in South America has recently been published by the same investigative group. During 1992 to 1994, mothers of 5324 anomalous infants and their matched controls were interviewed during the immediate postpartum period. Overall, no significant differences in frequency of NFP usage were observed between mothers of malformed infants (6.6%) and normal controls (5.7%). In summary, studies in liveborns resulting from NFP populations give no evidence for increased anomaly rate.