A prospective study of pregravid oral contraceptive use in relation to fetal growth
Dr L. Mucci, Department of Epidemiology, Harvard School of Public Health, 677 Huntington Avenue, Boston, Massachusetts 02115, USA.
Objective Because oral contraceptives are so widely used, any health consequences may have substantial public health implications. Whether pregravid oral contraceptives could affect subsequent pregnancies has not been adequately studied. The study objectives were to examine whether pregravid oral contraceptive use affects fetal growth and pregnancy hormone levels.
Design A prospective study of pregnant women followed through pregnancy.
Setting A major teaching hospital in Boston, USA.
Population Two hundred and sixty Caucasian pregnant women, with a mean age of 31, and a parity of no more than two. Seventy-nine percent of the women were pregravid oral contraceptive users.
Methods Exposure and covariate data were collected through structured questionnaires. Blood was drawn for hormonal analysis during the 16th and 27th gestational week. Information on pregravid oral contraceptive use included duration and recency of use, and oral contraceptive formulation. Multivariate regression models were used to examine the effect of pregravid oral contraceptive use on birth outcomes and the studied pregnancy hormones.
Main outcome measures Birthweight, placental weight, gestational age, pregnancy hormone levels of oestriol and progesterone at 16th and 27th gestational week.
Results Adjusting for confounders, pregravid oral contraceptive use increased birthweight (mean difference =+207.3 g, 95% CI =+77.6 to +337.1) and placental weight (mean difference =+64.9 g, 95% CI =+13.0 to +116.9) compared with never use. Women with prior oral contraceptive use had higher levels of serum progesterone (P= 0.002) and oestriol (P= 0.12) at the 27th gestational week measurement. The effect on birthweight, placental weight and hormones was stronger among those using oral contraceptives in the previous year and those using a high progestin/high oestrogen potency preparation.
Conclusions Pregravid oral contraceptive use is positively associated with fetal growth, and this effect may be mediated through oestriol and progesterone.
Oral contraceptives have markedly impacted both the social dynamics and health of women. Some of the health effects have been undesirable, but their frequency and severity seem to have declined with the introduction of newer preparations.1 In contrast, the beneficial health effects of oral contraceptive use are considerable and include the substantially reduced risk of ovarian and endometrial cancer.1,2
There are fewer data, however, on possible health effects of oral contraceptive use prior to pregnancy and subsequent birth outcomes.3–8 Many of these studies were conducted during the early 1980s, under the concern that pregravid oral contraceptive use could have negative health consequences on future birth outcomes. We undertook an investigation among a cohort of US pregnant women to examine the effect of pregravid oral contraceptive use in relation to subsequent pregnancy hormones and outcomes.
The study base for the present investigation was an international prospective study of pregnant women in Boston, USA and Shanghai, China. The study was undertaken to investigate whether in utero and perinatal differences between pregnant women from the two countries could explain observed differences in breast cancer rates.9 In the current investigation, we examine the effect of pregravid oral contraceptive use on subsequent pregnancy hormone levels and birth outcomes. Because of the low prevalence of oral contraceptive use among Chinese women (<1%), we relied on data from US women only.
Between March 1994 and October 1995, 402 eligible adult pregnant women were identified at the Beth Israel Hospital in Boston, Massachusetts, USA. As part of the study criteria, all women were Caucasian, between 18 and 40 years old and had a parity of no more than two. Women were not eligible if they had taken any hormonal medication during the index pregnancy, if they had a prior diagnosis of diabetes mellitus or thyroid disease or if the fetus had a known major anomaly. A trained health professional met all pregnant women coming for their first routine prenatal visit to the collaborating maternity clinic, ascertained whether the woman was eligible to participate, explained to her the objectives of the study and the requirements for participation and obtained informed consent. The procedures for the study were in accordance with the ethical standards for human experimentation established by the Institutional Review Boards at the Harvard School of Public Health and Beth Israel Hospital. The data remained blinded in the present analysis to maintain patient confidentiality.
Of 402 eligible women, 77 refused to participate, 9 were subsequently excluded because the index pregnancy was terminated through a spontaneous or induced abortion, 2 were excluded because of twin birth, 10 were lost to follow up after the initial meeting and 44 had missing data for the hormonal or reproductive factors evaluated in this analysis. Thus, 260 pregnant women were included in the present study.
Exposure, covariate and outcome data were collected through structured questionnaires, review of medical records and through blood samples. Details of the data collection are given in an earlier publication.9 The principal exposure variable for the present study was ever use of oral contraceptives prior to pregnancy. To further understand the mechanism by which pregravid oral contraceptive use might influence birth outcomes, we evaluated duration of use, recency of use and oral contraceptive formulation. Women were shown pictures of different oral contraceptive preparations to help identify the brand of oral contraceptives used. We categorised women according to the potency of progestin and oestrogen in the oral contraceptive preparations according to criteria followed by Schildkraut et al.10 Low potency progestin preparations were those described as low or intermediate based on a standard reference text, while all other preparations with known progestin levels were categorised as high potency. Oral contraceptive formulations containing 35 μg or less of ethinyl estradiol or its equivalent were classified as low oestrogen potency, while higher levels were classified as high oestrogen. Women were classified in one of the following five categories: high progestin/high oestrogen, high progestin/low oestrogen, low progestin/high oestrogen, low progestin/low oestrogen and never oral contraceptive use. We also examined the effect of monophasic versus bi-/triphasic preparations compared with never users. Among the 21 women who reported using two oral contraceptive brands with differing preparations, we coded them as phasic if either brand was phasic, and coded them as high potency if either brand was of high potency.
We examined several outcome variables: risk of low birthweight or pre-eclampsia, birthweight, gestational age at birth, placental weight and maternal serum levels of unconjugated oestriol (E3) and progesterone at 16 and 27 completed weeks of gestation. Birthweight and placental weight were measured at delivery by study collaborators. Low birthweight infants were those weighing less than 2500 g. Gestational age at delivery was estimated as the exact difference between the date of last menstruation and date of delivery. The attending physicians at Beth Israel Hospital used criteria provided by the International Classifications of Diseases, ninth edition (ICD-9), to diagnose pre-eclampsia (ICD-9 codes 642 E and 642 F). In the original study, seven compounds were measured in maternal blood around week 16 and 27 of gestation: E3, oestradiol, sex hormone binding globulin, progesterone, prolactin, growth hormone of pituitary origin and albumin. For the present study, we have a priori chosen to concentrate on E3 and progesterone given the dominant role they play in pregnancy. Unconjugated oestriol in 50 μL serum was measured with a time-resolved competitive solid-phase fluoroimmunoassay method (AutoDELFIA unconjugated oestriol kit; Wallac Oy, Turku, Finland). Progesterone in 25 μ L serum, diluted 1:8, was measured with a time-resolved competitive solid-phase fluoroimmunoassay (AutoDELFIA Progesterone kit; Wallac Oy). Both E3 and progesterone concentrations are expressed in ng/mL.
Crude and adjusted analyses were conducted to assess the independent effect of pregravid oral contraceptive use on each of the outcomes. Multiple logistic regression was used for binary outcomes (low birthweight and pre-eclampsia) to estimate adjusted odds ratios (OR) and 95% confidence intervals (CI) of ever oral contraceptive users versus never users. Multiple linear regression was employed for continuous outcomes (birthweight, gestational age, placental weight and hormones) to estimate adjusted mean differences in these outcome measures for oral contraceptive users versus never users. Because the distributions of E3 and progesterone were positively skewed, we used log-transformed values for these variables. In this way, partial regression coefficients, after exponentiation, express the relative change (i.e. percent) in hormone levels comparing oral contraceptive users to never users. We considered the following variables as potential confounders: age, marital status, education, pre-pregnancy body mass index, maternal height, weight gain during pregnancy, smoking, alcohol use, parity, gestational age at birth, gender of offspring and type of delivery. Because several of these variables had no confounding influence, they were not included in the final models. In the final adjusted analyses, we controlled for maternal age (<25, 25–29, 30–34, 35+), parity (1 vs 2), height (continuous), pre-pregnancy body mass index (continuous), exact gestational age at birth (continuous) and gender of the offspring (female vs. male). With respect to hormones, exact gestational age at blood draw was also controlled for (continuous). Adjusting for these factors controls for possible confounding effects generated by differences between the subcohorts of ever and never oral contraceptive users. All analyses were conducted using SAS Software version 8.2.
All 260 pregnant women in this investigation delivered a live born infant. Almost 79% of women had ever used oral contraceptives prior to pregnancy (Table 1). More than three-quarters of ever users had used oral contraceptives for at least two years, and 37% had used oral contraceptives within the year prior to conception. The frequency distribution of oral contraceptive formulation is also presented in Table 1.
Table 1. Pregravid oral contraceptive use among 260 pregnant women in Boston, USA, 1994–1995.
|Previous oral contraceptive use|
|Duration of use of oral contraceptives1|
|2 or more years||160||78.0|
|Recency of use1|
|Within 1 year of pregnancy||75||36.6|
|Oral contraceptive formulation1|
|High progestin/high oestrogen||5||2.4|
|High progestin/low oestrogen||27||13.2|
|Low progestin/high oestrogen||18||8.8|
|Low progestin/low oestrogen||128||62.4|
Maternal and newborn characteristics by oral contraceptive use are presented in Table 2. Women who had ever used oral contraceptives had lower body mass index. Otherwise, there was little difference in the frequency distribution of education, marital status, weight gain, smoking, gestational age and offspring gender among ever and never users of oral contraceptives.
Table 2. Maternal and newborn characteristics and reproductive factors among 260 pregnant women by oral contraceptive use, Boston, USA, 1994–1995. Values are expressed as n (%).
| 18–24||5 (2.4)||1 (1.8)|
| 25–29||67 (32.7)||9 (16.4)|
| 30–34||116 (56.6)||39 (70.9)|
| 35+||17 (8.3)||6 (10.9)|
| Married||198 (96.6)||53 (96.4)|
| Not married||7 (3.4)||2 (3.6)|
| High school graduate||43 (20.2)||12 (25.5)|
| College graduate||170 (79.8)||43 (75.5)|
|Pre-pregnancy body mass index (kg/m2)|
| 20 or less||47 (22.9)||15 (27.3)|
| 20–24||127 (62.0)||26 (47.3)|
| 25–29||25 (12.2)||12 (21.8)|
| 30+||6 (2.9)||2 (3.6)|
|Mother's height (cm)|
| 145–154||12 (5.9)||3 (5.5)|
| 155–164||77 (37.6)||25 (45.5)|
| 165–174||102 (49.8)||22 (40.0)|
| 175+||14 (6.8)||5 (9.1)|
|Weight gain during pregnancy (kg)|
| <10||89 (43.4)||24 (43.6)|
| 10–15||84 (41.0)||20 (36.4)|
| 15 or more||32 (15.6)||11 (20.0)|
|Smoking in pregnancy|
| Yes||12 (5.9)||2 (3.6)|
| No||193 (94.1)||53 (96.4)|
|Gender of offspring|
| Female||106 (51.7)||29 (52.7)|
| Male||99 (48.3)||26 (47.3)|
| <37||15 (7.3)||2 (3.6)|
| 37–42||184 (89.8)||52 (94.6)|
| 43+||6 (2.9)||1 (1.8)|
|Type of delivery|
| Caesarean||48 (23.4)||11 (20.0)|
| Normal delivery||157 (76.6)||44 (80.0)|
|Age at menarche|
| <12||31 (15.1)||10 (18.2)|
| 12||51 (24.9)||15 (27.3)|
| 13||69 (33.7)||17 (30.9)|
| 14 or older||54 (26.3)||13 (23.6)|
| 1||130 (63.4)||31 (56.4)|
| 2||75 (36.6)||24 (43.6)|
|Previous pregnancies only|
| Stillborn|| || |
| 0||205 (100.0)||54 (98.2)|
| 1+||0 (0.0)||1 (0.8)|
|Spontaneous abortions|| || |
| 0||167 (81.5)||47 (85.5)|
| 1+||38 (18.5)||8 (14.5)|
| Induced abortions|| || |
| 0||178 (86.8)||50 (90.9)|
| 1+||27 (13.2)||5 (9.1)|
| Age at first pregnancy|| || |
| Never pregnant before||130 (63.4)||31 (56.4)|
| <25||13 (6.3)||4 (7.3)|
| 25+||62 (30.2)||20 (36.6)|
A central issue in evaluating apparent consequences of pregravid oral contraceptive use is whether inherent fecundity differs as a function of oral contraceptive use. Data in Table 2 allow comparison of ever and never oral contraceptive users with respect to reproductive variables that reflect fecundity. Except for parity and maternal age at index pregnancy, there was little evidence of differences in reproductive characteristics between these two groups of women. Indeed, the somewhat higher parity among never users suggests higher fecundity in this group. All subsequent analyses were adjusted for parity and maternal age.
Table 3 shows crude and adjusted estimates of the effect of pregravid oral contraceptive use on birth outcomes as well as on pregnancy hormones. There was no evidence in these data that pregravid oral contraceptive use is associated with pre-eclampsia risk. In contrast, the likelihood of a low birthweight newborn was substantially reduced among pregravid users of oral contraceptives (adj OR = 0.1, 95% CI = 0.0 to 0.9). Moreover, mean birthweight was substantially higher (+207.3 g, 95% CI =+77.6 to +337.4) among pregravid oral contraceptive users compared with never users, after controlling for potential differences between the groups. Placental weight was also significantly higher (+64.9 g, 95% CI =+13.0 to +116.9) among women who previously used oral contraceptives. No association between oral contraceptive use and gestational age at birth was evident. Limiting the analysis to women who did not smoke during pregnancy did not alter the study findings (data available upon request).
Table 3. Effect estimates of pregravid oral contraceptive use on birthweight, pre-eclampsia, gestational age and placental weight among 260 pregnant women in Boston, USA, 1994–1995.
|Low birthweight (<2500 g)|
|Never oral contraceptive users||7.2|| ||Ref|| ||Ref|
|Ever oral contraceptive users||2.4||0.3||0.1 to 0.9||0.1||0.0 to 0.9|
|Never oral contraceptive users||5.5|| ||Ref|| ||Ref|
|Ever oral contraceptive users||7.3||1.1||0.4 to 3.5||1.2||0.3 to 4.2|
| ||Mean||Mean difference||95% CI||Mean difference||95% CI|
|Never oral contraceptive users||3353.3|| ||Ref|| ||Ref|
|Ever oral contraceptive users||3540.2||+186.9||+27.4 to +346.4||+207.3||+77.6 to +337.1|
|Gestational age at birth (weeks)|
|Never oral contraceptive users||39.9|| ||Ref|| ||Ref|
|Ever oral contraceptive users||39.9||−0.01||−0.6 to +0.6||+0.04||−0.5 to +0.6|
|Placental weight (g)|
|Never oral contraceptive users||540.6|| ||Ref|| ||Ref|
|Ever oral contraceptive users||601.2||+60.7||+6.8 to +114.6||+64.9||+13.0 to +116.9|
Pregravid use of oral contraceptives is examined in relation to maternal serum levels of E3 and progesterone at completed weeks 16 and 27 of gestation (Table 3). There were no important differences between pregravid oral contraceptives ever and never users with respect to the levels of the studied hormones at week 16 of gestation, although a suggestive increase of progesterone may deserve some attention. At week 27, however, both E3 and progesterone were elevated in the serum of pregravid users of oral contraceptives compared with never users (+8.1% and +12.9%, respectively).
To further elucidate the relation between pregravid oral contraceptive use and the studied outcomes, we examined duration of oral contraceptive use, recency of use and oral contraceptive preparation in relation to birthweight, placental weight and hormone levels at week 27 in Table 4. The effect of pregravid oral contraceptive use on birthweight, placental weight and progesterone levels during week 27 of gestation was stronger among women who had used oral contraceptives in the year prior to pregnancy than among those who had used oral contraceptives more distantly. For birthweight, the effect was stronger among women who had used oral contraceptives for two years or more compared with those who had used oral contraceptives for less time (adj mean difference +248.4 and +76.6 g, respectively, compared with never users). Duration of oral contraceptive use did not appear to be related to hormone levels. With respect to oral contraceptive preparation, women using preparations with high progestin/high oestrogen potency had substantially heavier babies, higher placental weights and more elevated E3 and progesterone levels at week 27. Moreover, monophasic preparations appeared to have a greater effect on birthweight and placental weight than bi- or triphasic preparations.
Table 4. Recency, duration and preparation of pregravid oral contraceptives in relation to birthweight, placental weight and pregnancy hormones in a cohort of 260 pregnant women in Boston, USA, 1994–1995.
|Recency of oral contraceptive use|
|Within 12 months prior to pregnancy||+250.3||+94.5 to +406.1||+92.1||+29.1 to +155.1||+13.9||+4.8 to +23.8||+15.6||+2.9 to +27.6|
|>12 months prior to pregnancy||+197.8||+59.2 to +336.3||+53.0||−2.2 to +108.2||+12.8||+4.8 to +21.5||+5.4||−4.2 to +16.1|
|Duration of oral contraceptive use|
|2 years or more||+248.4||+116.4 to +380.4||+65.2||+11.7 to +118.7||+13.1||+5.3 to +21.6||+7.6||−2.1 to +18.2|
|Less than 2 years||+76.6||−93.8 to +247.0||+63.9||−7.9 to +135.7||+12.1||+2.2 to +23.0||+9.6||−3.0 to +23.7|
|Preparation of oral contraceptives|
|High progestin/high oestrogen||+556.6||+204.9 to +958.2||+140.2||−21.7 to +302.1||+20.8||−2.3 to +49.4||+25.4||−5.0 to +65.6|
|High progestin/low oestrogen||+188.3||−15.4 to +392.0||+75.0||−15.6 to +165.6||+16.8||+4.7 to +30.2||+17.7||+2.1 to +35.7|
|Low progestin/high oestrogen||+251.2||+20.0 to +482.4||+63.9||−31.8 to +159.6||+15.4||+2.2 to +30.3||+10.5||−5.9 to +29.7|
|Low progestin/low oestrogen||+233.8||+96.3 to +371.3||+76.2||+18.3 to +134.0||+13.7||+5.6 to +22.3||+9.3||−0.7 to +20.4|
|Monophasic||+290.7||+149.3 to +432.1||+80.8||+20.0 to +141.5||+12.4||+4.2 to +21.3||+11.6||+1.1 to +23.2|
|Bi/Triphasic||+157.3||+7.0 to +307.6||+66.4||+4.5 to +128.4||+15.7||+6.8 to +25.4||+8.8||−2.1 to +20.9|
We found evidence that pregravid use of oral contraceptives is associated with reduction in the risk of low birthweight and an increase in average birth and placental weight. Because oral contraceptive use is not associated with gestational age, our results suggest that pregravid use of oral contraceptives is positively associated with fetal growth. The apparent effect of pregravid oral contraceptive use may be mediated through increases in serum levels of progesterone, particularly during the last trimester, and possibly through higher levels of E3, also during the last trimester. Progesterone and E3 play a central role in pregnancy physiology,11 and maternal hormones are strongly correlated with birthweight and placental weight.12,13 The indication in our data that high progestin/high oestrogen preparations have a greater effect on birth and placental weight, as well as on progesterone and E3 levels at gestational week 27, supports the hypothesis that pregravid use of oral contraceptives has discernible effects on the processes and outcomes of subsequent pregnancies.
There is a vast literature concerning possible adverse effects of oral contraceptive use,1 and on the established beneficial long term effects on ovarian and endometrial cancer risk.1,2 In particular, it is well documented that the protection afforded by oral contraceptives on cancer risk persists for a long time after stopping use.2,14 The postulated mechanisms by which oral contraceptives could influence health status include the disruption of ovulation,10 progestin-mediated apoptosis15 and changes in endogenous hormone metabolism.16
Fewer studies have examined the role of pregravid oral contraceptive use on the outcome of subsequent pregnancies.3–8 Vessey et al.4 and Rothman3 reported no association between pregravid oral contraceptive use and birthweight. Rothman, however, found that pregravid oral contraceptive use substantially increases the likelihood of subsequent twin pregnancies, suggesting that oral contraceptives can have long term pregnancy effects. While Polednak et al.7 found no overall effect of oral contraceptive use on birthweight overall, they did note a positive effect on birthweight among older women. Similarly, Alberman et al.5 noted a significant interaction between oral contraceptive use and maternal age. Finally, Peterson8 reported a lower risk of low birthweight births among prior oral contraceptive users compared with non-users.
There are important points to consider in interpreting these prior studies. First, these earlier studies did not control for confounding as extensively as we did in this investigation. Second, the earlier studies did not fully explore the effect of duration, timing or formulation of oral contraceptives in relation to birthweight, of which we found evidence of a dose–response. Third, in the Polednak study,7 comparison of pregravid oral contraceptive users was made not with women who had never used oral contraceptives, but with women who had not recently used them. Because we found an increased birthweight among women who had used oral contraceptives more than one year prior to pregnancy, the comparison group of Polednak would have diluted an effect of oral contraceptives. Finally, it should also be pointed out that Polednak et al.7 and Alberman et al.5 observed that an effect of pregravid oral contraceptive use on birthweight might be evident among older women, findings which are compatible with our data as most of the women in the present investigation were older than 30 years.
With respect to pre-eclampsia, Bracken and Srisuphan6 found no association with pregravid oral contraceptive use. In contrast, Thadhani et al.17 reported that pregravid oral contraceptive use increases the risk of pre-eclampsia but reduces the risk of gestational hypertension alone. Our findings are in agreement with those reported by Bracken and Srisuphan and point to no adverse of pregravid oral contraceptives on the occurrence of pre-eclampsia.
We have found no previous studies that have examined pregravid oral contraceptive use in relation to subsequent pregnancy hormones. Nevertheless, our results are not implausible given the established ability of oral contraceptives to regulate menstruation2 and to affect endogenous hormone metabolism.16 Thus, if the hormonal influence of oral contraceptives persists after discontinuation, this could be one possible mechanism by which pregravid oral contraceptive use could affect fetal growth.
Our prospective study was undertaken with a strict protocol. Selection bias could have been introduced if women taking oral contraceptives had an inherently better reproductive health. However, no evidence of differential fecundity was noted in our study population. We may also be concerned that the women who refused or otherwise did not participate in the study may differ from those who were included. However, this would only result in bias if the effect of oral contraceptives on fetal growth were different between the groups, a theoretical possibility that has no empirical support. We had sufficient statistical power (>80%) to evaluate the main effects of oral contraceptive use on continuous outcomes like birthweight and placental weight, as evidenced by the respective confidence intervals. We have controlled for several potential confounders and the oral contraceptive associations were robust. In addition, the dose–response relations observed for duration of use and oral contraceptive formulations argue against residual confounding as an explanation of the findings. Composition of maternal diet was not associated with birthweight in these data.18 The study population was fairly homogenous with respect to education, predominantly highly educated women. This should not compromise the generalisability of the results, unless socio-economic status was an effect modifier, which is unlikely on the basis of current knowledge about the biology and epidemiology of oral contraceptive use and its consequences.
Misclassification with respect to exposure and outcome variables is unavoidable. Specifically, misclassification of oral contraceptives, particularly oral contraceptives brand, may be of concern. Results from a validity study with similar methodology suggest that self-reported information on brand of oral contraceptive use has good validity.19 Moreover, any misclassification would be non-differential, and thus unlikely to account for the observed associations.
From a biological perspective, there have been reports of long term consequences of past use of oral contraceptives on hormone metabolism,16 menstrual regularity,2 and ovarian and endometrial cancer risk.1,2 It is also known that the order of pregnancies (first vs second) is related to pregnancy oestrogen levels, implying long term effects of the first pregnancy on the second.20–22
Although our results need to be confirmed in other investigations, the findings of this study suggest that pregravid oral contraceptive use may positively influence fetal growth in subsequent pregnancies, and point to changes in levels in progesterone, and conceivably E3, as possible mediating mechanisms. The potential clinical implications of these findings are twofold. First, because fetal growth is conducive to the health of the newborn, these findings suggest that pregravid oral contraceptive use may have similar effects. Second, recent evidence supporting a role of birthweight on risk of adult disease in the offspring23,24 indicates that pregravid oral contraceptive use could potentially modulate disease susceptibility in adult life.
This study was supported in part by grant number CA54220 from the National Cancer Institute/National Institutes of Health. The authors would like to acknowledge the contributions of Dr Loren Lipworth, who was actively involved in the study design and collection of data of the original cohort study.