In the United States each year, at least 1.5 million women use nonsteroidal anti-inflammatory drugs (NSAIDs) around the time of conception, implantation, and early embryonic development,1 – 3 making them the most common medication exposure reported in the first trimester.4 Nonsteroidal anti-inflammatory drugs are primarily used to relieve pain and reduce inflammation and their effects generally result from inhibition of cyclooxygenase (COX)-2. Recently, three population-based observational cohort studies have implicated first-trimester NSAID use as a risk factor for spontaneous abortions (miscarriage) using data obtained from pharmacy records or study interviews that limited NSAID exposure to a subset of NSAIDs (eg, ibuprofen, naproxen).5 – 7 However, studies have not examined NSAID exposure and risk for adverse pregnancy outcomes considering all forms of over-the-counter formulations in early pregnancy.
Women and their health care providers currently lack clear empirical evidence to inform clinical care regarding the consequences of NSAID use during pregnancy. Prior evidence linking NSAIDs to spontaneous abortion risk is inconclusive attributable potentially to recall bias, limited power, incomplete documentation of NSAID exposure or all of these for all participants. This includes lack of information about over-the-counter NSAID exposure that should more accurately represent typical NSAID use during early pregnancy. We used data from the Right from the Start study (2004–2010), a nonclinical, community- based pregnancy cohort, to examine NSAID use during the early first trimester of pregnancy as it relates to risk for spontaneous abortion. This study tests whether over-the-counter NSAIDs used early in the first trimester are associated with spontaneous abortion risk.
MATERIALS AND METHODS
Right from the Start is an ongoing community-based cohort that began enrolling in 2000. Over time, Right from the Start has been funded in three major phases (RFTS1, 2, 3) and has enrolled participants in Galveston, Texas; Memphis, Nashville, Knoxville, and Chattanooga, Tennessee; and the Research Triangle region (Raleigh, Durham, and Chapel Hill) in North Carolina. RFTS1 participants were excluded from the analyses (n=1,956) because over-the-counter NSAID use was not ascertained during the interview. Right from the Start participants are 18 years or older and did not use assisted reproductive technologies to conceive. Consent was obtained to review all records pertaining to the study pregnancy. Direct marketing and recruitment strategies have been described.8
Women who were not yet pregnant but trying to conceive could pre-enroll before pregnancy and were followed until a positive pregnancy test. To avoid overenrollment of subfertile women, nonpregnant participants in the study had to be attempting to get pregnant for fewer than 6 months (RFTS2) or fewer than 3 months (RFTS3). Women were eligible for up to 12 months of pre-enrollment. Our Right from the Start study participant median gestational age at enrollment since last menstrual period (LMP) was 42 days (interquartile range 35–52 days). Thirty-seven percent of women were pre-enrolled in the study and enrolled on confirmation of a positive pregnancy test. Participants completed an intake interview at enrollment and a computer-assisted telephone interview at the end of the first trimester. The computer-assisted telephone interview was conducted at a median gestational age since LMP of 98 days (interquartile range 95–103 days), providing information on history of bleeding or pain, medication use, and exposure to potential confounders in the time since the LMP. Follow-up was conducted to document outcomes. At enrollment a study transvaginal ultrasound was scheduled at a participating ultrasound site to assess embryonic development. The institutional review boards of Vanderbilt University, Nashville, Tennessee, and the University of North Carolina, Chapel Hill, North Carolina, approved this study.
Pregnancy outcomes were self-reported and abstraction of medical records was used to verify outcomes. Live births were linked to state vital records to assist in verifying the pregnancy outcomes for ongoing pregnancies. Spontaneous abortions were defined as a loss before 20 completed weeks of gestation. Those without losses included both live births and stillbirths, excluding ectopic pregnancies (n=9) and induced abortions (n=14). Gestational age was estimated from self-reported LMP. We have previously shown the overall accuracy of self-reported LMP in our cohort.9 Women could enroll in Right from the Start during more than one pregnancy, but only the first enrollment was included (n=251 subsequent pregnancies excluded). Right from the Start also includes information on prescription medications, including NSAIDs, in the first-trimester interview regarding prescription medications taken for pain, bleeding, and other reasons. However, number of women reporting prescription NSAID use was too small to perform statistical analyses (n=10). We, therefore, excluded prescription NSAIDs from the analysis.
Participants were queried about all medications in the intake and first-trimester interviews (Box 1). Both interviews included NSAID exposures during the periconceptional period (eg, from LMP through 6 weeks of gestation). The primary exposure was classified as any NSAID use compared with no NSAID use based on whether the participant reported NSAID use in either interview. Nonsteroidal anti-inflammatory drugs were further grouped by drug class, generic name, and brand name. The primary resource used to classify drugs was the Food and Drug Administration drug classification database.10 Other resources also used include Micromedex 2.0, Lexi-Comp ONLINE, Epocrates Online Premium, and DailyMed.11 – 14 Over-the-counter drugs not reported according to store brands and that could not be classified with the previously listed resources were identified with www.drugstore.com.15 We did not include acetaminophen use in the NSAID definition, with the exception of drugs that included acetaminophen and an NSAID as the active ingredients.
Maternal characteristics and obstetric history were also recorded. These included maternal age, height, weight, body mass index (calculated as weight (kg)/[height (m)]2), race or ethnicity, diabetes status, parity, gravidity, induced abortion history, study site, and smoking status (current or not current smokers). Information on these characteristics was obtained from either the first-trimester interview or in person during the study ultrasonography visits.
Analyses were conducted with STATA 11.0 statistical software. We used Cox proportional hazards survival models with variable gestational age at study entry to characterize the rate of pregnancy loss in relation to NSAID exposure (any compared with none) both unadjusted and adjusted for confounders. Allowing for variability of study participant gestational age at study entry will correctly estimate the risk of spontaneous abortion conditional on the fact that none of the participants had pregnancy loss before they were recruited into the cohort.16 Longitudinal data from each woman start at enrollment or after women report a positive pregnancy test if they pre-enrolled in the study and continues through 20 completed weeks of gestation, the occurrence of a pregnancy loss, or loss to follow-up. We used a two-sided α=0.05 significance level for all tests of statistical significance.
Candidate confounders included maternal age (years), body mass index, race or ethnicity (Caucasian [referent], African American, Hispanic ethnicity), income ($40,000 or less, $40,000–80,000 [referent], more than $80,000), diabetes status (no diabetes [referent] compared with any [type 1, type 2, gestational, or multiple]), parity (none [referent] compared with one or more), gravidity (none [referent] compared with one or more), induced abortion history (none [referent], one or more), study site (North Carolina [referent], Tennessee, Texas), and smoking status (not current [referent] compared with current). Candidate confounders were analyzed for independent association with both NSAID exposure and spontaneous abortion outcome. Those that were independently associated with NSAID exposure and spontaneous abortion outcome and that resulted in a 5% relative change in NSAID effect size estimates were retained in the model. No candidate met inclusion criteria. However, analyses adjusted for maternal age are also presented because this covariate was commonly included in multivariable models in previous studies of NSAID use and spontaneous abortion risk.5 – 7 Using Cox regression, we estimated adjusted hazard ratios (HRs) with 95% confidence intervals (CIs) for risk of miscarriage with NSAID exposure. Subanalyses were performed to assess whether the effect sizes varied by gestational age at time of spontaneous abortion (gestation less than 10 weeks compared with 10 weeks or more) and NSAID class. Additional secondary analyses were performed examining the relationship between total days of NSAID use and spontaneous abortion risk using Cox regression. Total days of NSAID use was calculated based on the total number of days reported across all NSAIDs an individual reported taking. Because individuals were asked the same set of questions in the intake and first-trimester interviews, we analyzed using the largest number of days reported across the two sets of interviews. To compare our findings with those from studies that have used different definitions for NSAID exposure, we also performed Cox regression analyses excluding aspirin from the NSAID exposure definition.
Finally, we examined developmental stage at ultrasonography among women who had a loss to assess whether ultrasonographic characteristics differed if they were exposed compared with unexposed to NSAIDs. Ultrasonographic characteristics examined included: the presence of fetal pole with a normal heart rate, fetal pole with an abnormal or no heart rate or an abnormal heart rate, the presence of a gestational sac or the presence of a gestational sac and yolk sac, and an empty uterus with a positive pregnancy test.
We included a total of 2,780 women who enrolled in the study between 2004 and 2010 (Table 1). The overall prevalence of early first-trimester NSAID use among women in Right from the Start was 43% (n=1,185). In the total sample, 367 had a spontaneous abortion (13%). Half of the losses occurred before the tenth week of pregnancy. Compared with nonusers, NSAID users were more likely to be white, have had no previous pregnancies, and have had no history of spontaneous abortion. In bivariate analyses, increased maternal age, gravidity one or more, African American race, having an income greater than $80,000, and having a history of spontaneous abortion were associated with an increased risk of spontaneous abortion (P<.05). The most common classes of NSAIDs taken were propionic acids (n=1,020, such as ibuprofen and naproxen) with similar use among NSAID users with (83%) and without (86%) a loss (Table 2). Women also reported taking salicylates (n=293, primarily aspirin), acetic acids (n=3, such as indomethacin), and enolic acid derivatives (n=1, piroxicam).
The unadjusted and adjusted models did not show an association between NSAID exposure and spontaneous abortion risk (unadjusted HR 1.01, 95% CI 0.82–1.24; adjusted HR 1.00, 95% CI 0.81–1.23) (Table 3). Analyses stratified by early and late losses showed no association between NSAID use and risk for early or late losses. We also examined spontaneous abortion risk excluding salicylates and did not observe an association (HR 0.91, 95% CI 0.72–1.15; adjusted HR 0.91, 95% CI 0.72–1.14). Additional analyses of total days of NSAID use and spontaneous abortion risk did not show evidence of an association (Table 3).
We further examined whether risk for spontaneous abortion by NSAID use differed by whether the interview when NSAID data were collected was conducted before or after the loss occurred. All women who had interviews after their loss (n=42) reported NSAID use. When information from ultrasonography is incorporated to assess stage of pregnancy development before loss, we observed no difference in stage of arrested development among losses across the spectrum from anembryonic gestation through normally developing early gestations (Table 4). A summary of the probability of loss by gestational age and NSAID exposure is shown in Figure 1.
Given the plausibility of NSAID effects on prostaglandin synthesis, embryo transport, implantation, placental development, maintenance of pregnancy, and prior evidence from epidemiologic data, we examined the relationship between periconceptional over-the-counter NSAID exposure and spontaneous abortion risk in a nonclinical cohort. We found no evidence of an association between NSAID use and increased risk for spontaneous abortion. Analyses stratified by NSAID class suggested that aspirin users (salicylates) were possibly at increased risk of spontaneous abortion; however, this association was not statistically significant. To compare our results with those from other studies that examined only the effect of nonaspirin NSAIDs, we also performed subanalyses excluding aspirin users and did not observe and increased risk for spontaneous abortion.7 However, we had less power to detect an association with aspirin use in our cohort relative to prior studies of aspirin use and spontaneous abortion risk.7 Also, we note that a small subset of women (n=42) were interviewed after their spontaneous abortion reported. All 42 women reported taking an NSAID, which may indicate an element of recall bias but would be expected to bias results toward finding an association.
The biological basis for suspecting a link between NSAIDs and risk of spontaneous abortion rests on the multiple stages in development that involve prostaglandin synthesis, particularly during early pregnancy when prostaglandins such as prostaglandin E synthases may be essential for establishing the implantation and early placentation but may be protective later. NSAIDs inhibit COX enzymes, which catalyze the formation of prostaglandins from arachidonic acid and thus reduce prostaglandin synthesis. Prostaglandins have important roles in both maintaining and achieving pregnancy. For example, COX-2 knockout mice (that lack COX-2) ovulate fewer oocytes have incomplete decidualization, have low fertilization rates, and fail implantation, all of which can be improved by administering prostaglandins.17 Altering prostaglandin levels has direct effects on conception, implantation, and maintenance of pregnancy in animal models.18 – 28 Because NSAIDs inhibit prostaglandin synthesis, they may influence embryo transport, implantation, placental development, and maintenance of pregnancy and have been shown to lead to embryonic demise in animal models.29 This suggests that NSAID use during early pregnancy is a plausible biological candidate for causing a spontaneous abortion.
Few studies have examined NSAID exposure and risk for spontaneous abortion (Table 5). Two studies examined prescription NSAIDs only. These include a Danish study that observed an increased risk of spontaneous abortion for women taking NSAIDs.6 This study, using national health care and pharmacy records, considered the effect of prescription NSAID use from 30 days before conception through the end of pregnancy on risk of congenital anomalies, low birth weight, preterm birth, and spontaneous abortion in primigravid women. One percent of the cohort had been prescribed NSAIDs. They observed a statistically significant association between NSAID use and spontaneous abortion for NSAIDs taken 1 week and up to 9 weeks before the loss.6 However, in subsequent re-examination of results adjusting for gestational age,30 the association was no longer statistically significant (NSAIDs taken 1 week and up to 9 weeks before spontaneous abortion, odds ratio 1.5–3.4, 95% CI 0.6–12.8). A second, more recent study of prescription NSAIDs conducted using the Quebec Pregnancy Registry observed that nonaspirin NSAID use was associated with an increased risk of spontaneous abortion, with an odds ratio of 2.4 (95% CI 2.1–2.8), using NSAID prescription data available in the registry for which both indication and duration of use are likely to differ from over-the-counter. Three percent of their cohort reported exposure to nonaspirin prescription NSAIDs.7
Only one prior study included nonprescription NSAIDs, and their analysis included adjustment for gestational age at entry.5 This study used self-reported NSAID use, including prescription and over-the-counter NSAIDs (ibuprofen, naproxen, Advil, Motrin, and Naprosyn) with a cohort of patients from Kaiser Permanente HMO in northern California.5 Nonsteroidal anti-inflammatory drug exposure included use between the LMP and a positive pregnancy test. The authors reported that NSAID use around the time of conception resulted in a 5.6% increase in hazard of spontaneous abortion compared with no NSAID use (95% CI 2.3–13.7). Five percent (n=53) of the cohort reported using NSAIDs during this gestational period.
In our study we collected data on all over-the-counter pain medications including medications to prevent menstrual cramps in anticipation of menses as well as medicine for headaches, aches, pains, and colds. We then identified all specific classes of NSAIDs in these medications. We used Cox regression and accounted for gestational age at enrollment to accurately model time at risk.5 However, we observed much higher use of NSAIDs compared with what was reported by other studies (43%). Over-the-counter NSAIDs may be more likely to be taken at low and intermediate dosages compared with prescribed forms of NSAIDs, which may explain why the studies of prescription NSAIDs observed an increased risk for spontaneous abortion, whereas we observed a null effect. There may also be other indications for NSAID use among women who take prescribed NSAIDs (eg, a chronic medical condition) that put them at increased risk of spontaneous abortion.
We observed a much higher exposure rate of NSAIDs in our cohort than was previously observed 5% for the Li and colleagues5 study, which also attempted to captured both prescription and nonprescription NSAID exposures. This is likely the result of differences in exposure assessment. We asked about any type of pain or cold medication and then determined whether it included NSAIDs, whereas Li and colleagues limited NSAIDs to a specific subset. As a result, Li and colleagues may have missed some women exposed to NSAIDs. Another difference is the timing of the interviews across our studies. Li et al interviewed participants immediately after pregnancy confirmation, capturing exposures between LMP and pregnancy confirmation, although in our study we obtained data from baseline screening interviews after pregnancy confirmation and from first-trimester interviews targeted at less than 13 weeks of gestation that specifically ask about use between LMP and 6 weeks. Finally, it may be the case that prescription NSAID use was responsible for the observed association with spontaneous abortion in the previous studies. Because exposure to prescription NSAIDs was rare in our study population, we excluded the 10 women who were known to have been exposed to prescription NSAIDs.
We did not observe an association between NSAIDs and spontaneous abortion risk; however, our findings have implications for women taking over-the-counter NSAID during pregnancy. In our cohort, assuming an α of 0.05 and a 13% prevalence of spontaneous abortions, we would have 81% power to detect an HR of 1.35, which is well below the risk estimate observed in prior studies. We show that typical over-the-counter NSAIDs taken by women during the first trimester of pregnancy do not put them at increased risk of spontaneous abortion. As a result of limited numbers, we could not directly address the previous findings that focused on prescription-based NSAIDs, which are more likely to be at higher dosages and reflect a different risk attributable to NSAID exposure than what we observed in our study. It is also possible that women did not indicate if they took a prescribed NSAID or that they did not distinguish between prescribed and unprescribed forms. Our rate of miscarriage, 13%, is consistent with most epidemiological studies that are looking at spontaneous abortion in the context of clinical recognized pregnancies. Further research is necessary to assess whether NSAID exposure before implantation or shortly after may cause a spontaneous abortion before clinical recognition of pregnancy. Finally, further understanding the role of dose, timing in gestation, and the potential role of recall bias may help to reconcile this with previously observed associations between NSAIDs and spontaneous abortion.
1. Kaufman DW, Kelly JP, Rosenberg L, Anderson TE, Mitchell AA. Recent patterns of medication use in the ambulatory adult population of the United States: the Slone survey. JAMA 2002;287:337–44.
2. Paulose-Ram R, Hirsch R, Dillon C, Losonczy K, Cooper M, Ostchega Y. Prescription and non-prescription analgesic use among the US adult population: results from the third National Health and Nutrition Examination Survey (NHANES III). Pharmacoepidemiol Drug Saf 2003;12:315–26.
3. Curhan GC, Bullock AJ, Hankinson SE, Willett WC, Speizer FE, Stampfer MJ. Frequency of use of acetaminophen, nonsteroidal anti-inflammatory drugs, and aspirin in US women. Pharmacoepidemiol Drug Saf 2002;11:687–93.
4. Olesen C, Steffensen FH, Nielsen GL, de Jong-van den Berg, Olsen J, Sorensen HT. Drug use in first pregnancy and lactation: a population-based survey among Danish women. The EUROMAP group. Eur J Clin Pharmacol 1999;55:139–44.
5. Li DK, Liu L, Odouli R. Exposure to non-steroidal anti-inflammatory drugs during pregnancy and risk of miscarriage: population based cohort study. BMJ 2003;327:368.
6. Nielsen GL, Sorensen HT, Larsen H, Pedersen L. Risk of adverse birth outcome and miscarriage in pregnant users of non-steroidal anti-inflammatory drugs: population based observational study and case-control study. BMJ 2001;322:266–70.
7. Nakhai-Pour HR, Broy P, Sheehy O, Berard A. Use of nonaspirin nonsteroidal anti-inflammatory drugs during pregnancy and the risk of spontaneous abortion. CMAJ 2011;183:1713–20.
8. Promislow JH, Makarushka CM, Gorman JR, Howards PP, Savitz DA, Hartmann KE. Recruitment for a community-based study of early pregnancy: the Right From The Start study. Paediatr Perinat Epidemiol 2004;18:143–52.
9. Hoffman CS, Messer LC, Mendola P, Savitz DA, Herring AH, Hartmann KE. Comparison of gestational age at birth based on last menstrual period and ultrasound during the first trimester. Paediatr Perinat Epidemiol 2008;22:587–96.
16. Dupont W. Statistical modeling for biomedical researchers: a simple introduction to the analysis of complex data. 2nd ed. Cambridge (UK): Cambridge University Press; 2009.
17. Lim H, Paria BC, Das SK, Dinchuk JE, Langenbach R, Trzaskos JM, et al.. Multiple female reproductive failures in cyclooxygenase 2-deficient mice. Cell 1997;91:197–208.
18. Bergh PA, Navot D. The impact of embryonic development and endometrial maturity on the timing of implantation. Fertil Steril 1992;58:537–42.
19. Duffy DM, Stouffer RL. The ovulatory gonadotrophin surge stimulates cyclooxygenase expression and prostaglandin production by the monkey follicle. Mol Hum Reprod 2001;7:731–9.
20. Fazleabas AT, Kim JJ, Srinivasan S, Donnelly KM, Brudney A, Jaffe RC. Implantation in the baboon: endometrial responses. Semin Reprod Endocrinol 1999;17:257–65.
21. Frank GR, Brar AK, Cedars MI, Handwerger S. Prostaglandin E2 enhances human endometrial stromal cell differentiation. Endocrinology 1994;134:258–63.
22. Kelly RW, Carr GG, Elliott CL, Tulppala M, Critchley HO. Prostaglandin and cytokine release by trophoblastic villi. Hum Reprod 1995;10:3289–92.
23. Kim JJ, Wang J, Bambra C, Das SK, Dey SK, Fazleabas AT. Expression of cyclooxygenase-1 and -2 in the baboon endometrium during the menstrual cycle and pregnancy. Endocrinology 1999;140:2672–8.
24. Kistner EO, Weinberg CR. Method for using complete and incomplete trios to identify genes related to a quantitative trait. Genet Epidemiol 2004;27:33–42.
25. Norwitz ER, Schust DJ, Fisher SJ. Implantation and the survival of early pregnancy. N Engl J Med 2001;345:1400–8.
26. Pall M, Friden BE, Brannstrom M. Induction of delayed follicular rupture in the human by the selective COX-2 inhibitor rofecoxib: a randomized double-blind study. Hum Reprod 2001;16:1323–8.
27. Sookvanichsilp N, Pulbutr P. Anti-implantation effects of indomethacin and celecoxib in rats. Contraception 2002;65:373–8.
28. Wilcox AJ, Baird DD, Weinberg CR. Time of implantation of the conceptus and loss of pregnancy. N Engl J Med 1999;340:1796–9.
29. Lala PK, Scodras JM, Graham CH, Lysiak JJ, Parhar RS. Activation of maternal killer cells in the pregnant uterus with chronic indomethacin therapy, IL-2 therapy, or a combination therapy is associated with embryonic demise. Cell Immunol 1990;127:368–81.
© 2012 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.
30. Nielsen GL, Skriver MV, Pedersen L, Sorensen HT. Danish group reanalyses miscarriage in NSAID users. BMJ 2004;328:109.