Overall, there were 3173 women (15.5%) who had been exposed to NSAIDs during the first trimester who reported their frequency of use and 1452 women (7.1%) who reported using NSAIDs “as needed” during a time period that included the first trimester. Of the women who reported their exposure frequency, 703 (22%) reported using aspirin; 2451 (77%) reported using ibuprofen; 475 (15%) reported using naproxen, and 39 (1.2%) reported using other NSAIDs. Of the women who reported using NSAIDs as needed, 218 (15%) reported aspirin use; 1240 (85%) reported ibuprofen use; 249 (17%) reported naproxen use, and 5 (0.34%) reported the use of other NSAIDs. In total, 693 of exposed women (15%) reported using >1 type of NSAID (aspirin, ibuprofen, or naproxen). There were 15,836 women (77%) who were not exposed to NSAIDs at any time during pregnancy.

Table 1 shows the characteristics of women according to NSAID-exposure groups, among control subjects only. Exposed women were fairly similar, regardless whether they reported their frequency of use but were more likely to take folic acid supplements, to drink alcohol or smoke during pregnancy, and to have a shorter time span between birth and interview than unexposed women.

Table 2 shows the patterns of NSAID exposure among cases and control subjects. Sixteen percent of all cases and 14% of all control subjects were exposed to NSAIDs during the first trimester. Approximately one-half as many case and control mothers reported NSAID use as needed (7.3% and 6.6%, respectively). Because most exposure to NSAIDs, regardless of case/control status and frequency of exposure, was through oral route and was single component use, we decided to focus on oral, single component NSAID exposure for the remainder of the analysis.

Table 3 shows the distribution of oral, single component aspirin, ibuprofen, and naproxen use for specific defect groups and control subjects. Women whose NSAID information was missing were included in the total number of cases but not in the exposed or unexposed group counts. Exposure rates were generally similar for most specific defect case groups and control subjects, with some exceptions. ORs and 95% CIs (both crude and adjusted for study site, race/ethnicity, and first trimester folic acid supplementation, smoking, alcohol, and fever) were calculated. We presented results for which adjusted ORs were <0.70 or >1.5 for at least 1 of the NSAID types.

Adjusted ORs for single component, oral ibuprofen, aspirin, and naproxen exposure in the first trimester are presented in Table 4 for the 9 defect types for which an elevated risk was seen. When specific defect types were considered, risk estimates were not consistent across the 3 types of NSAID exposure for these defects. However, ibuprofen, aspirin, and naproxen were each associated with increased risks for neural tube defects, albeit different neural tube defect types. Similarly, one or another of the oral cleft types was associated with ibuprofen, aspirin, and naproxen, although the cleft type varied by exposure. Amniotic band/limb body wall complex was associated with both aspirin and ibuprofen exposure. For pulmonary valve stenosis, the risk was confined to naproxen use. The strongest and most consistent associations were observed for anophthalmia/microphthalmia, with adjusted ORs of 3.0 (95% CI, 1.3–7.3), 1.9 (95% CI, 1.1–3.3), and 2.8 (95% CI, 1.1–7.3), for aspirin, ibuprofen, and naproxen exposure, respectively. The OR that compared any NSAID exposure to nonexposure for the anophthalmia/microphthalmia group was 2.1 (95% CI, 1.3–3.5). A possible 2-fold increased risk of transverse limb deficiencies was observed for naproxen use, with a lower 95% CI of 1.0. For all the birth defects that are presented in Table 4 with enough power to examine the “as needed” exposure group, relative risk estimates for women with “as needed” use were nearly identical to women who were classified as exposed.

We also ran this analysis that was restricted only to cases with isolated defects (Table 4) and found that the effect estimates were nearly identical to those for the nonrestricted cases, with the exception of anophthalmia/microphthalmia, for which the increased risk seemed confined to associated cases only. Most NSAID-exposed anophthalmia/microphthalmia cases had a variety of accompanying birth defects; the adjusted OR for any NSAID exposure and risk of anophthalmia/microphthalmia that was restricted to the 14 nonisolated cases was 2.8 (95% CI, 1.4–5.5). Because other NBDPS investigators have shown both fever and genitourinary infection to increase anophthalmia/microphthalmia risk,⁸ we estimated the OR for NSAID use after excluding women with such histories during the first trimester of pregnancy; the elevated ORs for non-isolated anophthalmia/microphthalmia and exposure to any NSAID during the first trimester remained (OR, 2.4; 95% CI, 1.5–3.8).

This multisite, population-based, case-control study found that 22.6% of women in the NBDPS reported the use of NSAIDs in the first trimester of pregnancy, of which 70% reported the frequency of their use and 30% reported using NSAIDs “as needed.” The most commonly reported NSAID exposures were aspirin, ibuprofen, and naproxen. Although most defects that were examined were not associated with NSAID use, small-to-moderate increased risks were estimated for ibuprofen, aspirin, and naproxen use in relation to anophthalmia/microphthalmia, amniotic bands/limb body wall defects, pulmonary valve stenosis, possibly transverse limb deficiencies, oral clefts, and neural tube defects, although the type of oral cleft and neural tube defect that was associated with exposure that varied across specific drugs. For example, ibuprofen use was associated with spina bifida; aspirin use was associated with anencephaly/craniorachischisis and encephalocele, and naproxen use was associated with encephalocele. Although we evaluated women who reported using NSAIDs “as needed” separately to examine whether the association may differ in this group, given the larger potential likelihood of exposure misclassification, we found no evidence of this difference in our analysis.

The NBDPS data have been used previously to examine the relationship between NSAIDs and muscular ventricular septal defects. Similar to this study, the authors found no association between NSAIDs and muscular ventricular septal defects during the first trimester (adjusted OR, 1.0; 95% CI, 0.64–1.6) or 1 month before pregnancy to birth (adjusted OR, 0.99; 95% CI, 0.67–1.5).⁹ However, that study did not examine birth defects other than muscular ventricular septal defects. Other previous analyses of NBDPS data that used earlier versions of the data set have seen associations between nonaspirin NSAIDs and gastroschisis among older women¹⁰ and between NSAIDs and amniotic bands and terminal transverse limb defects.¹¹ In the present study, we also reported a borderline increase in terminal transverse limbs with the use of naproxen and in amniotic bands with the use of aspirin and ibuprofen. Similar to our results, a previous study,¹² which used prospectively reported registry data, found an association between cleft lip with or without cleft palate and naproxen. That study was based on 4 exposed cases; this current study is based on 45 cases and thus able to produce a more stable effect estimate. Therefore, for oral clefts, neural tube defects, and amniotic bands/limb body wall complex, increased risks were observed for isolated defects, which suggests an effect on the development of those particular structures. In contrast, the association for anophthalmia/microphthalmos was confined to cases with accompanying defects and did not appear to be attributed to associations with neural tube defects, oral clefts, or amniotic bands. Of the 14 NSAID-exposed anophthalmia/microphthalmos cases, 7 had unilateral craniofacial anomalies that were suggestive of possible hemifacial microsomia. Early pregnancy ibuprofen exposure was associated with a possible increased risk of hemifacial microsomia in a previous study (OR, 1.7; 95% CI, 0.9–3.0).¹³

Strengths of this study included detailed clinical classification of outcomes and ability to control for a wide number of covariates. Our study is limited by the retrospective design and self-reported exposures, which could lead to differential misclassification between cases and control subjects. The use of NSAIDs during pregnancy was self-reported by the mother between 6 weeks and 24 months after expected due date. It is possible that mothers had trouble recalling their use of NSAIDs in pregnancy. Further, the data were collected after the outcome of pregnancy was known, and it is possible that women with offspring who had a birth defect were more or less likely to recall their use of NSAIDs than women who served as control subjects. Another potential limitation results from multiple comparisons. However, although it is possible that chance is the explanation for some of the observed associations, the fact that 20 of the 168 comparisons that were made were statistically significant suggests that at least some of them may not be due to chance.

In summary, the use of NSAIDs in early pregnancy does not appear to be a major risk factor for birth defects, although there were a few moderate associations between NSAIDs and anophthalmia/microphthalmia, amniotic bands/limb body wall defects, pulmonary valve stenosis, oral clefts, and neural tube defects. Because these associations have not been reported previously from other datasets (with the exception of naproxen and cleft lip with or without cleft palate), further studies with detailed data on timing, frequency, dose, and indication are necessary.