A large population-based cohort study found no adjusted association between first-trimester exposure to nonsteroidal anti-inflammatory drugs and major congenital malformations, offering cautious reassurance for early-pregnancy exposure at a time when analgesic counseling in pregnancy has become increasingly complicated.
The study followed recent regulatory attention to the possible neurodevelopmental risks associated with acetaminophen use during pregnancy. Although the study did not evaluate acetaminophen safety directly, its findings provide additional evidence for physicians weighing treatment options for pain, inflammation, and fever in early pregnancy. An accompanying Perspective published in PLOS Medicine cautioned, however, that the results should not be interpreted as evidence of broad nonsteroidal anti-inflammatory drug (NSAID) safety throughout pregnancy or across all fetal and neonatal outcomes.
In the retrospective register-based study, investigators evaluated pregnancies in Southern Israel from 1998 to 2018 using data from the Southern Israeli Pregnancy Registry. The initial data set included 267,301 births and elective terminations occurring during the study period. Following exclusions for factors such as multiple gestations, genetic or chromosomal anomalies, established teratogenic drug exposure, and missing data, 264,858 complete cases were included in the final analytic cohort. This cohort included singleton pregnancies resulting in live birth, stillbirth, or elective termination for suspected fetal malformations.
First-trimester NSAID exposure was defined using pharmacy dispensation records from the first day of the last menstrual period through the end of gestational week 13. Major congenital malformations were identified through linked clinical, hospitalization, and pregnancy-termination records through the child’s first year of life.
Overall, about 8% (n = 20,202) of the pregnancies had first-trimester NSAID exposure. Ibuprofen was the most commonly dispensed agent, identified in 5% (n = 13,627) of the pregnancies followed by diclofenac in 2% (n = 4,334) of pregnancies, and naproxen in 1% (n = 3,105) of pregnancies. Etodolac, indomethacin, piroxicam, and lornoxicam were less commonly dispensed.
Major congenital malformations were more common in NSAID-exposed pregnancies compared with in unexposed pregnancies in unadjusted analyses. Among NSAID-exposed pregnancies, about 8% (n = 1,651) were diagnosed with major congenital malformations compared with about 7% (n = 16,998) among unexposed pregnancies. However, after generalized full matching and adjustment for maternal and pregnancy characteristics, including clinical indications for NSAID use, the association was no longer statistically significant.
Null findings were also observed across organ-system categories, including cardiovascular, musculoskeletal, central nervous system, gastrointestinal, and genitourinary malformations. Drug-specific analyses did not identify statistically significant associations between ibuprofen, diclofenac, naproxen, etodolac, or indomethacin and major congenital malformations. However, indomethacin exposure was identified in just 287 pregnancies, and the matched adjusted relative risk for any major congenital malformation was 1.76. As a result, the investigators could not reliably determine a clinically meaningful indomethacin-specific risk.
The investigators also found no statistically significant dose-response association, but the dose findings should not be read as equally reassuring across all exposure durations. The prevalence of major congenital malformations was 8.2% among short-term NSAID users, defined as 1 to 7 defined daily doses; 8.6% among medium-term users, defined as 8 to 21 defined daily doses; and 9.3% among long-term users, defined as more than 21 defined daily doses, compared with 7.0% in the unexposed group. Adjusted analyses did not show a statistically significant association across these categories, but the investigators noted that the long-term point estimate was 1.24, indicating greater uncertainty with higher cumulative dispensing.
The study had several methodologic strengths. The investigators used directed acyclic graphs to guide covariate selection, generalized full matching to improve the balance between exposed and unexposed pregnancies, and G-computation to estimate adjusted relative risk. The inclusion of elective pregnancy terminations for fetal malformations was important because birth-only cohorts may miss severe anomalies detected in the prenatal stage. Follow-up through the first year of life also increased the likelihood of identifying major congenital malformations diagnosed following birth. The investigators made the analytical code publicly available on GitHub, adding transparency to the analysis.
Several limitations should guide interpretation. NSAID exposure was based on drug dispensation rather than confirmed ingestion. Some over-the-counter ibuprofen use may not have been captured, although sensitivity analyses suggested that plausible levels of unrecorded exposure were unlikely to materially change the results. The study also did not capture spontaneous abortions, a limitation that is clinically important because data on early-pregnancy NSAID exposure and miscarriage remains mixed, with some studies reporting increased risks and others reporting elevated but statistically insignificant associations. This limitation was also notable because the investigators cited prior findings on fetal NSAID exposure and spontaneous abortions as well as prior Soroka University Medical Center validation work used to inform assumptions about unrecorded over-the-counter exposure.
The study population also raised generalizability considerations. It included a large proportion of Bedouin women, including 77% of NSAID-exposed pregnancies and 53% of unexposed pregnancies, an imbalance consistent with observed differences in NSAID indications and drug-use patterns. The investigators adjusted for ethnicity, and they noted that the Bedouin population has distinct demographic and social characteristics, including historically higher rates of consanguinity. These features did not negate the findings, but they could be relevant when applying the results to other populations and health systems.
In the accompanying Perspective, authors Andrew S. C. Yuen and Kenneth K. C. Man described the findings as clinically meaningful and reassuring, but not definitive. They emphasized a persistent challenge in perinatal pharmacoepidemiology: analgesic use during pregnancy is often intermittent, sequential, and tied to evolving symptoms. In the study, exposure to other antipyretic or analgesic drugs, including acetaminophen and dipyrone, was incorporated into the analytic approach to help account for underlying indications for NSAID use.
That adjustment was not comparable to adjusting for a baseline factor such as maternal age or diabetes. If a patient uses acetaminophen or dipyrone after starting an NSAID, that later medication use may reflect persistent symptoms, worsening disease, or treatment switching, according to the authors. In that setting, adjusting for the later drug use can partly adjust away information that occurred after the exposure began.
In the study, the investigators recognized this concern and conducted additional sensitivity analyses addressing temporal ambiguity and this type of overadjustment related to other analgesic or antipyretic exposure. Those analyses provided additional reassurance, but they did not fully resolve the broader challenge of evaluating intermittent and sequential analgesic use during pregnancy as a single exposure.
The Perspective authors cautioned against extrapolating the results beyond the specific outcome studied. The study did not resolve questions about miscarriage, low birth weight, oligohydramnios, neurodevelopmental outcomes, or later-pregnancy NSAID risks. NSAID-related fetal renal and ductal risks later in pregnancy remain well recognized, particularly after 20 weeks’ gestation. The Perspective authors also called for future studies to examine the risks by drug class, including selective COX-2 inhibitors as well as by dose, duration, gestational timing, indication, concomitant treatments, and broader offspring outcomes.
The broader literature remains mixed. The Perspective authors noted that a previous nationwide South Korean cohort study reported modestly increased risks of major congenital malformations associated with early-pregnancy NSAID exposure and higher risks of low birth weight and oligohydramnios. The latter findings involved outcome domains the Israeli cohort did not examine, meaning the study could not address whether early NSAID exposure was associated with fetal growth or amniotic-fluid outcomes. Differences in exposure definitions, comparator groups, outcome ascertainment, populations, and health systems may partly explain the discordant findings.
For clinicians, the practical message was cautious reassurance, particularly for short-course first-trimester exposure. This study authors did not find evidence of a substantial increase in major congenital malformation risk following first-trimester NSAID exposure, but its findings should not be read as categorical reassurance about NSAID use throughout pregnancy or about outcomes beyond those examined in the analysis.
The counseling challenge remains that untreated pain, inflammation, and fever may also carry risks. In practice, the findings support individualized, indication-specific counseling in early pregnancy, with decisions grounded in shared communication about the trade-offs between drug exposure and undertreated maternal disease rather than categorical reassurance or categorical avoidance.
The study reported no specific funding and no competing interests. The accompanying Perspective also reported no specific funding. Yuen disclosed grants from the University College London Hospitals NIHR Biomedical Research Centre, the College of Mental Health Pharmacy, and the UK Turing Scheme outside the submitted work. Man disclosed grants from the CW Maplethorpe Fellowship, the European Union Horizon 2020 program, the UK National Institute for Health Research, the South Korea Ministry of Food and Drug Safety, the Hong Kong Research Grant Council, and the Hong Kong Innovation and Technology Commission, as well as consultancy from IQVIA and AstraZeneca, all outside the submitted work.
Source: PLOS Medicine, Perspective