To evaluate the association between non-steroidal anti-inflammatory drugs (NSAIDs) and miscarriage, which remains elusive in the actual literature, using modelling approaches to mitigate time-related bias.

Nationwide, population-based retrospective cohort study.

The study used data from the French National Mother-Child Register (EPI-MERES), developed from the French National Health Data System containing the health information (hospital admissions, drug dispensing, comorbidities, etc.) of about 99% French population.

4 857 907 pregnancies ended in a live birth, a stillbirth or a miscarriage from 2013 to 2019.

Exposure to NSAIDs from 2 weeks before conception to the 20th week of pregnancy.

Miscarriage (ie, spontaneous abortion before the 20th week of pregnancy) from the sixth week of pregnancy. The Cox regression with a time-dependent exposure that incorporated a 3-day lag was used to estimate HRs and their 95% CIs adjusted for maternal characteristics. The 3-day lag period allows for addressing protopathic bias.

In total, there were 163 666 (3,37%) miscarriage cases, and 349 294 (7.19%) pregnancies were exposed to NSAIDs. Unexposed pregnancies were used as the reference category in all analyses. In the main analysis, exposure to NSAIDs increased the risk of miscarriage (HR, 1.83; 95% CI 1.81 to 1.86). The effect of individual drugs was heterogeneous, with 7 of the 19 drugs evaluated shown to increase the risk (flurbiprofen had the highest risk (HR, 3.28; 95% CI 3.15 to 3.41) and naproxen the lowest (HR, 1.09; 95% CI 1.01 to 1.18)). In the sensitivity analyses, increasing the lag time decreased the estimated HR (from HR, 2.25; 95% CI 2.21 to 2.28 with no lag to HR, 1.56; 95% CI 1.54 to 1.59 with a 7-day lag). Overall, the risk of miscarriage remained statistically significant in all the analyses.

Our study found that early exposure to NSAIDs could increase the risk of miscarriage. This finding contributes to the body of evidence on their safety profile and may help inform future recommendations for their use in pregnant women.