Intended for healthcare professionals


NSAIDs during pregnancy and risk of miscarriage

BMJ 2004; 328 doi: (Published 09 January 2004) Cite this as: BMJ 2004;328:108

True risks or only suspicions?

  1. Benedetta Schiavetti, research fellow (schiavetti{at},
  2. Antonio Clavenna, senior research fellow,
  3. Rita Campi, statistician,
  4. Maurizio Bonati, head
  1. Laboratory for Mother and Child Health and CRIF-Regional Drug Information Centre Mario Negri Institute for Pharmacological Research, I-20157 Milan, Italy

    EDITOR—The finding of Li et al of an association between an increased risk of miscarriage and use of non-steroidal anti-inflammatory drugs (NSAIDs) relies on biased data.1 For the same cohort of women, previous analysis showed that exposure to magnetic fields was a risk factor for miscarriage,2 and this variable should have been included. Moreover, 103/170 women had already had a miscarriage at the time of the interview3; this information should also have been taken into account.

    With regard to the widespread use of NSAIDs4 a cohort study was performed on 1557 pregnant women (gestational age < 20 weeks) who contacted a drug information centre was performed, collecting follow up data on pregnancy outcome after delivery.5 The NSAID users were 281 (10% aspirin, 90% other anti-inflammatory agents) whereas 1276 used other drugs (11% psychotropic drugs, 10% systemic anti-infective agents), with an average of 2.2 drugs per woman (range 1-14).

    In addition to a significant association between miscarriage and exposure to NSAIDs (table), a significant trend became apparent (χ2t = 10.54, P = 0.0011) for the number of drugs taken since the last menstruation. A logistic regression analysis using a backward stepwise selection then showed a significant association between number of drugs taken and miscarriage (odds ratio 1.15, 1.06 to 1.25), but not for use of NSAIDs.

    Univariate analysis of spontaneous abortion risk in percentages (absolute numbers) related to characteristics of 1557 women studied

    View this table:

    Interpretation of data should include all possible confounding factors. In this case scant methodological approaches indicated exposure to NSAIDs early in pregnancy as a substantial risk factor for miscarriage, but only well designed studies and complete analysis can produce more concrete evidence.


    • Competing interests None declared.