Use of nonaspirin nonsteroidal anti-inflammatory drugs during pregnancy and the risk of spontaneous abortion

Dear Editor,

We would like to thank Dr Clark and colleagues for their letter on our paper published in CMAJ (1), and we thank CMAJ for giving us the opportunity to answer. We disagree with the majority of what Clark et al. have put forward but we agree that prescription medications do not necessarily indicate that women take their medications as was mentioned in our discussion section. However, our study defined NSAIDs exposure dichotomously and thus anyone taking at least one pill was defined as being exposed. Given that NSAIDs use is on an acute basis even when used for chronic conditions such as rheumatoid arthritis or lupus, it is very unlikely that women who took the time to go to the pharmacy and gave out of pocket money did not take at least one pill. Furthermore, Glover et al. (2) have shown that the majority of pregnant women who fill a prescription take it, and Daniels et al. (3) have shown that self-reporting measures of medication use does not provide accurate measurements of adherence when compared with electronic monitoring as was the case in our study.

Our study team has performed and is still performing many validation studies within the Quebec Pregnancy Registry (which includes data from the RAMQ, MED-ECHO, ISQ and a maternal self-administered questionnaire) to increase internal and external validity. Unlike what Clark and colleagues put forward, we feel it is important to do these studies to help us assess potential biases that might arise from our studies. This is why we are referring to these studies in our paper just published in CMAJ. The fact that Clark et al. have referenced a study performed by our team to assess generalizability of study findings and put it in the context of interval validity shows a clear misunderstanding of perinatal pharmacoepidemiology.

Data on maternal smoking and BMI is only available on a subset of women (n=3273) included in the Quebec Pregnancy Registry. This was acknowledged in our discussion section. It remains however that for these variables to bias our results, they need to confound the effect. Hence, for smoking and BMI to be confounders, they need to be associated with both the outcome (spontaneous abortion), and independently and separately the exposure (NSAIDs). Hence, Delaney et al. (4) have shown that smoking and BMI were not associated with prescribed NSAIDs use or over-the-counter (OTC) NSAIDs use. Our discussion clearly states that. With this information at hand, although data on smoking and BMI were missing from our analyses, it is very unlikely that this would have biased our estimates.

Clark et al. are suggesting that we should compare the obstetrical histories and pre-existing comorbidities of the subgroup of women who filled a prescription for NSAIDs and had a spontaneous abortions, and the group that filled a prescription NSAIDs and did not have a spontaneous abortion. Assuming that they would also like us to adjust for time of follow-up during pregnancy, which is essential to correctly answer their question, we would end up with the results we are presenting in our paper. Indeed, this is exactly what we have done. We have shown that there were differences in health care utilization and comorbidities between cases and controls, and NSAID users and non-users. These differences were adjusted for in our multivariate analyses.

Clark et al. are suggesting that OTC NSAID use was not considered in our study, and this seriously flawed our results. As was mentioned in our paper, prescribed OTC NSAIDs were considered. Indeed, the Registry includes data on prescription medication use (OTC included) during gestation and no data are available on OTC medication use without a prescription. In the case of NSAIDs, the only NSAID available OTC in Canada is ibuprofen. As highlighted in our CMAJ paper, although our study looked at OTC ibuprofen obtained with a prescription, it is true that the number of women who used ibuprofen OTC without a prescription could not be measured. However, there is no reason to believe that women who miscarried used more or less ibuprofen OTC without a prescription than those who did not miscarry as was wrongly insinuated, leading to non-differential misclassification. Hence, even if OTC ibuprofen use without a prescription was not measured in the study, the study estimate is an underestimate of the true effect, meaning that if we had had data on OTC NSAID medication use without a prescription, the estimate would have been higher. This was explained in our manuscript. Interpreting results differently is a misunderstanding of the results.

The aim of our study, or any other studies performed in pregnant women, is not to scare women but to better inform women and physicians on the potential risks of medication use during pregnancy. This is even more important given that 50% of women do not plan their pregnancy (where the majority of inadvertent exposures occur). We strongly believe that more data (as opposed to withholding of data) are needed regarding medication use during pregnancy, and that pregnancy does not make women unable to make appropriate decision regarding their health during this critical period. Good quality data on large cohorts of pregnant women empower women and do not scare them.

Anick Berard PhD and Odile Sheehy MSc University of Montreal and CHU Ste-Justine

References

1. 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 Sep 6. [Epub ahead of print] 2. Glover DD, Amonkar M, Rybeck BF, Tracy TS. Prescription, over-the- counter, and herbal medicine use in a rural, obstetric population. Am J Obstet and Gynecol 2003 ;188 :1039-1045. 3. Daniels T, Goodacre L, Sutton C, Pollard K, Conway S, Peckham D. Accurate assessment of adherence: self-report and clinician report vs electronic monitoring of nebulizers. Chest 2011 ;140:425-432. 4. Delaney JA, Biggs ML, Kronmal RA, Psaty BM. Demographic, medical, and

behavioral characteristics associated with over the counter non- steroidal

anti-inflammatory drug use in a population-based cohort: results from the Multi-Ethnic Study of Atherosclerosis. Pharmacoepidemiol Drug Saf 2011 ;20:83-89.

Conflict of Interest:

None declared

This is a very interesting result from a study of miscarriages. I am not in the medical field, but I think a lot, i am very science oriented. I just had a thought about a link between Autism and Ibuprofen. Everyone in the U.S. tries to blame our milk or vaccines on the increase of Autism, but has anyone thought of Ibuprofen, it has gained popularity in the last 20-25 years and so has Autism.

Thank you for letting me share my idea.

Conflict of Interest:

None declared