NSAIDs during pregnancy and risk of miscarriage
http://www.100md.com
《英国医生杂志》
EDITOR—Although exposure to magnetic fields during pregnancy was associated with miscarriage,1 it was not associated with use of non-steroidal anti-inflammatory drugs (NSAIDs). Therefore it was not a confounder, and adjustment for exposure to magnetic fields did not make any difference to the association between NSAID use and miscarriage. With regard to interviews after miscarriage, please refer to our previous response on bmj.com.2 The bottom line was: had we included only women who were interviewed before their miscarriage (prospectively ascertained), the association between NSAID use and miscarriage would have been even stronger.
If Schiavetti et al intended to disprove our findings with results from their study they failed to provide enough information to allow readers to evaluate their study. On the basis of the limited information they provided, they clearly did not have adequate data to address the relation between NSAIDs and miscarriage.
Firstly, their study was based on selected populations (volunteers who took drugs during pregnancy and called the centre).
Secondly, there was no information on indications for drug use and gestational age at entry for users of different drugs. Their use of logistical regression, which assumes entry at the time of conception for all study participants, could result in biased estimates. In addition, what was considered to be NSAID use was not clear.
Thirdly, they did not provide information on the correlation between NSAID use and the number of drugs used. If these two variables were highly correlated, putting both variables in the logistic regression model and letting the model decide which one was the risk factor for miscarriage would represent a misguided approach in data analysis and subsequent interpretation of results.
Finally, the fundamental weakness in their study was the comparison group. Users of other drugs could easily have a higher miscarriage rate than that in the general population simply because of their underlying conditions (depression, reproductive tract infections, etc) or the use of other drugs.
The striking contrast of a strong association between NSAID and miscarriage, and a lack of association between acetaminophen and miscarriage observed in our study provides strong evidence against the argument that NSAID-miscarriage association was due to generic association with any drug use. Such striking contrast supported by underlying biological plausibility has laid a solid foundation for the NSAID-miscarriage association.
De-Kun Li, epidemiologist
De-Kun.Li@kp.org
Liyan Liu, programmer analyst, Raxana Odouli, research associate
Division of Research, Kaiser Foundation Research Institute, Kaiser Permanente, 2000 Broadway, Oakland, CA 94612, USA
Competing interests: None declared.
References
Li DK, Odouli R, Wi S, Janevic T, Golditch I, Bracken TD, et al. A population-based prospective cohort study of personal exposure to magnetic fields during pregnancy and the risk of miscarriage. Epidemiology 2002;13: 9-20.
Li DK, Liu L, Odouli R. Authors' response to Questions regarding methodological details. BMJ 2003. http://bmj.bmjjournals.com/cgi/eletters/327/7411/368#36109 (accessed 11 Dec 2003).
If Schiavetti et al intended to disprove our findings with results from their study they failed to provide enough information to allow readers to evaluate their study. On the basis of the limited information they provided, they clearly did not have adequate data to address the relation between NSAIDs and miscarriage.
Firstly, their study was based on selected populations (volunteers who took drugs during pregnancy and called the centre).
Secondly, there was no information on indications for drug use and gestational age at entry for users of different drugs. Their use of logistical regression, which assumes entry at the time of conception for all study participants, could result in biased estimates. In addition, what was considered to be NSAID use was not clear.
Thirdly, they did not provide information on the correlation between NSAID use and the number of drugs used. If these two variables were highly correlated, putting both variables in the logistic regression model and letting the model decide which one was the risk factor for miscarriage would represent a misguided approach in data analysis and subsequent interpretation of results.
Finally, the fundamental weakness in their study was the comparison group. Users of other drugs could easily have a higher miscarriage rate than that in the general population simply because of their underlying conditions (depression, reproductive tract infections, etc) or the use of other drugs.
The striking contrast of a strong association between NSAID and miscarriage, and a lack of association between acetaminophen and miscarriage observed in our study provides strong evidence against the argument that NSAID-miscarriage association was due to generic association with any drug use. Such striking contrast supported by underlying biological plausibility has laid a solid foundation for the NSAID-miscarriage association.
De-Kun Li, epidemiologist
De-Kun.Li@kp.org
Liyan Liu, programmer analyst, Raxana Odouli, research associate
Division of Research, Kaiser Foundation Research Institute, Kaiser Permanente, 2000 Broadway, Oakland, CA 94612, USA
Competing interests: None declared.
References
Li DK, Odouli R, Wi S, Janevic T, Golditch I, Bracken TD, et al. A population-based prospective cohort study of personal exposure to magnetic fields during pregnancy and the risk of miscarriage. Epidemiology 2002;13: 9-20.
Li DK, Liu L, Odouli R. Authors' response to Questions regarding methodological details. BMJ 2003. http://bmj.bmjjournals.com/cgi/eletters/327/7411/368#36109 (accessed 11 Dec 2003).