LETTERS: Letters to the Editor
The authors respond:
We thank Burt et al1 and Kelly et al2 for their interest in our study.
Burt and colleagues1 argue that “positive” studies on the safety of antidepressants need special care in conduct and presentation due to potential high media attention. In line with that opinion, Kelly et al2 argue that SSRI use in pregnancy is a “controversial public health issue” and accordingly these studies should be presented with scrutiny. We disagree in that all studies, and not positive studies in particular, should be conducted with the highest possible quality. The awareness of possible media interest or controversy as benchmarks for how to conduct and present studies may, in the long run, threaten the credibility of research.
We are aware that we did not find a pattern of specific defects as mentioned by Burt et al1 (the only exception was the rate of malformations of the digestive organs, which seem to be increased among the offspring of SSRI users.) However, the finding of increased risk of malformations in children born to mothers with drug prescriptions during 2nd and 3rd pregnancy month compared with women with prescriptions during 1st trimester and one month before the conception date provides support for a causal effect, because the 2nd and 3rd pregnancy months constitute the critical period for most birth defects. This may represent a generalized serotonin-mediated effect rather than a specific effect from different SSRIs that do not share chemical structure.3
The lack of separation between major and minor malformations was questioned by Burt et al.1 We want to underscore the fact that we excluded congenital dislocations of the hip and undescended testes. Furthermore, we identified the 3 most prevalent congenital malformations in the offspring of women with a SSRI prescription in early pregnancy and did not find that the increased risk could be caused by minor malformations.
Another point of criticism from both groups was the uncertainty of compliance. Our analyses showed that many of the exposed women filled more than one prescription. Furthermore, drugs are paid for in part by the women and therefore there is likely to be a considerable degree of compliance in taking the pills that were purchased. Finally, noncompliance would likely bias the results by attenuating an effect.4
Burt et al1 and Kelly et al2 mentioned failure to control for maternal illness and other possible confounders as a limitation in our study. This is a well-established problem in nonrandomized observational studies of drug effects5—for example, the finding of an increased mortality among patients with cystic fibrosis treated with tobramycin for inhalation.6 To explore the influence of depression-related factors, we analyzed the risk of malformations among another group of women with depression—women exposed to non-SSRI antidepressants in early pregnancy. Finding no increased risk of malformations among these women suggests that the underlying disease is of less importance than the SSRI medications. Bonari et al7 reviewed the literature on perinatal risks after untreated depression during pregnancy. While there were suggestions of an increased risk of adverse pregnancy outcomes such as preterm delivery; no studies demonstrated increased risk of malformations.
Kelly et al2 noted that the rate of smoking in our study was nearly twice as high among the SSRI group as compared with the non-SSRI group. Although the evidence for smoking as a general teratogen is weak, we agree that smoking may be indicative of an unhealthy lifestyle. Our best options for examining the influence of the underlying disease in our setting were 1) to control for smoking and 2) to examine the risk among non-SSRI antidepressant users. The results of these analyses did not allow us to exclude a causal association.
Our findings should be discussed in a context of adverse effects of depression itself. However, we actually outlined the possibility of the underlying disease as a confounding factor, and already stressed in the conclusion, as well as in the abstract, that the observed association may not be causal. We believe that clinicians with the final responsibility for counseling individual patients have the full ability to combine their knowledge from research, clinical experience and understanding of the patient's situation into the best treatment. Thus in each counseling situation the absolute risk of malformations associated with and without drug use in pregnancy should be related to the benefits of drug use.
Department of Clinical Epidemiology Aarhus University Hospital Aarhus, Denmark firstname.lastname@example.org
Preben Bo Mortensen
National Centre for Register-based Research Aarhus University Aarhus, Denmark
Department of Preventive Medicine Vanderbilt University Medical Center Nashville, TN
Henrik Toft Sørensen
Department of Clinical Epidemiology Aarhus University Hospital Aarhus, Denmark
1. Burt V, Miller L, Einarson A. SSRIs and birth defects [letter]. Epidemiology
2. Kelly MB, Wisner KL, Cornelius MD. SSRIs and birth defects [letter]. Epidemiology
3. Briggs GG, Freeman RK, Yaffe SJ, eds. Drugs in Pregnancy and Lactation
. 7th ed. Philadelphia: Lippincott Williams & Wilkins; 2005.
4. Strom BL, ed. Pharmacoepidemiology
. 2nd ed. Chichester: Wiley; 1994.
5. Sørensen HT, Lash TL, Rothman KJ. Beyond randomized controlled trials: a critical comparison of trials with nonrandomized studies [review]. Hepatology
6. Rothman KJ, Wentworth CE 3rd. Mortality of cystic fibrosis patients treated with tobramycin solution for inhalation. Epidemiology
7. Bonari L, Pinto N, Ahn E, et al. Perinatal risks of untreated depression during pregnancy [review]. Can J Psychiatry