To the Editor:
In a recent publication, using the Niday Perinatal Database from 82 primary and tertiary centers in Ontario for 2005, El-Chaar et al found that assisted human reproduction (AHR) procedures increase the risk of birth defects in a retrospective cohort design (1). Their results are not only significant but also important and widely relevant in clinical practice all over the world. However, several issues in the study need closer scrutiny.
First, the outcome of birth defects remains loosely defined and allows a potential for misclassification bias, especially with multiple sources and sites of data collection.
Second, the baseline comparison of the groups is presented in an abbreviated table, and it would have been interesting to see the distribution of other variables in the different groups even though the authors mention that these were similar.
Third, we feel that socio-economic status would be a significant confounding factor since it would influence the decision of mothers opting for assisted reproduction procedures, and hence lead to a biased comparison.
Further, although loss to follow up is difficult to account for in any investigation, yet 609 records missing from a total of 1399 cases amounts to nearly 44% attrition. Were these pregnancies terminated preterm or did they end in spontaneous abortion?
Similarly, it would be interesting to know what happened to 16,708 (28%) pregnancies that were conceived naturally and spontaneously in the comparison group but were not accounted for in the final analysis.
Moreover, we feel that comparison of all pregnancies with AHR (including multiple births) with those of only singletons in pregnancies of natural conception makes the groups prone to show differences with respect to birth defects. Besides, several potential confounders like hypertension and other co-morbid illnesses have not been accounted for in the analysis.
Finally, it is unclear how babies with multiple birth defects were counted while describing the risk of different types of anomalies in pregnancies conceived by AHR.
Departments of aMedicine, bBiostatistics, c Pediatrics, and dGynecology
New Delhi, INDIA
1. El-Chaar D, Yang Q, Gao J, Bottomley J, Leader A, Wu Wen S, et al. Risk of birth defects increased in pregnancies conceived by assisted human reproduction. Fertility and Sterility. 2008 Oct 28.
Published online in Fertility and Sterility doi:10.1016/j.fertnstert.2009.02.089
The Authors Respond:
We thank Dr Goel and his colleagues for their interest in our article (1) and their comments. We appreciate the opportunity to reply to their concerns about our paper.
For the authors’ concerns about the definition of birth defects, the potential for misclassification is a limitation of studies using administrative data. However, since clerks who coded birth defects were unaware of this study, misclassification may have occurred randomly, which may actually attenuate the observed effect of IVF. IVF parents usually have higher socioeconomic status than parents with natural pregnancy; this could result in ascertainment bias because they may have better access to quality maternity care including prenatal diagnosis. On the other hand, higher socioeconomic status is a protective factor for most adverse pregnancy outcomes including birth defects. These opposite effects of higher socioeconomic status in IVF pregnancies may have cancelled each other.
The missing records that were not used in the analysis are not related to loss of follow up but due to incomplete record entry. This is likely to have occurred randomly as it was caused by the fact that some hospitals did not code non-mandatory fields into the database.
The study population for the paper consisted of all Ontario women with information about reproductive assistance recorded in the 2005 Niday Perinatal database. In this study, we compared the births by AHR (including multiple births) to pregnancies of natural conception (including multiple births). In future studies, we are interested in comparing the rates in singletons versus multiples births; however we would require a bigger study group in order to have a larger group in the multiple births group. As for obstetrical complications such as hypertension and other illness, we did account for those, by actually adjusting to exclude any case that had any of these factors reported; and the results were similar to those presented in the paper. The adjusted odds ratio for birth defects with AHR was 1.58 compared to an odds ratio of 1.55 without the adjustment for obstetrical complication.
We agree that it would help to display details of the cases of birth defects, and to sort out babies with multiple birth defects separately. This could be done by running the different variables through the database. However, the number is too small to make a difference, as those would be rare cases.
Many of the points that were brought forward by Goel et al. are limitations of administrative data; we hope that future studies by our group in this area would help to address some limitations of our study.
Darine El-Chaar, M.D.a
Qiuying Yang, M.D.a,b
Jim Bottomley, M.H.A.c
Arthur Leader, M.D.a,d
Shi Wu Wen, M.D.a,b
Mark Walker, M.D.a,b,c
aDepartment of Obstetrics and Gynecology and Neonatal Care
bOMNI Research Group
cOntario Perinatal Surveillance System
dThe Ottawa Fertility Centre
Ottawa Health Research Institute
The Ottawa Hospital
University of Ottawa
1. El-Chaar D, Yang Q, Gao J, Bottomley J, Leader A, Wen SW, et al. Risk of birth defects increased in pregnancies conceived by assisted human reproduction. Fertility and Sterility. 2008 Oct 28.
Published online in Fertility and Sterility doi:10.1016/j.fertnstert.2009.02.088