ART and major structural birth defects in the United States

Abstract

Listen

Translate article

Sir, We read with interest the study by Reefhuis et al. (2009) showing that assisted reproductive technology (ART) treatment is a risk for structural birth defects. The authors recognize that this risk may be over-stated because underlying subfertility may in itself be an important risk factor for such defects. In our meta-analysis of 19 studies (Rimm et al., 2004), where we found an overall risk of 1.29, we noted the failure of each of those 19 studies to use the most appropriate control group, namely infertile couples conceiving spontaneously. Responding to the concern we raised, Zhu et al. (2006) attempted to design an analysis with a control group made up of subfertile couples. They compared three groups of children: those born of fertile couples (Group A), those born of infertile couples who conceived spontaneously after 12 months (Group B) and those born of infertile couples who conceived after ART treatment (Group C). They found that when compared with Group A, singletons in Group B and in Group C both had a higher incidence of congenital malformations. The adjusted hazard ratios were 1.20 [95% confidence interval (CI): 1.07 –1.35] and 1.39, respectively (95% CI: 1.23– 1.57). When the singletons in Group C were compared with those in Group B, the adjusted hazard ratio was 1.17 (95% CI: 1.00– 1.36). Zhu et al. suggested that some of the determinants of infertility might share a common causal pathway with mechanisms that cause congenital malformations. We were surprised that the Reefhuis study presents no overall rate of malformations in women aged 35 and over. Apparently there is a very large difference in the age distribution of the ART group and the non-ART group as shown in Table 1 for those without major birth defects. Only 13% of the non-ART mothers were 35 and older, whereas 55% of the ART mothers were 35 or older. The authors failed to show the age of the ART and non-ART mothers with a major defect. When there is such a large difference in the age distribution between the two groups, adjustment for age is not a sensitive method of analysis. Stratifying by age is more sensitive because it takes interaction into account.