Register-based studies of infertility and the use of assisted reproductive techniques : possibilities and challenges

Abstract

The implementation of assisted reproductive techniques (ART) has enabled over 10 million births in the past 45 years. Concern for excess risks of adverse birth outcomes has largely, but not entirely, been mitigated by practice recommendations for single embryo transfer. Evaluations of long-term health are still relatively limited but increasingly warranted as the number of children conceived with ART that have reached adolescence and adulthood is growing.

In this thesis, we used nationwide Swedish register data to assess the long-term health of children conceived with ART, with the specific aim of disentangling the potential role of the intervention from its indication, i.e., underlying infertility and associated parental background characteristics. Linkage of multiple registers from different sources enabled prospectively recorded follow-up of sufficient length, consideration of a wide range of parental background characteristics, and low risk of selection bias. Focusing on outcomes for which evidence was limited due to few or poorly performed studies, we investigated attention deficit hyperactivity disorder (ADHD) and poor school performance in Study I; depression, anxiety, antidepressant use, and suicidal behavior in Study II; and childhood asthma in Study III. In addition to examining how much of the potential observed differences in these outcomes could be explained by parental factors, we also explored the potential influence of specific procedures such as intra-cytoplasmic sperm injection (ICSI) and frozen embryo transfer.

In Study I, we found that children conceived with ART were at lower risk of ADHD and they performed better in grade 9 (higher average grade and eligibility for upper secondary school) compared with all other children. Adjustment for parental background characteristics reversed the findings such that children conceived with ART appeared at a slight disadvantage. However, when restricting the contrast to children of couples with known infertility, no differences were seen in any of the outcomes. In Study II, we applied a similar approach to evaluate mental health outcomes, extending the follow-up period into early adulthood. Adolescents conceived with ART were not at elevated risk of depression or suicidal behavior compared to other adolescents (irrespective of parental infertility). Compared with all others, adolescents conceived with ART had a modestly elevated risk of obsessive-compulsive disorder, but this was greatly attenuated by adjustment for parental factors and not seen at all in comparison to adolescents born to couples with known infertility. In Study III, the previously reported link between ART and childhood asthma was put to even further scrutiny by including also a comparison of first-born maternal cousins. Once again, an observed small excess risk appeared to be largely explained by parental background factors. However, a modestly elevated risk of asthma in the first year of life seen only in children born after frozen embryo transfer warrants further confirmation. In Study IV, we shifted focus from casual investigation of children’s health to prediction of pregnancy complications in women undergoing ART, developing pre- and post-treatment prediction models for preeclampsia, placental complications (accrete, previa, and abruption), and postpartum hemorrhage. A wide range of predictors was considered, including demographic factors, prior medical history and medication use, and ART treatment cycle information. Several different model algorithms were implemented and cross-compared, including logistic regression, random forest, and gradient boosting. Still, the final model prediction performances were just above 60% at best (ranging from 54% to 62%). Across the different complications, the most influential predictors tended to be body mass index (BMI), parental age, treatment year, region of residence, infertility diagnosis, and type of embryo transfer (in additional attempts). While some gain may be made from further refinements of the predictor categorization, the most critical improvement for continued prediction efforts should entail expansion to consider also repeated clinical measurements in pregnancy.

In conclusion, these studies provide overall reassurance with regard to childhood asthma and the long-term neurodevelopmental and mental health of individuals conceived with ART. Some excess risks seen in relation to the general population appear to be largely driven by differences in parental characteristics, stressing the importance of considering the role of the underlying indication (and its associated risk factors) in ART safety studies. Further studies are required to combine treatment-related information with clinical measurements throughout pregnancy to improve the prediction of maternal complications and tailor a clinical protocol for this specific population at risk.