Maternal Spontaneous Abortion and the Risk of Attention-deficit/Hyperactivity Disorder in Offspring

A Population-based Cohort Study

Hui Wang; Fei Li; Maohua Miao; Yongfu Yu; Honglei Ji; Hui Liu; Rong Huang; Carsten Obel; Jun Zhang; Jiong Li


Hum Reprod. 2020;35(5):1211-1221. 

In This Article


Principal Findings

In this large population-based cohort study, we observed that maternal SA was associated with an 11% higher rate of ADHD in offspring, and the rate increased with the number of maternal SA, particularly for the firstborn child. Our findings suggested that the observed associations were independent of a number of factors, such as maternal socioeconomic status, type of SA, parental history of psychiatric disorders, pregnancy characteristics (maternal smoking status, infection, diabetes and hypothyroidism status during pregnancy) and birth outcomes (low birth weight, preterm birth, low Apgar score and SGA).

Previous studies have investigated the association of maternal history of SA with neonatal outcomes, such as preterm birth, low birth weight, stillbirth and small for gestational age in the subsequent pregnancies with short-term follow-up (Xiong et al., 2002; Weintraub et al., 2005; Bhattacharya et al., 2008; Klemetti et al., 2012; Gunnarsdottir et al., 2014; Makhlouf et al., 2014; Ahrens et al., 2016). One US study of 21 277 women reported that the risks of preterm birth, low birth weight and stillbirth in the next pregnancy were increased with the number of previous spontaneous abortions (Ahrens et al., 2016). Another Danish study of 619 587 women also showed that maternal history of two or more SAs was associated with increased risks of preterm birth, small for gestational age infants and stillbirth in subsequent pregnancy (Gunnarsdottir et al., 2014). Birth outcomes, such as preterm birth, low birth weight and low Apgar score, could be considered as potential mediators in the pathway from maternal SA and ADHD in offspring (Biederman, 2005). However, findings from our study showed that the elevated rates of ADHD in offspring were not attenuated by the exclusion of children with these birth outcomes. Mediation analyses further indicated that adverse birth outcomes could only explain a very small proportion of the overall associations. This may suggest an aetiological role of maternal SA on ADHD in offspring. SA has been proposed as a marker of genetic susceptibility to psychiatric disorders in mothers and offspring (Fergusson et al., 2008; Toffol et al., 2013); however, we did not observe an association between maternal SA after the childbirth and the ADHD risk in offspring, indicating that prenatal adverse environmental exposure could play an important role on long-term mental health including ADHD.

Our findings indicated that the association between maternal history of SA and rate of ADHD in offspring was greater in the firstborn child. A study in the USA (791 live-born offspring with 385 women) had also suggested that a history of SA was associated with a 4-fold risk of epilepsy in offspring, and the risk was particularly high in the firstborn child (Schupf and Ottman, 2001). Previous studies have showed that the risk of pregnancy complications was increased among primigravida women, compared with women with a previous live birth (Jivraj et al., 2001; Magnus et al., 2019). A live birth seems to reduce the negative outcomes of pregnancy losses prior to the childbirth (Egerup et al., 2016). Our data suggested that primigravida women with a history of SA gave birth to children with the highest risk of ADHD. We also observed a more pronounced increase of ADHD rate in children whose mothers with SA and comorbid parental psychiatric disorders, indicating an added influence of genetic susceptibility.

There may be two plausible pathways underlying our observations. First, foetal programming could be one possible pathway of maternal history of SA leading to ADHD in offspring. Women with a history of SA were more likely to experience elevated levels of stress in the next pregnancy (Fergusson et al., 2008), which could activate the hypothalamic–pituitary–adrenal axis of pregnant women (Tsigos and Chrousos, 2002). The activation of the hypothalamic–pituitary–adrenal axis may impair placental function and downregulate the activity of placental 11β-HSD-2, decreasing the protection of this 'barrier' (Seckl and Holmes, 2007). As a result, increased maternal cortisol can cross the placenta, leading to an increased cortisol level in the foetal circulation (Seckl and Holmes, 2007). Elevated levels of foetal cortisol might play a permanent detrimental role in foetal brain development, leading to an increased risk of all mental disorders including hyperactive, behavioural and emotional disorders later in life (Bale et al., 2010, Lewis et al., 2014). Another potential biological pathway could be hypoxic status in pregnancy resulting from previous SA (Renaud et al., 2011). Hypoxic conditions have adverse consequences on foetal brain development and are associated with an increased risk of ADHD in children (Getahun et al., 2013).

Strengths and Limitations of This Study

Our study has several methodological strengths. First, the comprehensive administrative database allowed us to access pregnancy information that was actually obtained in an objective manner, which reduced the chance of recall bias. Moreover, this study was based on the national register data with nearly complete follow-up, thus minimizing the possibility of misclassification bias. Second, the diagnosis of SA has been validated and confirmed in the Danish Patient Register and the distributions of women with one SA and ≥2 SAs are comparable with those from other studies (Lohse et al., 2010). Third, ADHD was identified from both hospital medical record and prescription database (Pottegård et al., 2012). The combination of hospitalization registers and prescription data allowed us to identify individuals with both mild and severe ADHD symptoms.

Several limitations need to be noted. First, even though we used population-based register data to capture maternal history of SA, misclassification should also be taken into account. SA is a common pregnancy outcome, but it is challenging to estimate the precise rate, especially for the earlier pregnancy losses when women may not even be aware that they are pregnant (Wilcox et al., 1988). But the misclassification of maternal SA is most likely to be non-differential, as the diagnosis of SA is made without any knowledge of the outcome (ADHD in offspring in this study). As shown in previous studies (Wilcox et al., 1988; Ahrens et al., 2016), this would lead to reduced statistical power and may bias the estimate towards the null. Findings from the probability sensitivity analyses indicate that only around 1% of change in the estimate was due to misclassification of SA. Second, previous studies have suggested that adverse pregnancy and neonatal outcomes were often worse following a second trimester SA (Edlow et al., 2007). However, we do not have detailed clinical information on the gestational age of pregnancy termination; thus, we were unable to distinguish between the association of SA occurring at different stages of pregnancy and the risk of ADHD. Third, even though we have adjusted for a wide variety of potential confounders, the residual confounding from unidentified confounders and unmeasured confounders, such as maternal inflammation response, is still possible, which might cause an overestimate of the true association. Nevertheless, additional adjustment for maternal infection during pregnancy did not change the results (results not shown). Using the calculated E-value, an unmeasured confounder would need to be associated with maternal SA and the ADHD risk in offspring by a magnitude of 1.46 or above and beyond the measured confounders to explain away the observed association, so it is unlikely that our association could be explained away by an unmeasured confounder (VanderWeele and Ding, 2017).