Lifestyle factors
Lifestyle factors such as exercise and nutrition exert a clear influence on maternal and fetal health across pregnancy and have been central to the investigation of fetal programming - mostly looking at cardiovascular and metabolic outcomes. Barker
et al. noted effects on fetal development of maternal diet based on relatively extreme circumstances such as famine [
41]. Maternal diet is critical to offspring growth rates and also has a programming effect on metabolic pathways. These mechanisms are thought to impose lifelong risks for the development of both diabetes and obesity [
42].
However, recent work suggests that maternal diet may also exert an influence on the biological systems that underpin future vulnerability to mental disorders [
43,
44]. Epidemiological evidence suggests that maternal and infant diet influences the risk for both emotional and behavioral disorders in childhood [
45]. Jacka
et al., for example, reported data from the Mother and Baby study of Norwegian mothers, showing that a higher intake of unhealthy foods during pregnancy predicted behavior problems among children after controlling for a range of confounders. Both maternal nutrition during pregnancy and lactation could be an influence [
46].
There is also an emerging body of evidence that maternal obesity during pregnancy is associated with the offspring’s subsequent mental health outcomes. Rodrigues
et al. found that maternal pre-pregnancy obesity was associated with child inattention symptoms and emotional difficulties [
47]. Van Lieshout
et al. conducted a systematic review of studies on maternal obesity up until 2011 and found that 8 out of 12 studies showed associations between maternal pregnancy obesity and offspring cognitive problems, attention deficit hyperactivity symptoms, eating disorders in adolescence and psychotic disorders in adulthood [
48]. Rodrigues suggested that while maternal adiposity at the time of conception may have a programming effect for child mental health, the possible mechanisms remain obscure [
47].
Mechanisms for the impact of lifestyle factors can be considered in terms of maternal mechanisms; placental mechanisms, where vascular and metabolic pathways converge in terms of placental function and would expect to be reflected in intrauterine growth retardation as a common pathway; and fetal mechanisms, such as epigenetic changes or differential fetal brain development in response to blood-borne factors that cross the placenta. Fetal pathways would also include fetal counter-regulatory responses to exposures, such as altered blood glucose or lipid ratios, and activation of hormonal signaling molecules such as leptin. While elucidation of the mechanism involved requires further consideration, there are plausible biological pathways involved and this emerging research strongly suggests that a range of lifestyle factors operating across pregnancy appear to influence the child’s subsequent mental health. The relative effects of nutrition, physical activity, obesity and other lifestyle factors are complex and may well interact. However, the evidence appears to be growing that maternal pregnancy and pre-pregnancy lifestyle factors do influence fetal development, and as such would become a modifiable target for prevention intervention.
Maternal depression and stress during pregnancy
It is well established that children are adversely affected across multiple domains when their mothers’ perinatal mental health is untreated or ineffectively treated [
49,
50]. Several lines of evidence suggest that perinatal exposure to maternal depression is associated with dysregulation of the child’s HPA response to stress, increasing risk for future stress-related disorders. A wide range of negative child outcomes following maternal depression in the postnatal period have been well documented and these include increased waking cortisol levels during adolescence [
51], larger amygdala volume and higher cortisol level at 10 years, higher levels of childhood emotional problems [
52], and higher rates of childhood and adolescent depressive symptoms [
53]. Infants of depressed mothers show more negative affect and lower sensitivity [
54,
55] and offspring may experience inadequate physical and verbal stimulation [
56].
The global definition of perinatal depression includes both prenatal and postnatal maternal depression and therefore does not allow a clear differential of the effects derived from the intrauterine
versus postnatal effects. Around 50% of women with postnatal depressive symptoms have also experienced depression during their pregnancy [
57,
58]. It is well established that postnatal depression reduces the sensitivity of the mother when interacting with her child and this results in poorer stress regulation and insecure attachments. A meta-analysis of seven studies found that the infants of depressed mothers also showed significantly reduced likelihood of secure attachment and raised likelihood of avoidant and disorganized attachment [
59]. Essentially, the putative mechanism here is the negative effects of
postnatal maternal caregiving in the context of maternal depression.
However, a fetal programming pathway for the transmission of maternal prenatal depression to offspring outcome is also a likely contributor that has been relatively neglected by developmental researchers. Animal studies have clearly shown that stress experienced by the mother
during pregnancy is associated with long-term neurobiological and behavioral effects on her offspring [
60]. Studies of prenatal maternal distress in humans show adverse child outcomes, which include symptoms of ADHD [
61], lowered cognitive performance and delayed language development [
62].
Maternal prenatal stress has an impact on her child’s physiological responsiveness to stress. Specifically, recent studies have found that maternal life stressors during pregnancy predict infant cortisol levels and reactive temperament [
63‐
65], and higher resting cortisol throughout the day in adolescence [
66]. It would appear that children born to stressed mothers have higher levels of cortisol, which follows from the disruptions to fetal stress biology previously described. Studies of the relationship between prenatal stress and child mental health have been reviewed recently by Glover [
67], van den Bergh
et al. [
68] and Räikkönen
et al. [
69], so here we refer to only a selection of larger studies.
Maternal prenatal stress in various forms is associated with a number of mental health disorders, but most previous research has been based on disaster records or retrospective assessment of prenatal stress. Khashan
et al. [
70] used two Danish national registries and found that maternal prenatal exposure to a family bereavement during the first trimester was related to a 67% increased risk of schizophrenia in offspring after adjustment for demographic confounders. Spauwen
et al. [
71] reported a small increase in risk of psychosis in adolescents whose mothers reported high levels of stress during pregnancy. Kinney
et al. used data from the national weather service and found that the prevalence of autism spectrum disorder increased markedly with the severity of a storm or hurricane if it was experienced during late pregnancy [
72]. Watson
et al. [
73] found that maternal prenatal exposure to a severe earthquake in China was associated with an increased risk of depressive symptoms in offspring, and this risk was more than double for male offspring exposed in the second trimester as compared to female offspring.
A number of large cohort studies have examined maternal anxiety and depression in pregnancy to prospectively predict child mental health outcomes. Loomans
et al. [
74] examined prenatal state anxiety and child outcomes at five years of age in a sample of over 3,000 mothers from the Amsterdam Born Children and their Development study. Maternal state anxiety measured at 16 weeks’ gestation was significantly associated with an increased likelihood of inattention or hyperactivity problems for boys (odds ratio = 2.39) but was not significant for girls. Using the Avon Longitudinal Study of Parents and Children, O’Connor
et al. [
65,
75] examined over 7,000 mother-child pairs and found that prenatal maternal anxiety measured at 32 weeks was a significant predictor of inattention or hyperactivity symptoms in boys at 48 and 81 months. However, maternal anxiety measured at 18 weeks’ gestation was not a significant predictor of inattention or hyperactivity scores in boys or girls. Drawing on the Mater University of Queensland Study of Pregnancy, Clavarino
et al. [
76] examined a sample of close to 4,000 mother-child pairs and reported that high prenatal maternal anxiety was associated with an increased risk for attention problems at 5 years that remitted by 14 years (odds ratio = 1.45) and with persistent anxiety problems from 5 to 14 years (odds ratio = 3.02). Robinson
et al. [
77] studied a sample of 1,700 drawn from the West Australian Raine Study. Women were asked at 18 and 34 weeks’ gestation if they had experienced major life stressors and then completed the Child Behavior Checklist when their children were two and five years of age. This study found that a higher number of stressful events was associated with a 23% increased likelihood of behavioral problems at age two and five and a 15% increase in the likelihood of emotional problems at age five.
The bulk of the evidence, and current practice in perinatal mental health, is concerned with addressing maternal antenatal depression and anxiety so as to improve the chances of more effective parenting postnatally. However, the findings emerging from fetal programming research suggest that child stress biology is probably being established during the intrauterine period and that preventions should be focused on preconception and pregnancy mental health and stress exposure of mothers.
Prenatal teratogenicity and neurodevelopmental toxicity
Prenatal teratogenicity originally referred to the risk of alteration in fetal development resulting in structural changes and malformations in offspring from the use of specific agents in the first trimester of pregnancy. This concept has been expanded to refer to a broader range of exposures across pregnancy, and to outcomes beyond malformations that include longer term child developmental and behavioral outcomes [
78]. For an agent to be considered a teratogen there must be a specific mechanism by which that agent alters fetal development, and these effects need to occur with particular timing of exposure during pregnancy and show a dose effect in relation to the outcome of interest [
79]. Recent work in environmental chemical exposures highlights the need to consider a broader category of neurodevelopmental toxicants. In particular, low-dose exposure to a number of chemicals with endocrine-disrupting properties is related to adverse neurodevelopmental outcomes in a non-dose-dependent manner [
80]. This departure from classical pharmacological models is predicted from the interplay between a complex and tightly controlled endogenous biological system and an exogenous chemical with biological effects outside of normal physiological boundaries. The elucidation of biological pathways does however remain a fundamental step in establishing a convincing case for causality in observed statistical associations between exposure and outcome.
Prenatal teratogenic exposures have been widely investigated, and well-documented associations with increased risk of emotional, behavioral and cognitive problems include environmental neurotoxicants like lead; substances of abuse such as alcohol, cigarettes and cocaine; and prescribed medications, such as the antiepileptic drug sodium valproate. The evidence for effects from other psychotropic medications is less well established [
21,
81‐
84]. Cigarette smoking has been found in up to 11.8% of pregnant women [
85], with 30.3% of women having some alcohol in pregnancy but only 2.7% having alcohol across all trimesters [
86]. Illicit substance use is likely to be lower, as is exposure to antiepileptic drugs. However, exposure to antidepressant medications was found in as many as 13.4% of pregnancies in one study from Tennessee in the US [
87]. Environmental chemical exposures, by contrast, can be near ubiquitous, underlining the population significance of even subtle neurodevelopmental toxicity. Consider for example that between 1976 and 1980, 77.8% of the US population had blood lead levels that were more than double the current threshold for reporting [
88].
There are established associations between exposure to maternal smoking in pregnancy and a range of pregnancy and child health outcomes, from growth restriction and preterm delivery to childhood respiratory illness. Increased incidence of childhood mental illnesses and symptomatology, specifically ADHD and conduct problems, have also been consistently observed [
89‐
93]. Even environmental tobacco exposure (passive smoking) is associated with adverse behavioral outcomes [
94‐
96], although a causal biological pathway has not been established. An association with childhood anxiety and depressive symptoms has been observed [
97], but not consistently [
93]. A number of recent studies have raised doubts about whether the observed associations between maternal smoking during pregnancy and childhood mental health are fully causal in nature or reflect, in part, shared genetic susceptibility. In the Avon Longitudinal Study, the effects size for paternal smoking was of similar magnitude to that of mothers [
98], and in a study of children born following assisted reproduction, the association between maternal smoking in pregnancy and childhood ADHD symptoms was greater in those where the child was genetically related to the mother [
99]. Shared inheritance does not, however, appear to account for all of the observed association between ADHD and prenatal tobacco exposure in other cohorts [
91,
93,
100]; a recent study from Taiwan explicitly tested and found evidence for a biological pathway that is dependent upon tobacco-related chemicals. Hsieh
et al. used genetic studies of children to demonstrate that the association between cord blood cotinine and childhood behavioral difficulties is modified by a genetic polymorphism in the metabolic pathway for smoking-related toxicants [
94].
Alcohol has also been associated with a range of teratogenic effects, from fetal alcohol syndrome to a broader fetal alcohol spectrum and later onset developmental and behavioral problems, such as low IQ, specific learning disorders, and internalizing and externalizing symptoms [
101,
102]. Of the other drugs of abuse, cocaine [
103,
104], marijuana, benzodiazepines and methamphetamine [
105] have each been associated with effects on neurodevelopment and later child mental health outcomes that appear to be independent of social factors [
106,
107]. These effects are subtler than earlier research in the area, and are not apparent until much later in child development and therefore referred as latent or ‘sleeper’ teratogenic effects [
78]. The evidence for opiate exposure is unclear [
103]. Epidemiological research in the area is complicated by covariance between substance use and social factors and, in the case of opiates, the small numbers of pregnancies affected and challenges of follow-up in this relatively chaotic social group.
More recently, longitudinal studies have started to focus on psychotropic medications including antidepressants, antipsychotics and mood stabilizers. Studies from North America have shown an increasing rate of antidepressant exposure in pregnancy, ranging from 7.6% to 13.4% in the US and 5% in Canada [
87,
108,
109]. In Australia, the rate has been shown in data from the Longitudinal Study of Australian Children to be around 2.1% [
110]. The difficulty in examining these agents for potential teratogenic effects is untangling the potential impact of the often serious maternal mental illnesses these agents are used to treat as well as co-morbid exposures that confound outcomes. For instance, one study examining malformation risk and antidepressant exposure found fetal alcohol syndrome was 10 times more likely in children exposed to antidepressants in pregnancy [
111].
The most rigorously studied psychotropic class is the antiepileptic drugs, which, in addition to being used in epilepsy, are used as mood stabilizers for treatment of bipolar disorder. These agents have previously been associated with an increased risk of specific structural teratogenicity, such as an increased risk of neural tube defects. There are now a number of rigorous, prospective, longitudinal studies that have followed children from pregnancy to school age to examine for neurodevelopmental and behavioral teratogenic effects [
112]. They have identified specific risks with exposure to specific agents and a dose effect. The antipsychotic medications, both typical and atypical, are not associated with a malformation risk but the literature for longer term effects is far too limited to draw any conclusions about child development outcomes [
113,
114].
There are now a number of large studies of antidepressant exposure in pregnancy and malformation risk but there is still no consensus as to whether there is a small increased risk of birth defects [
115]. Studies of longer term teratogenic risks are more limited and most have a small number of participants and short follow-up [
116]. However, while no studies to date have found an effect of exposure on global cognition, there are four studies that have found an increased risk of poorer motor development [
117‐
120]. It is important to balance these findings with a number of studies that have found that untreated depression is associated with poorer development, particularly language development [
121]. Given the increasing rate of exposure to this class of psychotropic medication, further studies are required that can robustly quantify the potential risks of exposure to balance against the harms of withholding treatments. Such studies ideally require longer follow-up, robust consideration of maternal depression and other confounding factors, and more robust child development measures in order to reach clear conclusions [
87,
108,
122,
123].
It can be concluded from studies on psychotropic medication that, to minimize effects on longer term child development, single agents should be considered when treating maternal mental illness in pregnancy, keeping doses as low as is feasible for effective treatment. There is an urgent need for further studies to delineate risks for specific agents so that more informed choices can be made [
124‐
126]. When considering the use, and hence exposure, to antidepressants in pregnancy, the issues relevant to child development and mental health outcomes are not just those that relate to exposure
per se but also to the impact of untreated maternal mental illness. There is mortality data in both the UK and Australia that suggest mental illness is a leading indirect cause of maternal deaths [
127]. In addition, untreated depression potentially impacts on the capacity of women to self-care in pregnancy, particularly in important areas that are increasingly associated with optimal fetal growth and development, such as nutrition and exercise. Finally, clinicians and patients need to consider the effects that remaining depressed may have on a woman’s capacity to enjoy motherhood, bond with her baby, and provide responsive and sensitive parenting to the child. All of these aspects of parenting also have a significant impact on short- and long-term child outcomes.
Environmental teratogens and neurodevelopmental toxicants differ from the above by the locus of control of the mother with respect to exposure, but are nevertheless an important modifiable risk factor for preventative mental health strategies. Lead exposure is the prototypic environmental neurotoxicant. Historical exposure was through use of lead in water pipes carrying drinking water, as a fuel additive, and in paints and certain toys. Lead persists in the environment and current household exposure is understood to be primarily from historical soil contamination and old paint, although lead does also continue to be used in a restricted form in hobby activities including soldering, pottery, collectables such as toy soldiers, certain artists’ paints, ammunition and fishing sinkers. Large longitudinal studies demonstrate that prenatal [
128‐
130], lifetime [
129,
131] and current exposure [
132‐
134] are each important [
135‐
137] for neurodevelopmental outcomes. The proposed biological mechanisms have supportive
in vitro evidence, including inhibition of N-methyl-D-aspartic acid glutamate receptors (the key molecule regulating synaptic long-term potentiation) and interaction with calcium ion signaling, with much wider implications. Although much research has focused upon outcomes of general cognitive ability and/or ADHD, there is well-replicated evidence for an association between lead exposure and adverse outcomes across broad ranging neurodevelopmental outcomes, including both behavioral and emotional symptomatology [
138‐
141]. The effect size is substantial. Froehlich
et al. estimate that in the US, 25.4% of ADHD in 8- to 15-year-olds is attributable to the low levels of exposure that persist today [
142]. Their analysis used data from the US National Health and Nutrition Examination Survey (NHANES) study, a cross-sectional study but whose design facilitates it having the statistical power to look at clinical outcomes rather than symptomatology.
The evidence base for lead exposure and neurodevelopmental outcomes is not matched anywhere else in the environmental chemical literature. Other environmental chemicals with established neurodevelopmental toxicity include methyl mercury and the polychlorinated biphenyls (PCBs). Environmental exposure to each has established associations with general cognitive function, although the relevance to broader mental health is less clear [
143]. Interestingly, the longitudinal data that does exist support an association with prenatal but not postnatal exposure, in support of the DOHaD hypothesis [
143]. The strongest evidence is for associations with ADHD and executive function deficits [
143]. Wider mental health outcomes have not been adequately studied to draw conclusions, although data would exist in the New Bedford cohort [
144,
145] that we have not been able to find in publication. Biological mechanisms are also unclear, although PCBs are potent endocrine disruptors and modeling supports thyroid function as a putative mechanism [
146].
No consistent specific pattern of deficits emerges that distinguishes neurodevelopmental toxicity due to lead from that due to mercury or PCBs. Observed outcomes in each span both cognitive and affective domains [
143], suggesting common developmental pathways of neurodevelopmental toxicity with relevance to mental health. Research to date does imply that cognitive performance and behavioral problems are more strongly associated with environmental toxicity than emotional problems, although it is not clear if this is due to greater sensitivity to neurodevelopmental toxicity, or greater sensitivity in the population-based methodology and assessment tools used to detect subtle effects at the population level.
Exposure to lead and PCBs are on the decline, and methyl mercury exposure appears to be stable [
147‐
150]. Yet there is substantial work still to be done to further reduce exposure to these established neurotoxicants, most notably lead, where the ongoing effects of exposure estimated from the US NHANES data mean that this must remain a key priority in the preventive mental health agenda [
142]. At the same time there is a growing body of modern chemicals - not previously evaluated for neurodevelopmental toxicity - that now raise concern as potential neurotoxins in need of further evaluation [
143,
151]. These include manganese and cadmium - bivalent heavy metal cations like lead and mercury - and many chemicals with endocrine-disrupting actions
in vitro similar to those of PCBs (for example, bisphenol A, phthalates, organochloride pesticides, organophosphate pesticides, brominated flame retardants and perfluorinated compounds). The robust evaluation of these chemicals is a substantial new opportunity in preventative mental health. However, beyond even this, there is an appreciation that the potential for neurodevelopmental toxicity is unknown for the vast majority of chemicals in ubiquitous modern usage [
151], emphasizing that there may be other substantial opportunities for mental health prevention, and that there is a need to prioritize understanding in this area.
Addressing environmental toxins, smoking, alcohol and illicit substance use in pregnancy are important to reduce the implications for child development and mental health outcomes. The latter three also have implications for pregnancy self-care and nutrition. However, there is evidence that both smoking [
152] and alcohol use [
111] in pregnancy are associated with depression in pregnancy. Therefore, interventions that address a broader approach to a healthy pregnancy may be warranted.