Depression in pregnancy

Treatment of depression in pregnancy, as with onset at any time, may include pharmacotherapy, electroconvulsive therapy (ECT) and psychotherapy. Lifestyle changes such as avoiding drug and alcohol use, taking up exercise and building social networks are also important, as is providing ongoing support and monitoring (Reference Dimidjian and GoodmanDimidjian 2009).

Anxiety disorders

Maternal anxiety during pregnancy has also been associated with adverse obstetric and neonatal outcomes (Box 2). Pregnancy anxiety, or anxiety regarding a current pregnancy, has been fairly consistently associated with shorter gestation and preterm birth, in rates comparable to other known risk factors such as medical and obstetric risks and smoking (Reference Dunkel Schetter and TannerDunkel Schetter 2012). Pregnancy anxiety has also been associated with higher rates of emergency Caesarean sections in one study (Reference Ryding, Wijma and WijmaRyding 1998) but not others (Reference Reck, Zimmer and DubberReck 2013), and longer duration of labour (Reference Reck, Zimmer and DubberReck 2013).

There are very few studies examining the relationship between actual anxiety disorders and perinatal outcomes. One well-designed study did not find associations between adverse perinatal outcomes and generalised anxiety disorder, panic disorder, social phobia or obsessive–compulsive disorder during pregnancy (Reference Andersson, Sundström-Poromaa and WulffAndersson 2004). Another did demonstrate an association between post-traumatic stress disorder and adverse outcomes such as spontaneous abortion and excessive fetal growth, although notably could not control for medication use in women with post-traumatic stress disorder (Reference Seng, Oakley and SampselleSeng 2001).

The pharmacological treatment of anxiety disorders includes antidepressants (e.g. SSRIs, SNRIs, TCAs) and benzodiazepines. The potential risks associated with taking these medications in pregnancy are discussed earlier.

Childhood development

There is mounting research concerning the childhood developmental risks associated with antenatal depression and anxiety and their pharmacological treatment (Box 2). It is difficult to interpret these studies as few distinguish between the timing of exposure to maternal mental illness, differentiate anxiety and depression or their combination, or consider maternal antenatal antidepressant or other medication use (and factors such as duration of use and adherence), and lifetime maternal depression. One review concluded that in relation to perinatal exposure to maternal depression or maternal anxiety and stress, antenatal exposure to maternal anxiety and stress had the most consistent and greatest negative effects on child and adolescent outcomes (Reference Brand and BrennanBrand 2009). These effects included reduced sustained attention and self-regulation in 15-year-olds and poorer cognitive/motor skills and receptive/expressive verbal language in toddlers (summarised in Reference Brand and BrennanBrand 2009). A large prospective study found an association between antenatal depression and developmental delay in 18-month-olds (Reference Deave, Heron and EvansDeave 2008).

There is one published study with contradictory findings. Reference DiPietro, Novak and CostiganDiPietro and colleagues (2006) found that maternal anxiety and depression were associated with higher motor and mental scores in 2-year-olds. Reference Brand and BrennanBrand & Brennan (2009) though, suggest that these unexpected findings may be attributable to the population studied (nonclinical, high-functioning pregnant women).

A recent review concluded that there was limited information on long-term growth, intelligence and behavioural outcomes in children exposed to SSRIs or TCAs antenatally (Reference ChaudronChaudron 2013), although a few studies have demonstrated impairment of fine motor skills in children who are exposed (Reference Gentile and GalballyGentile 2011). A comparison of fluoxetine, TCAs and controls did not demonstrate any difference in IQ, language development or behavioural development in toddlers and children aged 5–6 years (Reference Nulman, Rovert and StewartNulman 1997). However, depression duration was negatively associated with IQ and the number of maternal depressive episodes after delivery was negatively associated with language in a comparison of fluoxetine, TCAs and controls (Reference Nulman, Rovet and StewartNulman 2002).

Child behaviour and psychopathology

Teratogenic risks

Birth defects occur naturally in a significant number of neonates without exposure to medication in utero (Reference Ramos, Oraichi and ReyRamos 2007). Therefore, in discussions with women about the potential risks of using psychotropics in pregnancy, the background prevalence of major malformations (1.5–3%; International Clearinghouse for Birth Defects Monitoring Systems 2003) should also be presented. Doctors and pregnant women in particular are inclined to overestimate the teratogenic risk of using medication in pregnancy (Reference Sanz, Gómez-López and Martínez-QuintasSanz 2001). For example, the teratogenic risk of benzodiazepines was perceived to be 9.8% by physicians and 64.6% by pregnant women, although the actual risk was <5% (Reference Sanz, Gómez-López and Martínez-QuintasSanz 2001). This is of concern as pregnant women may be denied the benefits of relatively safe medication or be reluctant to take a recommended medication due to an excessively high perception of risk (Reference Sanz, Gómez-López and Martínez-QuintasSanz 2001).

As a result of such misperceptions, many women decide to suddenly stop their antidepressants when they discover they are pregnant, often with their physician’s recommendation or support. This is particularly relevant given significantly higher rates of relapse of depression on discontinuation of antidepressants in pregnancy (Reference Cohen, Altshuler and HarlowCohen 2006).

Suboptimal medical regimens

The exaggerated perception of risk may lead physicians and patients to discount the potential value of pharmacotherapy, by discontinuing important medications at times of significant risk of relapse or by using suboptimal medication regimens. In one study, two-thirds of pregnant women felt that it was unacceptable to take antidepressants in pregnancy, but only a third thought it would be unacceptable if they were not pregnant or were breastfeeding (Reference GoodmanGoodman 2009). It has been observed that antidepressant use tends to progressively decrease from the first (3.7%) to third (1.1%) trimester (Reference Ververs, Kaasenbrood and VisserVervers 2006). This trend has been attributed to patient and clinician concern about the link between exposure to antidepressants late in pregnancy and poor neonatal adaptation (Reference ChaudronChaudron 2013).

An additional concern is the undertreatment of depression in pregnancy, with reductions in the dosage of antidepressants to subtherapeutic levels in 8% of women in one study (Reference Ramos, Oraichi and ReyRamos 2007). This is despite the fact that the doses of some antidepressants must be increased later in pregnancy to maintain euthymia due to changes in maternal physiology. The reason for the use of lower dosages in pregnancy is unclear and no studies have yet demonstrated a dose-dependent relationship between exposure to antidepressants and outcomes for mother or baby (Reference ChaudronChaudron 2013).

Relational ethics

Relational ethics has been proposed as a means of guiding decision-making when the above situation arises (Reference MillerMiller 2009). The emphasis of relational ethics is that the well-being of the mother and baby are intertwined – ‘the maternal/fetal unit’ – rather than separate (Reference SeemanSeeman 2004). Reference SeemanSeeman (2004) argues that under this ethical framework, safety concerns for the mother/fetal unit should always take precedence over maternal autonomy alone.

Reference MillerMiller (2009) discusses a case vignette of a pregnant woman with bipolar disorder who acts on delusional beliefs and attempts to kill her fetus. Miller points out that the mother’s own mental health is likely to deteriorate if she harms her fetus, and therefore that treating the mother’s illness despite her refusal of treatment may be justified as much for the mother’s well-being as for that of the fetus. Of course, there is also the issue of whether the mother is actually able to make an autonomous decision given the influence of her psychotic symptoms (Reference MillerMiller 2009).

Major depression

Major depression may also limit decision-making capacity, as although the individual may understand the facts presented to them about admission to hospital or treatment, their decision-making may be influenced by negative thinking, hopelessness or negative self-appraisals such that they are unable to weigh those facts in the balance to come to a decision (Reference Coverdale, McCullough and ChervenakCoverdale 1997; Reference Callaghan and RyanCallaghan 2012).

Reference MillerMiller (2009) provides another case vignette of a woman with a history of major depression, who discontinues medication while trying to conceive and then relapses while pregnant. The woman declines to recommence medication for fear of harming the fetus, but raises concerns of being an inadequate mother and wanting to give the baby up for adoption, then kill herself. In this case the mother can articulate the risks to her baby of exposure to antidepressants as well as the risk to herself of remaining off medication, but she discounts the latter as irrelevant. The ethical conflict of patient autonomy over beneficence is again apparent. Reference MillerMiller (2009) suggests that even though reality-testing and cognition are intact, the symptoms of major depression may lead a mother to exaggerate the significance of risks to her fetus and devalue herself.

Substitute and supported decision-making

When depression significantly compromises decision-making, a substitute decision maker may be needed and a beneficence-based approach may be more relevant (Reference MillerMiller 2009). Another approach to this ethical dilemma is for the clinician to enhance the woman’s autonomy by assisting her to identify her values for the pregnancy (Reference Coverdale, McCullough and ChervenakCoverdale 1997). The clinician asks the patient what is important to her (now and prior to illness) about the pregnancy, the fetus and her own health. The patient could then be educated about her decision-making and how the depressive symptoms may influence this process. In this approach of ‘non-directive counselling’ the various alternatives are discussed, but no particular course of action is recommended by the clinician (Reference Coverdale, McCullough and ChervenakCoverdale 1997). Involvement of the woman’s spouse or family in the evaluation of her decision-making ability may also enhance her autonomy, particularly in assessing the influence of depressive symptoms on prior values and wishes.

Risk perception is influenced by how information is presented

Objective information (risk estimates) and subjective interpretation (the importance to the individual of various possible negative outcomes) are combined when making judgements about risk (Reference Bogardus, Holmboe and JekellBogardus 1999). The way in which the information is presented can alter risk perception, which in turn influences decision-making (Reference Wisner, Zarin and HolmboeWisner 2000). Thus, clinicians must be mindful to provide accurate information to patients (including the nature, degree and probability of risks of treatment and depression itself) that they can understand, despite the patient’s fear of potential adverse outcomes.

Clinicians may inadvertently collude with an anxious mother’s excessive fear about harming her fetus or depressive symptoms such as hopelessness or feeling overwhelmed, and opt not to recommend treatment in the perinatal period. It has been noted that societal ‘moral standards of sacrifice’ expected of mothers are already high and that pregnancy amplifies these expectations (Reference Lyerly, Mitchell and ArmstrongLyerly 2009). This may contribute to a tendency to overemphasise and weigh any risk to the fetus, no matter how remote or unclear, over considerations for the mother’s well-being (Reference Lyerly, Mitchell and ArmstrongLyerly 2009).

Omission bias

A related consideration is whether clinicians, in obtaining informed consent for treatment of a mental illness, place equal emphasis on the risks of treatment and of untreated mental illness. That is, was the clinician equally specific, clear and detailed in their discussion of the risk of untreated maternal depression on the fetus as they were in describing the potential risks of antidepressants (Reference MillerMiller 2009)? This is known as omission bias.

In general, studies of medical decision-making reveal that physicians worry more about risks of commission (actively doing something such as a procedure or prescribing medication) than of omission (not doing something) (Reference Kordes-de VaalKordes-de Vaal 1996). The underlying reason for this appears to be that physicians perceive themselves as more responsible for causing harm if the harm results from something they do, compared with something they do not do (Reference Kordes-de VaalKordes-de Vaal 1996). As the causal link between an omission and its consequences is perceived as less strong, an omission may be viewed as a less intended (or non-) decision than a commission (Reference Kordes-de VaalKordes-de Vaal 1996). Nonetheless, patients are not able to make accurate informed decisions when information is presented in a biased way. To prevent omission bias, clinicians should present the risks of withholding treatment as well as the risks of treatment (Reference MillerMiller 2009).