Safety of psychotropic drugs in pregnancy Reassuring findings on antidepressants and antipsychotics from the largest studies to date Hind Khalifeh NIHR research fellow, Clare Dolman service user researcher, Louise M Howard NIHR research professor Section of Women’s Mental Health, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, London, UK Mental illness in pregnancy is common—around 10% of women experience a major depressive illness or anxiety disorder, 1 and an increasing number of women with psychotic disorders are able to conceive owing to the decreased use of antipsychotics with prolactin raising properties. 2 Women who have severe mental illness in the perinatal period are at risk of considerable psychological morbidity, including suicide, 2 but they often discontinue psychotropics, 3 largely because of concerns about safety. 4 Evidence on the risks of drugs from observational data is limited and contradictory, with important methodological limitations due to bias, confounding, and small sample sizes. 5 In two linked papers in this issue, Furu and colleagues (doi:10. 1136/bmj.h1798 6 and Vigod and colleagues (doi:10.1136/bmj. h2298) 7 tackle some of these limitations by using novel methods to reduce confounding in large linked datasets, and thus provide a valuable addition to the evidence base on the safety of psychotropic drugs in pregnancy. Furu and colleagues used national data from the five Nordic countries to examine the association between maternal use of selective serotonin reuptake inhibitors (SSRIs) or venlafaxine in the first trimester and birth defects among 2.3 million live singletons, including 36 772 (1.6%) exposed infants. To account for familial or unmeasured environmental confounding, the study included a cohort of around 2000 sibling pairs who were discordant for maternal antidepressant use and for congenital malformations. In the whole cohort, and after taking into account key confounders, exposed infants had a small increased risk of any major birth defects and cardiac defects. In the sibling cohort, however, none of these outcomes was associated with exposure to SSRIs or venlafaxine. This suggests that the associations observed in the whole cohort may be due to confounding by familial or lifestyle factors, and points against a teratogenic risk from these drugs. Vigod and colleagues used linked healthcare data in a matched cohort of 1021 women who were prescribed antipsychotics during pregnancy and 1021 women who were not. The matched cohort was derived using a statistical technique called high dimensional propensity score matching, which aims to minimise unmeasured confounding. Results were also reported for an unmatched cohort of around 1200 women who used antipsychotics and 40 000 women who did not. In the unmatched cohort, women who used the drugs had an increased risk of several adverse outcomes, with a high prevalence of preterm birth (15%), gestational diabetes (8%), hypertension (5%), and large for gestational age infants (4%). In the matched cohort, however, none of these outcomes was associated with antipsychotic use. This suggests that the associations observed in the unmatched cohort were not due to antipsychotic use in pregnancy but rather to confounding. Of note were the absolute rates of these adverse pregnancy outcomes (considerably higher than in the general population of pregnant women) and the high rates of prepregnancy diabetes and hypertension in the cohort that used antipsychotics. Both studies have two key strengths; they are the largest comparative studies to date for their respective drugs, and they each contain a carefully chosen comparison group designed to deal with confounding. However, as the authors acknowledge, residual confounding, including confounding by indication, is likely. Moreover, the high rates of adverse outcomes in women using antipsychotics may be due to the presence of other risk factors, which may be misclassified or not recorded in administrative data, such as stressors associated with mental illness (for example, domestic violence), poor nutrition, high body mass index, and smoking. 8 Comprehensive assessment of physical and mental health and psychosocial risks in women with mental disorders is therefore vital. For women using antipsychotics, additional glucose monitoring including an oral glucose tolerance test is recommended by the recent update on antenatal and postnatal mental health from the UK National Institute for Health and Care Excellence. 5 When deciding on whether to use a psychotropic in pregnancy, the potential risks to mother and unborn baby need to be balanced against the potential benefit of treatment, including reductions in the risk of postnatal mental illness. There is no randomised controlled trial evidence on the efficacy of antidepressants or antipsychotics in pregnancy (and only emerging such evidence of the benefit of antidepressants on postnatal depression), 9 but observational studies suggest higher relapse rates among those who discontinue drugs, 2 10 especially Correspondence to: L M Howard louise.howard@kcl.ac.uk For personal use only: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe BMJ 2015;350:h2260 doi: 10.1136/bmj.h2260 (Published 13 May 2015) Page 1 of 2 Editorials EDITORIALS