Psychotropic medication use during pregnancy is a challenging issue.
Several Mental Elf blogs have covered this topic already. For example, Joanne Wallace blogged about a large cohort study that found that antipsychotic medication use was not associated with an increased risk of major congenital malformation. More recently, Flo Martin blogged about the findings from an umbrella review that revealed the dearth of high-quality evidence in this field. There are two things worth highlighting from these blogs. First, as Martin states in her blog, the evidence base for this topic relies almost exclusively on the findings from observational studies, because pregnant women are almost always excluded from randomised controlled trials. Second, as Wallace states in her blog: “For each individual women the decision to continue or discontinue psychotropic treatment encompasses a range of factors and for some this could be the hardest decision of their lives.”
To support expectant mothers in assessing the risk-benefit balance, we need to know the potential risks and benefits of antipsychotic medication use during pregnancy. Using the National Health Information Database of South Korea, Kang et al. (2025) have conducted an impressive array of analyses to progress our understanding of this important clinical question.
“For each individual women the decision to continue or discontinue psychotropic treatment encompasses a range of factors and for some this could be the hardest decision of their lives.”
Methods
The National Health Information Database in South Korea contains information from the single provider of mandatory health insurance for the entire national population. The study identified women who delivered live births at least two years after the first diagnosis of schizophrenia and were younger than 50 years of age (n = 3,026).
To control for potential confounders, particularly the severity of schizophrenia, the self-controlled case series method was used. In this simple yet elegant method, each individual acts as their own control. In this study, the incidence rate ratio (IRR) was calculated by comparing the rate of admission for psychosis six months postpartum with the rate during the period between two years before delivery and one year before delivery (the reference period).
The women were categorised into three groups:
- Non-users: those who did not use antipsychotics in the period from one year before delivery to 39 weeks before delivery (i.e. abstained from antipsychotics before becoming pregnant),
- Discontinuers: those who used antipsychotics in the period from one year to 39 weeks before delivery, but discontinued before delivery, or,
- Continuers: those who used antipsychotics in the period from one year to 39 weeks before delivery and continued taking their medication before delivery.
The authors then calculated the relative risk ratios (RRRs) for the IRRs for each group with the discounters as the reference group.
Results
Of 3,026 women with schizophrenia, most were either non-users (43.1%) or discontinuers (41.6%). Only 15.3% of women continued antipsychotic medication during pregnancy.
Overall, the IRR of admission for psychosis in the six-month postpartum period was 2.35 (95% confidence interval [CI] 1.99 to 2.78) compared to the reference period, meaning there was a significantly higher risk of admission postpartum across the whole sample. The risk was highest immediately after delivery (one to ten days after delivery, with IRR = 5.44), and it decreased over time until reaching a rate similar to the reference period at one to two years after delivery.
In terms of different medication categories, the incident rates during the reference period were highest among the continuers (144.7 per 1,000 person-years) compared to the discontinuers (118.3 per 1,000 person-years) and the non-users (29.9 per 1,000 person-years). However, the incident rates increased significantly in the six-month postpartum period for the non-users (128.5 per 1,000 person-years) and the discontinuers (277.9 per 1,000 person-years), but not for the continuers (191.3 per 1,000 person-years). The IRR of relapse was thus the highest for the non-user group (4.18, 95% CI 2.85 to 6.12), followed by the discontinuer group (IRR 2.34, 95% CI 1.87 to 2.91). The IRR was not significant for the continuer group (IRR 1.31, 95% 0.89 to 1.92). In other words, the risk of relapse was not significantly different for the continuer group either before or after delivery. Using the continuer group as the reference group, the IRR was significantly higher for the non-user group (RRR = 1.79, 95% CI 1.15 to 2.78) and significantly lower for the continuer group (RRR = 0.56, 95% CI 0.36 to 0.87).
The authors conducted seven subgroup analyses stratified by; (i) the age at delivery (younger than 34 or 34 and older), (ii) insurance premiums, (iii) mode of delivery (vaginal or Caesarean section), (iv) number of prenatal care visits (fewer than 13 or 13 and more), (v) diagnosis of schizoaffective disorder, (vi) comorbid depressive disorders and, (vii) comorbid mood disorders. Further, the study also conducted six sensitivity analyses, which are a method of testing how changes in assumptions or inputs affect the results of a statistical model. Here the tests were; (i) excluding admissions with antidepressant use, (ii) the reference period set from two to three years before delivery, (iii) excluding women admitted for psychosis within one year before delivery, (iv) restricting the analysis to women who used antipsychotic medication within the 13 weeks before delivery, (v) changing the length of the grace period for discontinuation to 14 or 60 days rather than 30 days, and (vi) excluding women who were admitted for more than seven days for a live birth delivery.
These subgroup and sensitivity analyses largely maintained the general pattern of the risk of relapse being higher for the non-user group and lower for the continuer group compared to the discontinuer group – although not all the sub-analyses demonstrated clear statistical distinctions as the primary analyses.
Finally, the authors also examined the timing of antipsychotic discontinuation over different timeframes (one year to 39 weeks before delivery, 39 to 26 weeks, 26 to 13 weeks, and 13 weeks to date of delivery). They found that the risk of relapse did not differ by the timing of antipsychotic discontinuation.
Most people were either non-users (43.1%) or discontinuers (41.6%). Only 15.3% of women continued antipsychotic medication during pregnancy.
Conclusions
The authors concluded:
In women with schizophrenia, antipsychotic continuation during pregnancy was associated with a reduced risk of postpartum relapse.
Strengths and limitations
As mentioned at the beginning of this blog, the evidence base for pregnant women is built on the findings from observational studies. In the current study, the authors used several statistical methods to address the potential biases and limitations. First, the study uses the nationwide database that captures almost everyone with schizophrenia who had live births in South Korea. Second, they utilised the self-controlled case series method to control for the severity of schizophrenia among women in the dataset. Third, the authors conducted multiple subgroup and sensitivity analyses to adjust for those with comorbid bipolar disorder and depressive disorder, as well as for the use of other psychotropic medications, including antidepressants and mood stabilisers.
Despite these strengths, there are several limitations worth noting. As the study was conducted in South Korea with presumably predominantly South Korean women, the generalisability of the findings to our practice in Australia (or any other non-South Korean country) may be limited. Particularly, we wondered how much impact either comorbid substance use disorder or psychotogenic substance use would have on the decision-making process among women in South Korea compared to Australia. Moreover, as the authors acknowledge, the study was not able to examine specific types of antipsychotic medication.
While it may not be realistic to explore every unique antipsychotic type, we wondered if there may be some difference between oral and depot formulations, for example. Related to this, as with most routinely collected health data, it is important to acknowledge that medication prescription does not always equal medication consumption. Finally, the outcome examined (i.e. hospital admission for psychosis) may not be fully reflective of the benefits and risks associated with the exposure. For instance, there might have been many women whose psychotic symptoms deteriorated during the study period, but the deterioration did not lead to hospital admission for various reasons.
Generalisability of findings may be limited by data being derived from one country: South Korea.
Implications for practice
Pregnancy can be a confusing time with multiple decisions thrust upon you with an ever-looming deadline. Some decisions can be paralysing, while other decisions can be life-changing. When as trainees, we had more than a few health professionals explicitly telling us that pregnant women are different; they are protected from mental illness by the aura of pending motherhood. Some even implicitly advised us to stop loading pregnant women with unnecessary pills. We now know that this is not true, and if anything, pregnancy is a vulnerable time for women with schizophrenia (Lefebvre et al, 2022). That said, from a clinician’s point of view, it is true that there is something different about pregnant women. Many of them are highly motivated by the impending motherhood to be the best parents that they can be. Often, this means they give up lots of things – cigarettes, alcohol and other drugs, as well as unhelpful relationships. Sometimes, the list also includes psychotropic medications.
There is often an almost knee-jerk reaction for many practitioners to advise cessation of psychotropic medications at pregnancy confirmation, with potentially significant adverse impacts on pregnant women’s mental health. It is difficult to make clinical decisions because all the existing evidence is, and will always be, based on someone else’s pregnancy and baby. You never know how your pregnancy will impact your baby. Unlike observational cohorts, you only have one baby in your study. At times, all we can do is respect the decision, monitor for as long and as closely as possible, and hope for the best for the mother and the baby. Sometimes we get through the pregnancy without a hitch. Other times, we need to make hard decisions. Wherever possible, shared decision making and close liaison with obstetric and midwifery colleagues are key. Ideally, the decisions are made with the mother, but sometimes they are made against her wishes.
Wherever possible, shared decision making and close liaison with obstetric and midwifery colleagues are key.
Pregnancy is a time in which consideration must be given to not only the pregnant woman sitting in front of you but also to her unborn baby. A prudent practitioner will also consider the postnatal period and breastfeeding considerations. In clinical practice, you tend to remember the ones that went wrong and forget about the ones that went well.
When discussing the risk-benefit balance of antipsychotic medication with pregnant women, we wonder about the availability bias that impacts our decision and the hindsight bias that influences our perception. To investigate the risks and benefits of medication use during pregnancy, routinely collected clinical data, such as the ones examined in the current study, are critical in improving our clinical practice. Some would argue that it is our ethical obligation to examine such data scientifically as doing so would result in a greater understanding and improved outcomes for mothers and their offsprings (Grzeskowiak et al. 2013). In making sense of the findings derived from such data, however, we need to be mindful of potential limitations when interpreting them.
As outlined in the limitations section above, the available data is limited to what has been collected. This means we may lose some nuances associated with many aspects of the study and relevance to the mother in front of you (e.g. study population may not capture the full spectrum of women with schizophrenia you see in your clinic, the medication exposure may not be fully reflective of how often and how much women take antipsychotic medications in real life, and the admission for psychosis may be an outcome that you want to avoid, rather than an outcome of interest).
Nevertheless, the findings from the current study give us one more piece of the puzzle to help us judge the balance. Antipsychotic medications do seem to help reduce the relapse of schizophrenia in pregnant women. The implications of a relapse of psychosis in a pregnant woman cannot be understated, with risks for poor obstetric outcomes as well the risk of harm to self and from others. Should the relapse continue in the postpartum, this can have significant impacts on the mother-baby relationship as well. It is important to remember that a well mother gives her the greatest chance to be the mother she hopes to be. While acknowledging that there is a bigger picture and we definitely need more research to put together a complete picture, here’s another piece that may be helpful in the conversation.
There is a bigger picture but here’s another piece that may be helpful.
Statement of interests
Shuichi is on the editorial board of the British Journal of Psychiatry, where the primary paper was published. He was not involved in reviewing or editing the paper.
Links
Primary paper
Kang K, Yang J, Yun B et al (2025) Antipsychotic Continuation during Pregnancy and Risk of Postpartum Relapse in Women with Schizophrenia: Nationwide Register-Based Study. The British Journal of Psychiatry, 1 – 8.
https://www.cambridge.org/core/journals/the-british-journal-of-psychiatry/article/antipsychotic-continuation-during-pregnancy-and-risk-of-postpartum-relapse-in-women-with-schizophrenia-nationwide-registerbased-study/14FBE96D35D76307454A33AAD6F1CBF2
Other references
Grzeskowiak L, Gilbert A, Morrison J. (2013) Methodological challenges in using routinely collected health data to investigate long-term effects of medication use during pregnancy. Therapeutic Advances in Drug Safety, 4(1), 27 – 37.
https://journals.sagepub.com/doi/10.1177/2042098612470389
Lefebvre A, Pouchon A, Bioulac S, et al (2022) Management of schizophrenia in women during the perinatal period: a synthesis of international recommendations. Expert Opinion on Pharmacotherapy, 23(11), 1337–1350. https://www.tandfonline.com/doi/full/10.1080/14656566.2022.2102421
Martin F. Psychotropic medication during pregnancy: new umbrella review finds no convincing evidence of adverse health outcomes for the baby, The Mental Elf, 18 Nov 2024.
Wallace J. Psychotropic medication in pregnancy: new evidence may help achieve a safe balance, The Mental Elf, 17 May 2016.
Psychotropic medication in pregnancy: new evidence may help achieve a safe balance
