No one in the world has greater proscriptions on her behaviour than a pregnant woman in the developed world. And research published today in the BMJ – and the mass of media coverage it will inevitably inspire – will add a new and potentially dangerous rule to the list.
The commotion I’m expecting is based on a study that suggests a link between a common class of antidepressants, known as selective serotonin re-uptake inhibitors (SSRIs), and birth defects. These findings are not new.
For almost a decade now, we’ve known about the potential association between paroxetine, a member of this drug class, and a slightly increased risk of minor heart defects in children when mothers in first trimester use this drug. But women suffering depression and hoping to start a family may face a greater harm if they abruptly stop taking their antidepressants.
A Difficult Test
The potential for congenital effects from drug exposure in pregnancy has concerned the general public since the early 1960s, when thalidomide was found to cause limb defects. We now tend to look for a causal drug every time a woman bears a child with some form of congenital anomaly, and forget that there’s actually a background rate of children who will be born with a defect of some kind.
In fact, between two and four Australian women for every 100 will bear a child with a minor or major birth defect.
This BMJ paper, no doubt unintentionally, plays to the fears raised by the thalidomide case. If you just read the abstract or try to work through its elegant statistical analysis, it’s tempting to accept the results from a paper published in a highly regarded journal at face value.
And its central claim is clearly sensational enough for the media to cover the research in good faith, even though the story may promote fear in a relatively vulnerable group of women.
But there are not only key flaws in this study. Similar birth registry studies that suggest a causal association between this class of antidepressants and birth defects are also problematic. Let me explain why.
The best way to scientifically determine a causal association between a drug and an adverse outcome is to run a controlled randomised trial with one group receiving the drug and a similar group not receiving it. But it’s ethically not appropriate to do this with pregnant women.
Instead, we have to rely on research where we observe outcomes of people with a common characteristic – pregnant women taking antidepressants, for instance – over time (cohort studies). Or studies that identify two existing groups of people with a differing outcome, and compare them on the basis of some supposed causal attribute (case control studies). Such studies examine data to find possible “red flags”.
But research like this is open to misinterpretation because it’s impossible to account for all the causes of an effect. If I said there was an association between “number of churches” in a city and its “crime rate”, for instance, you would rightly be sceptical.
But if I swapped “number of churches” to “city size”, and could still draw the same graph, only one of these associations is likely to be true. The problem with epidemiological research like this is that there could be any number of causes contributing to an observed effect.
A number of maternal diseases, such as diabetes and epilepsy, may increase birth defect rates, as can untreated depression. Indeed, babies born to women with untreated depression are at risk of prematurity, low birth weights and cognitive or behavioural difficulties.
Untreated depression also has other adverse maternal outcomes, such as developing postpartum depression and suicidality, increased risk of hospital admission and pregnancy complications, including pre-eclampsia.
Importantly, in the BMJ study, pregnant women with depression taking SSRIs in early pregnancy were not compared with untreated pregnant women with depression in early pregnancy. In fact, women with depression, anxiety, bipolar disorder or obsessive compulsive disorder but not reporting any antidepressant use were excluded from the study.
This means that we don’t have a true baseline risk from which we can compare a possible increased risk from SSRI use in pregnancy.
For a drug to be a “proven” cause of birth defects, several criteria must be met. It must:
Produce deformities in more than 2% to 4% of mothers exposed (that is, the incidence rate must be higher than baseline);
Produce a consistent pattern of deformities;
Be given in sufficient dosage (as this adverse effect is dose-related effect);
Be given at the precise moment the vulnerable fetal body organ is forming. Once the organ has formed, the fetus is at no greater risk from the drug than a child or an adult.
Much of this information is not accessible in data collected from pregnancy registries, which is what most studies rely on.
Where does this leave our now anxious woman who is taking an antidepressant and planning a pregnancy or already pregnant? She needs to have a frank and open discussion with her prescriber about the true benefits and risks of her antidepressant medication in pregnancy. Together, they can make a shared decision on how best to proceed.
The BMJ study, together with many other registry studies, suggest that of the various drugs known as SSRIs, sertraline appears to have a reasonable safety track record for pregnancy outcomes. In fact, it’s fairly similar to that of mothers not taking an antidepressant. This would be a good first option for women of reproductive age and first-time users of antidepressants.
But it may not work for every woman. An alternative SSRI or an antidepressant from another class may be needed to control moderate to severe depression in non-responders.
Reputable publications have a duty of care to provide a lay explanation of how research that may invoke unnecessary anxiety or controversy should be interpreted by the general public. They need to clearly explain the limitations of such research in terms that can be easily understood by everyone.
If this had been done when the Women’s Health Initiative study, which suggested hormone replacement therapy caused breast cancer, was first published in JAMA, we would not have had millions of women suffering recurrence of their menopausal symptoms by unnecessarily abrupt withdrawal of their medication before seeking medical advice.
We don’t want this to happen too with SSRI antidepressants. Women should remember that the risk of having uncontrolled depression is greater for baby and her than the small absolute increased risk of birth defect that may be associated with specific antidepressants.