A new study looking at the correlation of antidepressant use during pregnancy and the development of autism spectrum disorder (ASD) has been making headlines. While the results are likely significant, they are not as worrisome as the headlines may suggest.
The study: strengths and weaknesses
Overall the study design is solid. They followed 145,456 singleton full-term infants for a total of 904,035.50 person-years of follow-up. That is the strongest aspect of the study, its power. Typically when you capture large numbers you have to trade-off detail of information. As Lincoln might have said, you can capture a lot of information about a small number of people, or a small amount of information about a large number of people, but it is difficult to capture a lot of information about a large number of people.
The use of databases, especially in socialized countries, does help. In this case they used the ongoing population-based cohort, the Québec Pregnancy/Children Cohort. One compromise, however, is that they followed whether or not the mother filled a prescription for an anti-depressant. They did not capture whether or not the mother actually took the medication.
They counted as statistically significant any result in which the p-value was <0.05 and the 95% confidence interval (CI) did not cross 1.0. They did not correct for multiple comparisons, which does tend to exaggerate the statistical significance when multiple comparisons are made.
The results: A slight increase in the absolute risk of autism
In the mainstream media the results of the study have been widely reported as “Antidepressants during pregnancy increase risk of autism by 87%.” Whenever you see a percent change in risk the first question you need to ask is whether or not this is an absolute or relative risk.
If the risk of a disease increased from 5 to 10% that would equal a 5% increase in absolute risk and a 100% increase in relative risk. Often the relative risk is reported because it seems more dramatic.
In this case the 87% is obviously a relative risk, which equals an absolute risk of 0.87% (an increased risk from 1% to 1.87%, with a 95% CI of 1.15-3.04). That does not make as shocking a headline. But let’s dive a bit deeper.
The increased risk of developing ASD was for exposure in the second and third trimester only, not for exposure during the first trimester or in the year prior to getting pregnant (which had no statistically significant difference).
This is interesting because a 2011 study (although much smaller) found an increased risk of ASD from SSRI antidepressants in the year before delivery, with the strongest signal in the first trimester.
While we’re discussing previous studies, a 2015 systematic review found that for SSRIs:
The pooled crude and adjusted odds ratios of the case-control studies were 2.13 (95% CI 1.66-2.73) and 1.81 (95% CI 1.47-2.24) respectively.
These results are reasonably consistent with the current study. The current study also found no risk from other classes of antidepressants, just from SSRIs.
What does this change? Not much!
The current large study, added to the previous systematic review, does show a consistent increase in risk of developing ASD for mothers who use SSRI antidepressants specifically from 1% to 1.8-2.0%. These studies are correlational and alone do not establish cause and effect.
The current study controlled for confounding factors, including depression itself, as best they could. That SSRIs taken during pregnancy cause a modest increase in risk of ASD is plausible, however, and is a reasonable assumption to make for the purpose of clinical decision-making until further data sheds additional light on the question.
This is not a dramatic change in current practice. Antidepressants are already considered category C in pregnancy, which means:
Animal reproduction studies have shown an adverse effect on the fetus and there are no adequate and well-controlled studies in humans, but potential benefits may warrant use of the drug in pregnant women despite potential risks.
The current studies are observational, and not considered well-controlled, so I don’t think the category will change. Even if they are changed to category D, which means human trials show a risk, both categories allow for use in pregnant women if the benefits outweigh the potential risks.
There are recognized risks to untreated depression in pregnancy. Depressed mothers do not take as good care of themselves, may turn to drinking or smoking, and make seek less prenatal care.
As with all of medicine, decisions consider risk vs benefit (including the risk of not doing something). In this case pregnant women and their doctors need to consider the overall risk of not treating depression with medication vs the modest increased risk from the medication. For mild depression the risks of medication are probably not worth it, but for severe depression it probably is. Where to draw the line is a judgement call.
I do hope the reporting of this study does not cause pregnant women diagnosed with depression to stop their medication without consulting their doctors. The decision should be made with a full understanding of the risks on both sides.