Recently, I blogged on a study by El Marroun et al.,¹ that examined prenatal antidepressant exposure and autism risk.

The main limitation of this study (and other similar studies²) is the risk of confounding by indication. This type of bias is frequently encountered in observational studies examining drug effects. As a result of non-randomisation of treatment, the indication for the treatment may be related to the risk of future health outcomes.

Since you’d expect mothers to only take antidepressants if they are depressed, the big questions here are:

• Is prenatal depression associated with autism? OR
• Is prenatal depression associated with antidepressant medication?

An American group have attempted to tackle this problem with a retrospective observational study³. They also examined neurodevelopmental disorders more broadly, by looking at autistic spectrum disorders (ASD) and attention deficit hyperactivity disorder (ADHD).

Methods

• Children, aged 2-19 diagnosed with one or more pervasive neurodevelopmental disorder, were identified from electronic health records:
  • Children with ASD (n=1,377) were matched 1:3 with non-ASD controls (n=4,022)
  • Children with ADHD (n=2,242) were matched 1:3 to non-ADHD controls (n=5,631)
  • Children with both conditions were included in the ASD group
• The primary exposure of interest was antidepressant exposure at any time during pregnancy.
• To address the issue of confounding by indication, they included the presence or absence of maternal depression in their analysis, and examined the effect of pre-pregnancy antidepressant use.
• A more detailed analysis was conducted looking at
  • Antidepressant exposure times
  • Antidepressant exposure period
  • Proxy measures of depression severity (number of treatment visits, subtype of antidepressant, recent history of antidepressant treatment, psychiatric co-morbidity and antidepressant serotonin transporter affinity).

Results

ASD vs. controls

Before maternal depression was included in the analysis, antidepressant exposure was associated with ASD risk:

• Pre-pregnancy (OR 1.98, 95% CI, 1.41 to 2.58)
• In the 1^st trimester (OR 1.84, 95% CI, 1.11 to 3.00)
• In the 2^nd trimester (OR 1.81, 1.07 to 3.00)

After taking in to account maternal depression:

• Antidepressant exposure was still associated with ASD pre-pregnancy (OR 1.62, 95% CI, 1.17 to 2.23)
• But there was no longer evidence of an association (OR 1.18, 95% CI, 0.86 to 1.61) during pregnancy (or when proxy measures of depression severity were examined).
  

  After taking in to account maternal depression, there was no longer evidence of an association between antidepressant exposure during pregnancy and autistic spectrum disorders

ADHD vs. controls

Before maternal depression was included in the analysis, antidepressant exposure was associated with ADHD risk pre-pregnancy until delivery (OR 1.69, 95% CI, 1.25 to 2.25).

After including maternal depression:

• Exposure in the 1^st trimester was associated with ADHD (OR 2.03, 95% CI, 1.19 to 3.44)
• But there was no longer evidence of this relationship:
  • Pre-pregnancy (OR 1.18, 95% CI, 0.86 to 1.61)
  • In the 2^nd trimester (OR 0.98, 95% CI, 0.56 to 1.68)
  • In the 3^rd trimester (OR 1.29, 95% CI,  0.76 to 2.15)

Limitations

It’s worth highlighting that some biases may exist in this research:

• The authors acknowledge that this study is not an RCT, therefore confounding by indication cannot be completely eliminated.
• The authors mention that they may not have captured all cases of ASD or ADHD, concluding that this could bias the results towards negative findings.
• The prevalence of ADHD and ASD in America is increasing over time, and there is concern that that the diagnosis of ADHD is over-used by doctors. In this study, cases were identified using ICD-9 codes assigned following clinical assessment. 3% of these case notes were blindly reviewed by a neuropsychologist, who rated the ICD-9 codes as having a sensitivity (the ability to detect cases correctly) of 0.84. So some of the children diagnosed as having ADHD may not have it, which could result in an overestimate of the effect size.
• Other possible confounders, such as obstetric complications and co-morbid psychiatric illness were examined, but they did not examine co-morbidity or family history of neurodevelopmental disorders, which are also relevant.

Conclusions

In relation to ASD, the authors said that:

the apparent risk of depression treatment during pregnancy may actually reflect maternal psychopathology.

In relation to ADHD, the authors said that they:

observed persistent risk associated with antidepressant exposure, particularly during the first trimester.

Our understanding of the risks associated with antidepressant treatment in pregnancy on the unborn child is limited, partly by the unsuitability of pregnant women as RCT participants. Observational data is prone to problems with confounding, so not an ideal way of studying this problem either.  However, these researchers were undeterred as they recognised the importance of this clinical question, and attempted to address the problem of confounding by indication. In doing so, they provide evidence for an association between maternal depression and ASD rather than antidepressant exposure. Their findings also suggest that antidepressant exposure in early pregnancy and maternal depression may be associated ADHD.

Having read the earlier paper¹, it’s invited commentaries^4,5 and this study³, the only thing I am clear about is that no one should feel guilty about making a decision to take or not to take antidepressants in pregnancy. As in many areas of mental health, there is no single right answer. The decision to take or not to take an antidepressant is an individual one, but need not be made alone.

Links

1. El Marroun et al. Prenatal exposure to selective serotonin re-uptake inhibitors and social responsiveness symptoms of autism: population-based study of young children. The British  Journal of Psychiatry (2014) ; 205: 95–102.
2. Rai et al. Parental depression, maternal antidepressant use during pregnancy, and risk of autism spectrum disorders: population based case-control study. British Medical Journal (2013); 346.
3. Clements et al. Prenatal antidepressant exposure is associated with risk for attention-deficit hyperactivity disorder but not autism spectrum disorder in a large health system. Molecular Psychiatry (2014), 1-8.
4. Jones & McDonald. Living with uncertainty: antidepressants and pregnancy. The British Journal of Psychiatry (2014)205: 103-104.
5. Petersen et al. Prenatal exposure to selective serotonin re-uptake inhibitors and autistic symptoms in young children: another red herring? The British Journal of Psychiatry (2014)205:105-106.