The effect of ascertainment bias in evaluating gestational antidepressant exposure.

Several administrative database studies have reported on a positive association between first trimester exposure to paroxetine and ventricular septal defects (VSD). Using multiple source data we have shown that depressed women utilize significantly more health care resources, including ultrasound, echocardiogram and emergency room visits for their babies. Hence, there is much higher chance to identify VSD in their babies than among healthy controls. Moreover, paroxetine has been used more specifically than other SSRI for anxiety, further increasing the chance of  ascertainment bias.