Working Paper: NBER ID: w29439
Authors: Janet Currie; Esme Zwiers
Abstract: Using data on over 420,000 first time Dutch mothers, we examine the effects of postpartum antidepressant use on a wide range of maternal outcomes including further treatment for severe mental illness, labor market outcomes, and family formation. We exploit rules which state that Dutch general practitioners (GPs) must be available to make house calls to their patients. In practice many therefore use postal code boundaries to limit their practices. We instrument a postpartum woman’s receipt of antidepressants with the propensity to prescribe antidepressants to women aged 46 to 65 among GPs in her postal code. Ordinary Least Squares estimates suggest highly negative effects of postpartum treatment with antidepressants, but this is mainly due to selection into treatment. Instrumental variable estimates suggest that the marginal patient treated with postpartum antidepressants is much more likely to continue taking antidepressants long-term, with little evidence of effects on other outcomes.
Keywords: Postpartum Depression; Antidepressants; Maternal Outcomes; Health Economics
JEL Codes: I1; J19
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| postpartum antidepressant use (J13) | long-term antidepressant use (C41) |
| postpartum antidepressant use (J13) | treated for severe psychiatric disorders (I12) |
| postpartum antidepressant use (J13) | positive earnings (D33) |
| postpartum antidepressant use (J13) | living with child's father (J12) |
| postpartum antidepressant use (J13) | having another baby within three years (J13) |
| negative selection into treatment (C24) | postpartum antidepressant use (J13) |
| long-term antidepressant use (C41) | maternal outcomes (J13) |