Journal Article Annotations
2017, 3rd Quarter
Annotations by Tareq Yaqub, MD, and Priyanka Amin, MD
of the APM Women’s Health SIG
The finding: After implementing a four-step covariate model to control for confounders, there was no statistically significant increase in risk of offspring intellectual disability in women taking antidepressant medications during pregnancy. Therefore, previous associations made could be attributable to other confounding factors.
Strength and weaknesses: The study included a rigorous analysis of confounders (a four-step covariate analysis) in a large starting population of nearly 180,000 patients. It included an analysis of the RR (relative risk) at each step. The participants were selected from one national healthcare database (Sweden) which made follow-up relatively simple and accurate. The study was selective in who to include in the experiment group: 2 antidepressant dispensations were required to be included in the experiment group; this was done to suggest that women included were adhering to their medications. The study also included a clinical subsample of women who carried diagnoses of depression/anxiety but did not receive anti-depressants during pregnancy as one of the control groups to further adjust for underlying mental illness as a confounder.
The study did contain some limitations. It did not include information on the severity of the depression in the patients. The design also did not allow for a way to prove patient adherence. There was no exploration of dose-dependent effects of medications. In fact, no dosages were provided in the study. The supplemental information did not seem to include information on how the psychotropic medications were classified. Finally, the experimental group population could have been larger as mental illness diagnoses made in primary care specialties were not included if there was no specialty follow-up.
Relevance: Given this information, women with underlying depression/anxiety should not be deterred from continuing or starting antidepressant therapy while pregnant if they have concerns about teratogenic neurocognitive effects in the offspring. There does not appear to be an association between maternal anti-depressant use and intellectual disability in the offspring.
The finding: Panic disorder (PD) and generalized anxiety disorder (GAD) did not increase risk of maternal outcomes during delivery (including hypertensive disease in pregnancy, cesarean delivery, and preterm birth) nor increase risk of neonatal outcomes (including low birth weight, neonatal ventilatory support, or minor respiratory intervention) in models adjusted for substance use, age, body mass index, prior preterm birth, race/ethnicity, and maternal education. Mothers are more likely to require a cesarean delivery when taking benzodiazepines (OR 2.45, 95% CI 1.36-4.40, or an additional 200 cesarean deliveries per 1000 births).
Maternal benzodiazepine use was also linked to an increased risk of neonatal low birth weight (OR 3.41, 95% CI 1.61-7.26, or an additional 117 newborns per 1000 births) and increased risk of the newborn requiring ventilatory support (OR 2.85, 95% CI 1.17-6.94). Women were more likely to develop a hypertensive disease in pregnancy when taking a serotonin reuptake inhibitor (SRI) (OR 2.82, 95% CI 1.58-5.04). Maternal SRI use is associated with an increased risk of the newborn requiring minor respiratory support (OR 1.81, 95% CI 1.39-2.37). Gestation age shortens modestly with both benzodiazepines and SRIs, by 3.6 days and 1.8 days, respectively.
Strength and weaknesses: A major strength of the research is that data analysis accounted for multiple confounders that have impacts on maternal and fetal/neonatal outcomes. Also, data were collected prospectively over pregnancy and had a large sample size of over 2500 women. Another strength is that a structured, validated interview was used to make the diagnosis of GAD or PD.
A limitation of the study is that the authors modified the GAD diagnostic criteria to require only one month of symptoms instead of six months. It is possible that there may have been maternal or neonatal outcomes associated with an increased duration of symptoms. Similarly, PD was excluded in individuals with another health condition that could mimic PD symptoms, such as asthma, which likely excluded individuals who have PD given the comorbidity of these conditions. Additionally, confounders were self-reported, and women may not have chosen to disclose substance use, including benzodiazepine abuse.
Relevance: This study is one of the first to examine the impact of GAD, PD, SRIs, and benzodiazepines on maternal and fetal outcomes, controlling carefully for confounders. The outcomes of the study are helpful to discuss when having a risk-benefit conversation with a pregnant patient with GAD or PD when thinking about medication options, such as SRIs or benzodiazepines.