Analytic Considerations for the Review of Meta-analyses or Umbrella Reviews Studying Adverse Effects of Medication in Pregnancy
Along with Dr. Michael Bloch and Dr. Victor Avila-Quintero, we evaluate problems in meta-analyses including confounding by indication, the potential outcomes of measurement error or incomplete measurement, issues in the adjustment for potential confounders, and difficulties with measurement of both outcome and exposure in observational studies of medication exposure during pregnancy. The accompanying Figure illustrates how these phenomena can lead to a significant measurable association between SSRI exposure in pregnancy and ASD in offspring that is subject to confounding and bias. These concepts are of equal concern surrounding the association between antidepressant use and suicide attempt or completion in children and adolescents, but they have been well discussed elsewhere. This was published in JAMA Psychiatry (Oct 2019).