[Skip to Navigation]
Invited Commentary
August 2017

Use of Instrumental Variables Methods in Examining Psychiatric Readmissions

Author Affiliations
  • 1Division of Health Policy and Administration School of Public Health, University of Illinois at Chicago, Chicago, Illinois
JAMA Psychiatry. 2017;74(8):805-806. doi:10.1001/jamapsychiatry.2017.1669

In this issue of JAMA Psychiatry, Slade et al1 take on difficult questions about the effects of electroconvulsive therapy (ECT) on patient health outcomes. Specifically, the authors try to measure the causal effect of ECT on inpatient hospital readmission for psychiatric reasons among patients with severe psychiatric conditions. The challenge with this question is that patients who receive ECT are likely different from patients who don’t receive ECT over a range of factors that may also affect readmission risks. For example, ECT might be appropriate for patients with greater demonstrated compliance with treatments or patients with particular kinds of insurance. Such patients would likely have higher or lower readmission rates, even in the absence of ECT. To address concerns about biases, the authors propose instrumental variable (IV) statistical analyses.

Add or change institution