[Skip to Content]
[Skip to Content Landing]
Views 298
Citations 0
Comment & Response
September 2014

Fixed-Effects Models and Diagnosing Psychiatric Disorders

Author Affiliations
  • 1University of North Carolina at Chapel Hill
JAMA Psychiatry. 2014;71(9):1078. doi:10.1001/jamapsychiatry.2014.687

To the Editor D’Onofrio et al1 reported an increased risk for the development of offspring psychiatric disorders with advancing paternal age. For most of the disorders (except for attention-deficit/hyperactivity disorder [ADHD]), the curves for the fixed-effects models have a similar shape to the curves for the baseline (ie, crude) and adjusted models; however, in each case, the fixed-effects models showed much larger non-null effects. For ADHD, the fixed-effects models were the only ones that showed non-null (ie, odds ratio > 1) effects. These large and substantively meaningful differences between the fixed-effects models and the baseline and adjusted models raise concerns about the validity of the fixed-effects models.

First Page Preview View Large
First page PDF preview
First page PDF preview