Author Affiliations: Departments of Neurology (Drs Greenberg and Frohman) and Ophthalmology (Dr Frohman), University of Texas Southwestern, Dallas.
We appreciate the comments of Shirani and colleagues in their letter to the editor. At the core of this research is the ability of observational studies to accurately inform patients and physicians about the effect of treatments. Inherently risky to this approach, as acknowledged by Shirani and colleagues, is the potential for unintended bias to be present and unmeasured. One mechanism, which is used to control for variable baseline characteristics of patients, is the propensity score adjustment. In essence, this statistical correction compares 2 groups of patients—the group that ends up taking therapy and those that do not. The 2 groups can be matched for baseline characteristics and compared to ensure that no group of variables was responsible for treatment decisions. Unmatchable patients are excluded from further analysis.
Greenberg BM, Frohman E. Interferon Beta and Long-term Disability in Multiple Sclerosis—Reply. JAMA Neurol. 2013;70(5):651-653. doi:10.1001/jamaneurol.2013.1