Statistical Method Affects Observational Study Results

Investigators should 'be cautious' when choosing statistical methods to adjust for bias

WEDNESDAY, Jan. 17 (HealthDay News) -- The statistical methods that are used to analyze data from observational studies can have a large impact on the results derived from that data, researchers report in the Jan. 17 issue of the Journal of the American Medical Association.

Therese Stukel, Ph.D., from the Institute for Clinical Evaluative Sciences in Toronto, Canada, and colleagues used four analytic methods to remove the effects of selection bias in an observational study that assessed survival in 122,124 elderly patients with acute myocardial infarction who underwent cardiac catheterization.

After adjusting for overt or measured bias using either multivariable model risk adjustment, propensity score risk adjustment or propensity-based matching, the investigators found a 50 percent reduction in death rate from catheterization. However, adjusting for hidden or unmeasured bias using instrumental variable analysis, and using regional catheterization rate as an instrument, the drop in death rate from catheterization was only 16 percent.

The article "is an important reminder of the need for careful and rigorous approaches to observational data analyses," write Ralph B. D'Agostino, Jr., Ph.D., of Wake Forest University School of Medicine in Winston Salem, N.C., and Ralph B. D'Agostino, Sr., Ph.D., of Boston University, in an accompanying editorial. The investigators "must ensure that they have available what they consider to be the most important patient characteristics measured before treatment assignment."

Abstract
Full Text (subscription or payment may be required)
Editorial

Physician's Briefing