Author Affiliations: Division of Gastrointestinal and Endocrine Surgery, University of Texas Southwestern Medical Center, Dallas (Dr Livingston) (email@example.com); and Department of Medicine, Rush University School of Medicine, Chicago, Illinois (Dr McNutt). Drs Livingston and McNutt are Contributing Editors, JAMA
In Reply: Dr Finney and colleagues raise an important point regarding cross-level bias. All studies based on administrative data or clinical registries will be at risk for drawing invalid conclusions because of unmeasured confounding and misspecification errors. These data sources have limited amounts of information, and the codes used to identify diseases and procedures lack clear definition or specificity. When attempting to assess individuals aggregated into groups, an additional level of cross-level bias results from covariate heterogeneity across groups and effect modification.
Livingston EH, McNutt RA. Cross-Level Bias and Variations in Care—Reply. JAMA. 2011;306(19):2096-2097. doi:10.1001/jama.2011.1663