This Article  • Full text  • PDF  • Send to a friend  • Save in My Folder  • Save to citation manager  • Permissions  Citing Articles  • Contact me when this article is cited  Related Content  • Similar articles in JAMA  Topic Collections  • Statistics and Research Methods  • Prognosis/ Outcomes  • Alert me on articles by topic

	Since this article does not have an abstract, we have provided the first 150 words of the full text and any section headings.

To the Editor: Dr Pine and colleagues¹ studied the enhancement of claims data to improve risk adjustment of hospital mortality. However, when comparing risk-adjusted hospital outcomes, adjustment for variables that are documented inconsistently between hospitals may introduce bias even though this adjustment may improve the discriminative power of the predictive model. This concern was not addressed in the article.

As an example of the problem, assume that 2 hospitals have identical mortality rates as well as an identical risk structure of patients (case mix) for a given disease or procedure. The hospitals would be expected to perform identically when compared in crude or risk-adjusted analyses. However, for some risk-adjustment variables, identification and documentation of risks might not be consistent between these hospitals. Consider the American Society of Anesthesiologists (ASA) Classification score, which has been shown to be vulnerable to interobserver inconsistency.^2-3 Documentation of this score may be different between the . . . [Full Text of this Article]

Günther Heller, PhD, MD
guenther.heller@wido.bv.aok.de
Research Institute of the Local Healthcare Insurance
Bonn, Germany

Rainer Schnell, PhD
University of Konstanz
Konstanz, Germany