Should aprotinin continue to be used during cardiac surgery?

Jerrold H Levy*, George J Despotis and Edward Spitznagel

This article has no abstract so we have provided the first paragraph of the full text.

Mangano et al. present an observational, nonrandomized database study, which has been subjected to extensive statistical analyses. When using nonrandomized patients, however, determining the efficacy or safety of a therapy adequately can be problematic.¹ In observational studies, clinicians control treatment assignments, and can introduce bias by administering FDA-approved agents, such as aprotinin, preferentially over other agents whose efficacy and safety might not be as well established. Thus, the different treatment cohorts evaluated by Mangano et al. could have large differences in their observed covariates that could lead to biased estimates regarding treatment effects and potential complication rate.² Propensity scoring reduces bias but does not eliminate it. Consequently, sicker patients could receive different treatments.² In this study, higher-risk patients were unequally distributed between the treatment cohorts (see Supplementary Table 1 online).

ARTICLE NAVIGATION - ARTICLE TOOLS

• Full text
• Download PDF
• Send to a friend
• PDA services
• Permissions
• Rights and permissions
• Order commercial reprints

ISSUE NAVIGATION - THIS ISSUE

• Table of Contents
• Previous article
• Next article