[Skip to Content]
[Skip to Content Landing]
Comment & Response
August 14, 2013

Mortality After Perioperative β-Blocker Use in Noncardiac Surgery—Reply

Author Affiliations
  • 1San Francisco Veterans Affairs Medical Center, San Francisco, California
  • 2Cardiology Section, Veterans Affairs Medical Center, Denver, Colorado
  • 3Health Outcomes Program, Colorado School of Public Health, Aurora
JAMA. 2013;310(6):646. doi:10.1001/jama.2013.8522

In Reply Drs Mansi and Mortensen contend that inclusion of risk variables such as angina, recent myocardial infarction, congestive heart failure, or prior percutaneous coronary intervention may have introduced bias in our propensity score regression model due to possible unmeasured clinical characteristics associated with β-blocker exposure in patients with these conditions. They suggest that such patients should be excluded from the analysis, reasoning that exposed patients may represent those who received better care or were healthier at baseline, thus allowing them to tolerate β-blockers or conversely, that unexposed patients may have had contraindications to treatment or were treated at a lower level of care. Although a variety of methods have been reported in the literature to construct and use propensity scores, we applied the generally accepted approach to include the widest variety of potential covariates related to both treatment and outcome.1,2