In Reply:—
We greatly appreciate the insightful comments submitted to Anesthesiology regarding our recent clinical concepts and commentary and the accompanying editorial remarks of Kertai et al. 1,2All of them raise cogent and valid concerns on this complex and controversial topic.
Dr. Kempen has raised the issue of timing of beta blockade and its interactions with the clinicians existing “best” anesthetic practices and intraoperative opioid administration. We agree with much of his assessment, but we do maintain that the ability of the clinician to control hemodynamic stress and myocardial ischemia on emergence from anesthesia, even with liberal use of opioids, has always been recognized as being quite limited. This is one area in which liberal use of beta blockade is nearly universally accepted as a major advance in anesthetic management. Beta blockers likely (but not definitely) have additional or superior benefits on modulating at least several aspects of “plaque stabilization,”3although opioids may play a more active role in myocardial preconditioning. With regard to the interaction and “substitution” of these therapies with existing practices, the early work of Zaugg et al. 4using bispectral index guided administration of intraoperative atenolol identified intriguing hypotheses that form the basis of a large scale study currently funded by the National Institute of Aging on functional recovery after surgery in the elderly using beta blockade as an integral component of the entire anesthetic.*
Drs. Basler and Daniel are concerned regarding the recommendation of Kertai et al. 2to prescribe beta blockers to patients with one or more risk factors correlated with higher risk of cardiac complications. We certainly do not advocate an uncritical use of perioperative beta blockade and feel that use of a criteria of one risk factor alone is problematic given the potential for perioperative hypotension in the acutely beta blocked naïve patient or those on other antihypertensive medications (a complication that we all have anecdotally observed but appears to be rarely associated with any substantial adverse outcome).
A prime concern raised by Drs. Basler, Daniel, Leslie, and Devereaux is that the lack of adequately powered randomized controlled trials demonstrating the effectiveness of perioperative beta blockade precludes firm treatment recommendations, particularly in patients with coronary artery disease risk factors only. We particularly thank Dr. Leslie and Devereux for their specific mention of the Perioperative Ischemic Evaluation (POISE) trial with a target goal of 10,000 patients. We had alluded to several trials in planning or progress in our review but space constraints precluded presenting specifics. The first author of this response is familiar with the POISE study, given his attempts in advancing a proposal for a similarly powered large-scale randomized controlled trial in the Department of Veterans Affairs in 2002 (DVA Cooperative Study #534 Proposal, “Perioperative β-adrenergic receptor blockade in patients undergoing major noncardiac surgery,” Martin London, M.D., Kamal Itani, M.D., Co-Principal Proponents, Department of Veterans Affairs, Washington, DC), which was independently powered to 10,000 patients. Direct communication with the executive committee of POISE was instituted at that time to preclude duplication of efforts. Because the specifics of the POISE protocol remain confidential, we cannot directly critique it here. However, as noted by Leslie and Devereaux, POISE is evaluating effectiveness in “moderate-risk and high-risk patients.” Our efforts in the Department of Veterans Affairs were aimed at “low and moderate risk” patients, particularly those undergoing nonvascular surgery (given the large numbers of eligible patients and substantial logistical issues involved in widespread implementation of potentially lengthy periods of beta blockade). We believe that these are the patients that the majority of clinicians are most interested in.
In that effort, we were obliged to follow the model of the “large simple trial” (by the experienced “trialists” in Department of Veterans Affairs Cooperative Studies) that is so eloquently explained by Dr. Devereaux and Yusuf (both of whom are involved with the design and conduct of the POISE study) in a highly recommended review.5As noted in that review, not infrequently the results of small randomized controlled trials (as Mangano et al. 6and Poldermans et al. 7fall into the category of) are invalidated by larger “mega-trials” and or meta-analysis of small randomized controlled trials. After concerted efforts, the Department of Veterans Affairs effort was unsuccessful at the final evaluation step given the uncertainty (and thus lack of enthusiasm) of the Department of Veterans Affairs Cooperative Studies Evaluation Committee (comprised predominantly of internists and cardiologists) with regards to several key methodologic issues including the accurate assessment of short and long-term cardiac morbidity and mortality in the general surgical population critical to the sample size calculations (even using the Department of Veterans Affairs sophisticated federally mandated National Surgical Quality Improvement Project),8developing a “simple” perioperative treatment protocol particularly with background use of chronic beta blocker use estimated at 35–40% (and increasing annually) and the unavoidable interaction of a study patient with multiple perioperative care providers, “cross-over” issues given very high prevailing use of beta blockers for early perioperative treatment of hypertension and tachycardia, “intent to treat” issues related to cancellation or delay of surgery, and adequately defining a “nontreatment” arm given that a placebo trial in the United States at this time would be considered unethical given existing “small scale” data, “quasi-guidelines,” and newer revisions to the Declaration of Helsinki.9
The experience with this process, and the observation that not all internists or cardiologists encountered in a variety of realms (both research and clinical) consider this as important an issue as we do, leads us to believe that no single study (POISE included) is likely to provide widely accepted “guidance” and that careful analysis and comparison of outcomes of local practice patterns (preferably collaborative efforts by all clinicians providing perioperative care), as well as ongoing consideration of potential new therapies, will remain important factors to consider. The latter is particularly highlighted by the intriguing, but speculative, protective “associations” of statins, based on the uncontrolled, observational case control analysis of Poldermans et al. ,10a very recent long-term observational analysis of Kertai et al., 11and the unpublished data from figure 1 of the editorial by Kertai et al. 2As well, intriguing prospective, observational data on the independent impact of impaired baseline endothelial dysfunction measured in the brachial artery on outcome in vascular patients by Gokce et al. 12have been recently reported. Of note, the latter factor appears to not be influenced by beta blockade.13It would appear that “classic trialists” are somewhat at odds with the “progressive evidence-based medicine” specialists who have made sophisticated treatment recommendations based solely on analysis of existing prospective and retrospective observational data (and in the case of Boersma et al. 14and Kertai et al. 15by inclusion of their local “larger universe”of patients from which the randomized controlled trials were derived).16,17Publication of a meta-analysis by Stevens et al. shortly after our review went to press has provided a unique perspective.18Analysis of 11 beta blocker studies conducted from 1980–2000 comparing treatment with placebo or standard care suggests that beta blockers reduce intraoperative and postoperative ischemic episodes, reduce perioperative myocardial infarction (but only when trials with the highest frequency of previous myocardial infarction are included), and reduce perioperative death (but only when the data of Poldermans et al. 7with its high outcome rate are included).
As noted by Kertai et al. 2and in our recently published national survey of perioperative practitioners in the Department of Veterans Affairs,19this topic remains one that is closely watched and has a high rate of informal use by interested practitioners but likely a low use overall and lower still on a formal clinical pathway. The good news is that a multipronged attack strategy appears to be in an active state of engagement and should provide reasonably comprehensive results in the very near future.
* University of California, San Francisco, San Francisco, California. londonm@anesthesia.ucsf.edu
