Among the 30 million patients who undergo non-cardiac surgery yearly, in the United States, about 3 million have, or are at risk for, coronary disease. In this subgroup, peri-operative ischemia is a major risk factor for subsequent events. This study was designed to look at the effect of peri-operative beta-blockade in preventing ischemia during and immediately after surgery and, in particular, in reducing cardiac events during the two years after surgery.
There were some differences in baseline characteristics between the
two groups, most of them not statistically significant:
Most of the advantage from atenolol occurred during the first 8 months
(1 death in the atenolol group vs. 10 in the placebo group, 7 from cardiovascular
causes). Thereafter, the survival curves became parallel.
In multivariable analysis, the hazard ratio for atenolol of 0.5 (50% reduction in risk of death at two years) did not quite reach statistical significance (p=0.06), presumably because of the powerful predictive value of diabetes.
Although medications differed between the two groups at baseline and
during follow-up, none of these differences were significantly associated
with two-year mortality.
The effect of peri-operative atenolol on survival (90% in treated patients at two years vs. 79% for placebo) and event-free survival (83% vs. 68%) was unexpectedly large, according to the authors. How might a beta-blocker administered at the time of surgery decrease mortality over the following half year? The authors do not speculate directly, but they note that peri-operative ischemia has been associated with a significant increase in cardiac events over the two years following non-cardiac surgery, and that treatment with beta-blockers reduces peri-operative ischemia.
Although there were differences in baseline characteristics between the groups, the authors argue that these do not account for the results. They state that "a larger proportion of the atenolol-treated patients had cardiovascular disease before surgery", and that other variables (such as medications or diabetes) did not account for the results, either.
They conclude that peri-operative beta-blockade in patients with, or at risk for, coronary artery disease is safe, effective and cost-effective.
In an accompanying editorial (the text is available at the NEJM website), Drs. Kim Eagle and James Froehlich from the University of Michigan discuss the association between peri-operative ischemia and post-operative outcome. Surgery causes catecholamine surges and a prothrombotic state which can, in turn, produce disrupted atherosclerotic lesions. Such complicated lesions can lead to clinical events in the following weeks or months. Blocking the peri-operative catecholamine surge could interrupt this process and might account for the beneficial effect of beta-blockade that was found in this study.
They note that a number of differences in baseline characteristics which would bias towards poorer outcome in the placebo group were present in the study, and that in multivariate analysis only diabetes was a statistically significant predictor of survival. Nevertheless, the results of this study do provide evidence for the benefit of peri-operative beta-blockade, particularly in patients with coronary disease. In addition, they note that a detailed history, physical examination, EKG and chest x-ray are instrumental in risk-stratification before non-cardiac surgery and they emphasize the importance of using the peri-operative period to maximize preventive cardiac care.
The baseline characteristics (noted in the table above) do indicate a number of factors that appear to bias against the placebo group, but the authors of the article argue convincingly that these do not explain their results. One objection I have: the baseline numbers appear to show more prior cardiac disease in the placebo group, but in their discussion the authors state that there was more prior cardiac disease in the atenolol group. Since the categories listed under "history of cardiac disease" are not mutually exclusive, one has to assume that overlap took place in such a way as to explain this seeming discrepancy. This should have been clarified in the article.
The authors imply that the improved outcome, which occurred mainly in the six months following surgery, was due to a reduction in perioperative ischemia. What evidence is there that such a reduction actually occurred? No data is given on in-hospital events (apart from the six deaths). In the methods section of the article, however, the authors state that they "designed the study to permit the assessment of both in-hospital events ... and adverse cardiovascular outcomes during the two years after discharge". They supply a reference for in-hospital events in this study, which is only an abstract (Anesthesiology, 1994;81:Suppl A99).
In that abstract, it is noted that patients were followed with daily EKG's, Holter monitoring pre-op, intra-op and for 7-days post-op and with CPK-MB's on days 1 and 5. Ischemic episodes lasting over 1 minute on Holter were analyzed. Results: the numbers of patients with ischemic episodes on post-op days 0-2 were 17 in the atenolol group and 34 in the placebo group (p=0.008). However, atenolol "did not modify the severity of ischemic episodes" and there was "insufficient power in the study design to demonstrate a difference in myocardial infarction".
Is this degree of reduction in perioperative ischemia sufficient to explain the observed reduction in two-year mortality and cardiovascular events? Perhaps. Alternatively, might beta-blockade inhibit the initiation of a cascade of events (in part inflammatory?) that begin during surgery but then evolve on their own over the next few months? The atherosclerotic process is currently viewed as a very dynamic one. Some theories point towards an inflammatory (or even infectious) component in its progression. Such an inflammatory process might be activated by triggers present during stress, trauma or surgery and a role for the adrenergic system could be postulated (thanks to Dr. E. Rodney Hornbake of North Shore Hospital for an informative e-mail exchange on this topic).
Obviously, this is highly speculative. I was not able to find any major references indicating that beta-blockade would inhibit stress- or trauma-related inflammation. Nevertheless, an explanation is needed for the impressive late benefit seen in this study as a result of short-term adrenergic blockade at the time of surgery. Hopefully, more studies will confirm these results, analyze them in subgroups (patients with coronary disease and those only at risk for CAD) and shed some light on the mechanisms involved.
In the meantime, a more liberal use of peri-operative beta-blockers in patients with, and at risk for, coronary disease certainly seems reasonable.
December 9, 1996
The proportion of patients previously taking beta-blockers was 18% in the atenolol group and 8% in the placebo group, a meaningful difference reaching a p level < 0.02.
Could this also suggest a bias toward a selection of patients in the
atenolol group already expected to have a better outcome with perioperative
June 24, 1997
to the Editor about this article, from the NEJM website. A number of
criticisms of the study, which the authors rebut convincingly, in my opinion.
Submit a comment about this article
Return to Journal Club home