Diabetes and Not Lack of Treatment with Atenolol Predicts Decreased Survival after Noncardiac Surgery

Abstract

Veterans Administration Medical Center; Anesthesia (129); San Francisco, California 94121;awallace@best.comIn Reply:-Thank you for allowing me to respond to Dr. Leung's comments about the two studies. [1,2]In February 1998, Dr. Leung wrote to Anesthesiology with six criticisms of Anesthesiology and New England Journal of Medicine articles. These criticisms included suggestions that it was inappropriate to withhold study drug from patients who might have a side effect, questions concerning the statistics in the New England Journal of Medicine article, requests for more postoperative hemodynamic data, questions about the withholding of beta blockade, interactions of diabetes and atenolol, and the lack of women in the study. I responded to all of these criticisms in writing, and those comments were published in Anesthesiology. [3]Dr. Leung was dissatisfied with the results of the letter to Anesthesiology and has sent a second letter to Anesthesiology with more criticisms, including:1. “Their study did not provide statistical support for such a claim.”2. “… this indicated that the effect of atenolol was not different from that of placebo, i.e., no influence of atenolol on survival.3. “[Greek small letter beta]-adrenergic blocking agents have reduced incidence of morbidity and mortality can not be supported by the current studies.”4. “… the recent popularized use of perioperative beta blockade is based on misinformation.”5. “Physicians caring for such patients should reevaluate the validity of recommending perioperative beta blockade for improved survival.”We strongly disagree with Dr. Leung and her misunderstanding of the results of the two manuscripts. [1,2]The primary end-point of the atenolol [2]study was the reduction of perioperative myocardial ischemia by atenolol. Perioperative administration of atenolol for 1 week to patients at high risk for coronary artery disease significantly reduced the incidence of postoperative myocardial ischemia (P = 0.029). Furthermore, a secondary finding was that reductions in perioperative myocardial ischemia were associate with reductions in the risk for death at 2 yr (P = 0.025). The primary end-point of the atenolol long-term study [4]was the effect of atenolol on 2-yr mortality. In patients who have or are at risk for coronary artery disease who must undergo noncardiac surgery, treatment with atenolol during hospitalization reduced mortality (P = 0.019) and the incidence of cardiovascular complications for as long as 2 yr after surgery (P = 0.008). These two studies provide clear and statistically significant evidence to support the claims that perioperative atenolol reduced incidence of morbidity and mortality and improves survival in patients at risk for cardiac morbidity.Dr. Leung does not understand the interpretation of univariable and multivariable models presented in Table 2of the Mangano et al. manuscript. [4]Once statistical significance is established, it is acceptable to do a secondary analysis of the data to look for other predictors of outcomes. We demonstrated clearly that atenolol reduced the risk of death. Diabetes mellitus and ischemia increased the risk of death. In the multivariable model we demonstrated that diabetes was statistically significant, but once patients with diabetes were excluded there was not sufficient power to demonstrate the effect of atenolol. Thirty-one percent of the patients in this trial had diabetes. If one exclude 31% of the patients in a trial, power is lost. That in no way brings the results into question; it simply implies that if one excludes 31% of the patients, one also loses power to see the effect. Furthermore, Dr. Leung is incorrect in stating that P = 0.06 (6% type I error) is not significantly when P = 0.05 (5% type I error) is suddenly significant. It is simply the difference between a 5% and 6% chance of a type I error. Perioperative administration of atenolol reduces the incidence of perioperative myocardial ischemia, postoperative mortality, and cardiovascular complications.Dr. Leung's statement that the use of perioperative beta blockade is based on misinformation is incorrect. It is important to understand statistics and study design before making such a statement. Physicians should clearly evaluate any therapy. Many trials of beta-blocker use demonstrate benefits in nonsurgical patients, including the ISIS-1 (International Study of Infact Survival), [5]the MIAMI (Metoprolol in Acute Myocardial Infarction), [6]the MAPHY (Metoprolol versus Thiazide Diuretics in Hypertension), [7]and the ASIST (Atenolol Silent Ischemia Study)[8]trials. We added to this abundant literature and have clearly demonstrated that perioperative beta blockade improves survival in surgical patients at risk for cardiac morbidity.Arthur Wallace, M.D., Ph.D.Veterans Administration Medical Center; Anesthesia (129); San Francisco, California 94121;awallace@best.com(Accepted for publication November 11, 1998.)