Perioperative β-Blockade: Atenolol Is Associated with Reduced Mortality When Compared to Metoprolol

• Previous data suggest that perioperative β-blocker blockade reduces both short- and long-term postoperative mortality.

• However, at present, it is unclear whether there is a difference in perioperative mortality among β-blockers.

• The current study investigated 30-day and 1-yr mortality in patients managed with atenolol versus  metoprolol.

• Perioperative β-blockade using atenolol is associated with reduced short- and long-term postoperative mortality compared with metoprolol.

IN 1998, a protocol for perioperative β-blockade was developed and implemented at the San Francisco Veterans Affairs Medical Center (SFVAMC) based on criteria used in the original Atenolol study.1,2The Perioperative Cardiac Risk Reduction Therapy (PCRRT) protocol closely adhered to the clinical pathway set forth in the original atenolol study1,2with a few exceptions. IV metoprolol was substituted when IV atenolol became difficult to obtain. Clonidine was added to the PCRRT protocol in 2004 with the publication of “Effect of Clonidine on Cardiovascular Morbidity and Mortality after Noncardiac Surgery.”3Before the 1998 introduction of the PCRRT protocol as a local clinical guideline, only patients in randomized clinical trials received perioperative β-blockade by protocol; patients not enrolled in randomized clinical trials received medications solely based on prescribing physicians' clinical judgment. After the introduction of the PCRRT protocol, patients were started on perioperative β-blockers in the anesthesia preoperative clinic and when deemed appropriate by their attending anesthesiologist. Recommendations for continuing β-blockade were made to the responsible surgical teams. Compliance with the protocol was voluntary, and was unmeasured until the introduction of Surgical Care Improvement Project (SCIP-12) measures in 2007. A number of hospitals and hospital systems have adopted the SFVAMC Perioperative β-blocker protocol, PCRRT.

In 2010 we conducted an epidemiologic analysis of the pattern of use of perioperative β-blockade at the SFVAMC from 1996 to 2008.4The PCRRT study4entitled “Association of the Pattern of Use of Perioperative β-Blocker Use with Perioperative Mortality” divided patients into four groups by pattern of β-blocker use: None , Addition, Withdrawal , and Continuous . The None  group consisted of patients who did not receive β-blockers before, during, or after surgery. The Addition  group included patients who did not have β-blockers before surgery, but who received at least one dose of β-blocker medication after surgery as either an inpatient or outpatient. The Withdrawal  group consisted of patients who were on β-blockers before surgery, but did not receive a single dose after surgery. The Continuous  group included patients who were on β-blockers before surgery and who received at least one dose after surgery.

Epidemiologic analysis using logistic regression, survival analysis, and propensity analysis demonstrated that in patients meeting PCRRT protocol indications for perioperative β-blockade, Addition  was associated with a reduction in 30-day (OR 0.52, 95% CI 0.33–0.83, P = 0.006) and 1-yr mortality (OR 0.64, 95% CI 0.51–0.79, P < 0.0001). Continuous  was associated with a reduction in 30-day (OR 0.68, 95% CI 0.47–0.98, P = 0.04) and 1-yr mortality (OR 0.82, 95% CI 0.67–1.0, P = 0.05). Withdrawal  was associated with an increase in 30-day (OR 3.93, 95% CI 2.57–6.01, P < 0.0001) and 1-yr mortality (OR 1.96, 95% CI 1.49–2.58, P < 0.0001).4During analysis of the data for this study the following question was asked: Is there a difference in the mortality rate for different β-blockers? The most common inpatient β-blocker prescription was metoprolol (51,787 [75%]) and the second most common was atenolol (10,811 [16%]). The most common outpatient prescription was atenolol (64,617 [54%]) and metoprolol was second (39,296 [33%]). There is the suggestion in the cardiology literature that there is a difference in efficacy between atenolol and metoprolol.5The PCRRT protocol and Atenolol study1,2used atenolol for the oral drug because of once-daily dosing.1,2,4The PeriOperative ISchemic Evaluation,6,7Diabetic Postoperative Mortality and Morbidity,8,9and Metoprolol after Vascular Surgery10trials used metoprolol as the primary medication.

The current study is, therefore, an epidemiologic analysis of safety and efficacy, comparing patients managed with atenolol with those managed with metoprolol at a single hospital, the SFVAMC. Although a program for perioperative β-blockade, the PCRRT protocol, was in wide but nonmandatory use, and it recommended atenolol, the selection of β-blocker agent was not randomized and was based on physician preference. The current study tested the hypothesis that there was a difference between 30-day and 1-yr mortality rates in patients treated with atenolol and those treated with metoprolol.

After approval by the University of California, San Francisco Committee on Human Research and SFVAMC Research and Development Committee (San Francisco, CA), computerized records for all surgical patients at the SFVAMC from 1996 to September 2008 were extracted into a database. International Classification of Diseases (ICD) Codes, Current Procedural Terminology (CPT) codes, demographic information, problem lists, laboratory data, medication use, hospitalization, and mortality data were extracted from Department of Veterans Affairs Veterans Health Information Systems and Technology Architecture databases.

For a patient to be included in the current study, he or she had to be treated with perioperative β-blockade after surgery. The pattern of perioperative β-blockade could either be continuous  or additional  use. Additional  use patients were those who were not on β-blockers before surgery and did not receive any oral dosage of β-blockers before surgery, but who received at least one dose after surgery. Continuous  use patients were those who were on β-blockers before surgery and who received at least one dose after surgery. Analysis of medication use revealed that it is surprisingly common for patients to be switched from one β-blocker to another during perioperative care. To simplify analysis, a decision was made to exclude all patients who were switched from one β-blocker to another. Patients in the Inpatient  group were defined as those who had inpatient β-blocker use with a single dose of oral β-blocker, either atenolol or metoprolol, after surgery. Patients in the Outpatient  group were defined as those who had continuous β-blocker use with a single oral β-blocker, either atenolol or metoprolol (succinate or tartrate), as a preoperative outpatient and at least one dose of oral β-blocker as an inpatient after surgery. During the period of the study, IV atenolol use was extremely rare, in part, because of its unavailability. The transient inpatient IV use of atenolol versus  metoprolol was therefore not considered in determining the patient group assignment. There are many patterns of use of perioperative β-blockade that might be used, including different durations of therapy or dosing in the postoperative period. The definitions for the pattern of use, used in the current study, were based on previous work, and were decided on before study initiation. The chosen definitions categorize patients into a limited number of groups for analysis, and concentrate on prophylactic rather than therapeutic use.

Surgical patients were identified by the occurrence of a CPT code consistent with a surgical operation. Surgical procedures were then divided by CPT codes into cardiac surgery, vascular surgery, and all others. Cardiac surgery was defined as CPT codes 33510–33530, 33533–33545, or 33572. Vascular surgery was defined by a CPT code of 35500–35907, 35201–35390, 35001–35162, 34001–34490, or 34800–34900. Other surgical procedures were defined as a surgical CPT code not defined as cardiac or vascular surgery.

Demographic information, ICD-9 (International Statistical Classification of Diseases and Related Health Problems), and CPT codes from the problem list and discharge diagnosis were used to identify risk factors. Coronary artery disease (CAD) was defined by ischemic heart disease ICD-9–Clinical Modification (CM): 410–14 or previous or planned coronary artery bypass graft surgery defined by CPT codes 33510–33530, 33533–33545, or 33572, or previous angioplasty by CPT codes 92980, 92981, 92982, 92984, 92995, 92996, G0290, or G0291. Peripheral vascular disease (PVD) was defined by ICD-9-CM: 440, 442.84, 443 or a CPT code for a past or current operation of 35500–35907, 35201–35390, 35001–35162, 34001–34490, 34800–34900 or a lower extremity amputation 27880–27889, 27590–27598, or 28800–28825. Risk factors for CAD were defined from previous studies and the PCRRT protocol.1–3,11If a patient had two risk factors for CAD, they were assumed to be at risk. The age risk factor for analysis is based on the PCRRT protocol and was defined as older than 60 years based on the results of the previous epidemiologic work,11,12as well as the atenolol1,2and clonidine3studies. Diabetes mellitus was defined from any ICD-9-CM: 250.0–250.9, that included diabetes mellitus type I and II, or unspecified. It excluded gestational diabetes (648.8), hyperglycemia (790.6), neonatal diabetes mellitus (775.1), nonclinical diabetes (790.29), nephrogenic diabetes insipidus (588.1), and diabetes insipidus (253.5). Hyperlipidemia was defined by ICD9-9 code 272 disorders of lipid metabolism. Hypertension was defined as ICD-9-CM: 401–05 or an entry on a problem list. Smoking was defined as ICD-9-CM: 305.1 tobacco use disorder, v15.82 history of tobacco use, or 989.84 tobacco, or a problem list entry. Smoking excluded e869.4 second-hand tobacco smoke.

Preoperative outpatient medications, in-hospital medication, and postdischarge outpatient medications were identified by time of prescription relative to the date of surgery. β-Blockers included atenolol, metoprolol tartrate, and metoprolol succinate. Only oral or IV β-blockers were considered.

The study population was developed by analysis of patient risk factors and analysis of the β-blocker medication used in the perioperative period. For patients having more than one surgery, only the most recent surgery was studied. Patients who had multiple surgical procedures during a single day were excluded because of the improbability that an appropriate risk-adjustment model could be developed for multiple procedures in the same day. The study population was further restricted to those who were high-risk patients (with CAD, PVD, or two risk factors for cardiovascular complications) having inpatient surgery.

For the outpatient drug population, patients required a prescription before hospital admission, as well as an inpatient prescription after the day of surgery, of a single type of oral β-blocker (atenolol or metoprolol). This choice to study continuous β-blockade in patients receiving a single oral β-blocker was designed to remove the effects of drug switching, addition, or discontinuation. In a superset of this group, we analyzed patients who received only a single type of oral β-blocker during the inpatient hospital stay and the inpatient postoperative period, without regard to previous outpatient medications. As previously discussed, this choice was designed to remove the effects of drug switching in the inpatient group. In both groups, any use of IV β-blocker administration was ignored in determining the population. The IV atenolol was not available during the study period, so IV metoprolol represented 98% of IV β-blockade.

The prevalence of risk factors and raw 30-day and 1-yr mortality rates were compared using the chi-square test or Fisher exact test, as appropriate. Multivariable logistic regression was used to evaluate the combined effects of patient risk factors and β-blocker administered on outcome variables (30-day mortality and 1-yr mortality) and to examine the effects of confounding variables such as preexisting medical conditions (for example, age, sex, presence of known CAD, presence of known vascular disease, diabetes, hypertension, smoking, hypercholesterolemia, congestive heart failure, medication use). Kaplan–Meier curves were used to illustrate group survival, and Cox proportional hazards regression was performed for statistical analysis, with testing using likelihood ratios.

Propensity score matching was used to adjust for possible sources of bias or confounding between the metoprolol and atenolol groups due to different patterns of risk factors in these observational data. Propensity scores were developed using logistic regression based on medical risk factors, predicting the use of metoprolol versus  atenolol for continuous β-blockade. The predictors used to develop the score were age of the patient at the time of surgery, CAD, history of myocardial infarction (MI), history of congestive heart failure, PVD, diabetes, increased cholesterol concentration, hypertension, age older than 60 years, and smoking. Patient matching was based on propensity score using a greedy matching technique to facilitate best-matching while maintaining sample size.13In order to obtain as much diverse scoring as possible, all predictors were used regardless of the predictor's significance. The β-blocker administered was used to start the propensity matching. The probability of being on atenolol was used as the score. Patients were first matched based on their score up to the eighth decimal place. Patients not previously matched were then matched at the seventh decimal place, and so forth until the first decimal place. Overall fit of the propensity-matched logistic regression model was tested using the likelihood ratio test. Goodness-of-fit of the model to predict death in 30 days or 1 yr was assessed using the c-statistic and Hosmer–Lemeshow tests. Statistical significance of outcomes between metoprolol versus  atenolol groups was tested with several methods, including ordinary logistic regression, generalized estimating equations (GEE), or Cox survival analysis. Analyses on the matched populations were done using GEE regressions and Cox survival analysis. The GEE test takes into account the matched nature of the data.14A simple variance component was used for the correlation of the matched pair. For Cox regression, the model was analyzed using the matched pair as the strata.

For all analyses, a two-tailed nominal P  value of 0.05 or less was considered statistically significant. Data are presented as mean ± SD. Data are presented by procedure. All statistical analysis was performed using SAS 9.2 (SAS Institute Inc. Cary, NC).

There were 38,779 surgical procedures in 20,937 different patients (fig. 1). Although patients in 20,303 procedures (52%) received β-blockers, only 8,779 inpatient procedures (33%) received β-blockers continuously and 4,570 (15%) had β-blockers added at the time of surgery with withdrawal in 951 (5%). In addition, 4,625 procedures were excluded because these were either not the last procedure or they used a β-blocker other than atenolol or metoprolol, and 4,928 procedures were excluded because of switching between oral atenolol and metoprolol or because of missing data regarding the continuity of β-blockade. The final study population consisted of 3,787 patients who received exclusively either atenolol or metoprolol during their inpatient care (tables 1 and 2). When continuous exclusive use of a single β-blocker from outpatient to inpatient medications was required, the population dropped to 1,871 patients (tables 3 and 4). Most of these patients were men (97%), with an average age of 68 ± 10 yr. The final study population is a relatively high-risk population, with 36% having PVD, 62% CAD, 8% previous MI, and 14% congestive heart failure. The raw 30-day mortality is 3% with a 1-yr mortality of 12% (table 1).

Table 2shows patient characteristics based on inpatient β-blocker use of a single agent. Metoprolol was more commonly used (73%) than atenolol (27%). It appears the metoprolol-treated patients may have higher risk with more peripheral vascular disease, CAD, previous MI, and congestive heart failure. The atenolol-treated patients had more hypertension and hyperlipidemia. The 30-day and 1-yr mortality rates for patients treated exclusively during their inpatient care with atenolol (30-day, 1%; 1-yr, 7%) were lower than those treated with metoprolol (30-day 3%, 1-yr 13%) (P < 0.0001).

Table 3shows patient characteristics based on continuous outpatient and inpatient β-blocker use of a single agent. Analysis based on outpatient medication use, although for a smaller population, gives similar results to analysis based exclusively on inpatient medication use. Table 4shows analysis based on outpatient medication use, classifying the patients by atenolol or metoprolol use. Metoprolol has a higher 30-day (3%) and 1-yr (15%) mortality rate than atenolol (1% and 8%, respectively) (P < 0.001).

Logistic regression analysis for outpatient and inpatient 30-day and 1-yr mortality rates (table 5) shows improved survival by odds ratio (OR) in inpatients continuously treated with atenolol compared with those treated with metoprolol: inpatient 30-day (OR 2.7 [95% CI 1.4–4.9], P = 0.0017), inpatient 1-yr (OR 1.9 [95% CI 1.4–2.5], P ≤ 0.0001). When analysis is restricted to those patients who received one β-blocker continuously as both outpatient and inpatient, mortality at 1 yr was improved in the atenolol group (1.7 [95% CI 1.2–2.3], P < 0.002), although the 30-day mortality rate was not different between the two groups. To check the results of the analyses on observational data, analyses were done on the population matched by propensity scoring.

Survival analysis was based on the last surgery. Patients with more than one surgery on the same day were excluded. Analysis was restricted to high-risk inpatients as defined by the presence of CAD, PVD, or two risk factors. For the inpatient drug population (3,787), Cox survival analysis (table 5) shows that metoprolol had a hazard ratio (HR) of 1.8 (95% CI 1.4–2.3, P < 0.0001). Cox survival analysis based on the outpatient drug population (n = 1,871) similarly demonstrates the association of the specific preoperative choice of β-blockade, metopolol with a HR of 1.6 (95% CI 1.2–2.2, P = 0.0017) (table 5) compared with atenolol. Table 6summarizes the details of the Cox regression statistical models for propensity-matched and -unmatched populations of patients with continuous outpatient or inpatient β-blocker use.

Analysis was conducted on the matched 1,948 patient procedures based on PCRRT risk factors for the population taking into account only inpatient medications (table 7) and 1,170 patients based on outpatient medications (table 8). Propensity score matching using a greedy algorithm13was able to find matched populations for inpatient (table 7) and outpatient (  table 8 )  medications as demonstrated by standardized differences, D, less than 10 for predictor variables.

Analysis using GEE on the propensity-matched population confirms the results found based on analysis of inpatient medications for 30-day mortality. Metoprolol had an OR of 3.2 (95% CI 1.6–6.1, P = 0.0007) for 1-month mortality. For 1-yr mortality, metoprolol had an OR 2.0 (95% CI 1.5–2.8, P = 0.0001) (table 5) (fig. 2,) . GEE analysis on outpatient medications confirms the effect on 1-yr mortality with an OR of 1.5 (95% CI 1.0–2.2, P = 0.04) (table 5) (fig. 3 )  but is not statistically significant for 30-day mortality.

Cox regression on the propensity-matched inpatient group showed that metoprolol had a HR of 1.8 (95% CI 1.4–2.3, P = 0.0001). For outpatients, the HR for metoprolol was 1.6 (95% CI 1.1–2.2, P = 0.0085) (table 6).

In a recent retrospective observational study, we analyzed the association between patterns of use of perioperative β-blockade and perioperative mortality, finding that the addition or continuation of perioperative β-blockade based on PCRRT protocol indications (patients with known CAD, PVD, or two risk factors for CAD including age older than 60 years, diabetes, hypertension, hyperlipidemia, or smoking) is associated with a reduction in 30-day and 1-yr mortality. In the current study we extend these findings, studying only patients on addition  or continuous  perioperative β-blockade, and asking whether outcomes are affected by the choice of β-blocker drug used, i.e. , atenolol versus  metoprolol. We found that, in patients undergoing major inpatient surgery and meeting PCRRT criteria for β-blockade, use of metoprolol as the oral β-blocker is associated with increased 30-day and 1-yr mortality, compared with that of patients using only atenolol. This finding was true whether analysis was restricted to only the inpatient drugs used, or whether analysis included the chronic outpatient medications.

Any analysis of the association of perioperative β-blocker use with perioperative mortality must correct for the simple fact that patients coming to surgery, who are treated with β-blockers and are older, have significantly more perioperative risk factors than those not treated with β-blockers. For this reason, careful risk adjustment strategies are necessary in studying the efficacy of perioperative β-blockade and in comparing outcomes among β-blockers. In the current study, the logistic regression revealed that patients treated with metoprolol had higher 30-day and 1-yr mortality than patients treated with atenolol (tables 2, 4, and 5). However, analysis of the risk factor patterns in the two patient groups showed that patients receiving metoprolol had significantly more risk factors at baseline than did patients on atenolol (tables 2 and 4). To correct for this source of potential bias, the method of propensity matching was used. The patient population was winnowed to highly matched populations of patients receiving either metoprolol or atenolol for their continuous perioperative therapy (tables 7 and 8). Even after careful propensity matching removed most significant differences in risk factor prevalence between the two drug groups, mortality was significantly higher in the metoprolol group, both at 30 days and at 1 yr (tables 7 and 8).

β-Blockers have been reported to decrease post-MI mortality and sudden cardiac death since 1981, with original results appearing in three placebo-controlled studies: the Norwegian Multicenter Study on Timolol after Myocardial Infarction,15–17the American β-Blocker Heart Attack Trial (BHAT),18,19and the Göteborg Metoprolol Trial.20A few years later, two very large trials, the Metoprolol in Acute Myocardial Infarction (MIAMI) study21and the First International Study of Infarct Survival (ISIS-1),22,23which included 6,000 and 16,000 patients, respectively, showed that β-blocker therapy could reduce mortality within the first 2 weeks after onset of MI. Data from 24 postinfarction studies with long-term follow-up show an average 20% reduction in mortality over 2 yr. Pooled results of 28 short-term, randomized, placebo-controlled trials in which β-blockers were given intravenously shortly after onset of MI indicate an average 13% mortality reduction within 2 weeks. In the 16 studies in which the sudden cardiac death rate was reported, the beneficial effect of β-blockade was even more marked: a 34% average reduction of risk. However, not all studies with β-blockers have demonstrated a significant reduction in the incidence of sudden cardiac death, and not all β-blocker drugs may be equal. A decrease in sudden cardiac death post-MI has been demonstrated only for the more lipophilic β-blockers (timolol, metoprolol, and propranolol). Two of these lipophilic β-blockers, metoprolol and propranolol, have also been shown to prevent ventricular fibrillation after MI in clinical studies.24 

What could explain the differences in mortality observed between metoprolol and atenolol in the current study? The first hypothesis is that metoprolol was given to patients at higher preoperative risk, and that our efforts to eliminate this bias through propensity matching were unsuccessful. Tables 2 and 4show that patients on metoprolol have higher risk of PVD, CAD, previous MI, and congestive heart failure, but less hypertension, hyperlipidemia, and two or more risk factors. Logistic regression and propensity analysis demonstrate that after risk correction, patients treated with metoprolol continue to have a higher risk of mortality. Propensity matching using the greedy algorithm matched patients effectively on all risk factors measured, and in the matched group mortality was still higher in the metoprolol-treated patients. This result suggests either a true difference in mortality between metoprolol- and atenolol-treated patients, or the possible presence of systematic hidden confounding risk factors or other bias.

In the current study patients receiving metoprolol for continuous perioperative β-blockade were much more likely to receive IV β-blockade than were patients receiving atenolol. This result suggests a possible mechanism for different effectiveness of the two drugs: metoprolol tartrate (most metoprolol doses) wears off much faster than does atenolol, so patients have more opportunities for missed doses and more opportunities to “withdraw” from therapy on a daily basis. The finding that metoprolol-treated patients more frequently need IV β-blockers suggests that the perioperative hemodynamics in metoprolol-treated patients may more often be poorly controlled because of the shorter duration of action of metoprolol. An alternative hypothesis is that clinicians simply do not think to give IV metoprolol to atenolol-receiving patients when hemodynamics require treatment, which we strongly doubt.

There are clear differences between the metabolic half-life of atenolol and metoprolol tartrate. The risk of a missed dose and acute drug withdrawal increases with the twice-daily dosing required with metoprolol tartrate. A recent epidemiologic analysis in 37,151 patients who were receiving atenolol or metoprolol before surgery found results similar to those of the current study.25The two groups were similar in demographic characteristics, medical therapy, and type of surgery. There were 1,038 patients who experienced a MI or died. The mortality rate was significantly lower for patients receiving atenolol than for those receiving metoprolol (2.5% vs . 3.2%, P < 0.001).25The decreased risk with atenolol persisted after adjustment for measured demographic, medical, and surgical factors. Redelmeier et al.  25hypothesized that patients receiving metoprolol have higher risk because of the effects of acute withdrawal after missed doses with metoprolol.25The risk of a missed dose increases with twice-a-day dosing. The PCRRT study4demonstrated an increase in risk of death with perioperative β-blocker withdrawal (30-day OR 3.93, 95% CI 2.57–6.01, P < 0.0001) and 1-yr mortality (OR 1.96, 95% CI 1.49–2.58, P < 0.0001). The current study found results similar to those of Redelmeier et al.,  25with a lower risk in patients treated with atenolol. Both atenolol and metoprolol tartrate are dose equivalent as antihypertensive β-blockers at rest and during exercise, 1 h after intake.26Metoprolol tartrate was less effective than atenolol 25 h after dosing because of its shorter plasma half-life, requiring a twice-daily regimen for metoprolol in standard preparation.26,27Atenolol and metoprolol succinate can be considered equally effective in the treatment of mild to moderate hypertension and can both be dosed once daily.28Mean reduction in exercise tachycardia at 24 h after dosing was significantly greater with atenolol and metoprolol succinate than with metoprolol tartrate.29β-Adrenoceptor-blocking activity at 24 h was more variable with all formulations of metoprolol than with atenolol, which was explained by differences in metoprolol metabolism.29In subjects who were found to be “poor metabolizers” when tested with debrisoquine, greater bioavailability, half-life, and response to metoprolol were found.29These subjects maintained β-blocking activity at 24 h following metoprolol, whereas extensive metabolizers did not, even with sustained-release formulations.29The response to atenolol did not depend on oxidation phenotype.29Atenolol 100 mg once daily, metoprolol 100 mg twice daily and 300 mg once daily, were equivalent as β-adrenoceptor-blocking doses.27Thus, milligram for milligram, atenolol and metoprolol do not produce equivalent blockade on the cardiovascular system.27Atenolol and metoprolol succinate are equipotent following once-daily administration in patients with mild to moderate hypertension.30Many studies have found little difference between the cardiovascular effects of these drugs; for example, there were no differences in antianginal efficacy between the two β-blockers in a multicenter double-blind crossover trial carried out in patients with stable effort angina.31 

In addition to expected differences in duration of action, there may be other differences in metabolism or excretion of the β-blocker explaining differences in efficacy.32Metoprolol is metabolized by cytochrome CYP2D6, which is blocked by selective serotonin reuptake inhibitors (paroxetine or fluoxetine).33Atenolol is CYP2D6 independent.33Clinically important drug-drug interactions that are associated with impaired bioactivation of metoprolol might be more significant in clinical practice than the factors that cause drug accumulation of atenolol (renal insufficiency).33Compounds predominantly metabolized by polymorphic cytochromes such as CYP2D6 are affected by the metabolizer status, and competing compounds using the same metabolic pathways can strongly influence plasma concentrations (for example, metoprolol or carvedilol). Studies have found that the variability in β-adrenoceptor blockade at 24 h was much greater with each of the three metoprolol formulations than that with atenolol.34These differences could be explained by the variability in metoprolol metabolism.34Oxidation phenotype testing with debrisoquine showed there were six extensive metabolizers and two poor metabolizers of metoprolol.34The area under the curve, half-life, and response to metoprolol at 24 h were much greater in poor metabolizers.34In contrast, the response to atenolol was not influenced by phenotype.34Metabolically stable compounds that are predominantly excreted renally, such as atenolol, may be unfavorable because their plasma concentrations largely depend on renal function.35For this reason, it has been suggested that atenolol could be inferior to metoprolol because of its dependence on renal excretion.35 

However, there may be additional pharmacologic differences between atenolol and metoprolol. Some studies have found that heart rate during submaximal and peak exercise was significantly lower with atenolol.36,37Atenolol achieved greater β-adrenergic blockade than metoprolol, and this was associated with significant inhibition of vagal withdrawal during stress.36It has been suggested that peripheral blockade of β-adrenergic receptors may be more important than central blockade in preventing stress-induced vagal withdrawal in patients after MI.36There are minor differences in lipophilicity between atenolol and metoprolol, with very little blood-brain barrier penetration with atenolol, but some with metoprolol.38,39,40,41However, there was no difference in central nervous system-related symptoms between metoprolol and atenolol at therapeutically comparable dosages, indicating that the degree of lipophilicity may be of minor importance for the occurrence of such symptoms.42Metoprolol has fewer central nervous system effects on sleep, dreams, concentration, memory, energy, and anxiety than propranolol, but more effects than atenolol, which did not have any central nervous system effects.43 

The effect of β-blockers on lipoproteins is variable. One study found that atenolol did not affect lipoproteins, whereas metoprolol increased the very-low-density lipoprotein concentration in 75% of the patients.44Other studies have found that, compared with placebo values, neither metoprolol nor atenolol influenced total plasma cholesterol concentrations, but total plasma triglycerides increased slightly.45Low-density lipoprotein concentration remained unaltered, whereas high-density lipoprotein cholesterol and protein were significantly reduced by both β-adrenoreceptor blockers.45The ratio of total cholesterol to high-density lipoprotein cholesterol was increased by both drugs.45Patients taking β-blockers over many years may be exposed to increased vascular risk, despite other anticipated benefits from such therapy.45 

Another factor that must be considered is the possibility that the physiologic effects and toxicities of orally administered drugs may be different from those of IV-administered drugs. The current study demonstrated increased risk in patients who received IV β-blockers. The Global Utilization of Streptokinase and TPA (alteplase) for Occluded Coronary Arteries study (GUSTO-I) found that adjusted 30-day mortality was significantly lower in atenolol-treated patients, but patients treated with IV and oral atenolol treatment versus  oral treatment alone were more likely to die (OR, 1.3; 95% CI, 1.0–1.5; P = 0.02).46Subgroups had similar rates of stroke, intracranial hemorrhage and reinfarction, but IV atenolol use was associated with more heart failure, shock, recurrent ischemia, and pacemaker use than oral atenolol use.46The GUSTO-I study concluded that, although atenolol appears to improve outcomes after thrombolysis for MI, early IV atenolol seems of limited value.46The best approach for most patients may be to begin oral atenolol once their condition stable.46The GUSTO-I trial may have led to limited availability of IV atenolol and the predominance of IV metoprolol.

The current study has several limitations. We evaluated the results of the implementation of a protocol (PCRRT) for perioperative β-blockade at a single hospital from 1996 to 2008. Although there was a persistent effort during this period to increase utilization of the protocol with educational sessions, laminated protocols, a website,§academic detailing, computerized reminders in the hospital computer system, and feedback of medication compliance based on Surgical Quality Improvement Project measures. Protocol compliance was voluntary. Because the individual physician caring for each patient decided on use of the protocol, we cannot exclude the possibility that selection bias, perhaps based on unmeasured patient risk factors, played a role in decisions to use a specific β-blocker in individual patients. Given the results of our propensity-matching analysis, confounding bias is unlikely.4This study used medication prescription orders as evidence of β-blockade and not medication administration records. This analysis represents intention to treat rather than actual medication administration. It is unlikely that there is a systemic bias in the medication administration patterns of atenolol versus  metoprolol. There may however be differences in compliance based on a once-daily (atenolol) versus  a twice-daily (metoprolol tartrate) medication.

Another potential limitation of this study is that the principal methods used included retrospective epidemiologic analysis of computerized medical records rather than chart review. The data used for cardiac risk were derived from problem lists, discharge diagnosis, CPT codes, and ICD-9 codes. Some patients may not have fully completed problem lists, so it is possible that some risk factors were missed.47In addition, it is likely that the quality of risk factor data likely improved over the course of the study. The sensitivity of computerized diagnoses in Veterans Affairs patient files has, however, been judged to be greater than 80% in the administrative files, with specificity of 91–100% for common chronic health diagnoses such as CAD.48In addition, the principal outcome variable, death, is carefully identified by the Veterans Affairs and has high accuracy.48Nevertheless, any study that relies on epidemiologic analysis of retrospective data is limited by the quality of the computerized medical record data. It should be recognized that medical care has changed during the period of this study and there have been changes in β-blocker use over time in the current population,4including greater prevalence of use of chronic β-blockade,4and evolving greater use of metoprolol compared with atenolol. Our study is not, however, powered to examine outcome patterns related to evolving prescription patterns.

Another potential limitation of this study is that it excluded patients who had their oral β-blocker switched during the inpatient stay, but it ignored the use of IV drugs in the exclusion criteria for drug switching. We believe this is justified because, during the period of this study, there was almost no access to IV atenolol. When an IV β-blocker was needed, either metoprolol or esmolol was used. Esmolol constituted only 2% of the total inpatient β-blockers. We decided that use of IV metoprolol or esmolol did not constitute a switch requiring patient exclusion because there was essentially no access to IV atenolol. Excluding all atenolol-treated patients who received a dose of IV metoprolol would exclude all patients who were on atenolol who needed IV β-blockade. Excluding all atenolol patients who received even a single dose of IV metoprolol would profoundly cut the sample size, making it untenable to study the relative effects of metoprolol versus  atenolol in chronic use. We therefore analyzed the effect of the oral medications and did not exclude patients who received IV β-blockers. We propose that the use of IV β-blockers (metoprolol or atenolol) may best be considered as a marker of clinical hemodynamics that requires acute treatment, and therefore use of IV β-blockers may be considered as a secondary, although surrogate, outcome. During the study period intraoperative anesthesia care was documented on paper records and we do not have electronic records that would allow analysis of either intraoperative β-blocker administration or hemodynamic responses. Electronic anesthesia record-keeping systems should permit such analysis in the future.

The current study analyzed the use of atenolol and metoprolol, but did not evaluate the effects of other cardiovascular medications or their interactions with β-blockers. Future research may be able to evaluate the efficacy of other antiischemic medications and/or the interactions with other medications. Mortality analysis was limited to inpatient surgery that required admission to the hospital, because the mortality rate for outpatient surgery was too low (0.1%) to allow for analysis. The exact cause of death cannot be established accurately with retrospective epidemiologic analysis of medical records. The exact cause of death is frequently difficult or impossible to establish even with chart review. The current study reports all-cause mortality and did not try to establish cause of death. The current study is observational, and thus a causal link cannot be definitely inferred.49Although a large cohort was obtained using computerized medical records, the accuracy of data could not be verified as it would be in a randomized controlled study.50 

Epidemiologic studies can be quite large (the current study has 3,787 surgical procedures) and can sometimes identify small effects and rare or infrequent events. Although epidemiologic studies reflect the actual practice of medicine, they can only identify associations and not establish causality. The current study, using multivariate risk adjustment and patient group refinement through propensity matching, demonstrates a strong association between the use of atenolol and an improved survival rate compared with metoprolol for perioperative β-blockade.4Nevertheless, this result should be viewed as a hypothesis-generating study, best followed by a randomized controlled trial, not as a stark conclusion.

Perioperative β-blockade in high-risk patients has been a level 1 standard of care since 1996.51–54The guidelines for implementation have been revised and updated several times and continue to evolve.51–54The PCRRT protocol4is a guideline for implementation that has been adopted by a large number of hospitals and even hospital systems. It should be pointed out that although mortality rates were lower in patients treated with perioperative atenolol than in those treated with metoprolol, patients who were treated with any β-blocker had a lower mortality rate than those who did not receive β-blockade. Appropriate use of the PCRRT protocol, with either atenolol or metoprolol, is clearly associated with a reduction in 30-day and 1-yr mortality compared with no β-blockade.4 

In conclusion, the use of perioperative β-blockade, based on PCRRT protocol4indications (patients with known CAD, PVD, or two risk factors for CAD including age older than 60 years, diabetes, hypertension, hyperlipidemia, or smoking), is associated with a reduction in 30-day and 1-yr mortality. Perioperative use of atenolol is associated with a reduction in 30-day and 1-yr mortality compared with metoprolol.