Author + information
- Received January 4, 1999
- Revision received May 20, 1999
- Accepted June 28, 1999
- Published online November 1, 1999.
- Jersey Chen, BA∗,
- Thomas A Marciniak, MD∥,
- Martha J Radford, MD, FACC∗,‡,§,
- Yun Wang, MS§ and
- Harlan M Krumholz, MD, FACC∗,†,‡,§,* ()
- ↵*Reprint requests and correspondence:
Dr. Harlan M. Krumholz, Yale University School of Medicine, 333 Cedar St., P.O. Box 208025, New Haven, Connecticut 06520-8025.
We sought to determine the use and association with one-year mortality of beta-blocker therapy for the treatment of acute myocardial infarction (AMI) in elderly diabetic patients and to examine whether beta-blocker therapy was associated with increased rates of hospital readmission for diabetic complications traditionally associated with beta-blockers.
Although many randomized trials have demonstrated that beta-blockers are effective in reducing mortality after AMI, some experts are concerned about the use of beta-blockers in diabetic patients. Little is known about the effectiveness and complication rate of beta-blocker therapy after AMI for elderly diabetics in community practice settings.
We conducted a retrospective cohort study using the National Cooperative Cardiovascular Project, which contained data abstracted from hospital medical records of Medicare beneficiaries admitted with an AMI during 1994 and 1995.
Out of 45,308 patients without contraindications to beta-blocker therapy, 7.4% were insulin-treated diabetics and 18.5% were non-insulin-treated diabetics. Beta-blockers were prescribed at discharge for 45% of insulin-treated diabetics, 48.1% of non-insulin-treated diabetics and 51% of nondiabetics (p < 0.001). After adjusting for demographic and clinical factors, diabetics continued to be less likely to receive beta-blockers at discharge compared with nondiabetics (odds ratio [OR] for insulin-treated diabetics 0.88, 95% confidence interval [CI] 0.82 to 0.96; OR for non-insulin-treated diabetics 0.93, 95% CI 0.88 to 0.98). After adjusting for potential confounders, beta-blockers were associated with lower one-year mortality for insulin-treated diabetics (hazard ratio [HR] = 0.87, 95% CI 0.72 to 1.07), non-insulin-treated diabetics (HR = 0.77, 95% CI 0.67 to 0.88) and nondiabetics (HR = 0.87, 95% CI 0.80 to 0.94). Beta-blocker therapy was not significantly associated with increased six-month readmission rates for diabetic complications among diabetics and nondiabetics.
Beta-blockers are associated with a lower one-year mortality rate for elderly diabetic patients to a similar extent as for nondiabetics, without increased risk of readmission for diabetic complications. Increasing the use of beta-blockers in elderly diabetic patients represents an opportunity to improve the care and outcomes of these patients after AMI.
Many randomized clinical trials have demonstrated the efficacy of beta-blocker therapy in reducing mortality after acute myocardial infarction (AMI) (1). Some experts have raised concerns, however, about the use of beta-blockers in patients with diabetes mellitus (2). In diabetics on insulin therapy, beta-adrenergic blockade has been known to reduce patient sensitivity to symptoms of hypoglycemia (3)and delay recovery from hypoglycemia through glycogenolysis inhibition (4,5). In type 2 diabetics, beta-blocker therapy has been known to exacerbate glucose intolerance (5,6)and induce hyperosmolar hyperglycemic coma (7). The clinical guidelines for AMI established by the American College of Cardiology (ACC) and the American Heart Association (AHA) (8)list insulin-dependent diabetes as a relative contraindication to beta-blocker therapy. Hence, concerns regarding beta-blocker use in diabetics persist even though clinical trials have suggested that beta-blocker therapy may reduce long-term mortality to a greater extent in diabetics than in nondiabetics (9). Subgroup analyses from randomized controlled trials have found beta-blockers to be at least as effective in diabetics as nondiabetics (10–12). However, these trials enrolled a small number of diabetics and included few elderly patients, a group believed to be at higher risk for drug-associated hypoglycemia (13). The relationship between use of beta-blocker therapy and outcomes for elderly diabetics outside of clinical trials in the setting of community practice has not been well studied.
To evaluate these issues, we conducted a study to address the following questions: 1) Do beta-blockers improve one-year survival for elderly diabetic patients after AMI? 2) Are diabetic patients discharged on beta-blockers at higher risk for hospital readmission for diabetic complications associated with therapy? To answer these questions we used data from the Cooperative Cardiovascular Project (CCP), a Health Care Financing Administration (HCFA) initiative designed to examine patterns of care and improve the outcomes of Medicare beneficiaries with AMI.
The CCP sample was identified from hospital bills in the Medicare National Claims History File (form UB-92), which included all claims submitted for patients treated under fee-for-service plans with a principal discharge diagnosis code of AMI (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] 410), except those indicating a subsequent, nonacute treatment of AMI (fifth digit of ICD-9-CM code equal to 2) (14,15). Cooperative Cardiovascular Project sampling was conducted for an eight-month period between February 1994 and July 1995, except for the CCP pilot study states (Alabama, Connecticut, Iowa and Wisconsin), which were sampled for a four-month period between August and November 1995. The Health Care Financing Administration established two Clinical Data Abstraction Centers to abstract predefined variables from copies of hospital records, with monthly random reabstractions to monitor data reliability.
The study sample was restricted to patients 65 years of age or older with a confirmed AMI who were discharged alive. A confirmed AMI was defined as a discharge diagnosis of AMI and chart documentation of either a creatine kinase-MB fraction >0.05, lactate dehydrogenase (LDH) level more than 1.5 times normal and LDH1> LDH2, or two of the following three criteria: chest pain, a twofold elevation of the creatine kinase level or a new AMI on the official electrocardiogram report.
We included only the patient’s first admission during the sample period to avoid double-counting patients. We excluded patients who were subsequently transferred to another acute care institution, because of an inability to determine their discharge medications. We omitted three patients in whom mortality could not be confirmed or dated. We also excluded patients considered to have a terminal illness (chart-documented as unlikely to live more than six months) or metastatic cancer, because the focus of their treatment may not have been targeted towards a survival benefit. We replicated clinical trial criteria in order to define a study sample of patients eligible for postdischarge beta-blocker therapy (those without strong contraindications to beta-blockers) based on the AMI clinical guidelines established by the ACC/AHA in order to minimize the potential for residual confounding (8).
Mortality was ascertained from the Medicare Enrollment Database. Hospital readmissions were identified from Medicare Part A claims containing data on all hospitalizations billed to Medicare. Due to the time lag for the availability of the latest Part A data, this study was able to consider hospital readmission only within six months of discharge (the longest follow-up time period available for analysis). Medicare Part A data were not available for three of the CCP pilot states (Alabama, Iowa and Wisconsin) and Minnesota. Overall, 95% of the patients in the study sample were included in the readmission analysis. We examined hospital readmissions for acute diabetic conditions (uncomplicated diabetes, ketoacidosis, hyperosmolarity, diabetic coma, hypoglycemia [ICD-9 codes 250.0 to 250.4, 250.8 to 250.9]) and readmissions for diabetes with renal, ophthalmic, neurological or peripheral vascular complications (ICD-9 codes 250.4 to 250.7).
The principal independent variables of interest were prescription of beta-blockers at discharge and history of diabetes. We identified all patients with an oral beta-blocker as a discharge medication. We classified patients as insulin-treated diabetics (patients prescribed insulin at discharge), non-insulin-treated diabetics and nondiabetics based on chart review. Although we were unable to distinguish between type I and type II diabetes, we expect the majority of diabetics to have type II disease, given the advanced age of the cohort.
Because differences in baseline characteristics between diabetics and nondiabetics may bias the association between beta-blockers and survival or readmission, we selected a set of variables to control for potential confounding. These variables were selected on the basis of previous work, strength of association and clinical relevance, and they included the characteristics listed in Table 2along with the region of the hospitalization as defined by the U.S. Census (16). Hospital length of stay was coded as greater than 12 days (yes/no), the 85th percentile for length of stay. We classified the attending physician’s specialty by linking the Unique Physician Identification Number listed in Medicare Part A claims with a directory of physician-reported specialties maintained by the HCFA, as in a previous report (17). For variables with more than 3% missing values (i.e., prothrombin time and albumin), a dummy variable was created and included in the multivariate analyses.
We first compared the use of beta-blocker therapy and differences in baseline characteristics among insulin-treated diabetics, non-insulin-treated diabetics and nondiabetics. The chi-square test was used for categorical variables, and analysis of variance was used for continuous variables. We used bivariate and multivariate logistic regression analyses to determine if diabetics were less likely to receive beta-blocker therapy at discharge than nondiabetics. The analysis was repeated among patients who were not on beta-blocker therapy before admission for the AMI.
In the second part of the analysis, we estimated the association between the prescribed use of beta-blockers at discharge and one-year survival in an unadjusted model and after adjusting for potential confounders with a Cox proportional hazards model. The assumption of proportionality was evaluated graphically and found to be satisfactory. We then evaluated whether the association between beta-blockers and one-year mortality differed among diabetics and nondiabetics in a combined interaction model.
Finally, we evaluated whether beta-blocker therapy was associated with increased rates of hospital admission for diabetic complications. We used log-linear Poisson regression models to estimate the hospital readmission rate, adjusting for prescribed beta-blocker use and other potential confounders. Among patients who were treated and not treated with beta-blockers, we examined the rates of readmission for all diabetic complications as well as readmissions for acute diabetic events only.
Of the 115,015 eligible patients (65 years of age or older, confirmed AMI, survived the hospitalization, no terminal illness and not transferred), 69,707 patients (61%) had one or more strong contraindications to beta-blocker therapy (Table 1). Our final study sample included 45,308 elderly AMI patients with no relative contraindications to beta-blocker therapy other than diabetes.
Of the 45,308 patients in the study sample, 3,347 (7.4%) were diabetics who were prescribed insulin at discharge, 8,370 (18.5%) were diabetics who were not prescribed insulin at discharge and 33,591 (74.1%) were nondiabetics. Insulin and non-insulin-treated diabetics were younger on average than nondiabetics. As expected, diabetics had higher rates of hypertension, prior AMI, congestive heart failure, peripheral vascular disease and renal dysfunction compared with nondiabetics (Table 2).
Comparison of patients receiving and not receiving beta-blockers
Of the 45,308 patients eligible for beta-blocker therapy, 22,665 (50%) had a beta-blocker prescribed as a discharge medication. Characteristics associated with patients receiving a prescription for beta-blockers at discharge have been described in a previous article (18). Younger age, white race, hypertension and higher left ventricular ejection fraction were characteristics associated with higher receipt of beta-blockers. Of note, patients prescribed calcium channel blockers at discharge were less likely to receive beta-blocker therapy.
Prescribed use of beta-blockers by diabetes status
There were significant differences in the prescribed use of beta-blocker therapy by diabetes status: 45% of insulin-treated diabetics, 48.2% of non-insulin-treated diabetics and 51% of nondiabetics received beta-blockers (p < 0.001). However, these differences were mediated, in part, by differences in clinical characteristics between the three groups. After adjusting for demographic and clinical factors, both non-insulin-treated diabetics and insulin-treated diabetics continued to be less likely to be prescribed beta-blocker therapy (odds ratio [OR] 0.88, 95% confidence interval [CI] 0.82 to 0.96; and OR 0.93, 95% CI 0.88 to 0.98, respectively) than nondiabetics. Of the 36,794 patients not on beta-blocker therapy before admission, 16,006 (43.5%) were prescribed therapy at discharge. There were also significant differences in the initiation of beta-blocker therapy by diabetes status: beta-blockers were prescribed for 37.0% of the insulin-treated diabetics, 41.1% of the non-insulin-treated diabetics and 44.8% of the nondiabetics (p < 0.001). In the adjusted analysis, both non-insulin-treated diabetics and insulin-treated diabetics were less likely to receive beta-blockers at discharge (OR 0.95, 95% CI 0.89 to 1.01; and OR 0.90, 95% CI 0.82 to 0.99, respectively) than nondiabetics.
Association with survival
Overall, patients prescribed beta-blocker therapy at discharge had a one-year mortality rate of 7.7% compared with 12.6% for patients who were not prescribed the therapy (p < 0.001). Beta-blockers were associated with increased survival among insulin-treated diabetics (12.5% vs. 17.8%, p < 0.001), non-insulin-treated diabetics (8.9% vs. 15.2%, p < 0.001) and nondiabetics (7% vs. 11.3%, p < 0.001). The differences in survival observed were due in part to the higher severity of illness in diabetic patients.
In analyses that adjusted for demographic and clinical confounders, the prescribed use of beta-blockers was associated with lower risk of mortality for non-insulin-treated diabetics (hazard ratio [HR] = 0.77, 95% CI 0.67 to 0.89) and nondiabetics (HR = 0.88, 95% CI 0.81 to 0.95). While not statistically significant, the estimate of the effectiveness of beta-blocker therapy for insulin-treated diabetics was identical to that for nondiabetics (HR = 0.89, 95% CI 0.73 to 1.08). A combined regression model that included diabetics and nondiabetics found no significant interactions between beta-blocker effectiveness and type of diabetes.
Association with readmission
The six-month readmission rates for diabetic complications were similar both for patients who were and who were not prescribed beta-blockers among insulin-treated diabetics (2.3% vs. 3.5%, p = 0.06), non-insulin-treated diabetics (0.8% vs. 1.3%, p = 0.07) and nondiabetics (0.04% vs. 0.06%, p = 0.60). There was no significant increase in the risk of readmission for diabetic complications for patients prescribed beta-blockers in adjusted analyses (risk ratio [RR] = 0.72, 95% CI 0.44 to 1.19 for insulin-treated diabetics; RR = 0.81, 95% CI 0.51 to 1.28 for non-insulin-treated diabetics; and RR = 0.80, 95% CI 0.27 to 2.34 for nondiabetics).
The rate of readmission for acute diabetic events alone (hypoglycemia, hyperglycemia or diabetic coma) was not higher for patients on beta-blocker therapy versus those who were not among insulin-treated diabetics (1.2% vs. 2.3%, p = 0.02), non-insulin-treated diabetics (0.5% vs. 0.7%, p = 0.27) and nondiabetics (0.04% vs. 0.05%, p = 0.99). Beta-blockers were not associated with increased risk for readmission for acute diabetic events among insulin-treated diabetics (RR = 0.52, 95% CI 0.27 to 1.00), non–insulin-treated diabetics (RR = 0.82, 95% CI 0.43 to 1.56), and nondiabetics (RR = 1.14, 95% CI 0.35 to 3.70). In models that combined diabetics and nondiabetics, we did not find significant interactions between diabetes and beta-blockers for readmission for diabetic complications.
The prescribed use of beta-blockers was significantly associated with lower one-year mortality among non-insulin-treated diabetics, and although not statistically significant, beta-blocker therapy was found to have similar effectiveness for insulin-treated diabetics and nondiabetics. Neither insulin-treated nor non-insulin-treated diabetics were at increased risk of readmission for diabetic complications; both groups, however, were significantly less likely to be prescribed beta-blocker therapy compared with nondiabetics.
There are at least two explanations why the relationship between beta-blockers and lower mortality was not statistically significant for insulin-treated diabetics. One possibility is that our population was too small to demonstrate a survival benefit in this group. However, the fact that the effect sizes of beta-blocker therapy in insulin-treated diabetics and nondiabetics were similar suggests that treatment was beneficial. The lack of interaction between beta-blockers and insulin use also suggests that beta-blocker therapy is effective for insulin-treated diabetics.
Alternatively, it is conceivable that the benefit from beta-blocker therapy is smaller for insulin-treated diabetics. Beta-blockers and insulin have similar effects on metabolic mechanisms that may be protective after AMI, such as reducing serum levels of fatty acids. During times of ischemia, fatty acids and their metabolites have been implicated in aggravating myocardial injury due to their chemical toxicity and increased oxygen demand (19). Both beta-blockers (20)and insulin-glucose infusions (21)have been shown to decrease plasma-free fatty acid levels. Thus, it is possible that decreased benefit of beta-blockers observed in the insulin-treated diabetics of our study may be due to some degree of overlap of beneficial mechanisms. This would be consistent with the findings from the DIGAMI study, where in multivariate analysis, insulin appeared to be less effective in diabetic patients on beta-blocker therapy after AMI (22).
The finding that most diabetics appear to benefit from beta-blocker therapy after AMI is consistent with results from subgroup analyses of randomized controlled trials focusing on diabetic patients (10–12). All of these studies demonstrated that beta-blocker therapy was associated with greater relative reduction in mortality for diabetics compared with nondiabetics, although their cohorts were younger on average than those in the CCP. In the Beta-Blocker Heart Attack Trial (BHAT), where 55 out of 4,163 patients had diabetes, there was a 35% reduction in 25-month mortality among diabetic patients and a 25% reduction among nondiabetic patients (10). In the Norwegian Multicenter Timolol Study, 99 of 1,884 patients had diabetes. In a mean follow-up period of 17 months, timolol treatment was associated with reduction in cardiac mortality of 63% in diabetic patients and 34% in nondiabetic patients (11). In the Göteborg Metoprolol Trial, in which 120 of 1,395 patients were diabetic, the reduction of three-month mortality was greater in diabetics (58%) than nondiabetics (30%) (12).
Our findings are also consistent with other non-randomized cohort studies that suggest beta-blockers are effective in reducing long-term mortality from AMI (10,23).
Clinical studies have found beta-blockers to be generally safe in diabetic patients. In a prospective study of 150 insulin-treated diabetics, the incidence of loss of consciousness as a result of hypoglycemia was the same for those subjects receiving and not receiving beta-blocker therapy, approximately 1 out of 50 patients taking beta-blockers (24). Beta-blockers have also been found to be a safe treatment for hypertension in older diabetic patients. A recent study from the United Kingdom Diabetes Study Group found that beta-blockers reduced the risk of microvascular and macrovascular complications in type 2 diabetics, without an increased risk for hypoglycemic events or worsened serum lipid profiles (25). In another study of over 13,000 elderly diabetics who were concurrently using insulin or sulfonylureas, beta-blocker therapy was not significantly associated with higher rates of serious hypoglycemia requiring hospitalization (26).
Our findings are consistent with those of a study using the CCP data set that examined the use of beta-blockers in patients with relative contraindications (27); however, the estimates of the effectiveness of beta-blockers differ. The risk reduction associated with beta-blockers for diabetics in the study by Gottlieb et al. (27)was 36% (95% CI 31% to 40%), greater than that found for this study. However, the study by Gottlieb et al. (27)included patients with relative contraindications to beta-blockers (such as heart failure and chronic obstructive pulmonary disease) without adjustment for the severity of these conditions. For example, diabetic patients who received beta-blockers were less likely to suffer from severe heart failure. Because patients with severe heart failure were both less likely to receive beta-blockers and more likely to die than patients with less severe disease, the outcomes may be biased towards the finding that beta-blockers have benefit. In contrast, this study restricted the analysis to patients without contraindications to therapy, reducing subsequent confounding and obtaining estimates of beta-blocker benefit similar to those reported in the clinical trials (1).
This study has several limitations. First, findings from retrospective studies are subject to limitations due to their nonexperimental design. Although our analysis controlled for a wide range of demographic, clinical, laboratory and functional characteristics, we cannot exclude the possibility that the benefit associated with beta-blocker therapy was a result of residual confounding. Because our ability to identify contraindications to beta-blocker therapy and adjust for confounding risk factors was limited by the information documented in the medical charts, there may have been unknown factors associated with improved survival that were also associated with the use of beta-blockers. Second, although the use of beta-blockers was based on retrospective chart review, we do not have information regarding long-term compliance with therapy in the year after discharge and may have misclassified the long-term patterns of use in some patients. However, this misclassification would bias the relationship between the use of beta-blockers and improved survival towards the null. Third, our ability to determine whether patients were diabetic was based on documentation in the medical charts rather than from serum glucose or hemoglobin A1c values. We were able to indirectly confirm the validity of diabetes status from medication use, and in analyses where diabetes status was defined by use of hypoglycemic agents or insulin, our results were similar to analyses wherein diabetes status was based on chart-documentation. Fourth, while beta-blockers were not associated with higher risk for readmission, our study was unable to assess whether therapy was associated with episodes of hypoglycemia or worsening diabetic control that did not lead to hospitalization. Lastly, we cannot exclude the possibility that the diabetic patients who were able to tolerate beta-blockers were the ones who received therapy, thus selecting diabetics who would benefit the most from treatment.
The findings from this study, combined with those from clinical trials, suggest that beta-blockers are at least as effective for diabetics as for nondiabetics at no increased risk for diabetic complications. Even if the relative reduction in mortality from beta-blocker therapy were the same for diabetics and nondiabetics, beta-blockers would still have the potential to save more lives because the baseline mortality is higher in diabetics than nondiabetics (28). The results from this study may serve to alleviate physicians’ concerns about prescribing beta-blockers to diabetic patients after AMI and help ensure that more of these high-risk patients have the opportunity to benefit from their use.
We are indebted to all the health care professionals, hospitals, and organizations that contributed to the development and implementation of the Cooperative Cardiovascular Project, and to Ms. Maria Johnson for her outstanding editorial assistance.
☆ Mr. Chen is a Merck/American Federation for Aging Research Student Scholar in Geriatric Pharmacology and an American Heart Association Student Scholar in Cardiovascular Disease and Stroke. Dr. Krumholz is a Paul Beeson Faculty Scholar.
↵1 The analyses upon which this publication is based were performed under Contract No. 500-96-P549, entitled “Utilization and Quality Control Peer Review Organization for the State of Connecticut,” sponsored by the Health Care Financing Administration, Department of Health and Human Services. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products or organizations imply endorsement by the U.S. Government. The authors assume full responsibility for the accuracy and completeness of the ideas presented. This article is a direct result of the Health Care Quality Improvement Program initiated by the Health Care Financing Administration, which has encouraged identification of quality improvement projects derived from analysis of patterns of care, and therefore required no special funding on the part of this Contractor. Ideas and contributions to the author concerning experience in engaging with issues presented are welcomed.
- American College of Cardiology
- American Heart Association
- acute myocardial infarction
- Cooperative Cardiovascular Project
- Health Care Financing Administration
- lactate dehydrogenase
- Received January 4, 1999.
- Revision received May 20, 1999.
- Accepted June 28, 1999.
- American College of Cardiology
- Mills G.A,
- Horn J.R
- Deacon S.P,
- Karunanayake A,
- Barnett D
- Ryan T.J,
- Anderson J.K,
- Antman E.M,
- et al.
- Kjekshus J,
- Gilpin E,
- Cali G,
- Blackey A.R,
- Henning H,
- Ross J Jr.
- Gundersen T,
- Kjekshus J
- Malmberg K,
- Herlitz J,
- Hjalmarson A,
- Ryden L
- Bureau of the Census
- Day J.L,
- Metcalfe J,
- Simpson C.N
- Stanley W.C,
- Lopaschuk G.D,
- Hall J.L,
- McCormack J.G
- Malmberg K,
- Ryden L,
- Hamsten A,
- Herlitz J,
- Waldenstrom A,
- Wedel H
- Barnett A.H,
- Leslie D,
- Watkins P.J
- UK Prospective Diabetes Study Group