Relationship between tertiles of admission blood glucose (ABG) level and in-hospital and 60-day mortality (A) and 6-month and 1-year mortality (B): median (interquartile range), ABG level in milligrams per deciliter (to convert glucose to millimoles per liter, multiply by 0.0555): first tertile, 92 (84-97) [n=370]; second tertile, 113 (108-121) [n=376]; and third tertile, 147 (136-162) [n=376]). *In-hospital mortality-adjusted odds ratio compared with first tertile, 2.64 (95% confidence interval [CI] 1.27-5.91; P = .01). †60-Day mortality-adjusted hazard ratio compared with the first tertile, 1.52 (95% CI 1.01-2.29; P = .047).
Kaplan-Meier survival curves for patients without diabetes mellitus categorized by tertiles of admission blood glucose (ABG) levels (median [interquartile range] ABG level in milligrams per deciliter [to convert glucose to millimoles per liter, multiply by 0.0555]: first tertile, 92 [84-97] [n = 370]; second tertile, 113 [108-121] [n = 376]; and third tertile, 147 [136-162] [n = 376]) (P = .08 [log-rank test] for 60-day mortality; P = .76 [log-rank test] for 1-year mortality).
Barsheshet A, Garty M, Grossman E, Sandach A, Lewis BS, Gottlieb S, Shotan A, Behar S, Caspi A, Schwartz R, Tenenbaum A, Leor J. Admission Blood Glucose Level and Mortality Among Hospitalized Nondiabetic Patients With Heart Failure. Arch Intern Med. 2006;166(15):1613-1619. doi:10.1001/archinte.166.15.1613
The significance of admission blood glucose level in nondiabetic patients with heart failure (HF) is unknown. We examined the possible association between admission glucose levels and outcome in a large cohort of hospitalized patients with HF.
We analyzed the data of 4102 patients with HF, who were hospitalized during a prospective national survey. The present study focuses on a subgroup of 1122 nondiabetic patients with acute HF who were admitted because of acute HF or exacerbation of chronic HF.
In-hospital mortality was twice as high in patients with admission blood glucose levels in the third tertile (7.2%) compared with the first (3%) and second (4%) tertiles (P = .02). Furthermore, mortality risk was correlated with admission glucose levels; each 18-mg/dL (1-mmol/L) increase in glucose level was associated with a 31% increased risk of in-hospital mortality (adjusted odds ratio, 1.31; 95% confidence interval, 1.10-1.57; P = .003) and a 12% increase in 60-day mortality (adjusted hazard ratio, 1.12; 95% confidence interval, 1.01-1.25; P = .04). Admission blood glucose levels remained an independent predictor of in-hospital and 60-day mortality even after the exclusion of 315 patients (28%) with acute myocardial infarction and HF. The 6- and 12-month mortality rates were similar in patients with and without abnormal admission blood glucose levels.
Elevated admission blood glucose levels are associated with increased in-hospital and 60-day mortality, but not 6-month or 1-year mortality, in nondiabetic patients hospitalized because of HF.
The association between high blood glucose concentration and poor outcome has been shown in both diabetic and nondiabetic critically ill patients and patients with acute myocardial infarction.1- 7 Diabetes is an independent predictor of poor outcome in patients with heart failure (HF).8 However, the significance of high blood glucose concentration in nondiabetic patients with HF is unknown.9 Thus, we sought to investigate the consequence of high blood glucose concentration in hospitalized nondiabetic patients with HF and to test the hypothesis that elevated blood glucose levels on admission could be a prognostic marker in these patients.
Admission blood glucose (ABG) levels and baseline and admission characteristics of patients were extracted from the Heart Failure Survey in Israel (HFSIS) 2003 database. This survey, conducted in March and April 2003, included 4102 patients with a diagnosis of either acute, acute exacerbation, or chronic HF, who were admitted to 93 of 98 internal medicine departments and 24 of 25 cardiology departments in all 25 public hospitals operating in Israel. The criteria used for the diagnosis of HF were symptoms of HF (at rest or during exertion) and objective evidence of cardiac dysfunction at rest.10 Diagnosis of acute HF or exacerbation of chronic HF was determined by the attending physician based on history, clinical presentation (symptoms and physical examination), response to HF therapy, chest radiography, echocardiography, radionuclide studies, cardiac catheterization findings, and in-hospital course.10 Detailed data regarding patient characteristics, in-hospital course, management during hospitalization, prehospital and discharge medications, and diagnoses were collected and recorded on prespecified structured forms. Mortality, during the first year after index hospitalization, was assessed for 99% of patients by matching their identification numbers with the Israeli National Population Registry. The protocol was approved by the ethics committee at each of the participating hospitals.
Of 4102 patients, 2314 were admitted because of acute HF or exacerbation of HF. Patients with chronic stable HF admitted for noncardiovascular causes were excluded. Of these 2314 patients, 1192 (52%) with known diabetes mellitus were also excluded from the study. Thus, the final analysis included 1122 patients. Admission blood glucose levels were determined as the first blood glucose test on the day of admission as reported in medical records. Diabetes mellitus was defined by one of the following criteria: a history of diabetes mellitus obtained from medical records, an ABG level of 200 mg/dL or greater (≥11.1 mmol/L), or the use of antidiabetic agents (on admission or discharge). Left ventricular ejection fraction (LVEF) was determined by echocardiography. The LVEF data were collected by medical chart review and recorded only if echocardiography was performed within 1 year prior to or during hospitalization. The LVEF classes were classified as follows: normal, 50% or higher; mildly impaired, 40% to 49%; moderately impaired, 30% to 39%; and severely impaired, lower than 30%. Of 811 patients with LVEF data, echocardiography was performed in 500 (62%) during index hospitalization. The median (interquartile range [Q1-Q3]) timing of echocardiography was 0 months (0-3 months).
The end point of the study was all-cause mortality, which was obtained during follow-up either from the database itself (in hospital charts) or by matching the identification number of the patients with the Israeli National Population Register (7-day, 60-day, 6-month, and 1-year mortality). Efforts were made to ensure that the data on which the profiling is based are accurate and reliable. This includes standardizing the definitions of HF and data validation at 2 time points: first, during data entry by logical checks that were incorporated into the data entry interface and displaying error and warning signs to alert the data entry operator, and second, after data entry by batch checks that were conducted for missing values, data conflicts, and out of range values. These data conflicts and inconsistencies were resolved, and missing data were completed appropriately.
Continuous variables were expressed as median (interquartile range). Baseline characteristics of the groups (continuous data) were compared by use of the nonparametric Kruskal-Wallis test; categorical variables and frequencies were compared by means of the χ2 test. Kaplan-Meier survival curves were produced and compared using the log-rank test. To examine the relationship between ABG level and mortality, several models were conducted.
First, potential variables (identified in previously published studies9,11 as risk factors for mortality or clinical variables that were associated with in-hospital, 60-day, and 1-year mortality), were evaluated by univariate analysis and were selected based on clinical and statistical significance.
Second, multivariable analyses were conducted, whereby the number of variables selected was limited by the number of end-point events: 1 variable for every 10 end-point events. The values used to dichotomize the variables were taken from previous reports.9,11 The association between ABG level as a continuous variable and in-hospital mortality was assessed by multivariable logistic regression analysis adjusted for age, sex, New York Heart Association (NYHA) functional class III and IV, admission creatinine levels, and systolic blood pressure under 115 mm Hg.
Third, Cox proportional hazards regression analyses were performed for 60-day, 6-month, and 1-year mortality adjusted for glucose level (continuous), age (continuous), sex, NYHA functional class III and IV vs I and II, admission creatinine levels (continuous), systolic blood pressure under 115 mm Hg vs 115 mm Hg or above, sodium level less than 136 mEq/L vs 136 mEq/L or greater, hemoglobin level less than 10 g/dL vs 10 g/dL or greater, heart rate (continuous), and the following prehospital medication use: statins, β-blockers, and angiotensin converting enzyme inhibitors or angiotensin receptor blockers. Finally, the same multivariable models for in-hospital, 60-day, 6-month, and 1-year mortality, adjusted for all of the above parameters respectively plus LVEF class, were used for 811 patients with LVEF data. All analyses were performed using SAS software, version 8.2 (SAS Institute Inc, Cary, NC).
We identified 1122 nondiabetic patients who were admitted because of acute HF or exacerbation of chronic HF and divided into tertiles on the basis of their ABG levels (370 patients were in the first tertile, 376 were in the second tertile, and 376 were in the third tertile). Age, sex, and proportion of patients with ischemic heart disease were similar in the 3 subgroups (Table 1). The prevalence of hypertension, dyslipidemia, and chronic obstructive pulmonary disease were more frequent in patients with ABG levels in the second and third tertiles. In addition, admission heart rate and the number of patients with acute myocardial infarction were higher in patients with ABG levels in the third tertile. Prehospital use of statins was more frequent in patients with high ABG levels (Table 1).
At hospital discharge, prescription of diuretics, calcium channel blockers, and α-receptor blockers was similar among the 3 glucose tertiles. However, the use of statins was significantly higher in patients with high ABG levels: 23.1%, 36.6%, and 37.1% for first, second, and third tertiles of ABG level, respectively (P<.001). In addition, the use of angiotensin-converting enzyme inhibitors or angiotensin receptor blockers (61.9%, 67%, and 68.1%, respectively) and β-blocker agents (53.2%, 58.7%, 54.3%, respectively) tended to be greater in patients in the second and third ABG level tertiles (P = .30). According to inclusion criteria, no patients were given antidiabetic medications on admission or discharge.
By univariate analysis, in-hospital mortality was 2 times higher in patients with an ABG level in the third tertile (7.2%) compared with the first (4%) and second (3%) tertiles (P = .02) (Figure 1). Sixty-day mortality assessed by univariate analysis was higher by more than 35% in the third tertile (15.5%) compared with the first (11.4%) and second (10.7%) tertiles of ABG level (P = .10) (Figure 1). The adjusted odds ratio (OR) for in-hospital mortality was 2.64 (95% confidence interval [CI], 1.27-5.91; P = .01) in the third tertile of ABG level compared with the first tertile. The adjusted hazard ratio (HR) for 60-day mortality was 1.52 (95% CI, 1.01-2.29; P = .047) in the third tertile compared with the first tertile of ABG level. However, 6-month and 1-year mortality was similar in patients in all 3 tertiles of ABG level (Figure 2). In addition to ABG levels, age, NYHA functional class III and IV, admission creatinine levels, and systolic blood pressure under 115 mm Hg were independent predictors of in-hospital mortality (Table 2). Age, female sex, NYHA functional class III and IV, admission creatinine levels, systolic blood pressure under 115 mm Hg, sodium level less than 136 mEq/L, hemoglobin level less than 10 g/dL, and heart rate were independent predictors of 60-day mortality (Table 3). Predictors of 1-year mortality are listed in Table 4. Ejection fraction class was an independent predictor of 1-year mortality but not short-term mortality.
Adjusted analysis of continuous glucose levels showed that an 18-mg/dL (1-mmol/L) increase in blood glucose level was associated with an increased in-hospital mortality of 31% (adjusted OR, 1.31; 95% CI, 1.10-1.57; P = .003) and by an increased 60-day mortality of 12% (adjusted HR, 1.12; 95% CI, 1.01-1.25; P = .04).
After the exclusion of 315 patients (28%) who presented with HF in the setting of acute myocardial infarction, the association between ABG level and both in-hospital and 60-day mortality remained significant (35% [adjusted OR, 1.35; 95% CI, 1.09-1.68; P = .006] and 14% [adjusted HR, 1.14; 95% CI, 1.00-1.30; P = .05], respectively) (Table 5).
A similar association was observed in 811 patients with LVEF data. In this subgroup, after adjustment for LVEF class, elevated ABG levels were associated with higher short-term mortality. Mortality risk increased with each 18-mg/dL (1-mmol/L) increase in ABG level: 37% for in-hospital mortality (adjusted OR, 1.37; 95% CI, 1.10-1.71; P = .006) and 12% for 60-day mortality (adjusted HR, 1.12; 95% CI, 0.98-1.28; P = .11) (Table 5).
The major new finding of the present study suggests that elevated ABG level is associated with increased in-hospital and 60-day mortality in nondiabetic patients hospitalized because of HF. This association is independent of traditional risk factors for adverse outcome of HF such as age, hemoglobin level, renal function, and functional class. Our findings may have clinical implications as a potential new marker for early risk stratification and detection of a high-risk subset of patients admitted because of HF.
While several studies have documented the relationship between hyperglycemia and mortality in nondiabetic patients with acute myocardial infarction or ischemic heart disease,1- 7,12,13 the correlation between admission hyperglycemia in nondiabetic patients with HF and mortality is unknown. A recent study has shown that an 18-mg/dL (1-mmol/L) increase in glucose level was associated with a 4% increase in 2-year mortality risk in patients with acute myocardial infarction.6 In the present study, we show that elevated ABG levels are associated with increased short-term mortality in patients with and without acute myocardial infarction.
The present study is among the first to report a possible association between ABG levels and adverse outcome in nondiabetic patients with HF. In the Acute Decompensated Heart Failure National Registry (ADHERE) of patients hospitalized with a primary diagnosis of acute HF (33 046 hospitalizations), admission hyperglycemia was not tested as a potential predictor of in-hospital death.9 To our knowledge, only 2 previous studies have attempted to investigate the relationship between hyperglycemia and clinical outcome in nondiabetic patients with HF. Suskin et al14 showed that in nondiabetic patients with HF, impaired fasting glucose level was associated with more severe symptoms, worse NYHA functional class, and a shorter 6-minute walk distance. However, in contrast to our study, there were no significant differences in the number of HF hospitalizations or mortality rates between those with normal and those with elevated plasma glucose levels. Newton and Squire15 analyzed data of 528 patients hospitalized with HF and found that elevated glucose level was associated with increased all-cause mortality. In survivors of the index admission, the relationship was stronger for patients not classified as diabetic.
High ABG levels could be a marker of high-risk patients with excess stress response mediated by neurohormonal system activation, particularly cortisol and cathecholamines.2 Furthermore, impaired myocardial performance results in the activation of compensatory neurohormonal systems, including activation of the sympathetic nervous system, with the degree of sympathetic activation being proportional to the severity of ventricular dysfunction and degree of HF.16,17 Activation of the sympathetic system not only increases insulin resistance but also decreases the release of insulin from the pancreatic beta cells and increases hepatic glucose production by stimulating both gluconeogenesis and glycogenolysis.18,19
However, hyperglycemia may contribute to exacerbation of HF by several independent mechanisms. First, hyperglycemia inhibits production of nitric oxide and increases the production of reactive oxygen species in endothelial and vascular smooth muscle cells, thus impairing endothelial function.20,21 Second, hyperglycemia may be a marker of insulin deficiency, which is associated with increased lipolysis and excess circulating free fatty acids, which are toxic to ischemic myocardium and may cause damage to myocyte membranes, calcium overload, and arrhythmias.22 Third, hyperglycemia alters expression and function of sarcoendoplasmic reticulum calcium–adenosine triphosphatase (SERCA) and alters cardiac structure through posttranslational modification of extracellular matrix, both leading to impaired relaxation and increased ventricular stiffness, which exacerbates diastolic dysfunction independently of left ventricular systolic dysfunction.23- 26 Finally, hyperglycemia enhances platelet-dependant thrombosis and can accelerate atherosclerosis.27,28
We found an association between elevated ABG level and short-term (up to 2 months) but not long-term (up to 1 year) mortality in nondiabetic patients with HF. The reason for this versatile association is unknown. From our data, it is clear that nondiabetic patients admitted because of HF present a high-risk population with a 1-year mortality of 30%. In the present cohort, patients with a high ABG level died earlier (within 2 months). It is possible that hyperglycemia facilitates death in patient with HF at risk. Another possible explanation may be the differences in medical therapy after hospital discharge. The use of β-blockers, angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, and particularly statins was significantly more frequent in patients with an elevated blood glucose level. Thus, patients with a low ABG level were treated less frequently with evidence-based medications and were therefore less protected. Alternatively, hyperglycemia may be a confounder that reflects another factor of disease severity that facilitates early death in patients with HF.
Study results can be influenced by differences in disease assessment and documentation patterns at participating institutions. We do not have data on hemoglobin A1clevels and therefore were unable to comment on the differential impact of chronic vs acutely elevated blood glucose level. In addition, we have no data on neurohormones or brain natriuretic peptide levels, which could have facilitated further clarification of the relationship between acute stress, elevated blood glucose level, and HF severity. Even after adjustment, elevated blood glucose level can still be a surrogate for an unidentified confounder; thus, a causal relationship between elevated glucose levels and mortality remains unclear.
In this cohort of hospitalized nondiabetic patients with HF, elevated ABG level was associated with significant increased in-hospital and 60-day mortality. Our finding calls attention to a new prognostic marker that could be used for early risk stratification and management of patients with HF at hospital admission. It is possible that better glucose control may improve prognosis in hyperglycemic patients with HF, as demonstrated in acute myocardial infarction and critically ill patients.29- 32 Further research is needed to determine whether hyperglycemia is a marker or cause of adverse outcome and whether immediate, tight glycemic control would confer benefit and improve survival in these high-risk patients.
Correspondence: Jonathan Leor, MD, Neufeld Cardiac Research Institute, Sheba Medical Center, Tel-Hashomer 52621, Israel (firstname.lastname@example.org).
Accepted for Publication: May 15, 2006.
Financial Disclosure: None reported.
Funding/Support: The Heart Failure Survey in Israel 2003 was supported by the Israel Center for Disease Control; the Israel Medical Association; and Teva, Pfizer, MSD, Aventis, Medtronic, Dexxon, Levant, Medisson, Neopharm, Novartis, and Schering-Plough.
Disclaimer: The contents of this article do not reflect the views or policies of the Israel Center for Disease Control or the Israel Medical Association.
Acknowledgment: We thank all the physicians and nurses at the participating hospitals for their help during the course of the study.