Anemia for Risk Assessment of Patients With Acute Coronary Syndromes
Article Outline
In patients admitted with acute coronary syndromes, those with anemia are at higher risk. However, current risk score systems do not take into account the presence of anemia. The impact of anemia on mortality was studied, and its incremental predictive value was evaluated. Demographic, clinical, and biologic characteristics at admission, as well as treatments and mortality, were recorded for 1,410 consecutive patients with acute coronary syndromes. The incremental value of adding anemia information to risk score evaluation was determined using changes in the appropriateness of Cox models when anemia was added. Anemia was detected in 381 patients (27%). They were older, had more co-morbidities, had higher Global Registry of Acute Coronary Events (GRACE) risk scores, received fewer guideline-recommended treatments, and, as a result, had 4-fold higher mortality. When included in a prediction model based on the GRACE risk score, anemia remained an independent predictor of mortality. The addition of anemia improved both the discriminatory capacity and calibration of the models. According to the GRACE risk score, the population was divided into 4 groups of different risk levels of <1%, 1% to <5%, 5% to <10%, and ≥10%. The addition of anemia to the model made it possible to reclassify 9%, 43%, 47%, and 23% of patients into the different risk categories, respectively. In conclusion, our data confirmed that anemia was an independent predictive factor of mortality and had incremental predictive value to the GRACE score system for early clinical outcomes.
The increased risk of worse short- and long-term clinical outcomes associated with low hemoglobin has previously been documented in various conditions.1, 2, 3, 4, 5, 6, 7, 8, 9 Low hemoglobin or anemia was frequently observed in patients with acute coronary syndromes (ACSs),10 particularly (up to 45%) in elderly patients with ACSs.9 In patients with ACSs, anemia was associated with higher risks of in-hospital and 30-day mortality.11, 12 Recent guidelines for the management of non–ST-elevation ACSs consider risk stratification as an important step and recommend the use of established risk scores, such as the Global Registry of Acute Coronary Events (GRACE) risk score,13 to categorize patient risk at admission.12, 14 The aim of this study was to assess the impact of anemia on in-hospital and 30-day mortality in patients admitted for acute myocardial infarction and determine the incremental predictive value of anemia to established prognosis parameters.
Methods
The study population was part of the Registre Franc Comtois des Syndromes Coronariens Aigus, an ongoing prospective ACS registry, that included all patients admitted for ACSs in any of the 10 cardiology centers in the region of Franche-Comté in Eastern France with a population of 1.2 million inhabitants. All patients gave informed consent. ACSs included ST-elevation myocardial infarction and non–ST-elevation myocardial infarction according to the European Society of Cardiology universal definition published in 2000.15 A dedicated team of data managers was available to assist with completion of the data.
Automated analyzers were used for hematologic measurements. Baseline hemoglobin was available for 98% of patients. Anemia was defined according to the World Health Organization definition (admission hemoglobin <13 g/dl in men and <12 g/dl in women).16 Variables for estimating the GRACE risk score (age, history of chronic heart failure, history of myocardial infarction, heart rate, systolic blood pressure, ST-segment depression, creatinine, increased cardiac enzymes, and in-hospital percutaneous coronary intervention) were prospectively recorded, as well as demographic data, previous medication and diseases, clinical presentation, and treatment given during hospitalization. In-hospital mortality was recorded and survivors were contacted at 1 month through telephone contact or a scheduled consultation to assess survival (all causes of death were considered).
Categorical variables were presented as number of cases and percentage; non–normally distributed variables, as median and interquartile range; and continuous variables, as mean ± SD. The 1-month survival probability was presented using Kaplan-Meier curves stratified on anemia. The association between variables and mortality was assessed using the Cox proportional hazard model, including all variables potentially related to mortality or using only the GRACE score. To assess the incremental value of adding anemia information, we compared changes in the appropriateness of these models when anemia was added. As recommended by Cook,17 the different approaches used were (1) changes in the Bayes information criterion (BIC) and Akaike information criterion (AIC); lower values for the BIC and AIC indicated better fit; (2) changes in indexes of calibration (Hosmer-Lemeshow p value) and discrimination (C statistics); (3) graphically comparing the observed prevalence of mortality for each decile in the entire population; and (4) estimating the rate and appropriateness of patient reclassification.
Patients were classified according to GRACE score into 4 risk categories of <1%, 1% to <5%, 5% to <10%, and ≥10% in-hospital mortality rates. Risk was then recalculated using the GRACE score plus anemia information, and patients were reclassified according to the new results. The appropriateness of the reclassification was verified by comparing observed with predicted mortality obtained using the GRACE risk score, both with and without the inclusion of anemia. The reclassification was considered appropriate if the predicted mortality using the GRACE score plus anemia was closer to observed mortality than the predicted mortality using the GRACE score without anemia. All tests were 2 sided, and p <0.05 was considered significant. Analyses were performed using SAS software, version 9 (SAS Institute Inc., Cary, North Carolina).
Results
During 12 months (December 2005 to December 2006), 1,610 patients were admitted with a final diagnosis of acute myocardial infarction, 1,410 of whom had complete data and 30-day clinical follow-up. The distribution of hemoglobin was normal (Figure 1), with a mean of 13.5 ± 1.8 g/dl and median of 13.7 (interquartile range 12.4 to 14.8). Anemia was detected in 381 patients (27%).
Figure 1.
Distribution of hemoglobin (hemo) at admission in the study population.
Patients with anemia were older and had more previous co-morbidities, had more cardiovascular risk factors, and more frequently had renal dysfunction. They presented more frequent hemodynamic instability, with low systolic blood pressure, higher Killip class, more cardiogenic shock, higher GRACE risk score, and higher troponin, B-type natriuretic peptide, and high-sensitivity C-reactive protein. A higher proportion of patients with anemia had multivessel disease and underwent coronary artery bypass grafting. Non–ST-elevation ACSs were more frequently observed in patients presenting with anemia.
Last, they less frequently received guideline-recommended treatments (Table 1, Table 2). In-hospital and 30-day mortality rates were 4 times higher in anemic patients (12.4% vs 2.8% for in-hospital and 16.3% vs 4.3% for 30-day mortality). Figure 2 shows Kaplan-Meier survival probability curves at 1 month according to the presence of anemia.
Table 1. Comparison of clinical characteristics according to the presence of anemia
| Variable | Overall Cohort (n =1,410) | Anemia (n=381) | No Anemia (n=1,029) | p Value |
|---|---|---|---|---|
| Non–ST elevation myocardial infarction | 845(60%) | 262(69%) | 583(57%) | <10−4 |
| Men | 926(65%) | 231(61%) | 695(68%) | 0.015 |
| Age⁎ (yrs) | 70±14 | 75(13%) | 66(13%) | 0.001 |
| Diabetes mellitus | 330(23%) | 126(33%) | 204(20%) | <10−4 |
| High blood pressure | 794(56%) | 253(66%) | 541(53%) | <10−4 |
| Hypercholesterolemia | 664(47%) | 167(44%) | 497(48%) | 0.13 |
| Current smoker | 359(25%) | 102(27%) | 257(25%) | 0.24 |
| Previous myocardial infarction | 263(19%) | 93(24%) | 170(17%) | 0.0007 |
| Killip class | <0.0001 | |||
| I | 1,202(85%) | 286(75%) | 916(89%) | |
| II | 127(9%) | 65(17%) | 62(6%) | |
| III | 50(4%) | 19(5%) | 31(3%) | |
| IV | 31(2%) | 11(3%) | 20(2%) | |
| Previous coronary angioplasty | 197(14%) | 58(15%) | 139(13%) | 0.41 |
| Previous coronary bypass | 70(5%) | 30(8%) | 40(4%) | 0.002 |
| Stroke | 79(6%) | 29(8%) | 50(5%) | 0.005 |
| Peripheral vessel disease | 174(12%) | 73(19%) | 101(10%) | <10−4 |
| Glomerular filtration rate (ml/min/1.73 m2) | ||||
| >60 | 899(64%) | 166(44%) | 733(71%) | <10−4 |
| 30–60 | 431(31%) | 164(43%) | 267(26%) | <10−4 |
| <30 | 80(5%) | 51(13%) | 29(3%) | <10−4 |
| Fasting glucose (mmol/L) | 6.4±1.8 | 6.6±1.4 | 6.2±1.4 | 0.02 |
| B-Type natriuretic peptide⁎ (pg/ml) | 287(104–891) | 730(283–1727) | 196(81–542) | <10−4 |
| High-sensitivity C-reactive protein⁎ (mg/dl) | 6.5(3–23) | 16.2(4.7–73) | 5.2(2.1–15) | <10−4 |
| Serum creatinine (μmol/L) | 107±61 | 129±82 | 99±51 | <10−4 |
⁎p Values from t test. |
Table 2. Comparison of clinical characteristics, treatments, and outcomes according to the presence of anemia
| Variable | Overall Cohort (n=1,410) | Anemia (n=381) | No Anemia (n=1,029) | p Value |
|---|---|---|---|---|
| Troponin T release | ||||
| At admission | 7.7±26 | 8.2±22 | 7.4±28 | 0.61 |
| At 4 h | 25±60 | 30±72 | 21±52 | 0.02 |
| At 24 h | 28±82 | 33±125 | 26±63 | 0.74 |
| Heart rate (beats/min) | 80±19 | 81±20 | 78±19 | 0.006 |
| Admission systolic blood pressure (mm Hg) | 135±30 | 131±26 | 137±29 | 0.002 |
| Left ventricular ejection fraction (angiography) | 55±13% | 53±14% | 57±13% | 0.02 |
| No. of diseased vessels | 0.0006 | |||
| 0 | 90(6%) | 19(5%) | 71(7%) | |
| 1 | 617(44%) | 133(35%) | 484(47%) | |
| 2 | 375(27%) | 107(28%) | 268(26%) | |
| 3 | 328(23%) | 122(32%) | 206(20%) | |
| Cardiogenic shock | 32(23%) | 16(4%) | 16(1.6%) | 0.003 |
| GRACE score | 133±29 | 146±33 | 126±27 | <10−4 |
| Aspirin | 1,378(98%) | 361(95%) | 1,017(99%) | <10−4 |
| Clopidogrel | 1,357(96%) | 356(93%) | 1,001(97%) | <10−4 |
| Glycoprotein IIb/IIIa inhibitors | 419/737(57%) | 105/240(44%) | 314/497(63%) | <10−4 |
| Anticoagulants | ||||
| Unfractionated heparin | 396(28%) | 164(43%) | 232(26%) | <10−4 |
| Enoxaparin | 803(57%) | 178(47%) | 625(61%) | <10−4 |
| Fondaparinux | 529(38%) | 105(28%) | 424(42%) | <10−4 |
| Angiotension-converting enzyme inhibitors or angiotensin receptor blockers | 1,230(87%) | 315(83%) | 915(89%) | 0.02 |
| Statins | 1,327(94%) | 343(90%) | 984(96%) | 0.002 |
| β Blockers | 1,103(78%) | 253(66%) | 850(83%) | <10−4 |
| Coronary angiography | 1,108(79%) | 335(88%) | 773(75%) | <10−4 |
| Reperfusion (ST-elevation myocardial infarction) | 420/565(74%) | 85/119(71%) | 335/446(75%) | 0.41 |
| Early invasive strategy (non–ST-elevation myocardial infarction) | 586/737(80%) | 155/240(65%) | 431/497(87%) | <10−4 |
| Coronary bypass | 68(5%) | 25(7%) | 41(4%) | 0.02 |
| Mortality | ||||
| In-hospital | 61(4.3%) | 47(13.3%) | 14(3.9%) | <10−4 |
| 30-day | 81(5.7%) | 61(18.5%) | 20(5.8%) | <10−4 |
Figure 2.
Kaplan-Meier curves of 30-day survival probability according to the presence of anemia at admission.
Anemia was an independent predictor of mortality, both in-hospital and at 30 days. Cox multivariable analysis showed that this relation persisted after adjustment for age, co-morbidities, condition at admission, and treatments used. When included in a prediction model based on the GRACE risk score, anemia remained an independent predictor of mortality, and irrespective of the model, the hazard ratio associated with anemia ranged from 2.1 to 2.3 for in-hospital and 30-day mortality (Table 3). The addition of anemia improved both the discriminatory capacity and calibration of the models, with increased C statistic, decreased BIC and AIC, and higher p value for the Hosmer-Lemeshow test (Table 4). Figure 3 shows the plot of observed versus predicted 30-day mortality (by deciles of risk estimation) of the GRACE score alone and the GRACE score combined with anemia information.
Table 3. Cox multivariable regression on in-hospital and 30-day mortality for anemia and components of the Global Registry of Acute Coronary Events score and fitting prediction of models
| Variable | In-Hospital Mortality | 30-Day Mortality | ||
|---|---|---|---|---|
| Hazard Ratio | p Value | Hazard Ratio | p Value | |
| Anemia | 2.1 | 0.001 | 2.3 | 0.001 |
| Age (/yr) | 1.016 | <0.001 | 1.02 | <0.001 |
| Serum creatinine (/μmol/L) | 1.004 | <0.001 | 1.004 | <0.001 |
| Non–ST- vs ST-elevation myocardial infarction | 0.40 | <0.001 | 0.39 | <0.001 |
| Admission systolic blood pressure (/mm Hg) | 0.98 | 0.008 | 0.99 | 0.002 |
| Aspirin use | 0.48 | 0.07 | 0.71 | 0.36 |
| β Blocker use | 0.37 | <0.001 | 0.46 | <0.001 |
| Angiotensin-converting enzyme inhibitor use | 0.59 | 0.04 | 0.56 | 0.009 |
| Coronary angiography during index hospitalization | 0.57 | 0.02 | 0.59 | 0.02 |
| Measures of fit of models | Without Anemia | With Anemia | Without Anemia | With Anemia |
| BIC | 978 | 958 | 1,404 | 1,269 |
| AIC | 2,207 | 1,173 | 2,929 | 1,502 |
| C Statistic | 0.853 | 0.862 | 0.851 | 0.855 |
| p Value (Hosmer-Lemeshow) | 0.19 | 0.53 | 0.21 | 0.94 |
Table 4. Cox multivariable regression for in-hospital and 30-day mortality for anemia and Global Registry of Acute Coronary Events (GRACE) score and fitting prediction of models
| Variable | In Hospital Mortality | 30 Day Mortality | ||
|---|---|---|---|---|
| Hazard Ratio | p Value | Hazard Ratio | p Value | |
| Anemia | 2.27 | <0.001 | 2.3 | <0.001 |
| GRACE risk score | 1.032 | <0.001 | 1.026 | <0.001 |
| Measures of Fit | Without Anemia | With Anemia | Without Anemia | With Anemia |
| BIC | 1,125 | 1,115 | 1,338 | 1,238 |
| AIC | 2,085 | 1,106 | 2,062 | 1,072 |
| C Statistic | 0.833 | 0.845 | 0.837 | 0.854 |
| p Value (Hosmer-Lemeshow) | 0.18 | 0.81 | 0.11 | 0.47 |
Figure 3.
Plot of predicted versus observed mortality (percent) for visualization of the reliability of the model in each decile of risk estimation. (Top) GRACE risk score alone, (bottom) GRACE risk score plus anemia risk estimation.
According to the GRACE risk score, the population was divided into 4 different risk groups of <1%, 1% to <5%, 5% to <10%, and ≥10%. Average predicted mortality rates by group were 0.8%, 1.2%, 6.0%, and 21.5%. The addition of anemia to the model led to reclassification into different risk categories of 9%, 43%, 47%, and 23% of patients, respectively (Table 5). The observed 30-day mortality in these groups was compared with the prediction, and the reclassification was mainly appropriate because the estimation of mortality was closer to actual mortality, particularly in the higher risk categories. The added prognostic value of anemia is further illustrated in Figure 4, showing that in patients with a high GRACE score, anemia allowed further discrimination. In particular, in patients in the third and fourth quartiles of GRACE score, those with anemia had a 2-fold risk of mortality compared with those without anemia.
Table 5. Comparison of observed and predicted risk (30-day mortality) using the Global Registry of Acute Coronary Events score model with and without anemia information and percentage of patients reclassified into different risk categories
| Model With Anemia | Model Without Anemia | |||
|---|---|---|---|---|
| 0%–<1% n=253 | 1%–<5% n=343 | 5%–<10% n=399 | ≥10% n=332 | |
| 0%–<1%(n = 325) | 230(91%) | 95(28%) | 0(0%) | 0(0%) |
| Actual death rate (%) | 0.8% | 0.9% | 0% | 0% |
| 1–<5% (n = 333) | 23(9%) | 197(57%) | 113(28%) | 0(0%) |
| Actual death rate (%) | 0% | 2.5% | 2.6% | 0% |
| 5%–<10% (n = 337) | 0(0%) | 51(15%) | 211(53%) | 75(23%) |
| Actual death rate (%) | 0% | 3.9% | 5.2% | 8% |
| ≥10% (n = 332) | 0(0%) | 0(0%) | 75(19%) | 257(77%) |
| Actual death rate (%) | 0% | 0% | 13.3% | 25.2% |
| Reclassified | 9% | 43% | 47% | 23% |
| Appropriately | 0 | 95(28%) | 188(47%) | 75(23%) |
| Inappropriately | 23(9%) | 51(15%) | 0 | 0 |
Figure 4.
Thirty-day mortality rates according to quartiles of GRACE risk score and the presence of anemia.
Discussion
Our data confirmed that anemia was an independent predictive factor for mortality, had incremental predictive value to the GRACE score system for early clinical outcomes, and allowed reclassification of a large proportion of patients into different risk categories.
The higher risk linked to anemia can be explained by the decrease in oxygen delivery to the myocardium18 and resulting increase in myocardial oxygen demand.19 Furthermore, it can be partially accounted for by patient characteristics. Anemic patients were often older and more frequently had co-morbidities. In the large cohort study of Sabatine et al,11 patients with anemia had an odds ratio of 2.26 for 1-month mortality or congestive heart failure. In our study, patients with anemia had significantly more co-morbidities, which could explain the higher mortality, but the impact of anemia was significant and important even after adjustment. Hazard ratios associated with anemia were 2.1 and 2.3 for in-hospital and 30-day mortality, similar to that reported by Sabatine et al,11 respectively. An additional explanation for the increased mortality risk is the higher risk of bleeding in patients with anemia.20, 21 Previous bleeding may be responsible for low hemoglobin in anemic patients, or as in our population, it may also present more predictors of bleeding, such as renal dysfunction,21, 22 older age, female gender, or hemodynamic instability.22 Even in the absence of bleeding, patients with anemia received more transfusions, which could also increase mortality.23 In addition, patients with anemia received only suboptimal treatment because they often received fewer antithrombotic treatments.20 This was also observed in our population, in which patients with anemia were less often treated with antiplatelets and anticoagulants. Last, anemia was also shown to be a predictor of contrast-induced nephrotoxicity after coronary angioplasty,24 which, combined with the older age, higher bleeding risk, and impaired renal function, may have limited the use of invasive procedures in these patients, as found in our study.
Use of the GRACE risk score is recommended in all patients with ACSs using current guidelines. In our study, the estimated risk provided by the GRACE score was close to the observed mortality rate, but the addition of anemia to the GRACE score led to further improvement in risk estimation. More importantly, risk reclassification showed that a large proportion of patients were better categorized after the introduction of anemia.
The addition of any variable to a risk scoring system increases its complexity and may discourage physicians from using it, even when recommended. Nevertheless, hemoglobin and anemia are data that are usually already available at admission in patients with ACS. Consideration of anemia at admission may not only improve ischemic risk stratification, but may also serve for estimation of bleeding risk and choice of antithrombotic treatment.12
Despite the great attention given, this study had several inherent limitations associated with cohort studies. The use of actual hemoglobin for risk assessment would likely be even better because it would make it possible to consider the increased risk associated with very high hemoglobin,11 but at the price of increased complexity of risk assessment.15 In addition, the GRACE risk score was designed to estimate the risk of in-hospital mortality or congestive heart failure, not 1-month mortality.
References
- Hemoglobin level is an independent predictor for adverse cardiovascular outcomes in women undergoing evaluation for chest pain: results from the National Heart, Lung, and Blood Institute Women's Ischemia Syndrome Evaluation Study. J Am Coll Cardiol. 2004;43:2009–2014
- . Effect of anaemia and cardiovascular disease on surgical mortality and morbidity. Lancet. 1996;348:1055–1060
- . Anemia and diabetes in the absence of nephropathy. Diabetes Care. 2005;28:1118–1123
- . Anemia is common in heart failure and is associated with poor outcomes: insights from a cohort of 12 065 patients with new-onset heart failure. Circulation. 2003;107:223–225
- . Outcomes of acute coronary syndrome in a large Canadian cohort: impact of chronic renal insufficiency, cardiac interventions, and anemia. Am J Kidney Dis. 2005;46:845–855
- . Anemia is an independent risk for mortality after acute myocardial infarction in patients with and without diabetes. Cardiovasc Diabetol. 2006;5:8
- . Effect of anemia on hemorrhagic complications and mortality following percutaneous coronary intervention. Am J Cardiol. 2007;99:1513–1517
- . Anaemia as a contributor to morbidity and mortality in congestive heart failure. Nephrol Dial Transplant. 2005;20(suppl 7):vii11–vii15
- . Blood transfusion in elderly patients with acute myocardial infarction. N Engl J Med. 2001;345:1230–1236
- . Abnormal haemoglobin levels in acute coronary syndromes. QJM. 2006;99:851–862
- . Association of hemoglobin levels with clinical outcomes in acute coronary syndromes. Circulation. 2005;111:2042–2049
- Guidelines for the diagnosis and treatment of non-ST-segment elevation acute coronary syndromes: the Task Force for the Diagnosis and Treatment of Non-ST-Segment Elevation Acute Coronary Syndromes of the European Society of Cardiology. Eur Heart J. 2007;28:1598–1660
- . Predictors of hospital mortality in the global registry of acute coronary events. Arch Intern Med. 2003;163:2345–2353
- ACC/AHA 2007 guidelines for the management of patients with unstable angina/non-ST-elevation myocardial infarction—executive summary: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Revise the 2002 Guidelines for the Management of Patients With Unstable Angina/Non-ST-Elevation Myocardial Infarction) developed in collaboration with the American College of Emergency Physicians, the Society for Cardiovascular Angiography and Interventions, and the Society of Thoracic Surgeons Endorsed by the American Association of Cardiovascular and Pulmonary Rehabilitation and the Society for Academic Emergency Medicine. J Am Coll Cardiol. 2007;50:652–726
- . Myocardial infarction redefined—a consensus document of the joint European Society of Cardiology/American College of Cardiology Committee for the Redefinition of Myocardial Infarction. Eur Heart J. 2000;21:1502–1513
- . Nutritional Anemias: Report of a WHO Scientific Group. Geneva: World Health Organization; 1968;
- . Use and misuse of the receiver operating characteristic curve in risk prediction. Circulation. 2007;115:928–935
- . Effect of a reduction in blood viscosity on maximal myocardial oxygen delivery distal to a moderate coronary stenosis. Circulation. 1986;74:1085–1092
- . Acute dilutional anemia and critical left anterior descending coronary artery stenosis impairs end organ oxygen delivery. J Trauma. 1996;41:416–423
- . Bleeding complications in patients with anemia and acute myocardial infarction. Am J Cardiol. 2005;96:1379–1383
- . Development and validation of a prognostic risk score for major bleeding in patients undergoing percutaneous coronary intervention via the femoral approach. Eur Heart J. 2007;28:1936–1945
- . Predictors of major bleeding in acute coronary syndromes: the Global Registry of Acute Coronary Events (GRACE). Eur Heart J. 2003;24:1815–1823
- . Relationship of blood transfusion and clinical outcomes in patients with acute coronary syndromes. JAMA. 2004;292:1555–1562
- . A simple risk score for prediction of contrast-induced nephropathy after percutaneous coronary intervention: development and initial validation. J Am Coll Cardiol. 2004;44:1393–1399
PII: S0002-9149(08)01877-8
doi:10.1016/j.amjcard.2008.10.023
© 2009 Elsevier Inc. All rights reserved.
