Author + information
- Received January 12, 2004
- Revision received February 24, 2004
- Accepted March 11, 2004
- Published online August 4, 2004.
- Eugenia Nikolsky, MD, PhD⁎,†,
- Eve D. Aymong, MD, FACC⁎,†,
- Amir Halkin, MD⁎,†,
- Cindy L. Grines, MD, FACC‡,
- David A. Cox, MD, FACC§,
- Eulogio Garcia, MD∥,
- Roxana Mehran, MD, FACC⁎,†,
- James E. Tcheng, MD, FACC¶,
- John J. Griffin, MD, FACC#,
- Giulio Guagliumi, MD**,
- Thomas Stuckey, MD, FACC††,
- Mark Turco, MD, FACC‡‡,
- David A. Cohen, MD, FACC§§,
- Manuela Negoita, MD⁎,†,
- Alexandra J. Lansky, MD, FACC⁎,† and
- Gregg W. Stone, MD, FACC⁎,†,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Gregg W. Stone, Cardiovascular Research Foundation, 55 East 59th Street, 6th Floor, New York, New York 10022
Objectives We sought to investigate the impact of anemia in patients with acute myocardial infarction (AMI) undergoing primary percutaneous coronary intervention (PCI).
Background The prognostic importance of anemia on primary PCI outcomes is unknown.
Methods In the Controlled Abciximab and Device Investigation to Lower Late Angioplasty Complications (CADILLAC) trial, 2,082 patients of any age with AMI within 12 h onset undergoing primary PCI were randomized to balloon angioplasty versus stenting, each ± abciximab. Outcomes were stratified by the presence of anemia at baseline, as defined by World Health Organization criteria (hematocrit <39% for men and <36% for women).
Results Anemia was present in 260 (12.8%) of 2,027 randomized patients with baseline laboratory values. Patients with versus without baseline anemia more frequently developed in-hospital hemorrhagic complications (6.2% vs. 2.4%, p = 0.002), had higher rates of blood product transfusions (13.1% vs. 3.1%, p < 0.0001), and had a prolonged (median 4.1 vs. 3.5 days, p < 0.0001) and more expensive (median costs $12,434 vs. $11,603, p = 0.002) index hospitalization. Patients with versus without anemia had strikingly higher mortality during hospitalization (4.6% vs. 1.1%, p = 0.0003), at 30 days (5.8% vs. 1.5%, p < 0.0001), and at 1 year (9.4% vs. 3.5%, p < 0.0001). The rates of disabling stroke at 30 days (0.8% vs. 0.1%, p = 0.005) and at 1 year (2.1% vs. 0.4%, p = 0.0007) were also significantly higher in patients with anemia. By multivariate analysis, anemia was an independent predictor of in-hospital mortality (hazard ratio, 3.26; p = 0.048) and one-year mortality (hazard ratio, 2.38; p = 0.016).
Conclusions Anemia at baseline in patients with AMI undergoing primary PCI is common, and is strongly associated with adverse outcomes and increased mortality.
Early and late prognosis of patients with acute myocardial infarction (AMI) is principally dependent on three major variables: left ventricular function, susceptibility to ventricular arrhythmias, and residual ischemia after reperfusion (1–3). Reduction in the hemoglobin concentration may compromise oxygen supply to infarcted or ischemic myocardium, which may promote arrhythmias, worsen hypotension, and increase infarct size (4–6). Two prior studies in patients with AMI treated conservatively suggest that early and late mortality may be increased in patients presenting with anemia (7,8). No data are available regarding the impact of anemia on outcomes of patients undergoing primary percutaneous coronary intervention (PCI) for AMI. Therefore, we examined the database from the large, multicenter randomized Controlled Abciximab and Device Investigation to Lower Late Angioplasty Complications (CADILLAC) trial to determine the effect of anemia on early and late outcomes of patients with AMI undergoing primary PCI.
The CADILLAC protocol, inclusion and exclusion criteria, and principal results have been reported in detail elsewhere (9). Briefly, 2,082 patients of any age with AMI within 12 h onset undergoing primary PCI were randomized to balloon angioplasty versus stenting with the Multilink stent (Guidant Corp., Santa Clara, California), each with versus without abciximab (Centocor, Malvern, Pennsylvania). With regard to hemorrhagic risk, patients were excluded with a history of bleeding diathesis, major surgery within the preceding six weeks, gastrointestinal or genitourinary bleeding within six months, cerebrovascular event within two years, or any permanent residual neurological defect, a history of leukopenia, thrombocytopenia, hepatic or renal dysfunction, or recent administration of a thrombolytic agent. Anemia at baseline was not an exclusion criterion, however, and baseline laboratory values were not required to be available before enrollment and performance of PCI.
The primary composite end point included death from any cause, reinfarction, repeat target vessel revascularization (TVR) as a result of ischemia, or disabling stroke after the index procedure. The definitions of the primary end point have been previously described (9). Anemia was defined using World Health Organization (WHO) criteria as a hematocrit value at initial presentation <39% for men and <36% for women (10).
Clinical follow-up at 30 days, 6 months, and 1 year were performed in all patients. All events were adjudicated by an independent clinical events committee. Medical care costs were assessed for all U.S. patients using previously validated methodology (11). Detailed resource utilization data were recorded for all revascularization procedures, and procedural costs were based on measured resource utilization and 2001 unit costs. Nonprocedural hospital costs were estimated by converting hospital charges to costs according to hospital and cost-center specific cost-to-charge ratios. Costs for inpatient and outpatient physician services were calculated using the 2001 Massachusetts Medicare fee schedule.
Categorical variables were compared with the Fisher exact test. Continuous variables are presented as medians with interquartile ranges, and were compared using the Kruskal-Wallis test. Survival data was estimated by the Kaplan-Meier method and compared by the log-rank test. Multivariate analysis of predictors of death at several time periods was performed using Cox proportional hazards regression with stepwise selection using entry and exit criteria of p < 0.1. The candidate variables entered in the model included age, gender, diabetes mellitus, hypertension, hypercholesterolemia, current smoking, history of prior myocardial infarction or coronary artery bypass graft surgery, Killip class ≥2 creatinine clearance, left anterior descending artery as an infarct vessel, triple-vessel disease, treatment with abciximab or stent, time from the symptom onset to the first balloon inflation, and baseline anemia (as a binary variable) or baseline hematocrit (as a continuous variable). Separate models were then generated adding in baseline medication use. All analyses were two-sided, and significance was established at the 0.05 level.
Baseline clinical characteristics
A baseline hematocrit before PCI was available in 2,027 (97.4%) of the 2,082 patients enrolled in the CADILLAC trial, 260 (12.8%) of whom had anemia. As seen in Table 1,patients with anemia were older, more frequently female, and had smaller body mass index. Pre-existing diabetes mellitus, hypertension, peripheral arterial disease, and past gastrointestinal bleeding were all significantly more common in patients with anemia. Congestive heart failure at the time of presentation was also more frequent in anemic patients.
Procedural and in-hospital results
Pre- and post-PCI lumen dimensions, procedural success rates, and final Thrombolysis In Myocardial Infarction (TIMI) flow grades were similar in patients with and without anemia (Table 2),as were peak creatine phosphokinase levels (median 1,339 vs. 1,413 U/l, respectively, p = 0.13). Patients with versus without anemia more frequently developed moderate or severe hemorrhagic complications in-hospital (6.2% vs. 2.4%, p = 0.002), and had higher rates of blood product transfusions (13.1% vs. 3.1%, respectively, p < 0.0001). Partly as a result, the index hospitalization was significantly prolonged in patients with anemia (4.1 [3.0, 6.5] vs. 3.5 [2.8, 5.1] days, p < 0.0001), and in-hospital costs were significantly greater ($12,434 [$10,158, $16,412] vs. $11,603 [$9,587, $14,631], p = 0.002).
As seen in Table 3and Figure 1,rates of mortality (all-cause, cardiac, and noncardiac) and disabling stroke were markedly higher in patients with baseline anemia compared with those without anemia, during the in-hospital phase and at 30 days and 1 year. Mortality was also higher in patients with baseline anemia between the time of discharge and one year (6.2% vs. 2.4%, p = 0.001). In patients with anemia, the majority of disabling strokes at one year were ischemic (1.7%), and the minority (0.1%) were hemorrhagic. In patients without anemia, stroke was hemorrhagic in 0.1% of patients, ischemic in 0.1% of patients, and unclassified in 0.1% of patients.
Mortality was significantly increased in both men and women with anemia (Fig. 2A);the increase in disabling stroke with anemia was significant in women, and a similar trend was present in men (Fig. 2B). There were no differences in the rates of reinfarction or recurrent ischemia necessitating repeat TVR between anemic and nonanemic patients.
Considering baseline clinical and angiographic variables, anemia was identified as an independent predictor of in-hospital and one-year mortality by multivariate analysis (Table 4).Similarly, when baseline hematocrit as a continuous variable rather than anemia was introduced into the regression model, hematocrit was also a significant predictor of one-year mortality (hazard ratio = 1.06 [95% confidence interval 1.01 to 1.12], p = 0.042). When added to these models, neither nadir hematocrit value, absolute decrease in hematocrit, or blood transfusions correlated with mortality. There was no statistically significant interaction between baseline hematocrit and other variables (age, gender, prior myocardial infarction, prior coronary bypass surgery, hypertension, diabetes, hypercholesterolemia, smoking, peripheral vascular disease, congestive heart failure, and creatinine clearance) when specifically assessed with an interaction term in either in-hospital or in one-year multivariable models.
At admission, patients with anemia were less likely to be treated with aspirin than those without anemia. At discharge, 30-day, and 1-year follow-up, significantly fewer patients with baseline anemia were treated with aspirin, beta-blockers, and statins (Table 5).While more patients with anemia were prescribed angiotensin-converting enzyme inhibitors or receptor blockers at discharge, use of this class of drugs was similar in patients with and without anemia at 30 days and 1 year. When admission medication use was added to the multivariate models, baseline anemia continued to be an independent predictor of mortality during both the in-hospital period (hazard ratio = 3.36 [95% confidence interval 1.10 to 11.20], p < 0.05) and at 1 year (hazard ratio = 2.38 [95% confidence interval 1.18 to 14.80], p = 0.016).
Prior studies have suggested a link between baseline hematocrit and adverse outcomes in patients with cardiac disease and AMI treated conservatively. In a retrospective analysis of 78,974 Medicare beneficiaries, lower baseline hematocrit values were associated with higher 30-day mortality rates after AMI; only 10% of patients were treated by PCI in this series, however (7). Similarly, one-year mortality was higher in anemic patients in a report from the Myocardial Infarction Data Acquisition System (8). The association between baseline hematocrit and outcomes after primary PCI have not been previously examined.
In the present large trial of contemporary mechanical reperfusion strategies in AMI, anemia, as defined by WHO criteria, was common, occurring in 12.8% of patients. As patients with hemorrhagic diatheses and recent bleeding were excluded from the CADILLAC trial, the true incidence of anemia in a more general population of patients presenting for primary PCI is likely to be higher, especially given the increasing incidence of anemia as the population ages (7,12,13). Patients with anemia had an increased frequency of comorbid risk factors known to worsen prognosis after AMI, including advanced age, female gender, and diabetes mellitus (14,15). Despite similar rates of procedural success and infarct size, patients with baseline anemia had a significantly more complicated hospital course, including greater rates of hemorrhagic complications and blood product transfusion requirements. Hospitalization was significantly prolonged in patients with anemia, and costs were significantly greater. Most importantly, baseline anemia was strongly associated with an approximate three-fold increase in in-hospital, 30-day, and 1-year mortality. The impact of anemia on early and late death was present in both genders (though was especially profound in women), despite the different gender-based thresholds for anemia as established by WHO (10). Anemia was also strongly associated with an increased risk of disabling stroke, especially in women, a finding not previously reported.
The reasons for the worse outcomes in anemic patients with AMI treated with primary PCI are likely multifactorial. Anemia may by itself confer excess risk; decreased blood oxygen levels may be arrhythmogenic, directly exacerbate myocardial ischemia, activate the sympathetic nervous system, and result in cerebral hypoperfusion (4–6). Alternatively, anemia was a covariate with other high-risk features known to reduce survival after AMI. Patients with anemia were older, more commonly women, and had a higher prevalence of cardiovascular risk factors including diabetes and chronic renal insufficiency, factors known to worsen the prognosis of patients after AMI and primary PCI (14–19). Peripheral arterial disease, which is a marker of diffuse atherosclerosis (20,21), was more prevalent in patients with anemia. Clinical congestive heart failure was also significantly more common at presentation in patients with anemia, although left ventricular ejection fraction was not measured to be different. Even after multivariate adjustment for the important confounders, anemia remained an independent predictor of in-hospital and one-year mortality. However, multivariate models cannot entirely account for the baseline differences, and adjusted effects may still be influenced by residual confounding. Additional clinical measurements not included in this analysis (e.g., certain comorbidities or inflammatory markers) may also interact with anemia.
Differences in pharmacologic treatment may also have contributed to the worse late prognosis in patients with anemia. Specifically, anemic patients were less likely to be treated with aspirin, beta-blockers, and statins, all of which are known to prolong survival after AMI (22–25). When differences in medication use at the time of admission were considered in the multivariate model, aspirin continued to be an independent determinate of mortality both in-hospital and at one year. These models are unable to correct for postdischarge medication use, however, as approximately one-half of the deaths in patients with anemia occurred before hospital discharge. Notably, almost one-fifth of patients with anemia (18.8%) were not receiving aspirin at one-year follow-up, presumably due to their increased incidence of preadmission gastrointestinal bleeding, their greater rate of hemorrhagic events during the index hospitalization, and possible concerns about recurrent bleeding. Considering the impact of anemia on late survival in hospital survivors, the decreased mortality between the time of discharge and one year in patients with a reduced baseline hematocrit may be, at least partly, explained by the lower rate of aspirin prescription at discharge. Of note, prior studies demonstrating increased late mortality in patients with anemia and cardiovascular disorders did not examine the extent to which differences in medication usage may have contributed to the worse prognosis with this condition (8,26).
This post-hoc analysis was not prespecified and should, thus, be considered hypothesis-generating, and complementary to large, prospectively collected observational databases. The causes of anemia in the patients in our study were not known, and the prognostic importance of anemia may vary with different etiologies. The indications for and timing of blood product transfusions were not protocol-specified but, rather, left to local standards of care; whether a policy of early routine blood transfusion in anemic patients would improve their prognosis is unknown. Finally, no information was available regarding the correction of anemia at follow-up, or whether additional hemorrhagic events occurred.
Conclusions and clinical implications
Baseline anemia in patients with AMI undergoing primary PCI should be recognized as an important risk factor for prolonged hospitalization, increased hospital costs, and adverse short-term and late outcomes, including death and disabling stroke. Medications known to improve survival in AMI patients, including enteric-coated aspirin, beta-blockers, and statins should not be withheld from this high-risk population if at all possible. Further study is warranted to determine whether early red blood cell transfusions would improve the prognosis of patients with anemia and AMI undergoing primary PCI.
The authors are grateful to Prof. Stuart J. Pocock for his statistical review of this manuscript.
- Abbreviations and acronyms
- acute myocardial infarction
- Controlled Abciximab and Device Investigation to Lower Late Angioplasty Complications trial
- percutaneous coronary intervention
- Thrombolysis In Myocardial Infarction
- target vessel revascularization
- World Health Organization
- Received January 12, 2004.
- Revision received February 24, 2004.
- Accepted March 11, 2004.
- American College of Cardiology Foundation
- Berning J.,
- Steensgaard-Hansen F.V.,
- Appleyard M.
- Stone G.W.,
- Grines C.L.,
- Browne K.F,
- et al.
- Levy P.S.,
- Kim S.J.,
- Eckel P.K,
- et al.
- Stone G.W.,
- Grines C.L.,
- Cox D.A,
- et al.
- WHO Scientific Group
- Cohen D.J.,
- Taira D.A.,
- Berezin R,
- et al.
- Dallman P.R.,
- Yip R.,
- Johnson C.
- Fleming D.J.,
- Jacques P.F.,
- Tucker K.L,
- et al.
- Chung M.K.,
- Bosner M.S.,
- McKenzie J.P.,
- Shen J.,
- Rich M.W.
- Mak K.H.,
- Moliterno D.J.,
- Granger C.B,
- et al.
- Anderson R.D.,
- White H.D.,
- Ohman E.M,
- et al.
- Leng G.C.,
- Lee A.J.,
- Fowkes F.G,
- et al.
- Lee A.J.,
- Lowe G.D.,
- Woodward M.,
- Tunstall-Pedoe H.
- A randomized trial of propranolol in patients with acute myocardial infarction. I. Mortality results. JAMA 1982;247:1707–14.
- Horwich T.B.,
- Fonarow G.C.,
- Hamilton M.A.,
- MacLellan W.R.,
- Borenstein J.