Preoperative anemia is a risk factor for postoperative morbidity and in-hospital mortality in cardiac surgery. However, it is not known whether treatment of anemia before cardiac surgery by administering recombinant human erythropoietin (rhEPO) plus iron improves postoperative outcomes and decreases red blood cell transfusions in these patients. In 1998 a collection of consecutive data for patients who underwent valve replacement was initiated and the inclusion criterion was anemia. Treatment with rhEPO was given at a dose of 500 IU/kg/day every week for 4 weeks and the fifth dose 48 hours before valve replacement. During each rhEPO session, patients received intravenous iron sucrose supplementation. The intervention cohort (2006 to 2011) included 75 patients and the observation cohort was composed of 59 patients who did not receive any treatment (1998 to 2005). Multivariable logistic regression analysis showed that administration of combined therapy was independently associated with decreased postoperative morbidity (odds ratio [OR] 0.13, 95% confidence interval [CI] 0.03 to 0.59 p = 0.008) and in-hospital mortality (OR 0.16, 95% CI 0.28 to 0.95 p = 0.04) after adjusting for logistic European System for Cardiac Operative Risk Evaluation score, type of intervention, time of cardiopulmonary bypass, and year of surgery. Individually, this treatment also decreased postoperative renal failure (OR 0.23, 95% CI 0.06 to 0.88, p = 0.03). Rate of red blood cell transfusion decreased from 93% in the observation cohort to 67% in the intervention cohort as did days of hospitalization (median, 15 days, 10 to 27, versus 10 days, 8 to 14, respectively, p = 0.01 for all comparisons). In conclusion, administration of intravenous rhEPO plus iron in anemic patients before valve replacement improves postoperative survival, decreases blood transfusions, and shortens hospitalization.
Two randomized trials of anemic patients with coronary bypass surgery or valvular heart disease have demonstrated the efficacy of preoperative recombinant human erythropoietin (rhEPO) administration to decrease red blood cell (RBC) transfusion. It is not known whether this treatment improves postoperative outcomes in these patients. The most recent guidelines published in 2011 by the Society of Thoracic Surgeons and the Society of Cardiovascular Anesthesiologists include administration of rhEPO plus iron in anemic patients with level of evidence B, class IIa before cardiac surgery with the objective of decreasing RBC transfusion. Consequently, the main aim of this study was to investigate whether the combined therapy of intravenous rhEPO and iron before valve replacement (VR) improves postoperative mortality and morbidity. As a secondary objective, we evaluated the extent to which RBC transfusions were decreased in these patients.
In January 1998 the department of cardiology of Hospital Mar initiated the collection of consecutive data for a registry of patients undergoing cardiac surgery. Of 894 patients who underwent VR, only anemic patients (≥18 years old) undergoing elective heart valve surgery were included. The institutional review board of the ethics committee approved the study protocol and patients provided written informed consent. These patients were divided into 2 cohorts. The first cohort was involved in an observation study (from January 1998 to December 2005) of anemic patients whose results were previously published. This study showed increased postoperative morbidity and in-hospital mortality compared to nonanemic patients. After these results, we prospectively investigated the administration of rhEPO and iron to correct the anemia before VR and improve postoperative outcomes. This study ranged from January 2006 to 2011 and involved the second or intervention cohort.
All patients in the intervention cohort received treatment with intravenous rhEPO (Epoetin-Beta, Neo-Recormon, Roche, Germany) and intravenous iron sucrose (Venofer, Vifor-Uriach, Spain) before VR (nonrandomized). This intervention started 1 month before VR and rhEPO was given at a dose of 500 IU/kg/day every week for 4 weeks and the fifth dose 48 hours before surgery. During each rhEPO session, patients received iron supplementation (maximum dose 200 mg/day in a drip infusion for 2 hours). Dose of iron administered was calculated using the following formula: iron deficiency (milligrams) = body weight (kilograms) × (target hemoglobin [130 g/dl] − actual hemoglobin) × 0.24 + 500. Control of hemoglobin (Hb) was scheduled midway through the intervention and therapy was stopped when the Hb level reached normal values. Anemia was defined as Hb <13 g/dL for men and <12 g/dL for women according to World Health Organization criteria. Exclusion criteria included emergency surgery, isolated coronary artery bypass grafting, anemia from intestinal bleeding, and patient refusal of treatment. Clopidogrel was withheld 10 days before cardiac surgery, and vitamin K antagonist was replaced by enoxaparin 5 days before surgery.
All patients were operated on by surgeons at our reference hospital (Hospital de Sant, Pau, Barcelona, Spain) without significant changes between these periods. They used a standard cardiopulmonary bypass technique with moderate hypothermia (body temperature 33°C to 35°C). RBC transfusion was administered according to the protocol when Hb was <7 g/dl during or after cardiac surgery. Moreover, the surgical team had no knowledge of which patients were anemic or had received combined therapy before VR.
Clinical characteristics before surgery (risk factors, co-morbidities, and affected valves), aortic cross-clamping time and cardiopulmonary bypass time, number of RBC units, among others, were recorded. Definitions of risk factors and postoperative complications have been described elsewhere. Preoperative variables studied included classic cardiovascular risk factors and history of stroke, myocardial infarction, and atrial fibrillation or flutter. Functional New York Heart Association class III or IV and ejection fraction <50% were considered risk factors. Chronic obstructive pulmonary disease was defined as a risk factor when forced expiratory volume in 1 second was <75% or if there was a diagnosis previously made by a physician. All patients underwent coronary angiography with the exception of men <45 years old and women <55 years old without risk factors. For each patient, we calculated the European System for Cardiac Operative Risk Evaluation score (EuroSCORE). Hb levels were measured at the start of surgery, before initiation of cardiopulmonary bypass, every 15 minutes during cardiopulmonary bypass, and at the end of VR. Blood transfusion included those RBC units transfused in the operating-room and in the postoperative period until discharge. We divided the number of RBC transfusions received into 3 groups: no transfusion, 1 to 5 RBC units, and ≥6 RBC units. Total blood loss included bleeding in the operating room plus blood loss from thoracic drainage.
Postoperative morbidity included major adverse cardiovascular events such as heart failure, permanent cerebral vascular accident, renal failure, perioperative myocardial infarction, cardiac tamponade, reoperation, severe infection (sepsis, pneumonia, or mediastinitis), prolonged ventilation (≥24 hours), thrombosis or dysfunction of the prosthesis, and endocarditis as described previously. These major adverse cardiovascular events were analyzed as a composite outcome and separately.
Continuous variables were expressed as mean ± SD and categorical data were expressed as real number and percentage. Univariate analysis was performed using Student’s t test or Mann–Whitney U test according to distribution characteristics of variables. Categorical variables were analyzed with chi-square test or Fischer’s exact test when appropriate. Univariate analysis was used for identified risk factors with a potentially confounding effect on composite major adverse cardiovascular events and in-hospital mortality. Variables that achieved marginally statistical significance (p <0.15) in univariate analysis were selected to be included in multivariable logistic regression. Backward modeling was used to assess the independent association between combined therapy (rhEPO and iron) administration and risk for in-hospital mortality or composite major adverse cardiovascular events. Variables were removed 1 by 1 if their exclusion did not modify significantly the likelihood ratio statistics of the model. When removal of any variable changed the estimated parameters of the remaining variables by >15%, it was considered a confounding effect and that variable was retained in the model regardless of its statistical significance.
Sensitivity analysis was performed to evaluate the rhEPO and iron effect adjusted for confounders related to major differences between groups in univariate analysis. So we matched rhEPO groups for age, gender, EuroSCORE, and Hb before administration of this treatment. In these new groups multivariable logistic regression was carried out to assess the independent association between preoperative treatment and composite major adverse cardiovascular events or in-hospital mortality.
In the 2 different statistical risk adjustment methods we evaluated model discrimination using the c-index and calibration using the Hosmer–Lemeshow statistic. No sample size/power was calculated before the study. This is an ongoing cohort and we expect to increase the cohort in the future. Our data had a power of 80% to detect a significant odds ratio (OR) of ≤0.30.
All statistical tests with 2-sided p values <0.05 were considered statistically significant. Statistical analysis was performed with R, a language and environment for statistical computing (R Foundation for Statistical Computing, Vienna, Austria).
We enrolled 134 patients divided into 2 cohorts. The observation cohort included 59 patients and the intervention cohort contained 75 patients. This combined therapy increased the mean Hb level from 11.2 ± 1 g/dl at baseline to 12.6 ± 0.9 g/dl before VR (p <0.001; Table 1 ) and no side effects related to treatment were observed. Clinical characteristics, risk factors, and co-morbidities between the 2 cohorts are presented in Table 1 .
|Variables||All Patients||Observation Cohort||Intervention Cohort||p Value|
|(n = 134)||(n = 59)||(n = 75)|
|Age (years)||72 ± 9||71 ± 8||73 ± 10||0.16|
|Women||89 (66%)||42 (71%)||47 (63%)||0.30|
|Initial hemoglobin (g/dL)||11 ± 0.9||10.9 ± 0.9||11 ± 1||0.12|
|Preoperative hemoglobin (g/dL)||11.8 ± 1.3||10.9 ± 0.9||12.6 ± 0.9||<0.001|
|Diabetes||40 (30%)||12 (20%)||28 (37%)||0.03|
|Hypertension||95 (71%)||41 (70%)||54 (72%)||0.75|
|Hypercholesterolemia||64 (48%)||23 (39%)||41 (55%)||0.07|
|Current smoking||19 (14%)||9 (15%)||10 (14%)||0.16|
|New York Heart Association class III or IV||71 (53%)||35 (59%)||36 (48%)||0.19|
|Ejection fraction <50%||14 (11%)||5 (9%)||9 (13%)||0.37|
|Atrial fibrillation||31 (24%)||12 (22%)||19 (27%)||0.56|
|History cerebral vascular accident||4 (3%)||1 (2%)||3 (4%)||0.62|
|Previous cardiac surgery||8 (6%)||4 (7%)||4 (5%)||0.99|
|Serum creatinine >1.5 mg/dl||18 (14%)||8 (14%)||10 (14%)||0.98|
|Chronic obstructive pulmonary disease||57 (59%)||23 (70%)||34 (53%)||0.11|
|Body surface area (m 2 )||1.7 ± 0.2||1.67 ± 0.1||1.72 ± 0.2||0.15|
|Weight (kg)||68 ± 12||67 ± 12||70 ± 10||0.28|
|Pulmonary hypertension >60 mm Hg||23 (22%)||10 (31%)||13 (18%)||0.12|
|EuroSCORE||9 ± 3||8.2 ± 2.6||8.9 ± 2.9||0.15|
|Aortic occlusion time||73 ± 29||62 ± 27||80 ± 29||0.001|
|Extracorporeal bypass time||110 ± 40||114 ± 37||110 ± 42||0.12|
|Mitral||24 (18%)||17 (29%)||7 (9%)|
|Aortic||90 (67%)||35 (59%)||55 (73%)|
|Double||20 (15%)||7 (12%)||13 (17%)|
|Coronary bypass||34 (25%)||11 (19%)||23 (31%)||0.11|
|Blood loss, median (quartiles 1–3)||1,210 (810–1,680)||1,240 (790–1,800)||1,210 (810–1,680)||0.61|
The main difference between the 2 cohorts was a longer aortic cross-clamping time in the intervention cohort (80 ± 29 vs 62 ± 26, p = 0.001; Table 1 ). However, the nadir on cardiopulmonary bypass was significantly greater in this cohort (7.2 ± 1.1 vs 6.3 ± 1.2 g/dl, p <0.001) as shown in Figure 1 . Consequently, 33% of patients in the intervention cohort were not given any RBC units compared to only 7% in the observation cohort (p <0.001; Figure 2 ). Moreover, there was a significant correlation between nadir of Hb and RBC transfusions (r = 0.56, p <0.001).
Postoperative outcomes in unadjusted analysis showed a significant decrease in composite major adverse cardiovascular events in the intervention group (40% vs 79% observation group, p <0.001). Individual adverse postoperative outcomes that decreased significantly were acute renal failure (29% vs 54%, p = 0.004), severe infection (8% vs 24%, p = 0.01), heart failure (25% vs 44%, p = 0.01), and prolonged ventilation (15% vs 29%, p = 0.02). We did not observe any differences between cohorts for other postoperative complications.
In univariate analysis, in-hospital mortality was lower in the intervention cohort compared to the observation cohort (9% vs 23%, p = 0.04), which represents a 13% decrease in absolute risk of mortality. In the intervention group there were 7 deaths (5 women and 2 men): 5 deaths were secondary to heart failure (2 because of right ventricular dysfunction). The fifth death was due to an inability to come off pump in the operating room, and the remaining death was due to a dysfunction of the aortic prosthesis when the patient came off pump. In the control cohort there were 14 deaths (12 women and 2 men). Causes of death were severe infection in 5 patients, heart failure in 4, permanent stroke in 2, reintervention for bleeding in 2, and respiratory failure in 1 patient.
These improvements in morbidity and mortality in the intervention group were reflected in a significantly shorter postoperative hospital stay (intervention cohort, median 10 days, quartiles 1 to Q3 8 to 14, vs observation cohort, median 15 days, 10 to 27, p = 0.01).
Preoperative clinical characteristics, risk factors, and EuroSCORE (8.4 ± 2.8 vs 9.4 ± 2.6, p = 0.17) were similar in patients who died and in those who survived. However, patients who died had a longer cardiopulmonary bypass time (137 ± 65 vs 104 ± 32 minutes, p = 0.002) and received more RBC units than those who survived. All patients who did not receive any transfusions survived (28 patients), whereas 11 of 21 patients who were given ≥6 RBC units died (p <0.001). There were no differences in Hb levels during the operation between patients who survived and those who died, except for the nadir of Hb. Patients who survived had a significantly higher Hb nadir than those who died (6.9 ± 1.2 vs 6.3 ± 0.9 g/dl, p = 0.01).
In contrast, in nonanemic patients the mortality rate decreased during the study time, but to a lesser extent than in anemic patients: mortality in the first period (1998 to 2005) was 7% and decreased to 4% in the second period (2006 to 2011, p = 0.13).
Risk factors for postoperative adverse outcomes identified in univariate analyses were entered in multivariable logistic regression as confounding variables. After adjusting for several co-morbidities and year of surgery, treatment of anemia before VR was independently associated with a significant decrease in composite major adverse cardiovascular events and in in-hospital mortality. Furthermore, when postoperative complications were analyzed separately, this therapy proved to have an independent association with decreased postoperative renal failure as presented in Table 2 .