AJCC
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

American Journal of Critical Care. 2009;18: 124-131 doi:10.4037/ajcc2009193
Copyright © 2009 by the American Association of Critical-Care Nurses.
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Respond to This Article
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Right arrow Take the CE Test
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Engoren, M.
Right arrow Articles by Arslanian-Engoren, C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Engoren, M.
Right arrow Articles by Arslanian-Engoren, C.

CE Article

Long-term Survival in the Intensive Care Unit After Erythrocyte Blood Transfusion

By Milo Engoren, MD and Cynthia Arslanian-Engoren, RN, PhD, ACNS-BC. Milo Engoren is an anesthesiologist and intensivist in the Departments of Anesthesiology and Internal Medicine at St Vincent Mercy Medical Center and a clinical associate professor in the Department of Anesthesiology at the University of Toledo Health Sciences College in Toledo, Ohio. Cynthia Arslanian-Engoren is an associate professor in the School of Nursing at the University of Michigan in Ann Arbor.

Corresponding author: Milo Engoren, MD, 2213 Cherry St, Toledo, OH 43608 (e-mail: engoren{at}pol.net).


   Abstract
Top
 Abstract
Methods
 Results
 Discussion
 References
 
Background Erythrocyte blood transfusions are commonly used in intensive care units, yet little is known about their effects on long-term survival.

Objective To determine the effect of erythrocyte blood transfusion in intensive care units on long-term survival.

Methods Retrospective analysis of a prospectively collected database of 2213 patients admitted January 27, 2001, to April 30, 2002, to the cardiac, burn, neurological-neurosurgical, and combined medical-surgical intensive care units in a tertiary care, university-affiliated, urban medical center. Further analysis was done on a case-control subgroup (n = 556) formed by matching scores on the Acute Physiology and Chronic Health Evaluation (APACHE) II and propensity scores.

Results Although transfusion was univariably associated with increased risk of death at all 3 times (0–30, 31–180, and >180 days after admission to the unit), multivariable adjustment with Cox modeling showed that transfusion had no association with mortality for the first 2 intervals (0–30 and 31–180 days), but was associated with a 25% lower risk of death (hazard ratio, 0.75; 95% confidence interval, 0.57–0.99; P = .04) in patients who survived at least 180 days after admission to the unit. In the case-control patients, after correction for APACHE II risk of death and propensity to receive a transfusion, transfusion had no association with mortality for the first 2 intervals, but was associated with 29% lowered risk of death (hazard ratio, 0.71; 95% confidence interval, 0.50–0.99; P=.046).

Conclusion Blood transfusion was associated with a decreased risk of late (>180 days) death in intensive care patients.

Notice to CE enrollees:A closed-book, multiple-choice examination following this article tests your understanding of the following objectives:
  1. Describe the effect of erythrocyte blood transfusion in the intensive care unit on long-term survivability.
  2. Discuss the differences between the Cox proportional hazard modeling and the Kaplan-Meier survival curve.
  3. Integrate findings of the research study into nursing practice in the intensive care unit.
To read this article and take the CE test online, visit www.ajcconline.org and click "CE Articles in This Issue." No CE test fee for AACN members.

Although erythrocyte transfusions are commonly used in intensive care units (ICUs) to treat anemia, support hemodynamic status, minimize ischemic organ damage, and prevent mortality,1–3 evidence of the effectiveness of such transfusions is conflicting. Researchers have reported both increased1,2,4 and decreased5,6 short-term mortality with transfusion. In an observational study1 of 3534 patients from 146 ICUs in western Europe, 28-day mortality was increased in the patients who received transfusions. This increased risk persisted after organ dysfunction and propensity to receive a transfusion were controlled for.1 Similarly, in an observational study2 of 4892 patients from 284 ICUs in the United States, mortality was higher in the patients who had received a transfusion. In a Canadian study4 of 838 patients randomized to a restrictive or a liberal transfusion policy, 28-day mortality did not differ, but overall mortality was higher in the younger and less critically ill patients.

Conversely, in a retrospective study5 of 78 974 Medicare beneficiaries 65 years or older who were hospitalized with acute myocardial infarction, transfusion improved 30-day mortality if a patient’s hematocrit was less than 30% when the patient was admitted to the hospital. Similar improved results were found in another retrospective study,6 in which transfusion was associated with decreased ICU mortality. These beneficial effects were most pronounced in patients with anemia, high scores on the Acute Physiology and Chronic Health Evaluation (APACHE) II, or a cardiac diagnosis.6 Most recently, a retrospective, propensity-matched analysis of the Sepsis Occurrence in Acutely Ill Patients study indicated that 30-day survival rates were higher in patients who had received transfusions than in those who had not.7 These different effects on short-term mortality may depend on the patient’s age, severity and type of illness, and the hemoglobin level that triggered transfusion.

The effects of transfusion may be longer lasting, however, and few studies have examined this subject. In 2 studies8,9 in cardiac surgical patients, long-term survival was decreased in patients who received perioperative transfusions. However, in a study10 of patients undergoing surgery for hip fractures, researchers found no association between transfusion and long-term survival. These limited studies may have little applicability to ICU patients, who are often more critically ill than are cardiac surgery or hip surgery patients. The purpose of this study is to evaluate the effects of transfusion in the ICU on long-term survival.


   Methods
Top
 Abstract
 Methods
Results
 Discussion
 References
 
This retrospective study of a prospectively constructed database was approved by the appropriate institutional review board and was carried out according to the ethical standards set forth in the Helsinki Declaration of 1975. The ICU database was reviewed for all patients admitted to the cardiac ICU, the burn ICU, the neurological and neurosurgical ICU, and the combined medical-surgical ICU at St Vincent Mercy Medical Center between January 27, 2001 (the initiation of the database) and April 30, 2002. Although the ICUs predominantly accept the types of patients described in the names of the units, each unit serves overflow patients from the other 3 units. No cardiac surgical patients are cared for in these ICUs, and many of the data and variables collected for the ICU database are not collected on the cardiac surgery patients, so those patients were not included in this study.


Forty-four percent of ICU patients are transfused with little evidence of the effect on long-term survival.

 

The database (Table 1Go) included demographic data, admitting service and diagnoses, laboratory values at admission, and APACHE II scores (calculated from the worst physiological variables within 24 hours after ICU admission) and the resultant predicted risk of death. Data on processes and procedures performed on these patients during their stay in the ICU also were included. The database was compiled by a specially trained ICU registered nurse and was completed concurrently with each patient’s ICU stay. The database had been recently updated with dates of death (through January 23, 2007) where applicable (http://ssdi.rootsweb.com). The minimum follow-up was 4.74 years and the maximum was 5.99 years. For patients who were admitted more than once to any ICU during the course of the data collection, only the first admission was used for analysis; however, the occurrence of readmission was used as a possible variable in mortality models.


View this table:
[in this window]
[in a new window]

  		Table 1 Patients’ characteristics and processes of carea

 

Guidelines for transfusions include hemoglobin less than 8 g/dL and anemia with coronary artery disease.

 

Patients who had received packed erythrocytes, which were routinely leukoreduced, were compared with patients who had not received erythrocytes. Whole blood transfusions are not used at St Vincent Mercy Medical Center. General guidelines for erythrocyte transfusions are blood loss greater than 20% of blood volume, hemoglobin level less than 8 g/dL, anemia with coronary artery disease, chronic obstructive pulmonary disease or cerebrovascular disease, as well as normovolemic anemia with signs and symptoms.

The {chi}2 test and the Fisher exact test were used to compare categorical variables. A t test was used to compare normally distributed continuous variables. Cox proportional hazard modeling was used to determine the predictors of mortality. In Cox modeling, the risk of death is assumed to be constant across the time interval,11 so a Kaplan-Meier survival curve of all patients was inspected to determine where changes in risk of death occurred. Then each time interval with constant risk of death was analyzed separately by using Cox modeling.

Models were analyzed further by using a case-control method, in which the control patients (who did not receive a transfusion) were matched to the case patients (who did receive a transfusion) with respect to APACHE II scores and propensity to receive a transfusion.12,13 Because the decision to administer a transfusion was not random and physicians’ and patients’ choices determined who received a transfusion, propensity analysis was used to find equivalent patients to match those patients who had received a transfusion. The propensity to receive a transfusion or not to receive a transfusion was calculated by using a nonparsimonious binary logistic regression model. All variables in Table 1Go except for predicted risk of death based on APACHE II scores were entered into the binary logistic regression. This process produced a number between 0 and 1, which is the likelihood or propensity for a transfusion to be administered. Then patients who had received transfusions were matched to patients who had not received transfusions in a 2-step process. First, patients were matched by APACHE II scores predictive of risk of death to within 1 percentage point. Second, they were matched by propensity score, by using a greedy algorithm that found the closest match within a tolerance of ±0.01. If a match could not be found that met both criteria, that patient who had received a transfusion was excluded from further analysis. The Cox models were then repeated on this subpopulation of matched patients who had and had not received a transfusion.

All Cox models were constructed by using the 25 variables from the database (Table 1Go). Variables were entered into the Cox models by using forward selection with P < .05 and a 95% confidence interval required for statistical significance (SPSS 13.0, SPSS, Inc, Chicago, Illinois).

The power analysis was based on 50% mortality during the 5-year follow-up and 20% of patients receiving a transfusion in the ICU. In order to detect a 20% difference in the hazard ratio while achieving a power of 80% with a 2-sided significance level of .05, 2200 patients were needed for analysis.14


   Results
Top
 Abstract
 Methods
 Results
Discussion
 References
 
Of the 2213 patients, 404 (18%) received transfusions. A total of 1042 patients (47%) were dead at follow-up. Compared with patients who lived, patients who died were more likely to have received blood transfusions. They were also sicker: they were more likely to be tracheally intubated on arrival in the ICU and to have central venous catheters, pulmonary artery catheters, hemodialysis, continuous venovenous hemofiltration, and mechanical ventilation. Patients who died were also more likely to have a cardiac arrest in the ICU or to be reintubated. They were older, had lower scores on the Glasgow Coma Scale, and had higher serum levels of urea nitrogen and creatinine but lower hemoglobin levels. As expected, they also had higher APACHE II scores and thus higher risks of dying in the hospital (Table 1Go).

Kaplan-Meier survival analysis showed 2 elbows in the survival curve, at 30 and 180 days, where the risk of dying changed (Figure 1Go). A total of 366 patients (17%) had died by 30 days (Table 2Go). Compared with patients who lived, patients who died were more likely to have received transfusions: 101 of these 366 patients (28%) had received blood transfusions (P < .001). However, when Cox modeling was used to control for the differences in morbidities and processes of care between patients who received a transfusion and patients who did not, transfusion was no longer a predictor of death within 30 days of admission to the ICU (Table 3Go). Another 198 of the surviving 1847 patients (11%) died between 31 and 180 days (Table 2Go). Again, patients who died were more likely to have received a transfusion (49 of 198, 25%; P < .001). After other factors were corrected for, Cox modeling revealed no association between transfusion and mortality (Table 4Go). Of the remaining 1649 patients, 478 (29%) died after 180 days (Table 2Go). Of these 478 patients, 126 (26%) had received blood transfusions (P < .001). After other risk factors were corrected for, Cox modeling showed that blood transfusion was associated with a 25% lower risk of death (Table 5Go, Figure 2Go).


Figure 1
View larger version (18K):
[in this window]
[in a new window]

  		Figure 1 Kaplan-Meier survival curve of all 2213 patients shows 2 elbows, at 30 days (thin arrow) and 180 days (thick arrow), where the risk of dying changes. Vertical hash marks represent censored data. Actual follow-up of patients ranged from 4.74 to 5.99 years (1730–2187 days).

 

View this table:
[in this window]
[in a new window]

  		Table 2 Status of patients (dead or alive) from arrival in intensive care unit to follow-up

 

View this table:
[in this window]
[in a new window]

  		Table 3 Cox model of risk of dying for the 366 of 2213 patients (17%) who died within 30 days of admission to the intensive care unit (ICU)

 

View this table:
[in this window]
[in a new window]

  		Table 4 Cox model of risk of dying for the 198 of the surviving 1847 patients (11%) who survived at least 30 days and then died by 180 days after admission to the intensive care unit

 

View this table:
[in this window]
[in a new window]

  		Table 5 Cox model of risk of dying for the 478 of the surviving 1649 patients (29%) who survived at least 180 days after admission to the intensive care unit and then died after 180 days

 

Figure 2
View larger version (17K):
[in this window]
[in a new window]

  		Figure 2 Cox model of all 1649 patients who survived at least 180 days after admission to the intensive care unit, shown at the mean of the covariates (254 patients who received a transfusion, dashed line; 1395 patients who did not receive a transfusion, solid line; P = .04).

 

Patients transfused were more likely to die, but were also sicker.

 

We then used propensity matching to correct further for the fact that transfusion was not a random event, but was controlled by the physicians and was at least partially dependent on a patient’s medical condition, which might also be a predictor of death, and was not based solely on hemoglobin level. Of the 404 patients who received a transfusion, 278 (69%) were each matched to a patient who had an equal APACHE II score predictive of the risk of death and an equal propensity for receiving a transfusion but who did not actually receive one. Those patients who received a transfusion and the control patients with whom they were matched were similar (all P > .05) for all variables. However, patients included in the case-control part of the study had higher APACHE II risks of dying (mean, 28%; SD, 20%; vs mean, 19%; SD, 22%; P < .001) and were older (mean age, 64 years; SD, 16; vs mean age, 59 years; SD, 18; P < .001) than the patients who were not included.

Among the 278 case-control matched pairs, patients who received a transfusion had mortality rates for the first 30 days after ICU admission similar to those of patients who did not receive a transfusion (deaths: 52 who received a transfusion vs 67 who did not; P = .12). In those patients who survived at least 30 days after ICU admission, mortality rates, again, did not differ for the next 150 days (days 31–180, deaths: 31 who received a transfusion vs 36 who did not; P = .07). However, among those patients who survived at least 180 days after admission to the ICU, transfusion was associated with fewer deaths (deaths: 63 patients who received a transfusion vs 74 who did not; P = .008). Cox modeling indicated that transfusion was not associated with mortality for the intervals from 0 to 30 days and 31 to 180 days. However, in those patients who survived at least 180 days, transfusion was associated with a 29% lowered risk of mortality: 0.71 (95% confidence interval, 0.50–0.99; P = .046; Figure 3Go).


Figure 3
View larger version (17K):
[in this window]
[in a new window]

  		Figure 3 Cox model of propensity-matched patients who survived at least 180 days after admission to the intensive care unit, shown as the mean of the covariates (195 patients who received a transfusion, dashed line; 175 patients who did not receive a transfusion, solid line; P = .046).

 

   Discussion
Top
 Abstract
 Methods
 Results
 Discussion
References
 
We found that administration of a blood transfusion to patients in the ICU was associated with improved survival in those patients who survived at least 180 days after admission to the ICU. This result was confirmed when matching of APACHE II and propensity scores was used to control for differences in the characteristics of the patients.

Although we found no studies that evaluated the effect of transfusion on long-term survival in ICU patients, some investigators did evaluate long-term survival in surgical patients and others evaluated short-term survival in ICU populations. Our results of no effect of transfusion on short-term (0–30 and 31–180 days) survival in ICU patients are consistent with results of studies in orthopedic patients10,15,16 and of some studies in cancer patients.17,18 However, in other studies in surgical cancer patients and in studies of patients undergoing cardiac surgery, the risk of mortality was increased, for at least 1 year, in patients receiving a transfusion.8,9,19

Our ICU population contained no cardiac surgery patients, and thus those studies in cardiac surgery patients are not directly comparable.8,9,19 Our results of decreased mortality in patients who received a transfusion are similar to results from a recent study.7 Vincent et al7 found a decreased risk of death within 30 days of ICU admission (hazard ratio, 0.73; 95% confidence interval, 0.59–0.90; P = .004) in a propensity-matched group of ICU patients with sepsis. In our study, mortality did not decrease until at least 180 days after ICU admission, and unlike Vincent et al,7 we found no decrease in mortality in the first 30 days. This difference may be related to the greater statistical power in the study by Vincent et al (1642 propensity-matched patients vs our 556) or to the use of patients with the single disease category of sepsis compared with our diverse population of patients.

One advantage of our study is the long-term follow-up, a minimum of 4.74 years, compared with 30 days in the study by Vincent et al.7 The deleterious effects of transfusion may be limited to specific types of patients (eg, patients who have had cardiac surgery and some patients with cancer) and not occur in orthopedic patients or in the diverse population of patients in our study. Conversely, the beneficial effects of transfusion may also be limited to specific types of patients (eg, patients with sepsis or ischemic heart disease). Because our study did not have sufficient statistical power for us to evaluate this topic by diagnostic groups or ICU type, further study is recommended.

In both our total population and the group of case-control matched patients, transfusion was associated with an improvement in survival that occurred only after 180 days and had no effect on survival for 0 to 30 and 31 to 180 days. This result differs from the results of an observational, propensity-matched study1 in which transfusion was associated with increased 28-day mortality (22.7% vs 17.1%; P = .02) and from the results of the randomized study by Hebert et al,4 in which the risk of death within 30 days of admission to the ICU was increased in patients who received a transfusion. However, the association reported by Hebert et al was not present in patients who were more than 55 years old or had APACHE scores greater than 20. In subgroup analysis of patients with severe ischemic heart disease, Hebert et al3 found that transfusion was associated with a higher, but statistically nonsignificant, absolute survival rate.

A total of 30% of our patients were admitted to the coronary care unit; almost all of these had ischemic heart disease. Many of the patients admitted to other ICUs were elderly and may have had ischemic heart disease, which is common in ICU patients.20 Differences in proportions of patients with ischemic heart disease may account for the differences in our results. Patients with anemia in the ICU are at risk for adverse cardiac events,21 and studies in both cardiac surgery and coronary angioplasty patients have indicated that even small elevations in cardiac enzyme levels are associated with increased late mortality that persists for several years.22,23


Transfusion may be beneficial in myocardial ischemia, but may be neutral or harmful without it.

 

Results of 2 observational studies5,6 support the benefits of transfusion. Wu et al5 found that transfusion was associated with a significant reduction in mortality if the hematocrit at admission was 30% or less. Hebert et al6 found that patients with a cardiac diagnosis had a dose-dependent lower mortality associated with transfusion. However, these results were contradicted by those of another observational study,24 in which transfusion was associated with increased mortality in patients with acute coronary syndrome if the nadir hematocrit was greater than 25%.

These conflicting results may reflect different transfusion triggers or characteristics of patients. Transfusions may be harmful at a high transfusion level, but may be beneficial if delayed until the hemoglobin level is lower. Similarly, transfusion may be beneficial in patients at risk for myocardial ischemia but may have no effect or be harmful in patients without myocardial ischemia. Further study is needed to clarify these conflicting results. Further research is also needed to determine why the beneficial effects in our study do not occur for at least 180 days after ICU admission. We cannot compare our long-term results with results of these other studies because survival beyond 60 days was not evaluated in those studies.


Blood transfusion was associated with a decreased risk of late (>180 days) death in ICU patients.

 

Our study has several limitations. First, the database does not include any information about hemoglobin levels other than at admission. In particular, we do not have the hemoglobin levels that precipitated transfusions or the nadir hemoglobin levels in patients who did not have transfusions.

Second, we did not include transfusions given before or after the patients’ ICU stays. Although other transfusions may introduce an unknown effect, this limitation is also present in other studies of transfusion in the ICU.1,4 Third, because our study was retrospective, we could only find associations and not show causation. Other factors not included in this analysis that are associated with both transfusions and improved survival might explain the improved survival rates we found. Although we confirmed our findings by using propensity matching, propensity analysis has unknown confounders that may affect both transfusions and mortality.12,13 Fourth, this study was conducted in a single medical center. Transfusion practices at St Vincent Mercy Medical Center may differ from practices in other centers. In particular, our transfusion rate (18%) was much lower than that reported by Vincent et al1 (37%) and by Corwin et al2 (44%), even though the APACHE II scores in our patients were similar to or slightly higher than those in the patients in those studies. Our mean hemoglobin level at admission (11.4 g/dL; SD, 4.6), however, was similar to the levels (11.0 g/dL; SD, 2.4 and 11.3 g/dL; SD, 2.3) reported by those 2 groups. Because we do not have pretransfusion or "trigger" hemoglobin levels, we cannot say how much of our lower transfusion rate is related to less blood loss in the ICU, resulting in higher nadir hemoglobin levels, or to acceptance of lower hemoglobin levels before transfusion in our ICU.

A strength of the study is our strong results according to 2 different statistical techniques: traditional Cox modeling and Cox modeling of a case-control population matched for APACHE II score and propensity for transfusion.

In conclusion, we found that transfusion is associated with improved survival in patients who survive at least 180 days after ICU admission.


   ACKNOWLEDGEMENT
 
This study was presented at the 20th Annual Congress of the European Society of Intensive Care Medicine, Berlin, Germany, October 7–10, 2007.

FINANCIAL DISCLOSURES
None reported.

eLetters
Now that you’ve read the article, create or contribute to an online discussion on this topic. Visit www.ajcconline.org and click "Respond to This Article" in either the full-text or PDF view of the article.

To purchase electronic or print reprints, contact The InnoVision Group, 101 Columbia, Aliso Viejo, CA 92656. Phone, (800) 899-1712 or (949) 362-2050 (ext 532); fax, (949) 362-2049; e-mail, reprints{at}aacn.org.


   REFERENCES
Top
 Abstract
 Methods
 Results
 Discussion
 References
 

  1. Vincent JL, Baron JF, Reinhart K, et al. Anemia and blood transfusion in critically ill patients. JAMA. 2002;288(12): 1499–1507.[Abstract/Free Full Text]
  2. Corwin HL, Gettinger A, Pearl RG, et al. The CRIT Study: anemia and blood transfusion in the critically ill—current clinical practice in the United States. Crit Care Med. 2004; 32(1):39–52.[CrossRef][Medline]
  3. Hebert PC, Wells G, Martin C, et al. A Canadian survey of transfusion practices in critically ill patients. Transfusion Requirements in Critical Care Investigators and the Canadian Critical Care Trials Group. Crit Care Med. 1998;26(3):482–487.[CrossRef][Medline]
  4. Hebert PC, Wells G, Blajchman MA, et al. A multicenter, randomized, controlled clinical trial of transfusion requirements in critical care. Transfusion Requirements in Critical Care Investigators, Canadian Critical Care Trials Group. N Engl J Med. 1999;340(6):409–417.[Abstract/Free Full Text]
  5. Wu WC, Rathore SS, Wang Y, Radford MJ, Krumholz HM. Blood transfusion in elderly patients with acute myocardial infarction. N Engl J Med. 2001;345(17):1230–1236.[Abstract/Free Full Text]
  6. Hebert PC, Wells G, Tweeddale M, et al. Does transfusion practice affect mortality in critically ill patients? Transfusion Requirements in Critical Care (TRICC) Investigators and the Canadian Critical Care Trials Group. Am J Respir Crit Care Med. 1997;155(5):1618–1623.[Abstract]
  7. Vincent JL, Sakr Y, Sprung C, et al. Are blood transfusions associated with greater mortality rates? Results of the Sepsis Occurrence in Acutely Ill Patients study. Anesthesiology. 2008;108(1):31–39.[Medline]
  8. Engoren MC, Habib RH, Zacharias A, Schwann TA, Riordan CJ, Durham SJ. Effect of blood transfusion on long-term survival after cardiac operation. AnnThorac Surg. 2002; 74(4):1180–1186.[Abstract/Free Full Text]
  9. Koch CG, Li L, Duncan AI, et al. Transfusion in coronary artery bypass grafting is associated with reduced long-term survival. AnnThorac Surg. 2006;81(5):1650–1657.[Abstract/Free Full Text]
  10. Johnston P, Wynn-Jones H, Chakravarty D, et al. Is perioperative blood transfusion a risk factor for mortality or infection after hip fracture? J Orthop Trauma. 2006;20(10):675–679.[CrossRef][Medline]
  11. Hosmer DW Jr, Lemeshow S. Applied Survival Analysis. New York, NY: John Wiley & Sons; 1999.
  12. Rosenbaum PR, Rubin DB. The bias due to incomplete matching. Biometrics. 1985;41(1):103–116.[CrossRef][Medline]
  13. Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med. 1997;127(8):757–763.[Abstract/Free Full Text]
  14. Schoenfeld DA. Sample-size formula for the proportional-hazards regression model. Biometrics. 1983;39(2):499–503.[CrossRef][Medline]
  15. Halm EA, Wang JJ, Boockvar K, et al. Effects of blood transfusion on clinical and functional outcomes in patients with hip fracture. Transfusion. 2003;43(10):1358–1365.[CrossRef][Medline]
  16. Carson JL, Duff A, Berlin JA, et al. Perioperative blood transfusion and postoperative mortality. JAMA. 1998;279(3):199–205.[Abstract/Free Full Text]
  17. Nathanson SD, Tilley BC, Schultz L, Smith RF. Perioperative allogenic blood transfusions: survival in patients with resected carcinomas of the colon and rectum. Arch Surg. 1985;120(6):734–738.[Abstract/Free Full Text]
  18. Garau I, Benito E, Bosch FX, et al. Blood transfusion has no effect on colorectal cancer survival: a population-based study. Eur J Cancer. 1994;30A(6):759–764.[CrossRef]
  19. Kuduvalli M, Oo AY, Newall N, et al. Effect of perioperative red blood cell transfusion on 30-day and 1-year mortality following coronary artery bypass surgery. Eur J Cardiothorac Surg. 2005;27(4):592–598.[Abstract/Free Full Text]
  20. Lim W, Qushmaq I, Devereaux PJ, et al. Elevated cardiac troponin measurements in critically ill patients. Arch Intern Med. 2006;166(22):2446–2454.[Abstract/Free Full Text]
  21. Nelson AH, Fleisher LA, Rosenbaum SH. Relationship between postoperative anemia and cardiac morbidity in high-risk vascular patients in the intensive care unit. Crit Care Med. 1993;21(6):860–866.[Medline]
  22. Engoren MC, Habib RH, Zacharias A, et al. The association of elevated creatine kinase-myocardial band on mortality after coronary artery bypass grafting surgery is time and magnitude limited. Eur J Cardiothorac Surg. 2005;28(1):114–119.[Abstract/Free Full Text]
  23. Ricciardi MJ, Davidson CJ, Gubernikoff G, et al. Troponin I elevation and cardiac events after percutaneous coronary intervention. Am Heart J. 2003;145(3):522–528.[CrossRef][Medline]
  24. Rao SV, Jollis JG, Harrington RA, et al. Relationship of blood transfusion and clinical outcomes in patients with acute coronary syndromes. JAMA. 2004;292(13):1555–1562.[Abstract/Free Full Text]




This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Respond to This Article
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Right arrow Take the CE Test
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Engoren, M.
Right arrow Articles by Arslanian-Engoren, C.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Engoren, M.
Right arrow Articles by Arslanian-Engoren, C.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS