Minimizing Selection Bias Eliminates Transfusion’s Apparent Mortality Link in Acute MI Patients

Download this article's Factoid (PDF & PPT for Gold Subscribers)


After rigorously accounting for substantial clinical differences between patients with acute myocardial infarction (MI) who do and do not receive transfusion for anemia, the apparent association of transfusion with increased in-hospital mortality disappears, according to a study published in the August 26, 2014, issue of the Journal of the American College of Cardiology. In fact, the treatment may increase survival.

The paper is the latest in a long line of observational studies that showed opposite results, according to Sunil V. Rao, MD, of the Duke Clinical Research Institute (Durham, NC). In a telephone interview with TCTMD, he said all have been fraught with selection bias and all ended with pleas for a definitive randomized trial to finally determine whether, in this setting, transfusion should be used liberally or conservatively.

For the current study, investigators led by Adam C. Salisbury, MD, MSc, of Saint Luke’s Mid America Heart Institute (Kansas City, MO), analyzed data from 34,937 acute MI patients treated at 57 hospitals participating in the Cerner Health Facts database between 2000 and 2008.
Overall, 5.1% of patients received at least 1 packed red blood cell transfusion during hospitalization. These patients were older and had lower hemoglobin values throughout their hospital stay, longer hospitalization, and greater burden of in-hospital complications. They were also less likely to undergo diagnostic angiography or PCI or to receive medical therapies such as aspirin, thienopyridines, ACE inhibitors, and angiotensin receptor inhibitors.

In-hospital mortality was about twice as high among patients who received a transfusions as those who did not (OR 2.05; 95% CI 1.76-2.40). However, due to substantial differences in key baseline clinical characteristics between the groups, the great majority (91%) of patients could not be matched based on their propensity for transfusion.

Moreover, among the 3,108 patients able to be propensity matched (n = 1,121 transfused and n = 1,987 not transfused), red blood cell transfusion was associated with reduced in-hospital mortality (adjusted OR 0.73; 95% CI 0.58-0.92).

In additional analyses, there was an interaction between nadir hemoglobin values and mortality. Transfusion was linked to a trend toward lower mortality among patients with nadir levels below 9 g/dL, while values of at least 11 g/dL were associated with increased mortality (OR 6.28; 95% CI 2.12-18.6). After adjustment for center and patient characteristics, the trend toward lower risk for a value below 9 g/dL persisted, but the risk for patients with values of at least 11 g/dL was attenuated (adjusted OR 1.88; 95% CI 0.40-8.78). 

Variability in Transfusion Use Reflects Uncertainty About Best Strategy

Transfusion practice varied widely across hospitals. The adjusted transfusion rates ranged from 3.1% to 14.5% (median OR for transfusion 2.0; 95% CI 1.5-2.5), indicating a 2-fold variability across hospitals for randomly selected patients with identical clinical characteristics.

“The optimal threshold for blood transfusion during AMI remains a subject of debate,” the authors say, although they note that their finding of improved outcomes at a hemoglobin cutpoint below 9 g/dL is consistent with the results of several previous analyses.

They add that “[u]ntil additional data from randomized trials are available to guide practice, it seems reasonable to consider transfusion during AMI below ‘conservative thresholds.’ However, these data underscore significant uncertainty in the benefits and risks of transfusion, necessitating clinicians’ careful consideration of individual patient factors, which may influence the decision to provide a blood transfusion.”

Moreover, they say, the substantial variability among hospitals in the frequency of transfusion “likely reflects clinical uncertainty regarding the benefits and risks of transfusion during AMI and represents an important target for future research.”

In an accompanying editorial, Robert W. Yeh, MD, MSc, of Massachusetts General Hospital, and Neil J. Wimmer, MD, MSc, of Brigham and Women’s Hospital (both Boston, MA), call the study results and the authors’ discussion “a somewhat scathing rebuke of much of the previous literature on the role of transfusions among patients with coronary artery disease.”

They observe that physicians are highly selective in the patients they transfuse, and most databases cannot capture the factors governing their decisions, nor can multivariable regression analysis or propensity matching adequately control for confounding. After more than 12 publications on this issue, Drs. Yeh and Wimmer say, “we are no closer now to understanding the optimal way to treat anemia or bleeding in these patients than we were a decade ago.” 

Dr. Rao agreed, noting that thus far 2 small randomized studies have come to opposing conclusions: the CRIT trial (Cooper HA. Am J Cardiol. 2011;108:1108-1111) favored a conservative transfusion strategy and the MINT trial (Carson JL. Am Heart J. 2013;165:964-971) suggested a liberal approach.

Large Randomized Trial Needed But Difficult

Researchers face 2 main obstacles to conducting a large randomized trial, Dr. Rao said. First, transfusion rates in acute MI have declined substantially because of the current emphasis on bleeding avoidance—including growing use of radial PCI—and signals of harm from transfusion. Second, most patients do well with PCI, and the only ones considered for transfusion are generally old and frail with multiple comorbidities, which makes them difficult to enroll in a trial. 

“What the observed variability in transfusion practice means is that nobody knows what the right answer is,” Dr. Rao commented. 

In an earlier period when postprocedural bleeding was more common, clinicians often would just measure hemoglobin and if values were low transfuse patients regardless of their condition, he said. Today doctors “have gotten a little more sophisticated,” taking the overall clinical scenario into consideration, he suggested. 

First, clinicians try to reduce bleeding and preserve hemoglobin, Dr. Rao said. Then, they look at whether the anemia is causing symptoms, such as ongoing chest pain or heart failure, and use that to guide a decision regarding transfusion. 

Most important, he added, is determining the underlying cause of the anemia. In older cardiac patients that is often GI bleeding, and endoscopy can be performed either in the hospital or shortly after recovery, he noted.


1. Salisbury AC, Reid KJ, Marso SP, et al. Blood transfusion during acute myocardial infarction: association with mortality and variability across hospitals. J Am Coll Cardiol. 2014;64:811-819.

2. Yeh RW, Wimmer NJ. Blood transfusion in myocardial infarction: opening old wounds for comparative-effectiveness research. J Am Coll Cardiol. 2014;64:820-822.

Related Stories:


  • Dr. Salisbury reports receiving funding from the American Heart Association Pharmaceutical Round Table and David and Stevie Spina.
  • Dr. Yeh reports receiving funding from the National Heart, Lung, and Blood Institute and the Harvard Clinical Research Institute.
  • Drs. Wimmer and Rao report no relevant conflicts of interest.

We Recommend