Study Questions Harm of Blood Transfusions in AMI

August 21, 2014

KANSAS CITY, MO — A new study is casting some doubt on the mortality risks associated with acute-MI patients receiving a red blood cell transfusion, with researchers saying that previous studies showing blood transfusions increased the risk of death are likely flawed given how difficult it is to compare patients in observational studies[1].

In the newest study, which was led by Dr Adam Salisbury (St Luke's Mid America Heart Institute, Kansas City, MO) and published August 26, 2014 issue of the Journal of the American College of Cardiology, investigators observed a significant 27% relative reduction in the risk of in-hospital death among MI patients who underwent a blood transfusion in a propensity-matched analysis. While they are cautious in interpreting the data to suggest that blood transfusions reduce the risk of mortality, they say that previous reports suggesting harm might have been influenced by selection bias.

"This is something that's been controversial in the care of acute myocardial infarction patients for some time," Salisbury told heartwire . "I became interested in this after some of my own cases, where we had patients who suffered a heart attack and had been hospitalized. We opened their culprit vessel, but they also had residual coronary artery disease and residual symptoms. A couple of these patients, I remember distinctly, also had severe anemia. The question of whether or not transfuse them came up, and it was something that was hotly debated among the team."

For proponents of the transfusion, there was concern that withholding it would increase the risk of angina and result in decreased oxygen delivery to the heart muscle. For those against the transfusion, they cited papers suggesting that patients who received a blood transfusion had an increased risk of death. "We've struggled, not only in individual patient-care scenarios but also as a community of cardiologists, about the best approach with these patients," said Salisbury.

Existing Data Tough to Interpret

Despite previous studies showing that blood transfusions increased the risk of death, Salisbury said it is difficult to interpret the data because of the issue of confounding. Patients who receive packed red blood cells are usually sicker, and this has been difficult to sort out in observational studies conducted thus far.

The question of whether or not transfuse them came up, and it was something that was hotly debated among the team.

In their analysis of 34 937 acute-MI hospitalizations from 57 centers, 5.7% of patients received at least one red blood cell transfusion. As predicted, those who received the transfusion differed significantly from those who did not. For example, they were older, had lower hemoglobin values throughout their hospital stay, had more in-hospital complications, and had a longer hospital length of stay. They were also less likely to undergo angiography or PCI and were less likely to receive medical therapies such as antiplatelet agents, ACE inhibitors, or ARBs.

In the unadjusted analysis, mortality rates were significantly higher among those who received a transfusion (11.0% vs 5.7%; p<0.001). To address the imbalance in clinical characteristics, the researchers attempted to perform a propensity-matched analysis but were challenged on this front. Of the 34 937 patients in the analysis, they were able to match just 3108 patients, including 1121 who received a transfusion and 1987 who did not receive red blood cells. Compared with the unadjusted analysis, patients who received blood in the propensity-matched analysis had a 27% lower risk of death vs those who did not.

"One of the central messages of this paper, and one of the points we hope to drive home, is that you have to make sure you're comparing apples with apples in observational studies," said Salisbury. "As we showed, most of the patients who were transfused were nothing like those who weren't, including how low their hemoglobin went when they were in the hospital."

In terms of the data, Salisbury said the 27% reduction in mortality should be interpreted cautiously but "there certainly doesn't appear to be evidence of a three- to fourfold increase in mortality, which is what some of the prior studies have shown." Given this, he says the provision of blood is not a simple decision, in that there is no hemoglobin cut point that would make a transfusion automatic. Instead, physicians need to treat patients individually.

For Salisbury, an acute-MI patient who stands to benefit from a transfusion would be one with low hemoglobin levels in the setting of multiple coronary blockages that have not been revascularized with stents or surgery and who is still having ongoing evidence of ischemia. For those without symptoms, a low hemoglobin level should not be the sole factor in making a decision about whether to transfuse or not, he said.

For Dr Sunil Rao (Duke University Medical Center, Durham, NC), an interventionalist who was not involved in the study, the use of treatment strategies to prevent bleeding in the PCI setting are still a necessity, as most physicians would ultimately like to avoid the need for a transfusion. Such strategies include appropriate dosing when using unfractionated heparin and a glycoprotein IIb/IIIa (GP IIa/IIIb) inhibitor, if added. Bivalirudin (Angiomax, the Medicines Company) has been shown to reduce the risk of bleeding compared with heparin and GP IIa/IIIa inhibitors, although recent trials have challenged its advantage. Transradial PCI has also demonstrated a significant advantage over transfemoral PCI with respect to bleeding rates.

"I think trying to get patients out of the clinical situation where you would even consider a transfusion is a very firm recommendation," he told heartwire . "The question is really around those patients who, no matter what we do, are going to have a bleeding event and how to deal with that. The good news is that the population is shrinking because we're paying more attention to bleeding complications."

For Rao, strategies in place that "maintain an ischemic benefit," while minimizing the risk of bleeding, are still preferred, regardless of these new data suggesting transfusions might not be as harmful as thought.

Trouble With Observational Data

On the whole, Raopraised the analysis, saying the results also highlight the dangers of using observational analyses to guide clinical practice. "What it shows is that, depending on the database you use, the selection criteria you apply to the database, and the analytic techniques you use, you can basically get whatever answer you want."

To me, this says we really don't know what we're doing with transfusions.

He agrees with Salisbury and warns against overinterpreting the reduction in mortality observed in their trial but points out that researchers involved in previous transfusion studies, including the Conservative versus Liberal Red Cell Transfusion in Acute Myocardial Infarction (CRIT) and the Myocardial Ischemia and Transfusion (MINT) trials, also urged caution. These previous investigators all called for a randomized, controlled clinical trial. Mechanistic arguments can also be made by proponents and opponents of blood transfusions in the acute-MI setting and have been made in previously published papers, he added.

"To me, this says we really don't know what we're doing with transfusions," said Rao. "The only way we're going to resolve this is with a randomized trial, which I think is going to be very difficult to do." One of the reasons is that physicians are very wary of transfusing patients, given these previous observational studies, making randomization difficult. Data outside the cardiology setting, such as in critically ill patients, has also shown that tranfusions carry risks.

In an editorial[2], Dr Robert Yeh (Massachusetts General Hospital Boston) and Dr Neil Wimmer (Brigham and Women's Hospital, Boston, MA) say the study offers a "scathing rebuke" of the existing data on the role of blood transfusions in patients with coronary artery disease. Like Salisbury suggested, the transfused patients are often sicker in immeasurable ways compared with a patient who does not receive a transfusion.

These differences might not be adequately captured by various statistical means, including propensity matching or multiple regression analyses, which is why observational studies should be interpreted with care. As the editorialists point out, "The nonrandomized nature by which clinicians make treatment decisions—the thoughtful individualized approach that one expects from a knowledgeable physician—is precisely the factor that can doom an observational comparison that is unable to codify this thoughtfulness."

In the US, transfusion practices vary significantly across providers and institutions, note Yeh and Wimmer, likely because of the lack of strong evidence. Two small randomized trials testing liberal vs restrictive transfusion thresholds were recently conducted, with differing results, but the editorialists say the trials represent important attempts in understanding the optimal transfusion threshold for MI patients.

"Recent innovations in clinical-trial execution involving more 'pragmatic' designs (ie, streamlined studies that can leverage existing registry or electronic medical record infrastructure to capture patient baseline data and outcomes) may be ideal for a large yet relatively inexpensive randomized study of transfusion thresholds, a necessary quality for a study that is unlikely to find an industry sponsor," suggest Yeh and Wimmer.

The study was funded by a grant from the American Heart Association. Salisbury is funded in part by an award from the American Heart Association Pharmaceutical Round Table and David and Stevie Spina. Disclosures for the coauthors are listed in the article. Rao disclosed grant/research support from Ikaria and Abbott Vascular; consulting for Daiichi Sankyo/Lilly, Zoll, AstraZeneca, the Medicines Company, and Terumo; and being on the speaker's bureau for Abbott Vascular and the Medicines Company.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.