Transfusion Thresholds in the ICU: Is the Pendulum Swinging Back?

Aaron B. Holley, MD


November 03, 2016

Transfusion Thresholds

For many years, the findings of the Transfusion Requirements in Critical Care (TRICC) trial guided packed red blood cell (pRBC) administration in the intensive care unit (ICU).[1] This multicenter, randomized controlled trial (RCT) compared a conservative transfusion threshold (7.0 g/dL hemoglobin [Hgb]) with a liberal threshold (10.0 g/dL Hgb). The investigators found that outcomes were equivalent or improved (in certain subgroups) among the conservative approach group. Those with clinically significant cardiac disease showed equivalent outcomes, but the authors advised caution when applying their data to this subgroup.

The TRICC trial was designed to study only stable patients in the ICU. For specific subgroups, however, transfusion thresholds are more controversial. For example, let's consider patients with stable coronary artery disease (CAD) and acute myocardial infarction (AMI). I blogged about this back in 2013,[2] after a meta-analysis[3] and subsequent ACP Journal Club article[4] addressed transfusion and AMI. The meta-analysis included only one RCT, with a small sample. Otherwise, all studies were observational. The authors concluded that pRBC transfusion during AMI may actually be harmful, but acknowledged that a large trial was desperately needed.

Transition Thresholds in Patient Subgroups

Transfusion thresholds in stable CAD remain controversial. For many years, I extrapolated data from the TRICC trial and said that a threshold of 7 g/dL was appropriate for these patients. In TRICC, there were approximately 200 patients with stable CAD, and in subgroup analysis, there was no difference in outcomes between the two arms.[1]

Other subgroups must be considered. In 2001, Rivers and colleagues[5] showed that transfusion to a Hgb value of 10 mg/dL was beneficial for patients with septic shock. In fairness, transfusion to 10 mg/dL was one of many interventions for these patients, so any benefit attributable to this practice is difficult to quantify. Regardless, many authors recommended a target of 10 mg/dL for patients with septic shock.[6]

Another setting is acute bleeding. A recent RCT looked at trigger thresholds in patients with acute upper gastrointestinal bleeding.[7] The researchers found that a conservative threshold (7 g/dL Hgb) reduced mortality compared with a liberal threshold (9 g/dL Hgb). One caveat is that all patients in this study received gastroscopy and intervention to the site of bleeding (as appropriate) within 6 hours of presentation.

Updated Transfusion Guidelines

This month, JAMA published the American Association of Blood Banks (AABB) guidelines for pRBC transfusion and storage.[8] These guidelines update those published by the same organization in 2012.[9] The authors summarized findings across multiple subgroups using data published between 1950 and May 2016.

The guidelines include several interesting findings. Not surprisingly, they concluded that a conservative threshold of 7 g/dL was not associated with increased 30-day mortality. Although the absolute difference in mortality was 3 fewer deaths per 1000 patients, the 95% confidence interval ranged from 15 fewer deaths to 18 more deaths. There were no differences in outcomes other than mortality. Surprisingly, they did not find an increased risk for infection among patients in the liberal threshold group. This was in contrast to the results of a previously published meta-analysis.[9,10]

The guidelines compared Hgb levels of 7 g/dL and 8-9 g/dL and found no statistical difference between the groups. The caveat is that the patients studied were different. Most patients in the studies of 7 g/dL Hgb were critically ill, whereas those in the 8- to 9-g/dL Hgb groups were heterogeneous in degree of illness. Of note, patients undergoing orthopedic or cardiac surgery and those with stable CAD tended to be in the 8- to 9-g/dL Hgb groups. Even though an indirect comparison showed no difference, the AABB recommends a transfusion threshold of 7 g/dL for critically ill and hospitalized patients but 8 mg/dL for patients undergoing orthopedic/cardiac surgery and those with stable CAD.

The AABB was unable to provide recommendations for patients with AMI. Despite calls for a large RCT to answer this important clinical question,[2,3,4] only one additional RCT was performed between 2012 and 2016.[11] Data from this trial were combined with those from a small randomized trial published in 2011.[12] These studies included a total of 155 patients, with one showing increased mortality in the liberal threshold group and the other showing the opposite. Both studies reached statistical significance. When combined they showed a trend toward a benefit using a liberal threshold.

The AABB authors note that several other societies have issued recommendations for transfusion thresholds in the setting of AMI or acute coronary syndrome (ACS). These threshold recommendations are inconsistent and range from 7 g/dL to 10 g/dL Hgb.[8] The editorial that accompanied the AABB guidelines in this month's JAMA advocated for a liberal strategy in the setting of AMI/ACS.[13]

A Caution About Extrapolation

The editorial also mentions the "pendulum swing" that I refer to in the title of this commentary. Although it's true that the safety of a conservative Hgb threshold has been established for specific groups, it's also true that RCTs do not show adverse effects from pRBC transfusion. We know that such adverse effects exist (they are summarized nicely in the beginning of the guidelines), but they occur infrequently. Some patient groups are harmed by a conservative strategy, so clinicians should be careful not to extrapolate data beyond the specific population studied. Conservative strategies have become very popular, but if pRBC transfusions are not as harmful as once thought, the risk/benefit tradeoff is slightly shifted.

I plan to do as the guidelines say and use a threshold of 7 g/dL in the ICU, or 8 g/dL if the patient has a history of CAD. If there is evidence of ACS, I will go a little higher. Although the guidelines don't address patients with sepsis, I rarely push these patients to 10 g/dL, particularly because subsequent trials haven't confirmed the mortality benefit originally seen with early goal-directed therapy.[14] I don't have the guts to restrict my patients with active gastrointestinal bleeding—another group that isn't addressed by the AABB. We don't always obtain upper endoscopies within 6 hours of presentation. I would be careful in applying data from this trial to your patients if your facility doesn't reliably provide early intervention.


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.