A transfusion threshold for everyone

Evidence indicates restrictive thresholds are better for most inpatients.


It's been more than 20 years since the first evidence came out supporting restrictive blood transfusion strategies, Janice Zimmerman, MD, MACP, said during her ACP CME 30 session “Appropriate Transfusion of Blood Products in the Acutely Ill Patient.”

Yet, this less-is-more philosophy has only gradually expanded across inpatient practice. “Many clinicians thought that their patient was different and needed a higher hemoglobin because they were elderly, they had underlying cardiac disease or peripheral vascular disease, they had a neurological injury such as stroke, they had renal failure, an active GI bleed, or they were difficult to wean from a ventilator,” she said.

Photo courtesy of Dr Zimmerman
Photo courtesy of Dr. Zimmerman

By now, researchers have analyzed the effects of different transfusion thresholds in many of these patient populations, and Dr. Zimmerman, a critical care subspecialist and adjunct professor at Baylor College of Medicine in Houston, reviewed that data during her talk.

Transfusions are overall becoming less common in inpatient care, but their use continues to vary by patient characteristics and treatment location. “Certain groups of patients have a higher prevalence of transfusion, including older patients, blacks, and women. It is not surprising that patients with hematologic disorders also are more likely to be transfused. But geographical differences in transfusion are also found,” she said.

Dr. Zimmerman first tackled the question of whether elderly inpatients benefit more from a liberal or restrictive transfusion threshold. She cited the FOCUS trial, published by the New England Journal of Medicine (NEJM) in 2011, which compared transfusion thresholds of 10 g/dL and 8 g/dL of hemoglobin among patients with hip fracture and either history or risk of cardiovascular disease. “Note that the mean age was 81 years,” she said. “There was no difference in the primary outcome of death or inability to walk across the room. . . . This study suggests that a hemoglobin of 8 g/dL is adequate in this elderly population.”

Next up were cardiac surgery patients. In the TRACS trial, published in JAMA in 2010, researchers compared hematocrit thresholds of 30% and 24% (equivalent to hemoglobin levels of 10 g/dL and 8 g/dL, Dr. Zimmerman noted) in patients undergoing coronary artery bypass grafting. “They found no difference in the outcomes between groups. They did note that the risk of serious complications and death increased with every unit of blood that was transfused,” she said.

The subsequent TITRe2 trial, published in NEJM in 2015, did find higher mortality among cardiac surgery patients transfused at a hemoglobin level of 7.5 g/dL compared to 9 g/dL, but then the larger TRICS III, published in NEJM in 2017, found no difference in cardiac surgery outcomes between a restrictive threshold of 7.5 g/dL and liberal ones of 9.5 g/dL in the ICU and 8.5 g/dL on the ward. “They also analyzed 28-day mortality as a single variable and found it to be similar in both groups, laying to rest the concerns raised in the previous study,” said Dr. Zimmerman.

How about patients with acute myocardial ischemia? Available data on these patients come from two pilot studies, CRIT (published in the American Journal of Cardiology in 2011) and MINT (published in the American Heart Journal in 2013). Both compared hemoglobin levels of 8 g/dL and 10 g/dL as thresholds, but they had opposite results. “Interpretation of these two studies is limited due to the small patient numbers and different types of patients, but they do suggest that larger trials are reasonable,” said Dr. Zimmerman, who expects to see such trials, including the next phase of MINT, in the near future.

For patients with upper GI bleeding, a 2013 study (again in NEJM) found that a hemoglobin threshold of 7 g/dL was associated with improved survival compared to 9 g/dL. A similar subsequent comparison in a feasibility trial of 8 g/dL and 10 g/dL, published by The Lancet in 2015, failed to find a significant difference, but that might have been due to lower protocol adherence in the liberal group. “This may suggest that transfusion practice was already changing to a more conservative approach,” said Dr. Zimmerman.

Then there are transfusions for patients with septic shock, analyzed by the TRISS trial, a comparison of 7 g/dL and 9 g/dL published by NEJM in 2014. “Ninety-day mortality was the same in both groups, indicating that the lower transfusion threshold is safe,” she said.

However, oncology patients with sepsis may be different, according to a single center study that compared hemoglobin thresholds of 7 g/dL and 9 g/dL and was published in Critical Care Medicine in 2017. “The primary outcome of survival at 28 days was not significantly different, but a secondary outcome measure of survival at 90 days noted a trend in favor of the liberal transfusion strategy,” said Dr. Zimmerman. “More study may be needed in the septic oncology population.” Such research is currently under way in France, she noted.

There are still some other patient subgroups with insufficient data on the best transfusion threshold. “Unfortunately, no studies of blood transfusion strategies have been reported in patients with ischemic stroke,” Dr. Zimmerman said, adding that there are at least trials under way in patients with neurological injuries, such as trauma, subarachnoid hemorrhage, and intracranial hemorrhage.

Based on all this evidence, a group of organizations came together to support a 2014 Choosing Wisely recommendation to use a transfusion threshold of 7 g/dL in nonbleeding, stable ICU patients. The same standard should be applied to patients with sepsis, septic shock, or upper GI bleeding and those on prolonged mechanical ventilation, Dr. Zimmerman said.

For patients with cardiovascular disease, she said a higher threshold to transfuse at a hemoglobin level of 8 g/dL or in the presence of symptoms of anemia is advisable. That should also be the cutoff for elderly patients with hip fracture, Dr. Zimmerman recommended.

And finally, remember that the decision to transfuse is not always as simple as a number. “All guidelines recommend that clinical factors, such as symptoms, hemodynamics, and bleeding rate, be considered in addition to the hemoglobin level,” said Dr. Zimmerman.

The guidelines and the experts also support taking steps to reduce the need for transfusion, including minimizing blood loss with fewer phlebotomy orders and transfusing only one unit of blood at a time. “All of the studies we have just reviewed transfused only one unit of blood at a time and then reassessed hemoglobin levels,” said Dr. Zimmerman. “This is one practice change that everyone can adopt.”

She also encouraged physicians to learn more about patient blood management programs, which have been promoted by the World Health Organization. “These programs are patient-centered rather than centered on the blood products and involve multiple disciplines,” she said. “Interventions may include treatment of iron deficiency anemia, erythropoietin use, cell-saver techniques, restrictive transfusion strategies, and feedback to clinicians.” Dr. Zimmerman recommended the related AABB toolkit.