Red blood cell (RBC) transfusions can produce many adverse effects, but the safety of transfusions has increased dramatically. For example, the risk of transfusion-transmitted HIV or hepatitis C is now about 1 in 2,000,000 units transfused, and for hepatitis B about 1 in 205,000.

The overall rate of serious non-infectious RBC transfusion reactions is less than 1 in 1,000, with transfusion-related acute lung injury occurring in about 1 in 5,000, acute hemolytic reactions occurring in 1 in 6,000 to 33,000, and fatal reactions occurring in 1 in 250,000 to 600,000 transfusions. Less serious reactions such as fever are seen in about 1% of transfused patients.

In contrast, serious adverse effects of erythropoiesis-stimulating agent (ESA) therapy targeting hemoglobin to >11g/dL are in the 1 in 10 to 1 in 100 range, and include stroke and death. Consequently, attempting to aggressively use ESAs to prevent even one serious RBC transfusion reaction is likely to cause several strokes, cardiac events, and deaths.

When epoetin was introduced 1989, use was directed at dialysis patients with severe anemia, and the treatment goal was a hemoglobin of 10-11 g/dL. This led to virtual elimination of transfusion-dependent anemia, a 50% decline in red blood cell (RBC) transfusion rates, and contributed to a substantial decrease in patients with preformed human leukocyte antigen (HLA) antibodies.

Present management of anemia in CKD and dialysis is to initiate ESA therapy in most patients when hemoglobin is less than 10 g/dL and target hemoglobin to 10-11g/dL. More aggressive anemia treatment in the past reduced––but did not eliminate––transfusions. When the mean hemoglobin in dialysis patients was 12 g/dL in 2005, about 8% of dialysis patients still received transfusions each quarter. These transfusions occur overwhelmingly in the hospital during acute illness or major surgery.

Assessing the HLA risk

Accepting a higher risk of transfusion is increasingly recognized as wiser than aggressively increasing ESA or administering iron. Examples of where ESA use or dose escalation may be imprudent include recent myocardial infarction, stroke or transient ischemic attack; certain cancers; poorly controlled hypertension, or rising hemoglobin in response to the last ESA increase or iron administration. Administering IV iron is likely unwise in the setting of active or suspected infections. Compared to ESAs, transfusions are the preferred option in some cancers, hematologic disorders, and acute blood loss.

Consequently, greater use of transfusions in dialysis patients since 2010 is expected, appropriate, and may still be too low. Nevertheless, transfusions can induce human leukocyte antigen (HLA) antibodies, which are a concern for the minority of CKD patients being considered for transplantation.

The presence of an HLA antibody can preclude transplantation of a donor kidney expressing that HLA antigen. As there are many HLA antigens, a potential kidney recipient’s blood is tested for antibodies against a panel of antigens, and the degree of sensitization is expressed as the percent panel reactive antibody (PRA). The PRA test examines the presence of antibodies against approximately 40 unique antigens.

RBC transfusions are only one way HLA antibodies may be induced. More common inducers of HLA alloantibodies include multiple pregnancies, platelet and plasma transfusions, and prior organ transplants.

Modern screening techniques for HLA antibodies are much more sensitive than historical methods. Using such techniques, a study of 8171 transfusion donors found at least one positive HLA antibody in 1.0% of men never transfused, 1.7% of men previously transfused, 1.7% of nulliparous women and nulliparous, non-transfused women, and 24.4% of previously pregnant women. HLA antibodies directly correlated with number of pregnancies, rising from 17% in single deliveries to 35% with 4 or more deliveries. Use of a more sensitive cut-off for detection increased detection of HLA antibodies from 8.3% of nulliparous women to 45.4% of women with four or more childbirths. This lower cut-off yielded HLA antibody rates of 6.8% and 7.1% in transfused versus non-transfused men, respectively (p=079).

It is unclear whether the presence of low titers of donor-specific HLA antibodies, which would not normally preclude transplantation, affects subsequent rejection rate and graft survival. Early studies showed benefits from donor-specific transfusions prior to transplant, while a more recent analysis suggests the presence of donor-specific HLA antibodies associates with increased risk of rejection and graft loss.  

The mere presence of HLA antibodies is not a barrier to renal transplantation. Highly sensitized patients, with PRA of 80% or higher, may be very difficult to match or will require expensive and time-consuming desensitization protocols, although they receive priority points in the kidney allocation system. Patients with PRAs in the 20-79% range may have delays in finding a compatible kidney.

Can you avoid transfusions in transplant-eligible patients?

Is avoiding RBC transfusions necessary in all transplant-eligible patients? While desirable, it is not feasible. Patients with CKD develop serious medical problems that may warrant transfusion. ESA therapy may reduce the need for transfusions, but does not prevent them.

Additionally, while standard practice is to maintain hemoglobin about 10-11 g/dL, higher targets do not appear to offer meaningful transfusion reductions. In the largest trials, aggressively treating anemia to 13 g/dL or above only reduced transfusions by 12-53% compared to targeting hemoglobin in the 9-11.5 g/dL range, as shown in Table 1.

In hemodialysis patients, trials indicate one would need to treat 11-14 patients for one year to avoid one patient being transfused. Depending on patient type, the estimated cost of the extra ESA therapy would be $61,000 to $179,000. In stage 3 and 4 CKD patients where transfusions are less common, targeting hemoglobin above 13 g/dL yields dismal, statistically and clinically insignificant reductions in transfusions, compared to more prudent treatment.

Even in the placebo-controlled TREAT trial, 75.4% of patients randomized to placebo and given ESA only when hemoglobin was below 9 g/dL avoided transfusions over a 29 month period. While ESA therapy to a target of 13 g/dL reduced the transfusion rate from 25% to 15%, you would need to treat 25 patients for one year at a cost for ESA alone of greater than $300,000 to prevent one transfused patient.

All the above calculations are to avoid one person being transfused. But, if only 10% of transfused patients develop alloantibodies, then one would need to treat 10 times more patients to avoid one positive PRA, at the cost of increased ESA-related deaths and thrombovascular events.

Compared to placebo, aggressive ESA management in the TREAT trial did not even prevent multiple transfusions, which are more likely to cause high PRAs. One would need to treat 44 patients for 29 months to prevent one patient from receiving four or more days of RBC transfusions, while likely inducing one stroke and one other thrombovascular event.

It is possible the risks of ESA therapy are less in patients who are healthier than the diabetics randomized into the TREAT study, but healthier patients on placebo would also be less likely to require transfusions, thereby diminishing any benefit from ESA.

Managing the transplant-eligible patient

Based on the above discussion, it appears prudent to follow the label instructions and initiate ESA therapy only when hemoglobin is less than 10 g/dL after discussing the risks and benefits of ESA therapy with the transplant-eligible patient, and do not maintain hemoglobin >11g/dL. This will offer some transfusion mitigation, and is safer than more aggressive treatment.

Table 1

It is tempting to consider more aggressive anemia treatment in multiparous women, where the risk of transfusion-induced high PRAs is much higher. However, even if 50% of transfusions led to high PRAs in these women, one would need to treat more than 20 women on dialysis, or more than 50 women with CKD to prevent one high PRA.

It is also tempting to target an intermediate hemoglobin range of above 11 g/dL but below 13 g/dL. However, any transfusion reduction compared to targeting 10-11 g/dL would be smaller than that observed in the trials, and the intermediate target is likely less safe.

Most transfusions occur during serious acute illness or at the time of major surgery. One of the most effective ways to reduce RBC transfusions is to choose a lower hemoglobin trigger for transfusion. Trials in ICU patients, cardiac bypass patients, orthopedic surgery patients, and most recently, patients with gastrointestinal bleeding, have shown a low hemoglobin trigger for transfusion (generally <7 g/dL) is as safe, or safer, than a higher trigger of 9-10 g/dL.

Physicians in the United States may be too quick to transfuse. In the TREAT trial, the placebo arm transfusion rates were 29% in the North American (largely U.S.) patients and 16-19% in the rest of the world. Transfusion rates in the ESA-treated arm were similar across regions at 15%. ESA treatment only significantly reduced transfusions in the U.S. cohort.


Even with very aggressive management of anemia in CKD, one cannot avoid transfusions. RBC transfusions are significantly safer than perceived by many physicians. While RBC transfusions should be avoided in transplant-eligible patients whenever possible to minimize the risk of allosensitization, in practice this still means initiating ESA therapy when hemoglobin is less than 9-10 g/dL, and maintaining hemoglobin no higher than 11 g/dL. The supposed benefit of targeting hemoglobin above 11 g/dL is inconsistent with evaluation of the available evidence. Physicians can also reduce transfusions in transplant-eligible patients by adopting a low hemoglobin trigger for prescribing transfusion during acute illness.