Blood transfusion in the ICU (part 2): Cross-matching blood

Read part 1 (Overview) here

Blood transfusions are medications as well as IV fluids, and are associated with a spectrum of risks ranging from the infectious to the cardiovascular. However, as transfusion of banked blood is essentially a liquid organ transplant, it also conveys a risk of immune-mediated reactions. Just like for any organ transplant, much attention is therefore given to ensuring compatibility between donor products and their recipients. Let’s look at how this is done.

Product preparation

After being drawn, centrifuged, and mixed with citrate and preservatives, donor red cells are screened for infection (an imperfect process, which is why donors also undergo extensive screening questionnaires for infectious risk factors), and tested for ABO type, as well as non-ABO antigens.

In many cases, packed red blood cells are also leukoreduced, a process of filtration or washing that removes the bulk of the white blood cells from the infusate. These leukocytes serve no therapeutic purpose, and can potentially incite an inflammatory response by attacking the recipient (a sort of mild graft-versus-host disease); this phenomenon may be responsible for many of the mild febrile transfusion reactions we see. In some centers in the US, all banked PRBCs are leukoreduced by default, and in others they can be leukofiltered at the bedside during the transfusion process.

This is not to be confused with product irradiation, wherein the product is bombarded with enough ionizing radiation to completely deactivate any leukocytes present. This process is more thorough than mere leukoreduction, and is only necessary to prevent true graft-versus-host disease; such products are therefore only needed for heavily immunosuppressed recipients at risk for GVHD, such as those with hematologic malignancies or on certain chemotherapy regimens.

Recipient testing

Meanwhile, our recipient is undergoing their own testing.

Hemolytic transfusion reactions are the most severe transfusion reactions, causing massive hemolysis and rhabdomyolysis. These generally occur due to ABO incompatibility; if type A blood is given to a type O patient, the recipient’s anti-A antibodies attack the donor red cells, causing hemolysis and agglutination.

Proper ABO typing is therefore essential. This can be done quickly and accurately in modern labs, but the most common cause of error remains human (i.e. patient mixups), so current guidelines require drawing and running two blood types on each patient for confirmation—although a historical result in the hospital records may qualify as one. Rh typing (the “positive” or “negative” in a blood type) also occurs here.

Antibody screening comes next to identify non-AB antibodies present in the recipient’s blood. These include antibodies against the Rh antigens, but also against Kell, Duffy, Lewis, and other “optional” antigen systems, a veritable cornucopia of miscellaneous protein and carbohydrate markers that appear on on the surface of RBCs, all of which can stimulate an immune response in non-self patients. Unlike ABO antibodies, which occur naturally, these are learned immune responses; hence, they only develop after exposure to foreign blood, such as from prior blood transfusions or from fetal blood crossing the placenta. Patients with a history of multiple transfusions or pregnancies are therefore at the most risk to become alloimmunized and develop alloantibodies to these non-self antigens. (Pregnant women in whom placental disruption allows maternal-fetal mixing of blood can receive “Rhogam,” immunoglobulins to help prevent anti-Rh antibody formation and reduce the risk of future Rh alloimmunity.)

So, after ABO typing, an antibody screen is performed to detect these antibodies. The ideal recipient would have none, so the initial screening test is usually all-or-nothing, indicating only the presence or absence of clinically-significant anti-RBC alloantibodies. A negative test denotes a “naive” patient who can receive virtually any ABO/Rh-compatible blood. A positive screen, on the other hand, prompts more specific testing to identify the antibodies present. Not all of these antibody reactions are clinically significant, but many are, so a patient with (say) anti-K antibodies should generally only receive products lacking the K antigen. Patients who receive many transfusions tend to accumulate antibodies until it becomes very difficult to locate compatible blood.

Overall, this process constitutes the “type” (ABO typing) and “screen” (antibody screening) of a type and screen. What is a crossmatch?

Matching a specific unit of blood to a recipient is mostly done via the processes already described, i.e. ensuring ABO compatibility and the absence of non-ABO antibodies. But this is serious business, so after selecting a unit of red blood cells for potential transfusion, additional testing is done to confirm compatibility with the recipient.

Mostly, this is mostly a triple check for ABO matching. Since an ABO-mediated hemolytic reaction is rapid and dramatic, its absence can be confirmed by a simple mixing study. An “immediate spin crosmatch” involves mixing the patient’s serum with a sample of the donor blood, briefly centrifuging it, and inspecting it for hemolysis or agglutination. This can be done in minutes.

An even faster method is an “electronic crossmatch,” which essentially means trusting the computer. With two accurate blood types performed, and a rigorously tested and internally-validated computer system to ensure correct matching of ABO types, you simply trust the system to prevent clerical errors and goof-ups. This is only possible when strict internal controls are present and when the blood typing is straightforward.

A true, thorough crossmatch to rule out IgG reactions is necessary for patients with positive antibody screens and other complicating factors. This more complex process involves a mixing study with additional preparation and a prolonged waiting period for the slower IgG reactions to occur.


Crossmatching for plasma is a little simpler. FFP is ABO-matched similar to PRBCs, although in this case, we’re worried about donor antibodies attacking recipient red cells rather than vice versa. Hence, “universal plasma” is type AB (lacking any ABO antibodies), whereas “universal red cells” are type O (lacking any ABO antigens).

Generally, plasma does not require matching for non-ABO antibodies, because banked plasma does not contain any non-ABO antibodies; blood is screened at the time of donation and, if antibodies are present, is simply not used to produce plasma.


Platelet matching is even easier. ABO antigens are only expressed lightly on platelets, and other RBC antigens are not present at all. So while ABO matching can be performed, routine platelet transfusion is often not type-matched, and antibody screening is not needed either. (There is some controversy about this practice, and it may be that in an ideal world all platelet transfusions would be type-specific. However, platelets are always in shortage, and in many instances this just isn’t practical.)

Platelets do express other antigens, such as HLA, and some patients—particularly those with extensive transfusion history—will develop a diminished response to platelet transfusion (i.e. their post-transfusion platelet count increases only minimally). Platelets have a short half-life, so no transfusion lasts forever, and a good initial “bump” that later downtrends is often due to consumption or splenic sequestration. However, a negligible incremental response tested one hour after transfusion denotes true “platelet refractoriness” and is generally caused by immune response.

This is analogous to RBC alloimmunization, and a full workup for platelet refractoriness requires identification of anti-HLA antibodies and the less-common anti-HPA (human platelet antigen) antibodies. All of this is best done via consultation with your blood bank. Here’s a brief summary, however:

  1. The easiest approach is to give ABO-compatible platelets, which is straightforward and offers an improved response in some cases.
  2. Patients with HLA antibodies can receive antigen-negative platelets. In other words, once their specific HLA antibodies are identified, the blood bank is simply searched for platelet products that lack those antigens. This is similar to the process for PRBC transfusion in alloimmunized patients.
  3. In rare cases, true HLA matching can be performed, which is akin to the process for bone marrow donation. The donor registry is canvassed for a donor with a close HLA match, from whom a directed blood donation is then requested. (Family members might also be good candidates.) This takes a few days, is a major affair, and is usually only needed for severely refractory patients with numerous HLA antibodies.
  4. In some cases, serologic crossmatching can also be performed by directly testing the patient’s plasma against banked products, and may be easier than the full HLA-matching process.

In short, platelet refractoriness is complicated. Call your blood bank.


By this point, you should have a more nuanced understanding of what’s happening when you order that “type and screen,” the confirmatory typing, and the PRBC crossmatch. You should understand that FFP is simply ABO-matched and thawed, whereas platelets are usually not even matched, unless patients become resistant to them.

In Part 3, we’ll discuss thresholds and clinical situations when transfusion is indicated.