Blood still kills: six strategies to further reduce allogeneic blood transfusion-related mortality

Abstract

After reviewing the relative frequency of the causes of allogeneic blood transfusion-related mortality in the United States today, we present 6 possible strategies for further reducing such transfusion-related mortality. These are (1) avoidance of unnecessary transfusions through the use of evidence-based transfusion guidelines, to reduce potentially fatal (infectious as well as noninfectious) transfusion complications; (2) reduction in the risk of transfusion-related acute lung injury in recipients of platelet transfusions through the use of single-donor platelets collected from male donors, or female donors without a history of pregnancy or who have been shown not to have white blood cell (WBC) antibodies; (3) prevention of hemolytic transfusion reactions through the augmentation of patient identification procedures by the addition of information technologies, as well as through the prevention of additional red blood cell alloantibody formation in patients who are likely to need multiple transfusions in the future; (4) avoidance of pooled blood products (such as pooled whole blood-derived platelets) to reduce the risk of transmission of emerging transfusion-transmitted infections (TTIs) and the residual risk from known TTIs (especially transfusion-associated sepsis [TAS]); (5) WBC reduction of cellular blood components administered in cardiac surgery to prevent the poorly understood increased mortality seen in cardiac surgery patients in association with the receipt of non-WBC-reduced (compared with WBC-reduced) transfusion; and (6) pathogen reduction of platelet and plasma components to prevent the transfusion transmission of most emerging, potentially fatal TTIs and the residual risk of known TTIs (especially TAS).

Fig 1

Causes of ABT-related deaths as a proportion of all deaths reported to the US FDA or the United Kingdom SHOT in 2008. The number of deaths related to transfusion (US FDA), or in which transfusion was deemed to have a causal or contributory role (United Kingdom SHOT), is shown above each column. Only 1 death (secondary to a septic reaction to platelets) was deemed to have been caused by an ABT in the United Kingdom. In all other cases (including 2 cases of inappropriate transfusion and 1 febrile, nonhemolytic transfusion reaction), the transfusion was deemed only to have contributed to a patient's death.

Fig 2

The 3 leading causes of ABT-related deaths, along with all other causes of ABT-related deaths reported to the US FDA for the last 4 years (2005-2008). The figure shows the proportion of all deaths reported to the US FDA in 2005 to 2008 that was attributed to each cause of transfusion-related mortality. The actual number of deaths from each cause is shown above the corresponding column.

Fig 3

The leading causes of ABT-related deaths reported to the US FDA in 2001 to 2008 showing a comparison of the number of deaths attributed to these causes in 2001 to 2003 vs 2004 to 2008. For each time period, the figure shows the mean annual number of deaths deemed to be due to TRALI, TAS, or HTR.

Fig 4

A comparison of the number of deaths from TAS between 2001 to 2003 and 2004 to 2008 shown separately for all transfused products and for single-donor apheresis platelets. For each period, the figure shows the mean number of deaths deemed to be due to TAS each year based on the reports of transfusion-related fatalities made to the US FDA. Data are also shown for the earlier periods of 1976 to 1985 and 1986 to 1995, but the data for 1996 to 2000 have not been made available.

Fig 5

Comparison of the number of deaths from TRALI associated with the transfusion of FFP between 2005 to 2007 and 2008. For each year, the number of deaths reported to be due to TRALI is based on reports of transfusion-related fatalities made to the US FDA.

Fig 6

The number of deaths from TRALI for 12 years (1996-2008) of reporting to the United Kingdom SHOT system. For each period, the figure shows the mean annual number of deaths deemed to be due to TRALI based on the reports of transfusion-related fatalities made to the United Kingdom SHOT program. There were zero (0) fatalities from TRALI in 2008.

Fig 7

Estimation of the risk of TAS from platelet transfusions in the United States today. The depicted historical risk represents empirical data from settings, in which all febrile reactions to platelets were monitored and cultured. The estimates of current risk are based on the assumption that all whole blood–derived platelets distributed in the United States will soon be prestorage-pooled and cultured. For estimates of risk in the contemporary (October 2009) setting—when surrogate methods (rather than bacterial culture) are usually used to screen pooled whole blood–derived platelets for bacteria—see Vamvakas.

Fig 8

Effect of taking into account, for the diagnosis of TRALI, the laboratory results of donors tested for WBC antibodies. Because HLA antibodies are present in the serum of 24.4% of women who have previously been pregnant, and multitransfused patients are likely to have received some blood components from previously pregnant female donors, such recipients will be likely found to have received components from donors whose plasma contains WBC antibodies, regardless of whether TRALI is (or is not) present. Therefore, female donors and WBC antibodies are likely to be spuriously associated with TRALI when the results of such laboratory tests are considered in making the diagnosis. Such spurious associations may have been the origin of TRALI in some of the cases included in series compiled by blood suppliers on the basis of passive surveillance reports received from hospitals when the recipient's serum was no longer obtainable. For these reasons, the consensus criteria require that the diagnosis of TRALI be made clinically, without considering the results of laboratory tests. Provided that the diagnosis of TRALI has been made in this manner, the further designation of alloimmune TRALI should be reserved for cases of proven WBC incompatibility between an implicated donor and the relevant cognate recipient WBC antigen. In the case of common WBC antigens (and antibodies), even WBC incompatibility between an implicated donor(s) and the recipient can occur by chance. Thus, although such WBC incompatibility strengthens the belief in the correctness of the diagnosis in the appropriate clinical setting, it should not be used as the basis for the diagnosis.

Fig 9

Proportion of deaths from acute HTR reported to the US FDA secondary to ABO vs non-ABO HTRs in 2 periods separated by 20 years. The mean annual number of deaths recorded in each period is shown above each column.

Fig 10

Design and results of the TRICC RCT. There was no difference in the primary outcome (30-day mortality), but one of the mortality outcomes reported by the authors (in-hospital mortality) approached statistical significance (P ≥ .05). If this finding is confirmed by future RCTs, it would suggest that, for every 16.9 transfused critically ill patients, 1 might die because of the 3 excess RBC transfusions received per liberal (compared with restrictive) criteria.

Fig 11

Clinical outcomes analyzed as categorical variables in the TRICC RCT. For each comparison, the figure shows the OR of an adverse outcome in patients from the restrictive (compared with the liberal) transfusion strategy arm. Each OR is surrounded by its 95% CI. When the 95% CIs include the null value of 1, the corresponding OR is not statistically significant (ie, P > .05). Of 22 categorial comparisons reported by the authors, only the number of organs failing (>3 vs <3), septic shock, catheter-related sepsis, and pneumonia occurred less frequently in the liberal (compared with the restrictive) strategy arm. The differences were minimal, with only the difference in septic shock approaching a trend (P = .13). All other differences in categorical outcomes favored the restrictive strategy arm. Differences in any cardiac complication (P < .01), myocardial infarction (P = .02), and pulmonary edema (P < .01) were statistically significant, whereas the difference in ARDS (P = .06) approached significance. Five more outcomes were analyzed as continuous variables, and they all favored the restrictive strategy arm. Three of these differences were minimal, but the differences in adjusted MODS (P = .03) and for the change from baseline score (P = .04) were statistically significant.

Fig 12

Design and results of the cardiac surgery RCT of van de Watering et al. The authors enrolled patients into 3 arms (to receive prestorage-filtered WBC-reduced, poststorage-filtered WBC-reduced, or non–WBC-reduced RBCs), but the 2 WBC-reduced arms produced identical results and are combined here into one arm. The findings regarding the study's primary outcome (postoperative infection) were not significant, but the difference in the secondary outcome (60-day mortality) attained significance in both the 3-arm and the 2-arm comparisons. The absolute risk difference in mortality would indicate that, for every 23.3 transfused cardiac surgery patients, 1 might die because of the receipt of non–WBC-reduced (as opposed to WBC-reduced) RBCs. All platelets administered to patients enrolled in this trial had been WBC reduced.

Fig 13

Comparison of the annual number of TAS and TRALI deaths passively reported to the US FDA between 2005 and 2008 in association with single-donor vs pooled whole blood–derived platelets. Abbreviation: PWBD, pooled whole blood–derived.