For more than a quarter century, orthopaedic surgeons and their patients have been plagued with heightened concerns over complications from blood transfusions. The initial cause for increased caution, starting in the mid-1980s, was the increased awareness of bloodborne transmission of retroviral disease in the form of human immunodeficiency virus (HIV) and the new, near-concurrent ability to identify Hepatitis C separately from Hepatitis A and B1. The sense of burgeoning epidemics of both diseases and the incomplete ability of screening to prevent their transmission led to a new willingness to avoid transfusions at progressively lower hematocrits as well as an explosion of enthusiasm for preoperative autologous blood donations.
A second cause for concern, which grew more slowly over the next decade, was that of transfusion-related immunomodulation. Direct clinical measurements of components of the immune system being suppressed by allogeneic blood exposure have been demonstrated; these measurements have been coupled, mostly outside of the orthopaedic literature, with clinical reports of increased rates of infection and more frequent recurrence of tumors2. The most immediate concern for orthopaedic surgeons was the question of increased risk from transfusion for infections after insertion of prosthetic implants. Despite the frequent invocation of such risk as a reason to delay transfusion, the arthroplasty-specific literature has been varied and relatively underpowered on this question. As pointed out by Friedman et al. (one of the two articles discussed in this commentary), the articles that did address this question clinically have been single-center studies compromised by small patient group sizes.
In this edition of The Journal, two articles with large patient groups are published that specifically address the question of transfusion-related immunomodulation and the risk for infection after total joint arthroplasty. The issue of sufficient statistical power was addressed in each article by taking advantage of two very different large collections of clinical outcomes. The study by Friedman et al. uses a post hoc analysis of the data from the series of RECORD (Regulation of Coagulation in Orthopedic Surgery to Prevent Deep Venous Thrombosis and Pulmonary Embolism) trials, which were focused on the evaluation of the efficacy of rivaroxaban in the prevention of venous thromboembolic disease after lower-extremity total joint arthroplasty. The study by Newman et al. retrospectively looks, from a single large center’s experience over a decade, at a large series of patients who had undergone total joint arthroplasty.
At first glance, the conclusions might seem contradictory. Friedman et al. report a higher incidence of wound inflammation or infection in the patients receiving allogeneic transfusions. Newman et al., after risk adjustment, report no greater risk for postoperative infection with or without allogeneic transfusions.
However, because of the limitations of retrospective studies, the two studies do not differ as much in their outcomes as it might seem at first glance. By definition of being retrospective, these studies did not give the researchers the opportunity to define common important end points and data fields to be captured.
Data on infections were collected in the Friedman study; the participants in the RECORD groups reported on the occurrences of wound infection or inflammation, not all of which might meet the definition of a true infection. These occurrences, along with respiratory and urinary tract infections, were the negative outcomes reported as being influenced by allogeneic transfusions. However, the same study showed no difference in the groups for bone and joint infections, which might be interpreted as being the acute wound infections of greatest concern to the surgeon and patient. The exact definition of this outcome grouping is not given. The Newman study uses as an end point the surrogate of reoperation for suspected infection, even if cultures were negative. This end point excluded reoperations for presumed non-infectious reasons. The authors followed through on late outcomes of these select cases to try to reasonably mitigate the limitations of their database. When adjusted for other risks, this study also did not find any influence on acute wound infections from allogeneic transfusion exposure.
The patient populations of the two studies differ in terms of homogeneity and potentially confounding factors. The Newman study utilizes a single center with a single set of surgeons. The raw data, without adjustment, would argue for a deleterious effect from allogeneic exposure. More complicated cases and more ill patients both are at higher risk for infection and a need for transfusion; recognizing this, the authors used backward-stepwise logistic regression analysis to show that a greater number of units transfused and an American Society of Anesthesiologists score of >2, not the exposure to the allogeneic blood per se, are more accurate predictors of potential infection.
The Friedman study used similar techniques to try to parse out a larger number of potentially confounding factors, especially given its multicenter and international construct. The groups that received the allogeneic blood differed from the rest and tended toward including more women; more obese patients; those undergoing longer procedures; those with cardiac, renal, blood, and/or lymphatic system disorders; and those more likely to have had surgery outside of North America. There was not a common set of transfusion criteria. The authors felt that their statistical analysis eliminated the influence of the confounding variables. However, the international breadth of the study participants introduces potentially large differences in the local definitions of the end points. In addition, the wide range of diagnostic intensity used to identify and to confirm such end points as a urinary tract infection, an upper respiratory tract infection, or even a superficial wound problem is not reported and could range from clinical acumen with empiric therapy to advanced imaging and multiple specimens on culture that might contradict a physician’s first impression.
Perhaps the most important difference between the studies is that 92% of the allogeneic units transfused in the Newman study went through leukoreduction, removing the more antigenic and inflammatory white cells from the blood to be transfused. The prevalence of such treatment of the transfused allogeneic blood was recognized as not being available to the authors of the Friedman study.
Leukoreduction is felt to be an important enough cause of reduction of risk that it has been advocated for universal application with the potentially large cost savings from the reduction of complications to justify the cost of the added step3. From the Newman study, it appears that leukoreduction had already become a common practice for at least their center for almost a decade. It is potentially the most confounding factor in the study. However, if the extra step of leukoreduction has already become commonplace, the study gives surgeons who are affected by that transfusion practice information that is relevant to their current decisions.
Leukoreduction has been examined and was not found to be of benefit in autologous donations after total joint replacement4. However, if medical opinion has moved to the recommendation for routine leukoreduction for all allogeneic units on the basis of data outside of total joint arthroplasty, it might now be difficult to generate a study to examine the utility of this step prospectively. In this regard, the Friedman report, based on the more heterogeneous RECORD populations, might be the last, albeit clouded, glimpse of whether or not non-leukoreduced allogeneic transfusions have measurable influence on risk-adjusted acute wound infections after arthroplasty. The answer, at least for the complication of bone or joint infection, appears to be a muted “no” because the utilization rates of leukoreduction in that study are unknown.
At least in terms of an acute postoperative infection after arthroplasty, it appears that allogeneic transfusions, especially if leukoreduced, do not seem to have as large an effect as was feared. However, this issue is only one part of a much larger set of questions about the risks and benefits of transfusion. The utility of autologous donation has come under scrutiny in a number of studies that demonstrate the waste of such units and the lack of efficacy in non-anemic patients5,6. Neither of the current studies was designed to examine transfusion reactions, wrong units, transmission of disease, or volume overload. The studies were not intended to suggest appropriate trigger points for transfusions often needed to mitigate the risks of myocardial infarction, inadequate organ perfusion, and potentially death; a transfusion remains an important part of avoiding these very hard end points in those patients whose blood cell counts have dropped to low levels. Such trigger points have been suggested in other studies, but might need to be revisited in the face of an increasingly safer blood supply and the potential risk reduction introduced by the more common use of leukoreduction. Strategies regarding the reduction of intraoperative and postoperative bleeding are also part of the equation.
The need for the appropriate, balanced use of potentially life-saving transfusions after total joint arthroplasty should hopefully continue to become less common with better techniques and adjuvant means of obtaining hemostasis. Based on these two studies and the more common utilization of leukoreduction, the transfusion decisions that walk the line between the very real and potential complications from transfusion and the avoidance of morbidity and mortality from postoperative anemia need not be influenced by concerns over developing a postoperative acute wound infection.