Expansion of blood donor screening and improvements to laboratory markers have reduced the risk for HIV infection from as high as 1 in 100 units in some U.S. cities in the early 1980s [1] to approximately 1 in 680 000 units [2, 3] (Figure 1). Transfusion-related hepatitis has also almost been vanquished: Transmission rates for hepatitis C virus (HCV) has decreased from 1 in 200 units in the early 1980s to approximately 1 in 100 000 units today, and the risk for hepatitis B virus (HBV) infection has been reduced from 1 in 2100 units to 1 in 63 000 units [3, 5, 6].

View larger version (22K): [in this window] [in a new window]   Figure 1. Decrease in per-unit risk for transmission of hepatitis B virus (broken line), hepatitis C virus (dotted line), and HIV (solid line) by blood transfusion in the United States. The arrow shows the current risk for death from acute hemolysis [4] for comparison.

Despite these improvements, a zero-risk blood supply remains a popular goal. However, being on the flat part of the transfusion safety curve poses new problems. Do future "improvements" have unanticipated side effects that could offset their benefits? Can we improve transfusion safety while responding to demands to control health care costs? We considered ongoing efforts to improve transfusion safety and some of their potential consequences.

Established and Developmental Blood Safety Initiatives

Selection of Lower-Risk Donors

More attention to behavioral, medical, and demographic factors in donor selection improved transfusion safety even before specific laboratory screening tests were available [7, 8]. For example, increasingly specific questioning of donors about HIV risk led to a substantial decrease in transmission by transfusion before HIV antibody testing became available [1], and the rate of HIV seropositivity and seroconversion among volunteer blood donors today is 1% of the rate in the general U.S. population [3].

Detection of Blood-Borne Pathogens

Because not all infectious exposures are recognized or acknowledged, laboratory testing remains important and is becoming increasingly sensitive. Although surrogate markers for some infections (such as non-A, non-B hepatitis and AIDS) with limited efficacy were implemented or proposed in the mid-1980s, quantum reductions in transfusion-related risk accompanied implementation of virus-specific antibody tests for HIV (1985) and HCV (1990). Virus-specific antigen assays promise further improvements. For example, an assay for HIV p24 antigen, approved by the U.S. Food and Drug Administration in 1996, is projected to reduce HIV transmission by an additional 25% [9]. New tests based on polymerase chain reaction techniques are currently being developed and may reduce the infectious window even further [10].

Inactivation of Blood-Borne Pathogens

Because donor history and testing cannot eliminate all risk for transfusion-related infectious disease, viral inactivation techniques continue to be attractive. Solvent-detergent inactivation of lipid-enveloped viruses (including HIV, HBV, and HCV), which was developed to treat coagulation factor derivatives, has been adapted for liquid plasma and is used in Europe [11, 12]. Adding methylene blue to plasma is another successful viral inactivation technique used in Germany and Switzerland [13]. Techniques for cellular components remain in development as researchers seek ways to inactivate contaminating microbes while leaving intact the metabolic processes of blood cells.

Minimizing the Need for Allogeneic Transfusion

Avoiding allogeneic exposure is key to reducing transfusion risk. More conservative transfusion practices have developed in recognition of the fact that transfusion decisions must be tailored to an individual patient's clinical condition and not determined solely by laboratory data [14]. Autologous transfusions have also become more widely practiced. Preoperative autologous donation now accounts for approximately 5% of all red blood cell units transfused in the United States [15], and instruments for intraoperative red blood cell recovery are used more frequently.

Potential Side Effects of Blood Safety Measures

As the risk for transfusion-related infection approaches zero, the risks of blood safety measures themselves, once dismissed as trivial, become more important. The negative consequences of some measures may even outweigh the benefits.

Donor Selection

About 5% of prospective donors are turned down because of their answers to questions about medical history, demographic factors, or risk-related behaviors in the remote past. This loss is problematic given the marginal adequacy of the blood supply and the psychological effect of the deferrals on these (usually healthy) persons. Furthermore, characteristics that are associated with reduced risk for one infectious agent may be associated with increased risk for another. Concerns about the theoretical risk for transfusion-related transmission of the agent that causes Creutzfeldt-Jakob disease are illustrative. Because Creutzfeldt-Jakob disease usually presents in older persons, several blood collectors have proposed turning down older donors or diverting plasma from persons older than 50 years of age from derivative manufacture. However, older donors are the safest group with respect to recent infection with a blood-borne virus. Exclusion of older donors is projected to cause increases of 10% to 20% in the risk for infection with HIV, HCV, and HBV and to remove 20% of donated units from available inventories [16]. Thus, turning down older donors out of concern for a theoretical, unproven risk could adversely affect blood safety and availability.

Testing

More sensitive methods for viral detection also pose potential and real risks. Availability of more accurate blood tests for infections (especially HIV infection) may lead some high-risk persons to donate just to obtain this testing in a free, socially acceptable setting. This magnet effect could increase the frequency of infectious disease in blood donors [17]. Because residual risks in tested allogeneic units are a function of the incidence and prevalence of disease in donors and the accuracy of screening tests, the magnet effect of new tests for infectious disease could paradoxically result in a blood supply that is less safe. Furthermore, even small problems with the specificity of new donor-screening tests, particularly tests with very low yields and therefore low positive predictive values, can result in substantial losses of safe, repeat donors who will be replaced by relatively less safe first-time donors. Finally, consideration must be given to the multiple implications of the presence of an infectious disease marker. For example, the presence of viral antibody may indicate previous exposure and thus risk for infection, but it may also indicate the presence of neutralizing antibody. Testing donated blood for antibody to HCV dramatically reduced transmission of HCV by transfusion, but removal of plasma containing this antibody from pools used for the production of gammaglobulin products led to transmission of HCV to recipients of intravenous gammaglobulin for the first time [18, 19].

Viral Inactivation

Viral inactivation methods also carry potential risks. For example, the solvent-detergent process does not inactivate nonenveloped viruses, such as hepatitis A and parvovirus B19 [20, 21]. The clinical importance of nonenveloped viruses in blood donors is currently unknown. However, the risk for transmitting these types of viruses is amplified substantially by the solvent-detergent process, in which thousands of plasma units are pooled. According to one analysis, a nonenveloped virus that causes an AIDS-like syndrome would have to be present in the population only at an undetectably low level (1 in 71 000 000 donors) before all of the benefits of avoiding lipid-enveloped viruses were entirely negated [22]. Reductions in the projected pool size have been proposed but are unlikely to substantially increase safety [23]. Furthermore, any chemical inactivation method must be scrutinized for the toxic potential of residual decontaminants that may exceed the risk associated with viruses that are being inactivated.

Transfusion Alternatives

Blood donation by a healthy person is considered innocuous, but is this true for a patient whose coronary artery disease is so severe that bypass surgery has been planned? By using higher risks for infection than those currently present in allogeneic transfusion, one model [24] concluded that the risk for death from a donation reaction of only 1 in 101 000 for a patient awaiting bypass surgery negated all of the benefits of having autologous blood available. No study to date has been large enough to fully document the risks of preoperative autologous donation, but the risk for a reaction is so serious that hospitalization is required is 1 in 17 000 [25]. Using allogeneic blood may be the least risky approach for some patients. Even the drive to alter transfusion thresholds has come under scrutiny because of concern that failure to give a transfusion when indicated may result in morbidity or mortality [26].

Cost-Effectiveness of Safety Initiatives

Improvements in safety generally add new costs to the blood supply system. Because transfusion medicine faces the same economic pressures felt in other areas of medicine, interest in applying cost-effectiveness analysis to decisions about blood safety has increased.

Cost-effectiveness analysis can be used as part of an effort to optimize the health benefit to a population by comparing the net costs of an intervention with its net benefits. By defining the aggregate or average health benefits achieved from competing interventions in terms of common benefit units (often the quality-adjusted life-year) per unit of resource spent, interventions can be directly and objectively compared. These data may be useful in prioritizing health resource applications.

Many cost-effectiveness analyses have been done on blood safety issues. Several new tests that have resulted in substantial decreases in viral transmission are remarkably cost-effective: The introduction of HIV-antibody testing cost $3600 per quality-adjusted life-year, and alanine aminotransferase testing (as a surrogate for non-A, non-B hepatitis before specific testing for antibody to HCV was available) and specific testing for antibody to HCV testing actually reduced overall costs [5, 9]. In contrast, many recent and proposed safety initiatives do not measure up when compared with other medical interventions. Testing for HIV p24 antigen was projected to reduce the number of annual instances of HIV transmission by transfusion in the United States by up to 8 per year, but this intervention was predicted to cost more than $2 million per quality-adjusted life-year saved [9] (Figure 2). Solvent-detergent plasma in most applications would also have a cost-effectiveness far worse than the benchmark of $50 000 per quality-adjusted life-year often cited by policymakers [21]. In many situations, preoperative autologous donation has a cost-effectiveness of $500 000 to several million dollars per quality-adjusted life-year saved. These analyses have high-lighted the factors that are associated with more cost-effective application of preoperative autologous donation, such as situations in which transfusion is likely to be necessary and in which a reasonable postoperative longevity is expected [24, 32, 33]. However, when autologous donation is undertaken with a low likelihood of transfusion, cost is generated without offering commensurate potential benefit (beyond psychological comfort to the patient that may be equally achieved through careful explanation of current allogeneic transfusion risks).

View larger version (25K): [in this window] [in a new window]   Figure 2. Comparison of cost-effectiveness of transfusion safety interventions (striped bars) and other medical practices (white bars) [27-31]. ALT = alanine aminotransferase; anti-HCV = antibody to hepatitis C virus; HCV = hepatitis C virus.

Why are the cost-effectiveness estimates of many blood safety measures so dismal? To begin with, risks are now low, and there are far fewer opportunities than in the past to avoid viral transmission. Furthermore, many transfusion recipients have a significantly reduced lifespan because of illness [34]. Finally, the costs of these interventions are substantial and are incurred for every unit of blood, most of which are not contaminated. In summary, blood safety is on the asymptotic part of the safety curve and further efforts to improve safety are likely to have even worse cost-effectiveness.

Why Haven't Cost-Effectiveness Analyses Affected Blood Policy?

So far, enthusiasm for blood safety initiatives has not been affected by potential side effects or poor cost-effectiveness projections. The risks of these initiatives are certainly small and may seem only hypothetical. Furthermore, cost-effectiveness analyses are only tools with which to assist decision making; they cannot replace it because not all factors that enter into a decision are captured in the analysis framework. Standardization of analytical approaches may facilitate acceptance and application of cost-effectiveness analyses [35-38]. However, we believe that other, more specific reasons may explain why cost-effectiveness analyses of blood safety initiatives have not had a greater effect.

Perhaps the criteria for judging a blood safety improvement as cost-effective differ from those used for other preventive health measures, or perhaps cost-effectiveness analysis models for blood safety do not reflect the true desires and concerns of potential transfusion recipients. Clearly, HIV is perceived differently from other health threats, and cost-effectiveness models have not captured the peace-of-mind benefit of risk reduction. Surety of death as an outcome is associated with extreme avoidance, and fear of HIV transmission is understandable [39, 40]. Furthermore, viral transmission by transfusion is probably perceived as inequitable, and the true risk for a particular transfusion is unseen and unknown. By common theories of risk avoidance, it would be expected that transfusion would be regarded very warily by the public [41]. It is not surprising that the public is more concerned about transmission of HIV than about other complications of transfusion that have potentially fatal outcomes. For example, more than 1000 patients in the United States each year receive the wrong unit of blood [4], but definitive steps to reduce this risk are not being demanded. In addition, the "rule of rescue" [42] may apply: Both lay and professional groups show a preference for application of an intervention that avoids certain death for a small but definable and visible group over a larger (aggregate) benefit in the future distributed among a less-visible group [43, 44]. Such intangible values are difficult to capture in economic analyses. Furthermore, intervention by the U.S. Congress during the U.S. Food and Drug Administration's consideration of whether to require HIV p24 antigen testing of donated blood made it clear that increased protection from HIV transmission was politically necessary at any cost, rendering economic considerations irrelevant.

Public image issues are also at play. Blood safety decisions made a decade ago are now subject to retrospective analysis and criticism. Consequently, blood bankers and regulators wish to avoid potential criticism for failing to take new steps to improve safety. As a result of serious concerns about the magnet effect of using a test with poor sensitivity (such as hepatitis B core antibody testing) as a surrogate for AIDS risk [17, 45], most blood collecting agencies did not implement this option. Similarly, directly worded questions about sexual activity were approached cautiously on the advice that they might lead to donations by some at-risk persons who wanted to prove to others that they were not members of a certain group. A decade later, both of these decisions have been challenged as unnecessarily cautious [46, 47]. Blood bankers and regulators do not want to lose the credibility that they have worked hard to gain.

Cost-effectiveness data have sometimes contributed to decisions to discontinue outmoded tests, such as in a consensus conference panel recommendation to discontinue alanine aminotransferase testing [48, 49]. However, the same consensus panel recommended continued use of hepatitis B core antibody as a surrogate marker for HIV risk, despite similarly poor cost-effectiveness estimates [50].

Conclusions

The consequences of decisions to push toward a zero-risk blood supply are important to consider. In particular, care must be taken that efforts to improve transfusion safety do not undermine that safety through unintended side effects. Ultimately, society's resources are being used; therefore, society has the right to know the projected outcomes of these efforts and to decide how its resources are allocated. Although understanding small risks is difficult, public education about the current risks of allogeneic transfusion is one important step toward making rational decisions.

In a health care marketplace that is increasingly dominated by managed care and ever-tightening resources, decision makers are faced with critical choices among health care improvement options that pit improved blood safety against other worthwhile, effective interventions. When an intervention is undertaken despite poor cost-effectiveness predictions, the economic implications of the decision merit special scrutiny. Hospital administrators who are asked to pay for additional testing of donated blood in a time of declining, flat-fee reimbursements will be forced to reduce other services. If hospitals pressure their blood centers to absorb the cost of additional testing, cuts in other areas of their budgets may also have undesirable consequences, including a potential diminution of blood safety because of cuts in more effective safeguards.

Resolving these issues will be difficult, and physicians must remain involved in the discussion. Although physicians must continue to advocate the most effective care available for patients, they must also keep decision makers abreast of the effects of societal decisions that may have been made more on the basis of fear and emotion than logically defined benefits. Physicians must also widen the horizons of persons who are concerned about blood safety to consider all of the health threats faced during transfusion-including, for example, mistransfusion-so that limited resources are expended to increase overall safety to the greatest extent possible. The situation is too complex and the outcome too important to merely follow simplistic rhetoric that demands safer blood.

Dr. Birkmeyer: Departments of Surgery and Community and Family Medicine, Dartmouth-Hitchcock Medical Center, One Medical Center Drive, Lebanon, NH 03756.

Dr. Busch: Department of Laboratory Medicine, University of California, San Francisco, 270 Masonic Avenue, San Francisco, CA 94118-4417.