Transfusion hazard reporting: powerful data, but do we know how best to use it?

During the last 5 years, there has been a proliferation of national systems for collecting and collating hazard data on the transfusion of blood and components, a process broadly termed hemovigilance.1 This issue of TRANSFUSION contains a report on the first 4 years of data from the French Haemovigilance Network.2 Activities carried out under the hemovigilance umbrella vary in different countries. They broadly include collection of donor information such as the prevalence and cause of virus marker positivity, and adverse events in patients associated with transfusion of labile blood components, including errors. With sufficient resources, most of this is achievable, but what is less clear is how best to utilize this wealth of data. Clearly, the primary function of hemovigilance is as one contribution to the process of maintaining and improving transfusion safety. With publication of the French data, it is perhaps timely to reflect on what we specifically hope the collection of hemovigilance data will achieve, and how we envisage these goals being attained. It is interesting to consider why such a growth of hemovigilance activity is occurring at this time. The blood supply in the developed world has probably never been safer (variant CJD notwithstanding), and the HIV pandemic is dominating the blood safety agenda in the developing world. In some countries, such as France and Canada, hemovigilance grew out of government-driven reviews of transfusion organization following the HIV transmissions in the early 1980s. Within the 15 member states of the European Union, various articles and treaties from 1995 onwards have indicated a need for countries to establish hemovigilance systems.1 This has culminated in forthcoming legislation in the form of a European blood directive, legally binding within the European Union, in which Article 15 states that member states shall ensure that there is a system in place to collect, collate, and transmit information about adverse reactions and events related to the collection, testing, processing, storage and distribution of blood and blood components to the competent authority. The French Hemovigilance Network reported in the article by Andreu et al.2 was established in 1994, while in the UK the Serious Hazards of Transfusion (SHOT) initiative was launched in 1996. In 2001, the ISBT Haemovigilance Working Party noted that there were at least 20 countries in the process of establishing or considering a national hemovigilance system. Many different models of hemovigilance are emerging, reflecting the diverse way in which blood transfusion is organized throughout the world. Differences exist in both legal and medical aspects, and these will somewhat limit comparability of findings. For example, in Germany, hemovigilance is part of pharmacovigilance, because labile blood components are licensed medicinal products. In some countries, participation in hemovigilance is a legal requirement (France, some parts of US reporting requirements), while in others the scheme is voluntary (Denmark, the Netherlands). In the UK, there is no specific legal requirement until the European Blood Directive is adopted. In the last 3 years, however, two directives have gone out to hospitals from the UK Departments of Health, instructing hospitals to participate in SHOT.3, 4 The range of data collected also varies. For example, the French system collects data on every adverse event affecting the patient following transfusion, including minor rises in temperature. This results in approximately 7000 reports annually. Events are then categorized into 4 grades of severity, and 5 levels of imputability; that is, how likely it is that the reported event was related to the blood component. An impressive amount of data analysis is required, and it is noteworthy that 75 percent of reports are of the most minor grade of severity, and 20 percent are of doubtful or no imputability. In 50 percent of cases in the Andreu report, no definite cause for the reaction was identified. Some other systems, including SHOT, exclude minor fevers, and request only more serious reactions to be reported. This results in only 300 reports annually.5 It is as yet unclear which type of system is optimal in terms of prevention of serious harm to patients and cost-effectiveness. The types of hazards eligible for reporting also vary between national systems. For example, some systems collect only ABO incompatible transfusions. In SHOT, all errors which result in transfusion of the “wrong” component; that is, one intended for another patient, are reportable, even if the resulting transfusion is ABO compatible and no harm results. The rationale for this is that errors of all kind represent system failures, and there is much to be learned by their analysis.6 The same applies to “near miss” error reporting, because these events in transfusion have been shown to be at least four times more common than actual mistransfusions.7 In addition, some hemovigilance systems collect complications which may never harm the patient; for example, development of RBC alloantibodies. Another barrier to comparability between hemovigilance schemes is the lack of internationally agreed definitions. The terms new, lapsed, and regular donor mean different things to different people. Similarly, TRALI is one complication where an agreed definition would be very helpful, because there is no unique set of symptoms, nor a specific diagnostic test in patient or donor that is positive in 100 percent of cases.8 While SHOT asks for reports of TRALI, France does not—but does collect reports of volume overload. Andreu et al.2 comment that some reported cases of volume overload may actually have been TRALI, while in the 5th SHOT report (2000-2001),5 it is commented that some TRALI cases were uncertain, and may have been volume overload! The ISBT Working Parties on Granulocyte Serology and Haemovigilance are working together to propose agreed diagnostic criteria in this difficult area. Although hemovigilance information is relatively easy to collect, the denominator data essential to make sense of it is often lacking and is far more difficult to obtain. Blood centers and blood banks in the developed world usually have good data on the number of units of each component issued, but it is less easy to establish how many patients received these components, for what diagnoses, and how many different transfusion episodes were involved. For example, if 20 percent of all transfusion bedside errors occur at night, is this excessive, or does it simply reflect the fact that 20 percent of all transfusions occur at night? Studies are beginning to emerge that examine the circumstances in which blood is used, and the diagnoses and survival of transfused patients,9, 10 but much more is needed on this aspect of hemovigilance. Under-reporting is of course a key problem in any hemovigilance system, and Andreu et al.2 observe that in their “mandatory” system reporting rates varied fourfold between different regions. There are a number of reasons for incomplete data capture. First, the clinical complication and/or its relation to transfusion may not be recognized due to poor staff education; for example, fever and hypotension due to bacterial contamination. Second, staff may have a realistic fear of disciplinary action if they report that they have made an error. Promotion of a “no blame culture,” in which error reporting is positively encouraged, is critical if systematic defects in the transfusion chain are to be identified and dealt with. It has been estimated that getting the correct unit of blood to a patient involves more than 60 steps, and crosses several managerial boundaries!11 Systems could be simplified if weak links in the transfusion chain can be identified. So error reporting is to be encouraged, and the confidentiality of the reporting system needs to have the confidence of staff. Finally, some complications, such as hepatitis C-related cirrhosis, may take many years to manifest themselves, and the link with transfusion may be long forgotten. Questioning the patient about transfusion may not always be fruitful: many patients will be unaware that they have been transfused while under anesthesia. Hemovigilance systems have other inherent weaknesses. They can capture isolated transfusion incidents, but are not designed to assess a possible intrinsic effect of transfusion, such as immunosuppression. Prospective studies are required for this kind of analysis. Hemovigilance also has an inbuilt danger of focusing on the hazards of donor blood, while not collecting or assessing comparable data on alternatives such as autologous procedures, erythropoietin, or indeed the dangers of undertransfusion. Hemovigilance has also so far not encompassed the key question of appropriate use of blood components, itself an important arm of transfusion safety. Despite the differences and limitations mentioned earlier, certain key and consistent messages emerge from different national reports (France, SHOT, Denmark). First, in the developed world, transfusion error leading to ABO-incompatible transfusion has been clearly identified as a problem that needs attention. Solutions to transfusion error are challenging and resource heavy. For the short-term, the British Committee for Standards in Haematology has produced a useful practical guideline on blood handling and administration,12 which focuses on safe procedures in which all staff should be trained. For the longer term, extended use of bar code technology on the patient's wrist band and blood tube labels may be attractive,13 particularly if it can be proven to be cost-effective by also preventing drug administration errors. The second important finding is that the rate of virus transmission in the developed world is probably lower than it has ever been. It is impressive, though perhaps unexpected, to see in the Andreu report2 that there was not a single case of virus transmission reported to the French hemovigilance network in 4 years. What has emerged in both the French system and SHOT is the critical level of bacterial contamination of components, particularly platelets. Clinically apparent cases of bacterial contamination reported in the UK over 5 years (21) outnumber all virus transmissions combined.5 The French paper2 reports a reduction in cases over the 4 reporting years by a combination of better donor questioning, arm cleansing, and diversion of the first aliquot of the donation away from the main pack. Finally, the residual risks of immune complications remain a challenge. There seems little doubt that universal WBC reduction has reduced the incidence of febrile reactions,2, 14 transfusion-associated GVHD, and post-transfusion purpura.5 However, TRALI and anaphylactic type reactions are unaltered. In both the developed and the developing world, it is essential that decisions taken to improve blood safety are prioritized and appropriate for the environment: “aligning effort with risks.”15 Despite the high level of blood safety that the developed world enjoys, there are a host of initiatives that could still be taken, and where there are strong commercial interests operating: bar coding of patients, further virus genome testing, and pathogen inactivation of FFP and cellular components. The list never gets shorter. Clear decision-making pathways are essential, with a mechanism to feed hemovigilance data into the process. Every country is struggling with the major issues of how safe their blood supply “ought” to be, and how much health care resource it is legitimate to consume in this pursuit. It is interesting that the article on the French hemovigilance network gave no indication that collection of the data had led to specific recommendations, although initiatives had been taken to reduce bacterial risk. Each annual SHOT report contains a number of specific recommendations, although SHOT has no teeth to insist that they be implemented, or to monitor compliance. Other organizations within the UK National Health Service have to pick up the SHOT findings and consider what actions to take. Countries setting up hemovigilance schemes might therefore want to consider early on whether they have clear decision-making pathways for using the data to influence blood safety policy. One very positive effect of hemovigilance is to increase general awareness of transfusion among hospital staff. It is vital that hemovigilance data and recommendations can be fed back to all staff involved at any stage of the transfusion chain. The launch of an annual hemovigilance report provides an opportunity for an educational day on transfusion risks. A key development in the UK has been the establishment of multidisciplinary hospital transfusion committees to act as a focus for transfusion matters, including local development of safe blood handling procedures. There has also been a major rise in nursing or theater staff taking on a specialist “transfusion practitioner” role within hospitals, including hemovigilance reporting. A further use of hemovigilance data is in providing rapid alert of new hazards, a function that might be required if novel pathogen inactivation processes for components come on stream. Such an example was notification of the red eye syndrome resulting from the use of a particular type of WBC reduction filter.16 As has been seen in France with universal WBC reduction and programs to reduce bacteria in the blood supply, hemovigilance schemes are useful to monitor the effects of national changes to testing or processing policies. Finally, a good media handling policy for hemovigilance schemes is essential. Hemovigilance data can be good news, openly demonstrating a safe blood supply and a national commitment to transfusion safety. Production of an annual report can be an opportunity to interest professional journals and newsletters to run articles about transfusion. A proactive approach must be taken, however, with press releases and maybe even a press launch when hemovigilance reports are published. Any temptation to bury the findings should be resisted, as a damaging “blood scandal” is likely to result. The development of open national hemovigilance systems is good news, which we all welcome and embrace as a further tool in the transfusion armamentarium.