Systems for monitoring transfusion risk

The ultimate purpose of risk monitoring systems is to improve the quality and safety of transfusion therapy¹ through an improvement of transfusion standards and the preparation of specific guidelines by health care authorities, on the basis of data concerning the sequelae of blood transfusions.

Transfusion safety must be pursued in every stage of the procedure, starting from when a sample of blood is taken from the patient for pre-transfusion tests and from the donor at the time of collecting a unit of blood, up to the patient’s bedside at the time of transfusing the blood component.

The transfusion process can be represented diagrammatically as a Y⁵, in which one arm includes the procedures used to identify the patient correctly at the time of reception and the management of samples and requests; the second arm represents the production process, including the collection of blood, its processing, validation and storage and the certain identification of the unit of blood; and the meeting of the two arms represents the transfusion stage, including compatibility tests, distribution, transfusion and recording of transfusion events. This last stage involves the certain identification of the patient and the unit of blood. These processes are concatenated and an error in any of the links of the chain can continue to have effects until the end, causing harm to the patient or even his or her death.

The most frequent causes of transfusion incidents are due to errors in labelling and identification of the patient or blood component^2–8. Retrospective analyses carried out separately by Honig², Sazama³ and Linden⁴ on transfusion incidents occurring in the United States in the 1970s–1980s concordantly found that the incidence of fatal acute haemolytic reactions was one case every 600,000 units transfused. The observed incidence of transfusions of ABO incompatible blood in hospitals in New York State was one case every 33,000 units transfused and it was calculated that the incidence of transfusions to a patient different from the intended one, but, luckily ABO compatible, was one case every 12,000 units transfused⁴. An analysis of transfusion errors in New York State in a 22-month period in 1990–1991 showed that 43% (45/92) of the errors, of which 58% occurred outside the transfusion structure, were due to incorrect identification of the patient and/or the unit of blood⁴.

The cases of mistransfusions have changed over time, with the introduction and use of more sophisticated labelling and identification systems, as well a greater awareness of the risks of adverse events in transfusion medicine. Indeed, in the period 1990–1999, the incidence of mistransfusion of red blood cell blood components in New York State was lower than the incidences recorded by Honig², Sazama³ and Linden⁴ in the preceding two decades: for 9,000,000 units transfused, the incidence of mistransfusion was one case for every 14,000 units transfused. The incidence of fatal errors decreased to 1:1,800,000 trasfusions⁹. In the years 1994–1998, the French haemovigilance system¹⁰ recorded a mortality rate equivalent to that found by Linden, with one death for every 1,802,739 red cell products transfused.

Of the errors recorded by Linden, 56% occurred outside the transfusion structure (administration of a blood component to the wrong patient, errors in taking the blood sample for typing and/or compatibility tests or errors in filling in the request form), 29% were due to errors in the transfusion centre (tests on the wrong sample, errors in conducting the test, errors in recording data), and 15% of the errors were of a “mixed” nature, deriving from errors in both the transfusion centre and the ward (delivery error and mistaken identification in the ward, labelling error and lack of control in the ward).

The data concerning transfusion-related deaths, sent to the Food and Drug Administration (FDA) in seven different reports and reviewed by Myhre and McRuer⁵, revealed a high percentage of deaths, ranging from 20% to 75%, due to the transfusion of blood to the wrong patient.

These figures are, however, presumably considerable understimates^6,11, because of the understandable reluctance of doctors to notify errors, due to the fact that the concept of unacceptability of error is stressed during medical training¹². During investigations by health care authorities or the judicial system, the doctor usually finds that his responsibilities are considered in isolation, since such investigations are rarely directed at evaluating the circumstances in which the error occurred¹². Despite the fact that it has now been demonstrated that these circumstances are usually due to the system, it is common opinion that the main cause of poor health care is a bad doctor, who should, therefore, be expelled¹³. The awareness that a bad system and not a bad worker is the cause of most errors and harm, long accepted as a fundamental concept for improving safety in industry, is harder to accept in the field of medicine¹⁴. In consequence, the doctor tends to conceal or deny errors¹².

Furthermore, besides the data on misuse (errors of administration) of therapies, there is general paucity of information on their overuse (administration of treatments of little value) and underuse (lack of administration of necessary treatments), even though the quality of a treatment is intimately related to all three of these concepts. The reason for this is that harm caused by errors in prescribing a drug have a much greater emotional impact that that deriving from excessive or insufficient doses. As a result, authorities predominantly investigate errors of administration rather than under- or overtreatment and the doctor is more careful to avoid direct harm to the patient by an error of administration rather than by an inappropriate dose¹⁴. This could explain the lack of reports to the major transfusion surveillance systems of data related to insufficient or excessive use of blood components.

It is obvious that a haemovigilance system can only work correctly if, having overcome the reluctance to admit having made a mistake, all the secondary effects of transfusion therapy are reported to the appropriate authorities; not notifying a mistransfusion can have disastrous consequences because of the lack of corrective measures (for example, epidemics of infection because of the non-notification of donors at infective risk).

Systems for reporting non-conformities which do not attribute blame and are non-punitive (no fault/no fear confidential form of reporting)⁶, similar to those used in the aeronautic, nuclear and petrochemical industries, which are very effective in encouraging reports, can also be used in transfusion medicine. The confidential notification of adverse events by transfusion services to a neutral agency, which then makes them public each year in an updated list, is useful for the purpose of preventing such errors. Furthermore, the availability of the data enables a more precise definition of the problems that need to be solved to improve transfusion safety.

Obviously, this objective is easier to reach if notification, again in a confidential manner, of so-called near-miss events is requested, because these can be reported more freely, with less responsibility and fewer risks for the prestige of the worker. A near-miss is an error that is avoided in extremis, with dynamics very similar to those of incidents causing real and serious consequences. In the field of transfusion medicine, it is an error that, if not picked up in time, leads to a wrong blood group determination or in the collection, delivery or transfusion of a wrong or inappropriate blood component, but that is recognised before starting the transfusion. This type of error, which is well suited to analysis because it is much more frequent than a real, full error, provides a rich source of data which, given that it can be studied in depth without having to worry about the consequences had the error occurred, enables a search for the associated problems that caused it. Of course, a formal system of communicating near-misses aids understanding of the incident and constitutes an important resource for studying it in order to avoid its repetition or, at least, to be able to deal with it efficiently if is should occur again¹⁵.

Given the importance of recording near misses in the field of transfusion medicine, the Medical Event Reporting System for Transfusion Medicine (MERS-TM) project was designed and implemented in three large American Blood Banks (New York Blood Center, Oklahoma Blood Institute and Carter Blood Care in Dallas) and in three hospital transfusion services in the same cities, in order to estimate the incidence of these types of errors, and to analyse, classify, process and interpret them⁶.

The preliminary results of this “no-fault/no-fear” system for recording adverse events in the hospital in Dallas showed an increase in the notification of adverse events equivalent to that seen in industry, that is a 10-fold increase; numerically, this meant an increase from three reports a day prior to the introduction of the MERS-TM to thirty a day after it⁶. It is important to emphasise that this increase in the number of reported adverse events does not indicate an increase in the incidence of errors, but rather a heightened level of reactivity to the request to report data (increased detection sensitivity level - DSL), resulting from education concerning the value of reporting near-misses, swift feed-back and the lack of disciplinary action against the person reporting the incident.

The MERS-TM system for recording transfusion errors, supported by a grant from the National Institutes of Health, is currently active in 22 hospitals. The data collected reveal that the proportion of near-miss events, with respect to the total number of adverse transfusion events, is 90% and that the ratio between near-miss events and events with clinical consequences is 339:1. These data correlate well with those from industry, in which the ratio between near-miss events and incidents with harmful consequences is 300:1¹⁶.

Another form of collecting data concerning adverse events is that tried in a hospital in Atlanta, based on direct observation by appropriately trained technical staff who are sent to the patients’ bedside to watch the behaviour of the staff responsible for carrying out the transfusion therapy and to control their adherence to written procedures (Direct Observation System - DO)¹¹. This system of collecting data has been compared to the conventional system of event notification (Administrative Incident Report System - AIR).

The results obtained with the two monitoring systems in the same 18-month period were compared: the data from the DO concerned 202 transfusions in some selected wards of the hospital, whereas those from the AIR included all the wards in that hospital. There was a large difference in the number of reports between the two systems: 334 with the DO compared to only 52 with the AIR. Errors of identification of the patient or the unit were recorded with both systems, but the absolute number and frequency were higher with the DO than with the AIR (183 = 55% versus 9 = 17%). Cases of non-application or incorrect application of the identifying wristband were reported with the DO, whereas not even one such case was reported with the AIR.

The small number of notifications made using the AIR can be explained by the fact that this system cannot be considered no fault because it is focused mainly on the final step of the event rather than on the whole process; that is, the incident reported by the worker has already occurred and is the result of at least one behaviour that has deviated from a written procedure. There is, therefore, a higher likelihood that the person responsible for the error will be subject to disciplinary action. The DO, on the other hand, being a prospective monitoring system, picks up near-misses, that is, deviations from a written procedure before these evolve into an adverse event, and their notification is, therefore, associated with a lower risk of disciplinary action.

National haemovigilance systems

The risk monitoring systems described so far are limited to single, big health care structures with a large catchment area and a highly efficient organisation, not representative of the national reality and, therefore, of transfusion safety in smaller transfusion centres with fewer technological resources. However, the safety of transfusion medicine requires a national level of surveillance and regulation, including small structures.

It is easier to implement a national system of haemovigilance in countries in which transfusion activities are centralised in a few structures, usually separate from hospitals, rather than in countries in which the transfusion services are decentralised within hospitals¹⁷.

On the other hand, where transfusion centres are within hospitals (Italy and Argentina), the blood products are managed directly by specifically trained medical and technical staff, often supported by automation and computer technology which contribute notably to the safety of the process and prevention of human errors, as well as to the collection and storage of the transfusion data.

It is difficult to find this intra-hospital organisation in those countries (USA, England, France and Germany) in which the collection and preparation of blood components occurs in large structures separate from hospitals and in which the immunohaematology laboratories within the hospitals deal only with pre-transfusion investigations and the allocation and delivery of the blood components. It is conceivable that there is a higher incidence of errors of identification of patients and/or units of blood in such settings¹⁷.

There is no centralised haemovigilance system in the USA. The only obligation is to inform the FDA of transfusion-related deaths. The main cause of such deaths is transfusion-related acute lung injury (TRALI) (30%), followed by haemolytic reactions to a transfusion (16%) and bacterial contamination (16%)¹⁸. The risk monitoring systems in Europe are of two main types: the British system and the French one.

In the United Kingdom, reports from the whole country are monitored through a centralised institution called Serious Hazards of Transfusion (SHOT)¹⁹.

The SHOT project is a system for collecting voluntary, anonymous and independent notifications of adverse transfusion events, with an educational aim of building, through annual reports, an evidence base to support decisions, guidelines, clinical training and improvements in transfusion practices. It covers only serious incidents (grades 2–4) or potentially serious incidents without clinical consequences (mistransfusions of blood components and near-miss events), but does not include minor reactions (allergic or non-haemolytic febrile reactions). It reports the total number of the different adverse events, but not the incidence with respect to the number of blood components transfused.

In France there is a national system with two separate but parallel institutions, one with a regulatory function (Agence Francaise de Sécurité Sanitaire des Produits de Santé – AFSS-APS), the other with executive power (Etablissement Francais du Sang - EFS)^10,20. There is a legal obligation to notify all adverse events, independently of the severity and clinical consequences of the error (grades 0–4), but not near-miss events. Unlike the SHOT, this system explicitly reports the overall number of blood components distributed each year and, therefore, provides the incidence of adverse events with respect to the units transfused.

In Germany¹⁸, blood components are considered pharmaceutical products and monitoring of adverse events related to them is, therefore, covered by the pharmacovigilance system. Authorisation to produce blood components is granted by the Paul-Ehrlich-Institut (PEI) and every adverse transfusion reaction must be reported by the transfusion centres to this institute.

Transfusion errors are not, however, reported to the PEI, since in these cases the reaction is caused by an inappropriate use of the blood component rather than the properties of the product itself. National data on the incidence of ABO-incompatible transfusions are not available in Germany because the data related to incorrect use of blood components are managed within the hospitals in which the incidents occur.

In Italy²¹ there are data up to 2004 on positivity of viral markers in blood donors, collected by the Italian National Institute of Health (ISS) in the Surveillance of Transfusion-Transmissible Diseases (SMITT) programme.

These data cover 83.2% of the units donated. In 2004, the incidence of disease markers among 100,000 periodic donors was 2.1 for HIV, 2.1 for HBsAg, 2.3 for HCV and 6 for syphilis, whereas the incidence among 100,000 new donors was 16.6 for HIV, 214.2 for HBsAg, 205.7 for HCV and 89.9 for syphilis.

French and British haemovigilance: a comparison of the data

Although based on different ways of collecting data, the two major European haemovigilance systems do contain some elements useful for a comparison of the data collected, their interpretation and evaluation of the corrective actions taken. Data on fatal and non-fatal transfusion errors are collected in both countries, differing in this respect from the USA in which only fatal errors are notified.

The notification of transfusion errors in the United Kingdom is voluntary, whereas it is obligatory in France. Many fewer reports are made in England than in France (in 2001: 8.5 notifications/100,000 blood components distributed in the United Kingdom compared to 325.2/100,000 blood components distributed in France), of which 55–56% fell in the category “transfusion of the wrong blood to the wrong patient”, independently of whether or not there were clinical consequences. Given these two contrasting models, with completely different arrangements (voluntary or obligatory notification, notification of all transfusion events or only serious ones), it seems essential to have a standardised method of collecting data in order to be able to compare the incidences and causes of transfusion incidents and their corrective actions in the context of a “European haemovigilance network”.

Given that the SHOT does not report the incidence of events with respect to the total number of blood components transfused, a comparison between the British and French data can only be attempted by considering the absolute numbers of some serious transfusion events occurring in the two countries in similar periods and the fact that the populations are more or less the same, there being about 60,000,000 inhabitants in each country.

Transfusion-related acute lung injury

There were 20 definite cases of TRALI in France¹⁰ compared to 6 in the United Kingdom¹⁵ in 2005; while the numbers fell year by year in the United Kingdom from 2003 to 2005 (20, 12 and 6 in the 3 years considered), the opposite trend was seen in France (6 cases in 2002, 15 in 2003, 16 in 2004 and 20 in 2005).

According to the French Registry, the increase in cases is explained by improved diagnostic techniques. Given that cases of TRALI are more frequent after transfusions of platelet concentrates and units of plasma, according to the SHOT the decreasing frequency of TRALI in the United Kingdom can be the consequence of the exclusion, starting from 2003, of female plasma and platelet donors and, starting from April 2004, of donors transfused after 1980.

The incidences of TRALI due to different types of blood components can be derived from the French haemovigilance system, but not from the British one. These incidences are 1/312,249 (1 death) for red cell products, 1/30,296 (3 deaths) for apheretic platelet units, and 1/106,221 for plasma units¹⁰.

ABO incompatibility

Considering absolute numbers of transfusion events in the United Kingdom in the period 2003–2005, there was a reduction from 26 to 10 in the number of incidents due to ABO incompatibility following transfusions of red cells, but a marked increase in errors in identifying the recipient; on average, the number of deaths per year definitely attributable to ABO incompatibility was one (2 in 2002, 0 in 2003)¹⁵.

In France there was a less significant reduction in the number of incidents due to ABO incompatibility following red cell transfusions (from 34 in the 3-year period 2002–2004 to 29 in the 2-year period 2004–2005). However, in 2005, the number of incidents due to ABO incompatibility following red cell transfusions notified to the EFS was comparable to that reported to the SHOT (9 and 10, respectively). The average number of deaths due to ABO incompatibility in the period from 2000 to 2006 was just above one (2/year in 2000 and in 2003)²⁰.

In France the risk of ABO incompatibility in the period 2000–2004 was calculated to be 1:111,220 blood components transfused²⁰. In the period 1994–1998, the French haemovigilance system calculated that the incidence of major ABO incompatibility was 1:138,672 units of red cell concentrates transfused, with an associated mortality rate of 1:1.802.739²², equivalent to that reported by Linden for NewYork State ⁹.

Infectious risk

Cases of post-transfusion viral infection are extremely rare in both registries. The SHOT reported ten cases of post-transfusion infection due to HBV, two due to HCV and two due to HIV in the period 1995–2004¹⁵.

The French registry reported the residual infectious risk after the introduction of obligatory nucleic acid technology (NAT) testing for HIV, HCV, and HBV²³. In the years prior to NAT testing the incidence of these viral infections was 0.73, 1.16, and 1.81 per 10⁶ donations, respectively, whereas after the introduction of NAT testing these incidences decreased to 0.26, 0.17, and 0.42, respectively. It should, however, be noted that bacterial infections, caused mainly by platelet concentrates, were reported frequently to both haemovigilance systems, although the numbers have been progressively decreasing since 2000 to the present.

Among the strategies useful for reducing bacterial infections, the SHOT mentions only discarding the first 30 ml of blood and more accurate sterilisation of the skin at the time of collecting blood from the donor. The EFS attributes the reduction in the frequency of transfusion-related bacterial infections to information distributed by health authorities and recommendations on the procedure to adopt in the case of bacterial infection, the introduction of hygiene strategies to use during the collection of blood from donors, universal leucodepletion of all corpuscular blood components and the elimination of the first 35 ml of blood taken from the donor.

The EFS reports a bacterial infection rate of 1:224,000 blood components transfused and an associated mortality rate of 1:1,728,000. In 65% of the cases the bacterial infection was caused by platelet concentrates.

Platelet concentrates from pooled donations were responsible for one infection for every 25,000 units transfused, platelet concentrates from apheresis for one infection every 29,000 units transfused and red cell concentrates for one infection every 517,000 units transfused. The mortality rate for transfusion-related bacterial infections is 1:170,000 apheretic platelet concentrates and 1:5,000,000 red cell concentrates. There were no reports of deaths due to bacterial infections following transfusion of pooled platelet concentrates ²³.

Appropriateness of the transfusions

Neither of the two haemovigilance systems provides information on inappropriate transfusions of blood components, except in those cases in which such transfusions were the consequence of an error (for example, a transfusion given on the basis of an unreliable blood count); it is considered that clinical review is a more appropriate instrument than critical incident reporting for evaluating adherence to guidelines on the appropriate use of blood components¹⁶.

Concluding remarks

These data on transfusion safety can be drawn from the specialised literature, but are almost completely ignored in studies dealing with medical error in general because the frequency of adverse events is negligible in comparison with that of all medical incidents. In its epidemiological research of medical errors in hospitals in the USA, the Institute of Medicine (IOM)¹⁴, reported that there were at least 98,000 deaths due to medical errors among 33,600,000 patients admitted to hospital/year. The transfusion-associated mortality rate due to TRALI, haemolytic reactions and bacterial infections was calculated to be about 60–120 patients each year²⁴ for 12,000,0000–14,000,000 units transfused/year.

A comparison of the annual mortality rate due to medical errors and transfusion errors in the USA shows that the former is 800–1,600 higher than the latter. Although these comparisons indicate that satisfactory results have been reached in the field of transfusion safety, data from studies on transfusion risk management, showing a higher incidence of errors at the patients’ bedside^2–8, should induce a more widespread use of computerised systems for correct identification of recipients and blood components, not only in the immunotransfusion laboratory, but also in the ward.

Despite the not negligible incidence of transfusion errors at the patients’ bedside, there is little investment in systems to ensure the correct identification of the patient in the ward, even if there is study showing a tendency to an economic advantage from using systems that lower the transfusion risk from bedside errors in identification (reduction in legal expenses for mistaken identification which, in the USA, are about 725,000 dollars for a legal action involving a death, reduction in health care expenses for the sequelae of transfusion errors and reduction in insurance premiums) or that, at least, justify costs similar to others that, with a lower cost-benefit ratio, are supported in order to give greater transfusion safety (virus-inactivation of plasma, pre-depositing autologous blood for autotransfusion, diagnostic molecular biology tests)²⁵.

An efficient safety system should offer a 99.9% reduction in the incidence of fatal transfusion errors in wards, which is currently 1:1,800,000 units transfused, and that of non-fatal errors, which is currently 1: 14,000 units transfused.

Correct identification of the patient is of critical importance also, and above all, in other medical and surgical disciplines, considering that, according to a report from the British “National Patient Safety Agency”, errors in identifying patients caused at least 500 mistakes every week in treatment administration and diagnostic tests in 2006²⁶.

The benefits of computerised technology, such as bar codes and microchips, which tend to increase correct identification of patients, are not limited only to avoiding errors in the administration of transfusion or pharmacological treatments, but extend to preventing the wrong attribution of results of laboratory examinations. Thus, systems are being developed with the aim of avoiding mistakes in identification of patients both at the time of taking blood samples for analysis and at the time of administering the transfusion or drug treatment.

Various types and levels of action can be taken in the field of transfusion medicine:

1. bedside checks of the patient’s blood group and comparison with that of the blood component to be transfused;

2. use of identifying wrist bands;

3. use of “barrier” systems.

Besides all the strategies discussed above, it can be hypothesised that complete transfusion safety will be reached when methods to make red blood groups universal become available and when any contaminants of all types of blood components can be inactivated.

Table I.

National haemovigilance systems

- USA: obligatory notification to the FDA only of fatal transfusion-related events: - TRALI (30%), haemolytic transfusion reactions (16%) bacterial contamination (16%) - United Kingdom: voluntary reporting to the Serious Hazards of Transfusion (SHOT) only of serious transfusion incidents - France: obligatory notification to the Etablissement Francais du Sang (EFS) of all adverse events, regardless of their severity - Germany: monitoring adverse events is part of the duties of the pharmacovigilance system, managed by the Paul Ehrlich Institute (PEI); adverse transfusion events not related to the quality of the product are reported to the institute to which the transfusion centre belongs, so there are not national data on errors in the use of blood components - Italy: data are available on the presence of viral markers in blood donors; these data are collected by the Italian National Institute of Health (ISS): - cases per 100,000 periodic donors: 2.1 HIV, 2.1 HBsAg, 2.3 HCV, 6 syphilis, - cases per 100,000 new donors: 16.6 HIV, 214.2 HBsAg, 205.7 HCV, 89.9 syphilis.

Table II.

Comparison of the haemovigilance systems used in France (EFS) and the United Kingdom (SHOT).

EFS

- Notifications are obligatory - All adverse events - in 2001: 325.2 notifications/100,000 blood components distributed - Information on the incidence of adverse events with respect to the total number of blood components transfused

SHOT

- Notifications are voluntary - Only serious events - in 2001: 8.5 notifications/100,000 blood components distributed - No information on the total number of blood components transfused.