The epidemiology of transfusion: where blood goes and why we should care about it

B lood transfusion is one of the most commonly performed clinical procedures. The provision of blood for transfusion at the patient’s bedside is dependent on a complex series of events starting with recruitment of blood donors and the collection of blood, and it has much in common with supply chains in commerce: for example: the supply of coffee from the starting point of growing coffee beans to your cup for your breakfast or midmorning break. It requires a great deal of planning and logistical effort to ensure that the blood for transfusion (or your coffee) is safe and appropriate (“right product”) for the recipient (“right patient”) and is provided at the right location (“right place”) in a timely manner (“right time”). The essential activities in a supply chain must be performed efficiently and the cost of the product must be reasonable; otherwise alternative products will be sought. As a transfusion medicine physician with a limited knowledge of supply chains, I was interested when conducting some background reading for this editorial to find articles referring to “principles of supply chain management” in commerce and that one commonly quoted article included statements such as “listen to market signals and align demand planning accordingly”; “develop a technology strategy that gives a clear view of the flow of products, services, and information”; and “customize the logistics network to service requirements.” These principles seem logical and applicable to the supply of blood, but have proved difficult to implement. Recent editorials in TRANSFUSION have highlighted the lack of comprehensive information about the blood supply in the United States and the limitations that this inadequate database imposes on demand planning and the development of policy. The resilience and sustainability of the blood supply depend on reliable data, the key message being “you can’t manage what you can’t measure.” The provision of comprehensive data on the use of blood has proved particularly challenging to achieve in the United States, probably largely because of its relatively fragmented and competitive blood supply and delivery systems in contrast to countries with national blood services and health services. In this issue of TRANSFUSION, Karafin and colleagues on behalf of the National Heart, Lung, and Blood Institute Recipient Epidemiology and Donor Evaluation Study-III (REDSIII) describe a highly commendable initiative to provide comprehensive data on transfusion recipients in a subset of hospitals in the United States. In this article, the authors successfully used hospital electronic records to collect transfusion and patient data from 12 community and academic hospitals in four regions. Clinical data were collected for all inpatients during the study period and more comprehensive information was included for patients receiving a blood transfusion; data were also collected from outpatients receiving a transfusion or who had a type and screen or type and crossmatch. The data included patient age, sex, ethnicity, primary diagnosis, type of blood product, preand posttransfusion blood counts, and clinical outcomes. Transfusion reactions were also documented. Over 2 calendar years, 2013 and 2014, there were more than 641,751 inpatient encounters, 80,362 (12.5%) of which involved transfusion, most commonly red blood cell (RBC) transfusions (10.9%), followed by transfusions of platelets (PLTs; 3.2%) and plasma (2.9%). The most common diagnoses where blood was used were blood, infectious, cardiovascular, and gastrointestinal diseases, neoplasms, and injury. Transfusion incidence had a bimodal distribution with incidence peaks at 2 to 5 and 70 to 89 years. Inpatient mortality was higher in transfusion recipients (6.9%) than in transfused patients (1.4%) and was higher for encounters where PLTs or plasma units were transfused either alone or in combination with other products as reported previously; both of these findings reflect the fact that patients who receive transfusions, especially plasma or PLTs, are sicker than other patients and thus more likely to die. A median of 1 RBC unit was transfused per transfusion episode, and the median pretransfusion hemoglobin level was 7.6 g/dL, both reflecting the adoption of restrictive RBC transfusion strategies. Severe transfusion reactions such as transfusion-related acute lung injury (0.1 events per 1000 transfused encounters) and acute hemolysis (0.06 events per 1000 transfused encounters) were rare; more than 75% of reported reactions were either febrile nonhemolytic (6.4 events per 1000 transfused encounters) or allergic reactions (4.9 events per 1000 transfused encounters). The editorial does not represent the official position of any organization I am a member of. doi:10.1111/trf.14385