Three Kinds of Lies: Lies, Damned Lies, and Statistics
Two recent reports appeared separately in the ABCNewsletter. The Healthcare Cost and Utilization Project (HCUP) Statistical Brief, #149 from the federal Agency for Healthcare Research and Quality (http://1.usa.gov/14QoU6H) lists transfusion as the most frequent procedure performed in US hospitals during 2010, while the 2011 National Blood Collection and Utilization Survey (NBCUS) (www.hhs.gov/ash/bloodsafety/nbcus/) shows significant decreases in blood transfusions since 2008. I have received inquiries about the apparent discrepancy between the reports, but there is probably no conflict - they count apples and oranges.
HCUP tabulated what they could methodologically from standardized databases, while NBCUS elicited the specific information that blood bankers are most interested in. HCUP uses coding data from stays in short-term, non-Federal, general, and other hospitals, and should be quite accurate, provided that those institutions are coding and reporting appropriately. These data represent aggregation of ICD-9-CM codes for all autologous (including perioperative salvage) and allogeneic components, plus volume expanders and coagulation concentrates (undoubtedly resulting in inflation of transfusion rates) using HCUP Clinical Classification Software (CCS) to collapse the data into "meaningful groupings." Transfusion was reported during 91 stays/10,000 population during 2010, an increase of 126 percent since 1997. The authors cannot tell us how many allogeneic units are transfused during each episode. The data are therefore not useful in judging the volume of transfusion in the US. The comparator year (two years before TRICC was published) is ancient history in the context of blood management. The comparison also misses the peak of red blood cell (RBC) transfusion in the mid-2000s. Whether the patient populations from the two time frames are comparable is another confounder.
When these data were distributed, blood bankers were skeptical, being acutely aware of declining RBC and plasma use in our own systems during recent years. By contrast, NBCUS shows an 8.2 percent drop in RBC transfusions since 2008. While generally unpublished, it is hardly anecdotal to note that for many ABC members, the decline is greater than 8.2 percent and accelerating. Also, based on trends at hospitals served by Mississippi Valley Regional Blood Center and discussions with my colleagues, I suspect single-unit RBC transfusion (an anathema during training to those of us of a certain age) has caught on as the default strategy in many facilities. Anyway, it is possible that the population frequency at which inpatients are transfused is higher now than in the (medically remote) past, but the number of units to which they are exposed is falling - probably a good thing for patients. It is also a good thing for us to understand the two reports to articulate clearly the "state-of-the-art" in blood use to our varied constituencies.
Louis Katz, MD, Executive Vice President, Scientific, Medical, Technical, Quality, and Regulatory; LKatz@americasblood.org