Blood transfusion during hospitalization (Tx-hospitalization) in the U.S. is very common in medical practice. In-hospital blood transfusions are indicated for multiple medical, surgical and hemorrhagic conditions. However, descriptions of the patient populations receiving transfusions in U.S. hospitals are vague. The purpose of this study is to use nationally representative hospital discharge data to provide information regarding the patient populations and associated diagnoses that have caused the demand for blood transfusions in U.S. hospitals. Data were obtained through cross-sectional analysis of the 1997 through 2006 Healthcare Cost and Utilization Project (HCUP) Nationwide Inpatient Sample (NIS). The NIS is the largest, multi-year, all-payer inpatient care database that is publicly available in the U.S. The NIS contains data from 5 to 8 million unweighted hospital stays from about 1,000 hospitals sampled to approximate a 20-percent stratified sample of U.S. community hospitals. The NIS is drawn from the States participating in HCUP, which represent 90% of all hospital discharges in 2006. The NIS includes weights that allow calculation of reliable, representative national estimates of care in U.S hospitals. Blood transfusion was identified using the Clinical Classifications Software (CCS) procedure grouping 222. In 2006, 2.4 hospital hospitalizations included at least one blood product transfusion compared with 1.1 million in 1997. This 117% increase in Tx-hospitalizations was steady, with an average of 140,000 additional Tx-hospitalizations during each year. During that same period, the total number of U.S. hospitalizations increased 14%. Blood transfusions were among the top 10 procedures performed in all age groups. The elderly (>65+ years of age) accounted for the greatest number of hospitalizations with a transfusion with 1.4 million in 2006 compared to 622,000 in 1997. The most common type of transfusion was packed red cells, which accounted for about 90 percent of all types of transfusions (2.1 million of the 2.4 million transfusions in 2006). Coding of diagnosis at discharge was utilized to determine the conditions contributing to the increased utilization of transfusions. Transfusion was coded as a primary procedure during hospitalizations in which more invasive procedures were not performed. From 1997 to 2006, the fraction of hospitalizations with transfusion listed as the principle procedure was stable, and represented one-fourth of all Tx-hospitalizations. Within that group, the most common principle diagnoses were coded as anemia (15.8%), gastrointestinal hemorrhage (7.3%), CHF (5.5%), pneumonia (5.2%), and septicemia (5.0%). Cancers accounted for <1%. Transfusions were coded as a secondary procedure another invasive procedure occurred during the same hospitalization and represented the remaining three-fourths of cases. In that group, the most common principle diagnoses were osteoarthritis (8.3%), GI hemorrhage (6.5%), fracture of hip (5.1%), septicemia (4.3%), and complication of a device, implant or graft (4.2%). The most common principle procedures were UGI endoscopy (9.1%), hip replacement (6.0%), knee arthroscopy (6.0%) hip fracture repair (4.4%) and CABG (3.7%). In separate analyses, it was demonstrated that 20 percent of all hospitalizations for knee arthoplasty and 30 percent for hip replacement included a transfusion in 2006. In summary, hospital discharge data from the NIS provides critical and detailed clinical information on uses and trends of blood transfusions in U.S. hospitals. There has been a steady and significant increase in Tx-hospitalizations during the last decade that far exceeds the overall increase in hospitalizations (117% vs. 14%). A relatively small group of diagnoses including gastrointestinal bleeding and lower extremity arthroplasty represent one-third of the Tx-hospitalizations. Understanding the clinical demand for blood components in U.S. hospitals should facilitate optimal management of these limited and life-saving resources.

Disclosures: No relevant conflicts of interest to declare.

Author notes

Corresponding author

Sign in via your Institution