Summary
Background
Previous studies in upper gastrointestinal (GI) bleeding have reported inconsistent outcomes about packed red blood cell (PRBC) transfusion practices.
Aim
To assess whether PRBC transfusion is more likely to be dangerous in variceal bleeding than in non‐variceal bleeding due to concern of over‐transfusion leading to elevated portal pressure.
Methods
We used the Nationwide Inpatient Sample (1999–2018). We identified patients with upper GI bleeding using an algorithmic approach, categorising bleeding from non‐variceal or variceal sources. Our primary outcome was all‐cause inpatient mortality. To control for the severity of bleeding, we performed propensity matching of baseline features, including age, gender, the presence of shock, the need for ICU care and co‐morbidities. We also examined PRBC transfusion, inpatient mortality and hospitalisation rates for both populations.
Results
We included 10,228,524 upper GI bleeding discharges; 755,135 patients had variceal bleeding. After propensity matching, PRBC transfusion in variceal bleeders was associated with a 22% increase in inpatient mortality, whereas non‐variceal bleeders had a 9% increase in inpatient mortality. Compared to non‐variceal bleeders receiving blood transfusion, variceal bleeders had nearly four‐fold higher odds of inpatient mortality (propensity‐matched OR: 3.8; 95% CI: 3.7–3.8; p < 0.001). Notably, PRBC transfusion rates in both groups have declined since 2011, although it has remained higher in variceal bleeders. Mortality for upper GI bleeding has been declining since 1999.
Conclusions
Although decreased over the last decade, PRBC transfusion rates remain high for variceal bleeders. In addition, PRBC transfusion appears to be more detrimental in variceal bleeders than in non‐variceal bleeders.