I read the article by Restellini et al. with interest. The authors analysed 1677 patients with nonvariceal upper gastrointestinal bleeding, collecting clinical parameters (gender, age, comorbidity, ASA score, melaena, haematemesis, drug usage, etc.) and correlated these with rebleeding. After adjusting for confounders in multivariable analysis, they found transfusion of red blood cells within 24 h of presentation to be significantly and independently associated with an increased risk of rebleeding (OR: 1.8, 95% CI: 1.2–2.8, P = 0.005).
I am afraid that there are some points which need to be discussed to avoid misleading readers. Restellini et al.'s finding may be new, because using blood transfusion as a predictor for rebleeding has not been mentioned in recent consensus, ACG guidelines or review articles.[2-4] From a clinical point-of-view, doctors usually order blood transfusion for patients with low haemoglobin or unstable vital signs after resuscitation.
Recent ACG guidelines suggest a blood transfusion if haemoglobin is below 7 g/dL. In Restellini et al.'s observation, patients who received a blood transfusion may indicate a severe condition in this group therefore the rebleeding rate is theoretically high. It is unreasonable to regard blood transfusion as the cause of rebleeding.
There are some frequently used scoring systems for predicting prognosis which can be used to inspect this study. The Blatchford scoring system is widely used consisting of blood urea nitrogen, haemoglobin, systolic blood pressure, pulse, presentation with melaena or syncope, and presence of hepatic or cardiac failure. In the transfusion group of Restellini et al.'s study, more patients with lower haemoglobin (8.21 vs. 11.46 g/dL), more shocked patients at presentation (43.5% vs. 18.4%), more patients presented with melaena (76.1% vs. 59.4%), and more cases of comorbidity (2.7 vs. 2.3) were found.
Therefore, it is obvious that the transfusion group had higher Blatchford scores than those without transfusion. The higher the Blatchford score, the higher the rebleeding rate will be observed. If we checked with another frequently used system, the Rockall score, preadmission scores (age, shock comorbid illness) also indicate that the transfusion group was more severe than those without transfusion.
From an endoscopic point-of-view, stigmata of recent haemorrhage (SRH) over the ulcer base are linked to rebleeding. There were more patients with SRH in the transfusion group (45.6% vs. 24.6%), and therefore, more endoscopic therapies were required in this group (44.6% vs. 24.3%). Therefore, rebleeding rates will be higher in the transfusion group.
Thus, based on clinical data, the rebleeding rates are linked to the severity of bleeding within the patients rather than the blood transfusion itself. I agree with Restellini et al.'s suggestion that further studies are required before any firm conclusions can be drawn.