摘要Background Hemoconcentration may be an important factor that determines the progression of severe acute pancreatitis (SAP). In addition, it has been proposed that biomarkers may be useful in predicting subsequent necrosis in SAP. However, it is still uncertain whether hemodilution in a short term can improve outcome. We aimed to investigate the effect of rapid hemodilution on the outcome of patients with SAP.Methods One hundred and fifteen patients were admitted prospectively according to the criteria within 24 hours of SAP onset. Patients were randomly assigned to either rapid hemodilution (hematocrit (HCT) <35%, n=56) or slow hemodilution (HCT 235%, n=59) within 48 hours of onset. Balthazar CT scores were calculated on admission, day 7, and day 14, after onset of the disease. Time interval for sepsis presented, incidence of sepsis within 28 days and in-hospital survival rate were determined.Results The amount of fluid used in rapid hemodilution was significantly more than that used in slow hemodilution (P <0.05) on the admission day, the first day, and the second day. There were significant differences between the rapid and slow hemodilution group in terms of hematocrit, oxygenation index, pH values, APACHE II scores and organ dysfunction at different time during the first week. There were significant differences in the time interval to sepsis in rapid hemodilution ((7.4 1.9) days) compared with the slow hemodilution group ((10.2 2.3) days), and the incidence of sepsis (78.6%) was higher in the rapid group compared to the slow (57.6%) in the first 28 days. The survival rate of the slow hemodilution group (84.7%) was better than the rapid hemodilution (66.1%. P<0.05).Conclusions Rapid hemodilution can increase the incidence of sepsis within 28 days and in-hospital mortality. Hematocrit should be maintained between 30%-40% in the acute response stage.