Hemoconcentration as an early risk factor for necrotizing pancreatitis.

The aim of our study was to determine whether measurement of serum hematocrit during the first 24 h helps in distinguishing necrotizing from mild pancreatitis. From May 1992 to June 1996, a case-control study was performed with cases of patients with necrotizing pancreatitis. We selected as a control the next patient admitted with mild pancreatitis. There were 32 patients in each group. Logistic regression identified an admission hematocrit of > or = 47% and a failure of admission hematocrit to decrease at 24 h as the best binary risk factors for necrotizing pancreatitis. At admission, more patients with necrotizing pancreatitis than with mild pancreatitis had a hematocrit > or = 47% (11/32 vs 3/32; p = 0.03). At 24 h, 15 additional patients with necrotizing pancreatitis versus only one with mild pancreatitis showed no decrease in admission hematocrit (p < 0.01). Thus, by 24 h, 26 of 32 patients with necrotizing pancreatitis versus only four of 32 patients with mild pancreatitis met one or the other criterion (p < 0.01). The sensitivity and specificity at admission were 34% and 91%; at 24 h, 81% and 88%. Hemoconcentration with an admission hematocrit > or = 47% or failure of admission hematocrit to decrease at approximately 24 h were strong risk factors for the development of pancreatic necrosis.