I read with great interest the study by Meek et al1 in which they concluded that a serum glucose concentration of at least 150 mg/dL (8.3 mmol/L) is the best predictor of complications of gallstone pancreatitis. I have some reservations regarding their conclusions.
A variable that differs significantly between mild and severe pancreatitis is not necessarily a good predictor of severity.2 Serum glucose concentration is not a good predictor simply because its positive predictive value (41%) is low. This means that 60% of patients labeled severe according to this variable actually have mild pancreatitis. The false-positive results are too high.
Abu-Zidan F. Predicting Severe Pancreatitis. Arch Surg. 2001;136(10):1210. doi:10.1001/archsurg.136.10.1210