Hoffman Jerome R. M.D.; Jaber, Aileen Jadallah M.D.; Schriger, David L. M.D., M.P.H.Journal of Clinical Gastroenterology: August 1991 Article: PDF Only Buy Abstract We hypothesized that selective ordering of serum amylase in the emergency department (ED) is justified because (a) most patients with elevated amylase can be prospectively identified by characteristic clinical findings, and (b) the diagnosis of pancreatitis is usually predominantly based on clinical findings, since amylase is known to be neither sensitive nor specific for pancreatitis. The study population included 133 consecutive patients with a chief complaint of abdominal pain who had amylase drawn over a 2-week period at a university hospital ED. Patients with known major trauma were excluded. Emergency department and hospital charts were reviewed for selected clinical variables. The first part of our hypothesis was evaluated by comparing clinical characteristics of cases (elevated amylase) and controls; the second part was tested by comparing clinical findings and amylase in cases (patients diagnosed as having pancreatitis) and controls. We found that 17 patients with and 116 without elevated amylase were similar with regard to all clinical variables, and that no combination of findings could be used to predict elevated amylase. Amylase level was not predictive of an ultimate diagnosis of pancreatitis, which was, however, strongly related to classical clinical findings. Pancreatitis risk factors, epigastric pain and tenderness, radiation of pain to the back, and nausea and vomiting were each statistically more common in patients diagnosed as having pancreatitis (regardless of amylase) than in patients in whom pancreatitis was excluded despite elevated amylase; all patients diagnosed with pancreatitis had at least two of these. Thus, selective ordering of amylase on the basis of clinical characteristics fails to identify a large proportion of patients with elevated amylase. Routine ordering of amylase is not justified, however, if clinicians (as in this study) base diagnosis and treatment on purely clinical features. © Lippincott-Raven Publishers.