Original Article: PDF OnlyGreen Lee A. MD MPH; Becker, Mark P. PhDMedical Care: November 1994 - p 1086-1097 Buy Abstract The authors tested the “uncertainty hypothesis,” which holds that variations in rates of hospitalization or surgeries across small geographic areas reflect differences in physicians' decision making when confronting uncertainty. A small-areas variation analysis of suspected acute cardiac ischemia (ACI) admissions in northern Michigan was performed, and two de-mographically nearly identical towns differing by a factor of 3 in ACI admission rates were selected. Medical records of all patients evaluated in the emergency departments of these hospitals for suspected ACI in 1988 were abstracted retrospectively. Probabilities of ACI were objectively estimated using the Acute Cardiac Ischemia Time-Insensitive Predictive Instrument. Logistic regression of admission on patient characteristics, other illnesses, probability of ACI, and community revealed no difference in admission decisions between the two hospitals (odds ratio for community = 0.766, 95% confidence interval, 0.542–1.08, n = 787, P > .1). Nearly twice as many patients with ACI presented to the emergency department of the high-admitting hospital as to the low-admitting hospital. The authors conclude that, at least for ACI, population-based area discharge rates do not necessarily reflect case-based decision rates. Drawing inferences regarding physician decision making from discharge or claims datasets may lead to error. © Lippincott-Raven Publishers.