Modeling resource use with time equations: Empirical evidence

Abstract

We test whether cost-object-level characteristics significantly and predictably affect resource use within a single standardized activity. Using data on operating time for joint replacements, we find reliable effects for age, race, and gender after controlling for body mass index and other comorbidities. Modeling such granular variation is practically not possible in a classic two-stage system. Thus, this finding provides strong support for the use of time equations, as in time-driven activity-based costing, to model resource use. Supplementary analyses of hospital length of stay and gross charges show that, in addition to the expected direct effects, hospital-level characteristics moderate how patient-level attributes affect resource use.