Vantage: Exploring Variability in Inpatient Care Through Physicians’ Orders

Abstract

The inpatient care two seemingly identical patients receive for the same initial condition may be quite different. This variability in care affects patient outcomes and often increases costs. The goal of measuring care variability is to identify and prioritize services for care standardization. Cost variability is the current standard surrogate measurement, but is imperfect. Each patient group varies differently due to differences in severity of illness, comorbidities, and socio-economic status. A surrogate measure based on inpatient orders could provide unique insights, because inpatient care is largely enacted through orders. The authors extracted orders from Vanderbilt University Medical Center adult [18-64] inpatients with admissions between 07/01/2013 and 12/31/2016. The authors grouped the order descriptions into higher-level UMLS concepts to account for clinically redundant actions. For example, the authors grouped medication orders by primary ingredient(s) and administration route using RxNorm. The authors derived order statistics for each clinical domain (e.g. nursing, pharmacy, radiology). The authors created a measure for order variability by averaging some of their order statistics. They tested an order variability model of the actual length of stay to the expected length of stay ratio (a standard measure of inpatient quality) against a cost variability model. Both models were adjusted for covariates derived from the literature. The authors found that order variability had a significantly better adjusted-R2 than the cost variability model. Although challenging to work with, orders have the potential to elucidate questions regarding variability and quality.