How Does Workload Affect Test Ordering Behavior of Physicians? An Empirical Investigation

Abstract

We investigate the relationship between workload and the test ordering behavior of physicians, in an operational context where test ordering may act as a substitute for service process time. We define workload in two forms: the unfinished workload, that is, the number of patients waiting to be examined, and the finished workload, that is, the cumulative number of patients examined before a given patient in a work day. We hypothesize that physicians order more diagnostic tests at higher unfinished workload levels to speed-up, increase patient satisfaction and due to increased stress, and order fewer diagnostic tests at higher finished workload levels due to fatigue. Empirical analysis of data from a public training and research hospital provides strong evidence supporting both hypotheses: As the number of patients waiting increases from 0 to 28 patients, 2.3 more tests are ordered per patient on average, indicating a 22% increase. By contrast, as the finished load increases from 1 to 48 patients, 2.3 fewer tests are ordered per patient, corresponding to a 18% decrease. Our results suggest that workload smoothing mechanisms such as appointment systems alone may have significant cost saving implications. We also report the results of several robustness tests which confirm our results.