The impact of peer management on test-ordering behavior.

Abstract

Laboratory testing of hospitalized patients, although essential, can be expensive and sometimes excessive. Attempts to reduce unnecessary testing have often been difficult to implement or sustain. Use of peer management through a resource utilization committee (RUC) to favorably modify test-ordering behavior in a large academic medical center. Interrupted time-series study. Medical center with inpatient care provider order entry (CPOE) system and database of ordered tests. Predominantly housestaff physicians but all clinical staff (attending physicians, housestaff, medical students, nurses, advance practice nurses, and other clinical staff) at Vanderbilt University Hospital who used CPOE systems. The RUC analyzed the ordering habits of providers during previous years and made 2 interventions by modifying software for the CPOE system. The committee first initiated a daily prompt in the system that asked providers whether they wanted to discontinue tests scheduled beyond 72 hours. After evaluating this first intervention, the committee further constrained testing options by unbundling serum metabolic panel tests (sodium, potassium, chloride, bicarbonate, glucose, blood urea nitrogen, and creatinine tests) into single components and by reducing the ease of repeating targeted tests (including electrolyte, blood urea nitrogen, creatinine, and glucose tests; electrocardiography; and portable chest radiography). Pre- and postintervention volumes of tests; proportion of patients with abnormal targeted chemistry levels after 48 hours; rates of repeated admission, transfer to intensive care units, and mortality; adjusted coefficient of variation for test ordering; and length of stay. Voluntary reduction of testing beyond 72 hours (first intervention) decreased orders for metabolic panel component tests by 24% (P = 0.02) and electrocardiograms by 57% (P = 0.006) but not orders for portable chest radiographs. Prospective constraints on recurrent test ordering with panel unbundling (second intervention) produced an additional decrease of 51% for metabolic panel component tests (P < 0.001) and 16% for portable chest radiographs (P = 0.03). Incidence of patients with abnormal targeted blood chemistry levels after 48 hours decreased after the intervention (P = 0.02). Postintervention-adjusted coefficients of variation decreased for metabolic panel component tests (P = 0.03) and electrocardiography (P = 0.04). Rates of (adjusted) monthly readmission, transfers to intensive care units, hospital length of stay, and mortality were unchanged. Other activities occurring during the time period of the interventions might have influenced some test-ordering behaviors, and we assessed effects on only a limited number of commonly ordered tests. Peer management reduced provider variability by addressing the imperfect ability of clinicians to rescind testing in a timely manner. Hospitals with growing health care costs can improve their resource utilization through peer management of testing behaviors by using CPOE systems.