Increasing the appropriateness of outpatient imaging: effects of a barrier to ordering low-yield examinations.

Purpose: To determine the effect of a computerized radiology order entry system rule that prevented nonclinician support staff from completing orders for outpatient computed tomographic, magnetic resonance imaging, and nuclear medicine examinations that received initial low-yield decision support scores in the order entry system.

Materials And Methods: This retrospective HIPAA-compliant study was approved by the institutional review board; the requirement for informed consent was waived. The control group consisted of 42737 consecutive orders for examinations in which decision support was provided that were placed from April to December 2006. The study group consisted of 76238 consecutive orders that were placed from April to December 2007. During the latter time period, a new rule in the order entry system was implemented: Examinations that had low-yield decision support scores could not be scheduled when the orders were placed by nonclinician support staff. To schedule the blocked examinations, the responsible clinician was required to personally log in to complete the process. System event logs and records of outpatient imaging procedures were extracted, counted, and analyzed to determine which ordering sessions resulted in examinations being scheduled and performed and which sessions resulted in modified or cancelled examinations. Results were correlated with user status and decision support scores. The Cochran-Mantel-Haenszel technique was used to control for the status of the order initiator and to allow testing for significance of the effect of the intervention on the "fate" of ordering events.

Results: After the intervention, the proportion of total examination requests initiated by clinicians directly logging in almost doubled: from 11,243 (26.31%) of 4,737 to 41,450 (54.37%) of 76238 examinations (P < .001). The fraction of low-yield (decision support score, 1-3) examinations requested through the order entry system that were later scheduled and performed decreased from 2106 (5.43%) of 38,801 to 1261 (1.92%) of 65,765 (P < .001). This is in contrast to requests for examinations with higher initial decision support scores that were not affected by the policy change and were scheduled at the same rate (relative risk, 0.988) before and after the change.

Conclusion: A simple change in the business logic of the order entry system resulted in a substantially decreased rate of low-yield imaging examinations and a markedly increased percentage of tests personally ordered by clinicians.