2016 Volume 36 Issue 5 Pages 241-251
BACKGROUND: Due to the high number and variety of orders issued by health professionals, computerized physician order entry (CPOE) or hospital information systems (HIS) incorporate checking functions to prevent erroneous orders. Although these functions are widely implemented in Japanese hospitals, their clinical effectiveness remains unclear.
PURPOSE: To study the frequency of alerted orders and to review the effectiveness of the checking function in an HIS.
METHODS: We prospectively collected data on alerted orders and physician responses to system alerts from January to June 2015 in Yamaguchi University Hospital.
RESULTS: Among 1,836,209 orders in the study period, 144,423 alerts (7.7%) were generated. Major causes of ‘erroneous orders’ which require modification for processing included errors arising from department or system restrictions. Major causes of ‘order warnings’ (93,301 alerts) included redundant orders of drugs or laboratory tests (65.4%). Warnings were triggered in 702 (0.16%) of 452,487 prescription orders to alert prescribers to patient drug allergies or known history of adverse events to contrast agents. Among 95 prescription orders where label and dosage specified patient drug allergies were alerted, 39 (44.8%) orders were modified or cancelled, thus considered useful.
CONCLUSIONS: Low rates of modifications or cancellations for alerted orders implied suboptimal effectiveness of the current algorithm, with a risk of eventually inducing ‘alert fatigue’. Further multi-center studies are needed to validate these findings and optimize the check algorithm.
