ES Patterson, Ohio State University; ML Render, VA Cincinnati, Midwest Patient Safety Center; G Coyle, VA Martinsburg; DD Woods, Ohio State University; RI Cook, University of Chicago

Objectives: Nearly 20% of the adverse events described in the Harvard Medical Practice Study were drug related. Epidemiologic studies report 1-35% of hospitalized patients experience adverse drug events. The VHA has pioneered bar coding in medication administration (BCMA) to reduce medication error. BCMA is a software product that uses wristband barcodes to identify patients and verifies accurate medication names, doses, forms, and administration times by having nursing personnel scan unit dose medication barcode labels immediately prior to administration. In addition to new capabilities, new technologies often have unintended effects due to human computer interaction (HCI) deficiencies and mismatches between system designs and the structure of technical work, creating the potential for new paths to system failure. Human factors methodologies can be used to predict some of these new paths, allowing intervention prior to accidents. We asked: What new paths to medication administration errors, if any, does the introduction of BCMA create?

Methods: We conducted targeted ethnographic observations at two sites at one hospital prior to the introduction of BCMA (12 medication passes, 27 hours) and at three sites at two hospitals following the introduction of BCMA (12 medication passes, 32 hours). Ethnographic observation is a methodology derived from anthropology that emphasizes detailed data collection of observable human behavior. Our process data was coded into behavioral protocols to enable abstraction and comparison of patterns.

Results: Three unintended side effects resulting from design decisions in the BCMA system are likely to create new paths to medication adverse events (MAEs): 1) reduction in available data, 2) unobservable automated actions of the system, 3) poor usability of some documentation features. First, the BCMA interface does not provide information about pending and discontinued medication orders, which was available with the previous medication system. Without this information, nursing personnel can no longer double-check the status of medications intended for administration but automatically discontinued, not cosigned, or delayed at pharmacy, creating an opportunity for missed medications. Second, BCMA automatically performs some actions without alerting its users (e.g., deleting IV medications from active orders several hours after the scheduled administration time). These actions are surprising to users and could lead to missed or off-cycle medications. Third, poor usability of some documentation features contributed to uncertainty about whether medications that were not documented as administered in BCMA had actually been administered, leading to opportunities for missed medications or overdoses.

Conclusions: Several design choices in BCMA create the potential for new paths to medication administration errors. The risk of new types of adverse events may be reduced by system redesign, or changes to medication administration procedures, before adverse events occur and without losing the new core capabilities of the BCMA system. This research demonstrates how human factors knowledge of the impact of human-computer interaction design on human performance can advance safety in healthcare.

Impact: Bar coding technology is viewed as one source for improving safety in health care. Minimizing unexpected side effects using human factors methods allows us to maximize the technology's potential capabilities.