To the Editor

Dr. McLesky’s article “Bar Coding May Reduce Drugs Errors” was remarkable for its optimism about this technology. The few research-based examinations of the efficacy of these systems do not support such an optimistic view. Instead, they indicate that, like other complex information technology, bar coded medication administration (BCMA) is likely to create new problems, open new pathways to failure, and generate a new variety of counter-productive incentives in an environment already rather well-burdened with them.

One published study of the Veterans Administration BCMA system shows that BCMA application produces rigid sequencing of work, narrow “keyhole” computer generated views of work process, and potential for precise timing and tracking of bar code reading activities—all characteristics that poorly match the needs of workers in busy clinical settings.¹

For example, BCMA may increase both the pressure to produce records showing “on-time” delivery and the penalties for failing to do so. The time delays associated with ordinary order entry, processing, review by pharmacy, and medication “delivery” to the caregiver fluctuate with the tempo of operations. Confronted with interruptions or critical needs, caregivers may divert the drug from one patient to another. From a BCMA perspective, this is either wrong drug or wrong patient, and generates a time-stamp that is difficult or impossible to reconcile.

Similarly, apparent improved performance on “right time” measures may reflect work-arounds as much as real progress. We have been told of instances where nurses have photocopied patient wristbands and scanned them to generate “right time” documentation for the computer while actually delivering the medication at a different time. The incentives to do this are the difficulty in working with balky equipment, the rigid structure of work enforced by the information technology, and the potential for becoming a statistical quality control point outside “acceptable limits.” The important patient safety consequence is the loss of BCMA’s potential value in assuring the “right drug” delivered to the “right patient.” There are no data on how often this sort of work around is actually used; it is likely more common than we could be comfortable with.

Audits of administration time made possible by BCMA are a potent incentive for practitioners to adopt practices that increase their “on time delivery,” but undo the potential increase in safety from such systems. But what are the underlying drivers for the practices that we perceive to be unsafe? Is it laziness? Moral turpitude? Willful disregard of safety by capricious practitioners? We think not. Instead, the causes of this behavior include ongoing short-staffing of nurses, the lack of high quality support tools to manage work at the bedside, and so forth. None of these are addressed by the institution of BCMA.

Authors advocate eliminating timing data

Avoiding perverse incentives may mean sacrificing lower priority goals for higher ones. We have a concrete recommendation for proponents of BCMA. Throw away the timing data. The pressing motivation for BCMA was the desire to prevent wrong medications being given by providing computer-based matching of patient with drug. Make that the cornerstone of the BCMA safety program. Rather than have the computer record the time the band was “scanned,” have it record only that the drug was given sometime during the shift. Eliminate all automated recording of timing. Allow workers to enter the actual time if they choose to do so. Doing this will remove the incentive to “game” the system, and send a clear message about what really matters—that matching drugs and patients is vastly more important than small discrepancies in administration times.

Greatest value is insuring the right drug for the right patient

The introduction of BCMA into more widespread clinical use will reduce the incidence of some problems and create new ones. Careful design of such systems and deliberate avoidance of perverse incentives will increase the chances that they improve safety rather than hamper it.

Michael F. O’Connor, MD
Mark Nunnally, MD
Richard I. Cook, MD
Chicago, IL

References

1. Patterson ES, Cook RI, Render ML. Improving patient safety by identifying side effects from introducing bar code medication administration.
   J Am Med Inform Assoc 2000;9:540-53.