The impetus to create a paperless medication ordering system continues unabated since the late 1990s. CPOE systems are expected to affect the high proportion (56%) of medication errors that occur during prescribing (3). A medication error is defined as “a failure in the process of treatment that leads to, or has the potential to lead to, harm to the patient” (4). A goal of Institute of Medicine (IOM) was to have handwritten orders eliminated by 2010. However, currently only about 10% of hospitals have a CPOE system; nonetheless, this number had doubled in 10 years (5). The VA health administration is usually given as an example of a system that successfully implemented CPOE. Transition to a computerized medical record is not easy since old habits are hard to break. Consistency in enforcing the use of technology and having administrators and supervisors who themselves use and understand the system are key.

Designing good computerized systems should be a clear goal, because using automation that has been poorly designed and tested will only make it easier and faster to achieve undesirable results. Automation appears to work well for repetitive tasks while the human can spend their time with more complex tasks that require discernment, communication, cooperation, creativity, and flexibility (6). Humans are still needed to make complex decisions (7,8). These systems will avoid reliance on memory, recall, or vigilance. While the latter issues may not be totally eliminated, they should not be the main way to avoid errors and patient harm. In the planning process it is imperative to include as many end users as possible so that their opinions and desires can be incorporated into the process, and leave their “finger prints.” This nurtures the sense of “ownership” giving incentive to participate in the new process despite the unknowns and frustrations.

While CPOE eliminates the hassles of trying to decipher a prescriber’s handwriting, it can create new types of “problem orders.” For instance, errors due to “typos,” appearance of double dosing (9), drop down menu errors by selecting the wrong drug or route or dosage (3), and inflexible formats (10) may occur. Clinicians also may report “alert overload” if too many alerts are sent after an order is entered (11). Anecdotal evidence and reports show that the rate of ignoring or overriding these alerts increases with time (11). It is important to provide useful decision support information and to ensure that only clinically significant interactions or dosing errors require an override (12). One way to decrease problems with ignoring critical drug interactions is to force the prescriber to specify a reason to “override” the order; however, that alone should not be used by the pharmacist as a reason to dispense a medication without further review. In some cases, providers just enter several characters to “bypass” this alert. The technology is not advanced enough to recognize the randomly entered key strokes as nonsense. The availability
of a clinical pharmacist specialist in patient care areas has been shown to decrease medication errors in the intensive care unit and other settings (13,14,15).

As discussed in Chapter 12 of the Third Edition, the VA launched a major change in the control of medication by instituting BCMA for all inpatient medications. It has been several years since the initial trials were preformed at several VA hospitals, and the data collected have been used to improve the ongoing processes. In 2009, Mims et al. (16) reviewed the BCMA system and the quality monitoring program that identifies and corrects the problems discovered when implementing it within the VA. Their research determined that major problems of using BCMA included:

While the initial compliance data were as low as 91% for bar code accuracy (ability of scanner to read the bar code correctly), the data for the fourth quarter of fiscal year 2007 reported successful scans of 98% for the correct dose or correct patient. While the authors concluded that a quality-monitoring program that uses best practices had corrected the problems, a success rate of only 98% equals more that 40,000 errors per 1 million opportunities of identifying the wrong dose or wrong patient. This rate may be considered too high.

In an earlier article published by the Joint Commission (JC), 15 best practice recommendations for BCMA were listed to help non-VA hospitals to implement this system. The best practice list was the result of their study of nursing information specialists and 30 unstructured interviews with diverse stakeholders (17). Some of the recommendations are to create interdisciplinary committees, train all users, avoid double-documentation systems, and verify allergy information displayed in the computer prior to medication administration.

These recommendations were selected without the use of the models identified in Chapters 10 and 12 of the Third Edition of this book. The authors of this chapter suggest that it is important to verify the extent to which a problem is mitigated by following the recommendations provided by the JC article since these recommendations “might even create unintended consequences that generate new paths to failure” and that evaluation should be ongoing (17).

BCMA systems have been suggested as a way to decrease medication errors related to the administration phase by adding another system check to the ones already performed by the pharmacist and the nurse in a phase of medication use where errors may be less likely to be detected (18). The BCMA system incorporates the “five rights” that the nursing profession is familiar with right patient, right drug, right dose, right time, and right route. Nationally, the VA has reported a 75% decrease in the wrong medication being dispensed after implementation of BCMA (19,20). The baseline error rate was not reported. The use of BCMA in the VA started in 1999 and is currently on version 3.0. There are several errors that are possible to be introduced in the process; some of the most common errors are process workarounds such as performing steps out of sequence, adding steps that are outside policy, or omission of steps (i.e., manually entering a bar code instead of using the scanner, scanning medications in advance) (17,18). In some instances, scanning of multidose vials may be a problem (20). Similar to CPOE, education and testing by clinicians who will use the system are important steps in order to avoid underestimating the change that is required to adopt new technology (20). Teams should be multidisciplinary, and duplicate systems should not be used for more than 2 weeks to avoid duplicate orders and dosing or facilitating providers who will not make the switch (17). It is important to have management involved and to provide examples to employees (20).