Improving Patient Safety by Identifying Side Effects From Introducing Bar Coding in Medication Administration

7/25/2019 Improving Patient Safety by Identifying Side Effects From Introducing Bar Coding in Medication Administration

1/14

PATTERSON ET AL., Bar Coding in Medication Administration540

Research Paper

Improving Patient Safety byIdentifying Side Effects fromIntroducing Bar Coding inMedication Administration

A b s t r a c t Objective. In addition to providing new capabilities, the introduction of technolo-gy in complex, sociotechnical systems, such as health care and aviation, can have unanticipatedside effects on technical, social, and organizational dimensions. To identify potential accidents inthe making, the authors looked for side effects from a natural experiment, the implementation ofbar code medication administration (BCMA), a technology designed to reduce adverse drug events(ADEs).

Design. Cross-sectional observational study of medication passes before (21 hours of observationof 7 nurses at 1 hospital) and after (60 hours of observation of 26 nurses at 3 hospitals) BCMAimplementation.

Measurements. Detailed, handwritten field notes of targeted ethnographic observations of in situnurseBCMA interactions were iteratively analyzed using process tracing and five conceptualframeworks.

Results. Ethnographic observations distilled into 67 nurseBCMA interactions were classified into12 categories. We identified five negative side effects after BCMA implementation: (1) nurses con-fused by automated removal of medications by BCMA, (2) degraded coordination between nursesand physicians, (3) nurses dropping activities to reduce workload during busy periods, (4)increased prioritization of monitored activities during goal conflicts, and (5) decreased ability todeviate from routine sequences.

Conclusion. These side effects might create new paths to ADEs. We recommend design revisions,modification of organizational policies, and best practices training that could potentially mini-mize or eliminate these side effects before they contribute to adverse outcomes.

J Am Med Inform Assoc. 2002;9:540553. DOI 10.1197/jamia.M1061.

EMILY S. PATTERSON, PHD, RICHARD I. COOK, MD, MARTA L. RENDER, MD

Affiliations of the authors: Ohio State University, Columbus, Ohio(ESP); University of Chicago, Chicago, Illinois (RIC); University ofCincinnati College of Medicine, Cincinnati, Ohio (MLR); VAMidwest Patient Safety Center of Inquiry, Cincinnati, Ohio (ESP,RIC, MLR).

This research was supported by the Department of VeteransAffairs, Veterans Health Administration, Health Services Research

and Development Service (Project no. SAF 20-049). The views

expressed in this article are those of the authors and do not neces-sarily represent the view of the Department of Veterans Affairs.

Correspondence and reprints: Emily Patterson, PhD, 210 BakerSystems, 1971 Neil Avenue, Columbus, OH 43210; e-mail: .

Received for publication: 11/16/01; accepted for publication:4/16/02.


2/14

Background

Reducing the risk of adverse drug events (ADEs) is anational priority.1 Adverse events related to medica-tion errors are common, costly, and often preventable.The Harvard Medical Practice study found that 19%of adverse events were related to medications,2 a find-ing confirmed by the Colorado/Utah study ofadverse events during hospitalization.3 Medicationadministration in the acute care inpatient setting is acomplex system requiring coordination among physi-cians who order the medications, pharmacists whoverify and dispense the medications, and nursing per-sonnel who administer medications to the patients. Inan analysis of 334 medication errors from 11 acute

care wards, 39% of the problems were judged to occurduring physician ordering, 12% during transcriptionand verification, 11% during pharmacy dispensing,and 38% during nursing administration.4

Computerized support systems are advocated toreduce the risk of ADEs at each of these stages inmedication administration.5 These systems, includ-ing computerized physician order entry, automateddispensing systems, and bar code reconciliation, arebelieved to reduce reliance on memory, increaseaccess to information, and increase compliance withbest practice procedures.6 For example, Bates7

found an 86% decrease in intercepted serious med-ication errors associated with the introduction of acomputerized order entry system in three medicalunits at a large urban hospital. Similarly, medicationalerts,8 a bedside computerized prescribing cri-tiquing system,9 a decision support system for antibi-otic prescribing,10 and automated dispensing unitsintegrated with information about a patients med-ication profile11 were associated with improved pre-scribing outcomes or reduced medication errors.

The Veterans Health Administration (VA), one of the

largest health care systems in the United States with163 medical centers nationwide, is a leader in the useof medical informatics systems. Since 1996, the VAhas used an integrated electronic database for patientclinical information, the Veterans Health InformationSystems and Technology Architecture (VistA), whichenables staff to access administration and demo-graphic data in addition to radiology, laboratory, andpathology information from any computer on thenetwork. In 1997, the VA implemented computerizedorder entry (CPRS),12 which is integrated with theVistA database. A Graphical User Interface (GUI) forCPRS was subsequently developed and implement-ed to replace the original command line interface.

Recently, the VA internally developed bar code medica-tion administration (BCMA) software, based heavily onprototype software deployed at the Topeka VA medicalcenter in 1995. With BCMA, patients wear bar-codedwristbands, and unit dose medications are bar-coded inthe pharmacy before delivery to a patients medicationdrawer. Nurses scan the patients wristband to accessup-to-date medication order data in BCMA, which wasinput into CPRS by physicians and verified in VistA bypharmacists. When the nurse scans a medication, it isautomatically documented as administered at that timeif the bar code number matches the number for a med-ication displayed on the primary BCMA screen. If thereis a mismatch, whether it is a different medication, dose,form, or time, an immediate warning pops up as an

alert box. Implementation of BCMA and associatedhardware was mandated to be completed throughoutthe VA by 2000. The BCMA software is continuouslyrevised by a national development team, including asecond release planned for May of 2002 that is intendedto better meet the needs of the Intensive Care Unit (ICU)setting, with expanding IV medication functionalityand supporting documentation of STAT verbal ordersprior to pharmacy verification.

Even with the most effective systems, the introduction ofa new technology into a complex setting is never apanacea.13 In addition to providing new capabilities, new

technologies have unanticipated side effects on technical,social, organizational, economic, cultural, and politicaldimensions of work.14 Kaplan15 and the Institute ofMedicine report16 recommend a proactive approach topatient safety, including the injunction to examine newtechnologies . . . for threats to safety and redesign thembefore accidents occur. Even as barcode-enabled pointof care medication administration systems reduce the fre-quency of some failure modes,1719 the transformation ofthe medication administration process creates new,potentially predictable paths to adverse drug events. Weasked the question: What negative, unintended side

effects, if any, resulted from the introduction of BCMAthat might create new paths to ADEs? This paperdescribes negative, unanticipated side effects that wereidentified by targeted ethnographic observation prior toand following system implementation and have implica-tions for system redesign that can be incorporated priorto the occurrence of adverse events.

Methods

Bar Code Medication Administration

Nurses access BCMA software using a laptop perma-nently fixed to a wheeled medication cart and linked

541Journal of the American Medical Informatics Association Volume 9 Number 5 Sep/Oct 2002


3/14

by a wireless network to the VAs electronic databases.Data entered into BCMA via a bar code scanner areused to verify that the correct patient gets the orderedmedications in the correct dose and route at the correct

time. To identify the patient, the nurse logs into theBCMA software and scans a bar code representing thepatients identification number (ID) on his wristband.The computer then displays the medications that areDUE, based on the time the wristband is scannedand information in the patients active electronic record(Figure 1). As the bar code of each medication isscanned, if the information about the drug formulationmatches the displayed information, the system auto-matically records the medication as administered bythe nurse at the time of scanning. If the scanned infor-mation on the medication does not match, the nurse is

alerted to the discrepancy by a pop-up dialog box. Thenurse cannot override this alert but has the option oftyping in a bar code number for either a patients wrist-band or a medication in place of scanning. The scan-ning of bar coded wristbands is intended to reduce therisk of patients receiving medications intended foranother. Scanning each medication barcode is intendedto verify that the medication, dose, route, and adminis-tration time match what was ordered.

Setting

We observed medication administration on acutecare and nursing home wards of three VA hospitals

selected on the basis of BCMA implementationschedules and willingness to participate (Table 1).The hospitals varied in size from small (3000 acutecare admission per year) to moderate (9000 admis-

sions per year); hospitals 1 and 2 were universityaffiliated. In all hospitals at the time of observation,the same computerized physician order entry system(CPRS) had been used for at least 1 year before intro-duction of BCMA, and the mandated BCMA systemhad been in operational use on all three shifts for atleast 1 month. During the observations in hospital 2,full documentation on the paper MAR was main-tained in parallel with BCMA for the purpose of pro-viding a backup system, whereas in the other twohospitals documentation was maintained solely inthe BCMA software at the time of observation.

Sites selected hardware (scanners, ethernets, batter-ies, and computer terminals) from among national-ly recommended options. Staff personnel weretrained through national train the trainer work-shops, manuals to direct local training, and confer-ence calls with local experts describing the changesassociated with software updates. All nursesreceived four hours of hands-on training as well ashardware and software maintenance support fromInformation Resource Management (IRM). Thisstudy did not examine training, on-site support, orhardware or physical ergonomic issues associatedwith local decisions.


F i g u r e 1 Primary BCMAdisplay for a fictional patient.


4/14

Conceptual Framework

Ethnographic observations before and after the natu-ral experiment of introducing a new technology,BCMA, into a complex, sociotechnical setting wereused to identify new paths to failure in the complexprocess of medication administration in the acutecare inpatient setting. These observations were tar-geted in that data collection and analysis were guid-ed by five conceptual frameworks from the humanfactors literature base20:

1. Recognition-primed decision making (RPD)

2. Human-automation interaction

3. Workload

4. Authority-responsibility doublebinds

5. Mutual awareness

RPD is a model of expert decision making underuncertainty, in dynamic settings, and with high con-sequences for erroneous action.21 In contrast to theoptimal decision making model, in which carefulreview of all relevant factors in series results in theselection of the optimal choice, the RPD modelemphasizes that expert nurses make decisions aboutwhich medication to administer to which patientbased on past experience by selecting an early work-

able option that trades off conflicting goals anduncertainties. Thus, for nurses administering medica-tions, we were primed to observe (1) cues to assessthe patient state and the identity of the patient andmedications, (2) resolution of competing goals, and(3) strategies to reduce uncertainty.

Second, in humancomputer interaction22 terms,medication administration is a task that requires thenurse to interact with a computerized system withautomated features. Research on automation in avia-tion cockpits has revealed predictable failures,referred to as automation surprises,23 that occur

when the automation initiates actions that are sur-

prising to the human partner. The potential forautomation surprises is increased when three factorsconverge:

1. The automation can act on its own without direc-tion from its human partner (high autonomy andauthority).

2. There are gaps in users mental models of howtheir machine partners work in different situations.

3. There is weak feedback about the activities andfuture behavior of the machine agent (low observ-ability).

During the observations, if a nurse was surprised byactions initiated by the BCMA software, we collected

data about the action taken by the software, the dis-played information, the nurses interpretation of theautomated action, the vulnerabilities to failure creat-ed by the inability to anticipate that action, and theoutcome.

Third, workload is the perception of task demands bya human agent. It depends on performance criteria,task structure, task duration, and the complexity andvariability of task demands, among other factors.24 Inboth aviation and anesthesiology, research finds pre-dictable adaptations to reduce workload.25 From this

research base, we anticipated strategies to reduceworkload during peak times, including trading accu-racy for speed, reducing performance criteria, shed-ding tasks, deferring tasks, and recruiting resourcesfrom other personnel, and noted when these strate-gies increased the vulnerability to ADEs.

Fourth, the introduction of computerized supportsystems is often coupled with increased adherence toorganizational policies and procedures, increasingopportunities for authorityresponsibility dou-blebind situations, researched primarily in militarysettings.26 When procedures are not applicable, prac-

titioners must either deviate from the procedures to


Table 1

Characteristics of Observed Facilities

Beds No. Admits/Year Date BCMA Date CPRS

Hospital 1: acute care/oncology 155 9413 4/00 2/99

Hospital 2: acute care 99 3128 9/99 9/98

Hospital 2: nursing jome 460 3128 9/99 9/98

Hospital 3: acute care 80 3192 5/99 4/95


5/14

meet their goals, potentially increasing individualvulnerability to punitive or legal action, or follow theprocedures, potentially increasing the vulnerabilityto undesirable outcomes. When observed nursesencountered authority-responsibility doublebind sit-uations, we took detailed notes about their strategiesto resolve the situations, including tailoring proce-dures, rote rule-following, covert work practices, or

activities to reduce the risks of punitive action thatmight have undesirable effects.

The last conceptual framing is that medicationadministration is a distributed and cooperative activ-ity conducted in a social setting. Successful coordina-tion across distributed agents depends on a highlevel of mutual awareness, sometimes referred to ascommon ground. Common ground is defined asitems that are shared among human agents, includ-ing knowledge, goals, frames of reference, and beliefsabout the current and future state of affairs.27 Whenelectronic systems replace paper-based systems, acommonly observed side effect is the degradation ofcoordination as a result of privatizing informationthat was previously publicly accessible (e.g., elec-tronic flight strips28). For example, coordinationbetween controllers of the London underground isaided by lighted fixed line diagram displays that areviewable and easily referenced by physical gesturing.Replacing this display with individual workstationdisplays degraded coordination by reducing the abil-ity to use the display as a shared frame of reference.29

When a nurse was observed to be surprised byactions of other personnel, we interviewed oppor-

tunistically to see if design choices that reduce sup-

port for mutual awareness contributed to coordina-tion breakdowns.

Data Collection

One observer (EP), trained in ethnographic fieldobservations in complex settings,30 conducted allobservations before and after BCMA implementation

(Table 2). Each observation session lasted from one toseven hours and included observations during allparts of day, evening, and night shifts. To minimizethe effect of observations on behavior, we informedthe observed nurses that no one, including theirnurse managers, would be provided with informa-tion that could trigger a punitive response sinceobservations were de-identified for person, place,and time, and we did not collect demographic ormedication error data. Prior to BCMA implementa-tion, seven nurses were observed for a total of 21hours during 10 medication passes at one hospital.

Following BCMA implementation, 26 nurses wereobserved for a total of 60 hours during 23 medicationpasses at the three hospitals. We targeted timing ofthe observations to include observations immediate-ly after implementation, after initial problems wereresolved, and after the system was fully adopted intothe workflow (ranging from 1 week to 13 months).

In addition, the same observer also watched comput-erized order entry (2 physicians, 2 hours) and orderverification by inpatient pharmacists (5 pharmacists,2 pharmacy technicians, 10 hours) at hospital 1.Twenty complementary interviews were conducted

with practitioners, computer support personnel, and


Table 2

Observations Before and After BCMA Implementation

Hospital 1 Hospital 1 Hospital 2 Hospital 3 TotalPre-BCMA Post-BCMA Post-BCMA Post-BCMA Post-BCMA

Acute care

No. med passes 7 11 5 2 18No. nurses 6 13 5 3 21No. hours 14 33 13 3 49

Oncology

No. med passes 3No. nurses 1No. hours 7

Nursing home

No. med passes 5 5

No. nurses 5 5No. hours 11 11


6/14

nurse managers from the three hospitals. The focus of

the interviews was to elicit perceived barriers tointended system use and reactions to possibleredesign, organizational, and training interventions toreduce side effects. Finally, critical decision methodinterviews31 about a near-miss, wrong-patient casewere conducted with nurses, a nurse manager, and apharmacist to understand why the nurse using theBCMA software did not detect a patient mismatch.

During the observations, we recorded time-stamped,detailed, handwritten field notes of verbal and phys-ical behavior, targeted at the identification of nega-tive, unanticipated side effects, as well as information

captured in the electronic patient record.

Analysis

The data were iteratively analyzed using the concep-tual theoretical frameworks from the human factorsliterature to recognize and abstract emerging pat-terns (Figure 2). Process tracing protocols32 for all ofthe observations were created that detailed the activ-ities, grouped into several broad categories.Particularly interesting interaction sequences wererepresented and analyzed as mini-cases. The mini-cases were then grouped by emerging themes, andsupplementary data from the detailed notes based onthe themes were collated. Finally, the themes werematched to similar patterns documented in other

complex, high-consequence, sociotechnical domains


F i g u r e 2 Levels ofabstraction in the processtracing analysis.

What are negative side effects follow-ing the introduction of BCMA?

1) Recognition-Primed DecisionMaking

2) HumanAutomation Interaction3) Workload4) Authority/Responsibility

Doublebinds5) Mutual Awareness

Handwritten notesElectronic patient recordsInterview audio tapes


7/14

along with design, organizational, or training strate-gies to reduce the side effects, when available.

Results

Analysis of the process tracing protocols identified67 unique mini-cases, which were exclusivelygrouped into 12 categories based on a prominenttheme (Table 3).

Figure 3 provides an overview of a mini-case that haspoor usability33 as a prominent theme. In this case,workload on an acute care ward was unexpectedlyhigh as a result of unanticipated absences of staff andpatient factors (a coworkers emergency absence inthe middle of a shift, attendance of another staffmember at a meeting, discharge of a patient againstmedical advice, a new patient admitted from the

emergency department with continuing chest pain).

Using the computerized order entry system, a physi-cian ordered two medications to be given immediate-ly to treat the new patients chest pain (without oth-erwise informing the nurse of the order, which usedto be accomplished by a red color code on the outsideof a paper chart before the implementation of com-puterized order entry but is now communicated lessfrequently other than through the written orderbecause new orders cannot be highlighted in CPRS orBCMA). The nurse, anticipating that these medica-tions would be ordered, looked in BCMA but foundno cardiac medications (note that pending and dis-continued medications are not displayed). He thenlooked in the order entry system and found twopending medications. The nurse borrowed and

administered one of the two medications (taken fromanother patients medication drawer) and waited fora less critical medication to arrive from pharmacy. Hecould not document the medication administrationin BCMA since the medication was not yet displayed(medications need to first be ordered by physiciansand then verified by pharmacists prior to display),and only displayed medications could be document-ed as administered with the observed version ofBCMA. The nurse then turned his attention topreparing another patient for discharge. When thenurse later documented the medication administra-

tion, the administration time was automatically doc-umented as the time the medication was scanned,several hours after it was administered. This timewas not changed to reflect the actual administrationtime, partly because changing medication adminis-tration times required a long, confusing sequence ofactivities in a separate, command line interface. Poorusability was a prominent theme in this mini-case. Itwas difficult for the nurse to accomplish his goals inthis nonroutine situation because (1) pending anddiscontinued medication orders were not displayed,(2) medications that were not displayed could not bedocumented as administered, (3) it was not possible

to see changes to medication orders without firstopening a patient record, (4) it was difficult toundo actions, and (5) it was difficult to revise data-base information after it had been entered.

To illustrate the next step of the process-tracing analy-sis, several elements in this mini-case were re-repre-sented in more abstract, less domain-specific termi-nology that points to concepts drawn from the humanfactors literature. First, nursing expertise was broughtto bear in ways that resulted in deviations from aprocedural model of medication administration:

rotely and sequentially administering medications


Table 3

Sixty-seven Mini-cases Grouped by Category

Theme No.

Nurses had usability problems with the interface 15Data on the paper MAR conflicted with BCMA data 11BCMA detected a discrepancy between intended

and scanned medications 9Coordination issues with others 9Nurses believed medications were missing from BCMA 5Nurses were confused by automated BCMA actions 4Nurses requested missing medications from pharmacy

through BCMA 3Nurses received unexpected information updates

through BCMA 3BCMA was not accessible 3

Automated timestamp was different than theadministration time 2

BCMA did not detect discrepancy between intendedand actual patient 1

Other 2

F i g u r e 3 Example of mini-case.


8/14

displayed on an interface to a series of patients duringa medication pass.34 For example, knowledge andassessment of the risk of ischemic complications wereused to anticipate two pending medication orders andprioritize administering one medication over othertasks. Second, a goal conflict resulted in violation ofa best practice procedure.35 In this case, the goal offollowing the best practice procedure of having apharmacist verify a medication ordered by a physi-cian prior to administration (to increase order accura-cy and decrease the possibility of allergic reactions orinteractions with other medications, among other rea-sons) conflicted with the goal of intervening as quick-ly as possible to reduce cardiac risk, including thepotential for a code. Third, in choosing to meet per-

formance (medical) goals, an authority-responsibilitydoublebind situation occurred, in which the nurse feltthe responsibility to act without having the authorityto document the action.36 As a result, the medicationwas documented at a different time from when it wasadministered, which could be viewed as a covertwork practice. Fourth, the dynamics of the workloaddistribution encouraged task prioritization strategiesof delaying documentation until after a high-tempoperiod.37 When there was difficulty in documentinginformation, either because the automatically generat-ed data was incorrect or because medications were

not displayed, the nurses moved on rather than takethe time to ensure accurate documentation at thatmoment. Finally, system design choices created thepotential for coordination breakdowns.38 The relianceof the physician on BCMA to communicate a new,high-priority order for imminent administrationcould be viewed as a poor strategy, even though therewas no adverse outcome due to the nurses anticipa-tion of the order, because the software was notdesigned to actively highlight priority or new med-ication orders. In addition, the difficulty in correctingthe difference between the actual and documentedmedication administration time could enable physi-

cians, pharmacists, or other nurses to make incorrectinferences based on the data. For example, a nurse onthe next shift could delay medication administration.Similarly, the inability to document administration ofmedications not displayed creates the potential forcoordination breakdowns between multiple nurses(e.g., a covering RN could administer medicationsthat have already been given) and physicians (e.g., aphysician might unnecessarily increase the dose of amedication if he bases his inferences on an inaccurateadministration time).

Following similar re-representations of each mini-case based on the conceptual frameworks and guid-

ed by insights from the interviews, the data were ana-lyzed for recurring patterns across the cases. Thisiterative top-down and bottom-up data analysisrevealed five negative, unanticipated side effects fol-lowing the introduction of BCMA:

1. Nurses confused by automated removal of med-ications by BCMA

2. Degraded coordination between nurses andphysicians

3. Nurses dropping activities to reduce workloadduring busy periods

4. Increased prioritization of monitored activitiesduring goal conflicts

5. Decreased ability to deviate from routinesequences

Nurses Confused by Automated Removal ofMedications by BCMA

There were four cases in which nurses expressed sur-prise at automated actions initiated by the BCMAsoftware, which is the stereotypic reaction to a classicbreakdown in humancomputer interaction, referredto as an automation surprise.39 Two cases involvedIV medications that were automatically dropped

from the medication list a number of hours after thescheduled administration time. One was achemotherapy medication whose administration wasdelayed due to lack of an IV site, and another was apatient who received Ativan late because he was offthe ward for several hours. In another case, a nursewas surprised that an antibiotic was not available asa result of an automatic stop order. In all of thesecases, the nurses detected that the automaticallyremoved medications were intended to be given andeventually administered them, but their surpriseindicated that their expectation that the medication

would be displayed in BCMA was violated, and theyreported that this design decision created a newpotential path to missed medications.

Degraded Coordination between Nurses andPhysicians

A side effect of replacing the paper medicationadministration record (MAR) with BCMA appears tobe degraded coordination between nurses and physi-cians. During the observations, numerous coordina-tion breakdowns occurred between nurses andphysicians, some of which might not have occurredwith the previous paper-based system. Prior to



9/14

BCMA, physicians were observed to have quickaccess at the bedside to current medication adminis-tration information and nurses had immediate accessto information about pending and discontinuedorders on their paper MAR. Immediately after BCMAimplementation, most physicians were not aware ofhow to access medication administration informa-tion, and the display was restricted to fixed 7-daywindow (Saturday to Sunday). Although the CPRSsoftware was subsequently updated to provide

physicians with increased access to BCMA data, in afocus group conducted October 26, 2001, seven ofseven residents stated that they did not systematical-ly look at a patients medication orders in BCMAunless a question was asked because it is a time-intensive process. In comparison, they estimated thatwith a paper-based system at another hospital, theyreviewed the medication orders for each patient anaverage of 3 times/week.

Degraded coordination between nurses and physi-cians can lead to predictable new paths to adverseevents, including failing to detect erroneous medica-tion orders, verifications, or administrations, failingto renew automatically discontinued medications,failing to prioritize a STAT medication order overother activities, or failing to explain why laboratoryvalues are unusually high or low for an at-riskpatient.

Nurses Dropping Activities to Reduce Workloadduring Busy Periods

Nurses at all three hospitals consistently used strate-gies to increase efficiency that circumvented the

intended use of BCMA (Table 4). For example, we

observed that all nurses typed in at least one patientidentification number rather than scanning the wrist-band barcode.

Before the introduction of BCMA, system developerspredicted that an additional benefit of the systemwould be improved documentation efficiencythrough scanning barcodes rather than typing infor-mation.40 Nurses felt that there were efficiency bene-fits only when scanning was consistently reliable,

when there were no deviations from routine actionssuch as refused medications, and when paper-basedsystems were not used for parallel documentation. Inaddition, barcodes are normally scanned duringbusy, high workload periods. This characteristicworks against the commonly observed strategy inevent-driven, high-consequence domains, such astake-offs in aviation, to free mental and physicalresources during busy periods by shifting other tasksto less busy periods.41

Scanning wristbands to identify patients wasdropped more frequently than scanning medicationbarcodes, a finding also reported in a study of a dif-ferent barcode-enabled point of care system in a pri-vate institution.42 A variety of factors might explainthis finding. Obvious factors include that wristbandbarcodes did not scan as reliably as medication bar-codes43 and that wristbands could not be scanned insome cases (e.g., isolation patients, patients whoremoved wristbands because of swollen limbs or dis-comfort, particularly in long-term care).44 In addition,nurses uniformly believed that typing in a 7-digitnumber took less time than wheeling a large medica-tion cart into a room and scanning a wristband. In

addition to these factors, there were also more subtle


Table 4

Observed strategies to reduce workload during medication pass

Hospital 1 Hospital 2 Hospital 3

Not scanning wristband

Type in SSN X X XScan wristband stapled into MAR with photo XScan ID card XScan wristband on table next to patient X

Delaying scanning medications

Batch scanning medications for multiple patients before administration X XDelaying documentation of refused medications, medication effectiveness,

and medications that do not scan X X XScan opened or similar medications after administration X X XDocument previous administration by others (RN, NA, patients) X X X


10/14

barriers to scanning wristbands. Most nurses tried toavoid disturbing sleeping patients, particularly ifthey anticipated that a patient had no oral medica-tions during the medication pass. They tried to mini-mize interrupting discussions between patients, fam-

ily, and health care practitioners. In addition, severalnurses explained that they employed strategies toreduce distractions during medication preparation inorder to better detect inaccurate medication orders.When they scanned the patient wristband, patientsoften distracted the nurse by requesting an action orinformation.45 Finally, several nurses expressed con-cern that other nurses or physicians would log themout of BCMA if they left the computer console.

Increased Prioritization of Monitored Activitiesduring Goal Conflicts

Shortly after BCMA implementation, many nursesstated that they felt timeliness of medication admin-istration was greatly increased in priority. Similarly,nurses at every hospital were observed to be anxiouswhen required to type an explanation for late med-ications, generally 30 minutes before and 60 minutesafter the scheduled administration time. In addition,several nurses stated that they were concerned thatanalyses of timeliness of medication administrationby individual nurses would be much easier to dowith the electronic data than with handwritten data.Further research is needed to understand how anincrease in emphasis on timely medication adminis-

tration affects decision tradeoffs during goal con-flicts. One nurse stated that she believed that nurseswould be more likely to delay other activities duringthe medication pass that should not be delayed inorder to ensure timely medication administration.

Figure 4 depicts interruptions that were acted on dur-ing medication administration on a chemotherapyward before BCMAimplementation, including fixingrestraints on a patient at risk of falling.

Decreased Ability to Deviate from RoutineSequences

Computerized systems, such as BCMA, are ofteneffective at streamlining routine operations, makingthe system both more efficient and easier to antici-pate others actions and detect erroneous actions. A

frequent tradeoff from streamlining routine opera-tions is that non-routine sequences of activity becomemore difficult to perform. For example, prior toBCMA, although there was no taper dose feature inthe computerized order entry system, a physiciancould enter a description in a text box at the bottomof a single order (e.g., prednisone taper: start at 60 mgdecrease 10 milligrams every other day until at 10mg, then decrease to 5 mg for 2 days). Prior to BCMA,one pharmacist estimated that he would verify ataper dose order in less than 1 minute and pass thefree text note on to nursing personnel. After the intro-duction of BCMA, a pharmacist, in order to create theorder in a scannable format, was observed to take 17


F i g u r e 4 Interrup-tions during medicationadministration prior toBCMA.


11/14

minutes to break a taper dose order into daily ordersat the exact dose, a total of 14 new orders, and dis-continue the original order. In addition, the pharma-cist expressed the hope that the physician would notrewrite the order in the original format, as theobserved pharmacist explained had occurred in thepast when a physician did not understand why theorder had been changed. Although a taper order is aspecific example of an issue that can be resolvedthrough software enhancements, it illustrates theobserved pattern of decreased flexibility whenmachine algorithms critique human actions, becausethe vocabulary used in communicating with amachine is restricted. When the format is not wellknown or cumbersome to the practitioner, decreased

flexibility might lead to new vulnerabilities and thereduced ability to detect erroneous actions.

One of the greatest challenges with BCMA is how tosupport flexibility in the event that the system goesdown, either for scheduled maintenance or becauseof an unexpected loss of power. There appears to begreat variability in how institutions have adaptedBCMA to enable flexibility given the potentially highconsequences of being unprepared for the infrequentbut predictable eventuality of loss of power. Forexample, these adaptations have followed the intro-

duction of BCMA in several hospitals in order to pro-vide a backup system:

Maintaining documentation redundancy with thepaper-based MAR in parallel with BCMA

Printing out MARs every few hours for informa-tional but not documentation purposes

Creating an electronic backup by logging everychange to medication orders in a separate system

Reverting to a paper MAR on an executive decisionwithin 15 minutes of a power outage in parallelwith instituting periodic training and manuals tomaintain competency with the paper-based system

Capturing health summaries updated every hourto stand-alone computers on emergency power.

Discussion

Bar coding is advocated as a technology to reduceerrors in several health care settings.46 To improveproactively the effectiveness of the BCMA system inthe Veterans Health Administration, we identifiedfive negative, unintended side effects with the poten-tial to create predictable new paths to ADEs: (1) nurs-

es confused by automated removal of medications byBCMA, (2) degraded coordination between nursesand physicians, (3) nurses dropping activities toreduce workload during busy periods, (4) increasedprioritization of monitored activities during goal con-flicts, and (5) decreased ability to deviate from rou-tine sequences.

None of these side effects is inherent in barcode-enabled point of care technology. It is likely thatdesign revisions, modification of organizational poli-cies, and best practices training can proactivelyreduce each of these side effects before accidentsoccur. Automation surprises, for example, areknown to result from automated actions that are

strong in that they occur without immediately pre-ceding direction from a user combined with weakfeedback about past and future automation activities.The automation can be made less strong, such asby eliminating automated removal of medicationswithout prior approval from the human partner or byproviding the capability for users to restore automat-ically dropped medications. Alternatively, increasingthe feedback about automated actions, such as by dis-playing a grayed out medication to indicate that itwas removed from the display, would make it easierfor users to be aware of the change. Cooperative solu-tions, such as using pharmacy personnel to remindphysicians when orders have been automatically dis-continued, could also be explored.

A primary contributor to the degraded coordinationbetween the nurses and physicians appears to be theloss of quick, coded communication via artifacts thatcan be viewed at a glance. This functionality couldbe designed into BCMA or other software.Specifically, longshot visualizations47 could bedesigned that would allow both nurses and physi-cians to see at a glance when medication orders havechanged for multiple patients on a ward, similar to

the functionality provided by circular color codes onpatient paper charts representing STAT, discontin-ued, new order, and no change. In addition, the med-ication administration history window could beredesigned to allow physicians to quickly scan,search, and review medication orders for multiplepatients. Similarly, nurses could be provided quickeraccess to pending and discontinued medicationinformation to make it easier to anticipate orders anddetect erroneous actions. Finally, strategies adaptedto the paper MAR that had side benefits for coordi-nation, such as flagging pages to remember to doan action (that could then be seen by others wholooked at the paper MAR), could be supported in



12/14

BCMA. Design changes to improve coordination cangenerally be characterized as increasing the open-ness of the workspace, meaning to increase the abil-ity for health care practitioners to be aware of activi-ties being conducted by others. Open workspaceshave been shown to improve anticipation, synchro-nization of activities, detection of erroneous actions,and other forms of coordination.48

System redesign and best practices can also reduceworkload burdens to make it less likely that nurses cir-cumvent the system during busy periods as an adap-tation to increase efficiency. Circumventing the systemwill predictably lead to increased vulnerabilities tomis-identification of patients and medications as wellas inaccurate administration data. Obvious strategiesto reduce workload, such as decreasing staff/patientratios, might not always be possible, given the nation-al nursing shortage and increased acuity of hospital-ized patients. Alternatively, eliminating unnecessarysteps or shifting data entry to lower intensity timescould reduce the workload burden. For example,rather than typing the same reason for every late med-ication for a patient off the ward, data entry might bebatched or a default entry provided. In addition,improving the interface usability would decreaseworkload, such as by making it possible to documentrefused medications without exiting the system.49

Finally, the system could be designed to better fit thenatural workflow of providing patient care. For exam-ple, wristbands for patients could be scanned aftermedications have been prepared to minimize interac-tion between the nurse and patient during medicationpreparation, particularly when patients are sleeping orprone to interrupt the nurses.

The perception of increased emphasis on timely med-ication administration will likely drift over timebased on how organizations respond to medicationadministration data. If an organization builds incen-tive rewards based on medication administration

timeliness or punishes individual nurses for takingmore time than others, the perception will be rein-forced. If an organization clearly communicates thattimeliness is a lower priority than other goals, such asreducing patient falls, and restricts access to the timedata, the perception will be reduced. Although theorganizational context is likely to be the most impor-tant contributor to perception of timeliness as a pri-ority, redesign of BCMA might reduce this side effect.Medications that are particularly important toadminister at the intended time could be distin-guished from others on the display. Dialog boxes that

require documentation of why a medication is late

could be eliminated. Parameters defining on-timemedication administration could be enlarged.

Finally, the system could be redesigned to allow anincreased ability to deviate from a idealized work-flow. As specific deviations are identified, featurescan be designed to support them, such as by creatinga taper dose order feature in the computerizedorder entry system. In addition, flexibility can beincreased through generic features such as free textfields, undo, drag and drop, select, copy andpaste, and batch. Flexibility can also be increasedby allowing tailoring of options available in dialogboxes, making it easier to revise administration timedata and document changes in status from given torefused or held. Finally, features could be addedto support the ability to write free text information-only notes to document administration of medicationorders not viewed on the screen.

It is not surprising that negative side effects wereidentified after the implementation of BCMA. It isaxiomatic that technology change has far-reachingimpacts on the way in which work is conducted in afield of practice, even with the most successful tech-nologies.50 In opposition to the view that a change inmedium (e.g., computerized order entry instead ofhandwritten prescriptions) can be introduced with-

out greatly altering complex systems embedded insocial settings, we believe that little of the formerroles, strategies, and paths to failure are preservedacross the technology change boundary.51

This study illustrates how targeted observational stud-ies before and after naturally occurring points ofchange can be used to better understand processes oftransformation and adaptation.52 The points of changeform natural experiments where a change, the newtechnology, disturbs a system. Observers calibrated topatterns in human-machine performance watch howthe system re-settles into a new equilibrium state. The

point-of-change provides opportunities to learn howthe system actually functions and sometimes malfunc-tions by observing how practitioners accommodateand adapt to the technology change. The system isexamined to see how the technology transforms roles,judgments, and coordination, as well as how peopleadapt to the mix of new capabilities and new com-plexities. This information provides insight for organi-zational, design, and training post-conditions to makethe system more useful and reduce unintended, nega-tive side effects from the change.

In addition, observing at points of change can add to

the knowledge base of how changes impact cognition



13/14

and collaboration in a way that uniquely reveals thecomplex, inner workings of human-human interac-tions with a black box of technology. This knowl-edge enables better advance predictions of the effectsof a change, especially side effects such as new com-plexities and vulnerabilities. It also serves as a stimu-lus to envision alternative design concepts. Armedwith this knowledge, we can find higher leveragepoints for change and develop new concepts for sup-port. If we can better anticipate the reverberations ofnew technology prior to the process of change, newdirections can be pursued when intervention is lessdifficult and costly.

This study has several limitations. Our objective

was to identify proactively negative side effectsprior to the occurrence of adverse events. The studywas not designed to evaluate the effectiveness ofBCMA in reducing the frequency of predefined fail-ure modes at the three participating hospitals or toidentify beneficial side effects, such as more effi-cient and accurate requests for missing medicationsfrom pharmacy. Because our findings, as in anyanalysis of complex, interacting, multifaceted fielddata, depend on our conceptual frameworks, welikely did not find all side effects. For example, theintroduction of BCMA might have impacted insti-

tutional views of accountability for adverse drugevents. In addition, we observed primarily on acutecare wards and nursing homes at three institutions,and it is unclear how these findings would general-ize to other settings, such as intensive care units(ICUs) or institutions without 24-hour access topharmacy. In addition, the ability to generalizethese findings to other bar coding applications,such as software designed for the clinical laborato-ry or blood banking, remains a question to beaddressed by future research.

Conclusion

Side effects following the introduction of BCMA wereobserved that might create new paths to ADEs. Theseside effects have implications for system redesign,modification of organizational policies, and bestpractices training that might be implemented beforethe occurrence of adverse events. Investigation of thenatural experiment of implementation of new tech-nology can forestall potential accidents in the makingby identifying new sources of hazard and developingsystem redesign, organizational or training initiatives

to protect against them.

We thank the BCMA development team, led by Mr. Richard Singer,for their hard work and willingness to collaborate to improve thesoftware design, James Bagian, MD, for sharing his knowledge ofvulnerabilities with the use of BCMA; David Woods, Ph.D., forsharing findings on research on side effects of technology change;Steve Sawyer, PhD, for reviewing the paper; and Denis Coelho,Ph.D., for conducting pilot research.

References

1. Doing What Counts for Patient Safety: Federal Actions toReduce Medical Errors and Their Impact. Report of theQuality Interagency Coordination Task Force (QuIC) to thePresident, February 2000. Quality Interagency CoordinationTask Force. Washington, DC .

2. Leape LL, Brennan TA, Laird NM, et al. The Nature of AdverseEvents in Hospitalized Patients: Results of the HarvardMedical Practice Study II. N Engl J Med. 1991;324(6):37784.

3. Thomas EJ, Studdert DM, Newhouse JP, et al. Costs of medicalinjuries in Utah and Colorado. Inquiry. 1999;36:25564.

4. Leape L, Bates DW, Cullen DJ, et al. Systems analysis ofadverse drug events. JAMA. 1995;274:3543.

5. Bates DW, Cohen M, Leape LL, et al. Reducing the frequencyof errors in medicine using information technology. J Am MedInform Assoc. 2001; 8(4):299308.

6. Bates DW. Using information technology to reduce rates ofmedication errors in hospitals. BMJ. 2000;320:78891.

7. Bates DW, Teich JM, Lee J, et al. The impact of computerizedphysician order entry on medication error prevention, J AmMed Inform Assoc. 1999;6(4):31321.

8. Raschke RA, Gollihare B, Wunderlich TA, et al. A computeralert system to prevent injury from adverse drug events:

Development and evaluation in a community teaching hospi-tal. J Am Med Inform Assoc. 1998; 280(15):131720.9. Nightingale PD, et al. Implementation of rules based comput-

erised bedside prescribing and administration: interventionstudy. BMJ. 2000;320:7503.

10. Evans RS, Pestonik SL, et al. Acomputer-assisted managementprogram for antibiotics and other antiinfective agents. N EnglJ Med. 1998; 338(4):2328.

11. Borel JM, Rascati KL. Effect of an automated, nursing unit-based drug-dispensing device on medication errors. Am JHealth-Syst Pharm. 1995;52:18759.

12. Murff HJ, Kannry J. Physician satisfaction with two orderentry systems. J Am Med Inform Assoc. 2001;8(5):499509.

13. Barthelemy-Brichant N, Sabatier J, et al. Evaluation of fre-quency and type of errors detected by a computerized record

and verify system during radiation treatment. RadiotherOncol. 1999;53:14954.14. Anderson JG, Aydin CE. Theoretical perspectives and method-

ologies for the evaluation of health care information systems.In: Anderson JG, Aydin CE, Jay SJ (eds). Evaluating HealthCare Information Systems: Methods and Applications.Thousand Oaks, CA, Sage Publications,1994, pp 529.

15. Kaplan B, Brennan PR, Dowling AF, et al. Toward an infor-matics research agenda: Key people and organizational issues.J Am Med Inform Assoc. 2001;8:23541.

16. Kohn LT, Corrigan JM, Donaldson MS (eds): To Err is Human:Building a Safer Health System. Washington, DC, NationalAcademy Press, 1999.

17. Puckett F. Medication-management component of a point-of-care information system. Am J Health-Syst Pharm 1995;

52:13059.



14/14

18. Malcolm B, Carlson RA, Tucker CL, Willette C. VeteransAffairs: Eliminating medication errors through point-of-caredevices. Technical paper for 2000 Annual HIMSS Conference,

2000.19. Anderson S, Jensen R. Management Case Study: A multidisci-plinary approach to decreasing the occurrence of medicationerrors and variances. Presented at the American Society ofHealth-Systems Pharmacists Mid-Year Clinical Symposium,2000.

20. Roth EM, Patterson ES, Mumaw RJ. Cognitive engineering:Issues in user-centered system design. In Marciniak J (ed):Encyclopedia of Software Engineering, 2nd ed. New York,John Wiley & Sons, 2001.

21. Klein GA. Recognition-primed decisions. In Rouse WB (ed),Advances in Man-Machine Systems Research. Greenwich, CT,JAI Press, 1989, pp 4792.

22. Norman DA. The problem with automation: Inappropriatefeedback and interaction, not over-automation. Philosoph

Trans R Soc London. 1990;327:58593.23. Sarter NB, Woods DD, Billings CE. Automation surprises. In

G. Salvendy (ed): Handbook of Human Factors andErgonomics, 2nd ed. New York, Wiley, 1997, pp 192643.

24. Huey BM, Wickens CD. Workload Transition: Implications forIndividual and Team Performance. Washington, DC, NationalAcademy Press, 1993.

25. Woods DD, Patterson ES. How unexpected events produce anescalation of cognitive and coordinative demands. In:Hancock PA, Desmond PA (eds): Stress, Workload andFatigue. Hillsdale, NJ, Lawrence Erlbaum, 2001, pp 290304.

26. Woods DD, Shattuck LG. Distant supervision-local actiongiven the potential for surprise. Cogni Technol Work. 2000;2:8696.

27. Clark HH, Brennan SE. Grounding in communication. In

Resnick, Levine JM, Teasley SD (eds): Perspectives on SociallyShared Cognition. Washington, DC, American PsychologicalAssociation, 1991, pp 12749.

28. Harper RR, Hughes JA, Shapiro DZ. Harmonious workingand CSCW: Computer technology and air traffic control. InBowers JM, Benford SD (eds): Studies in Computer SupportedCooperative Work. Theory, Practice and Design. Amsterdam,North-Holland, 1991, pp 22534.

29. Heath C, Luff P. Technology in Action. Cambridge:,Cambridge University Press, 2000.

30. Hutchins E. Cognition in the Wild. Cambridge, MA, MITPress, 1995.

31. Klein GA, Calderwood R, MacGregor D. Critical decisionmethod for eliciting knowledge. IEEE Trans Syst ManCybernet. 1989;19(3):46272.

32. Woods DD. Process tracing methods for the study of cognitionoutside of the experimental psychology laboratory. In Klein G,Orasanu J, Calderwood R (eds): Decision Making in Action:Models and Methods. Norwood, NJ, Ablex PublishingCorporation, 1993, pp 22851.

33. Rubin J. Handbook of Usability Testing: How to Plan, Design,and Conduct Effective Tests. New York, John Wiley & Sons,1994.

34. Feltovich PJ, Ford KM, Hoffman RR. Expertise in Context:Human and Machine. Cambridge, MA, MIT Press, 1997.

35. Vicente KJ. Cognitive Work Analysis. Hillsdale, NJ, Lawrence

Erlbaum, 1999.36. Woods DD, Shattuck LG. Distant supervisionlocal action

given the potential for surprise Cogni Technol Work. 2000;

2:8696.37. Woods DD, Patterson ES. How unexpected events produce anescalation of cognitive and coordinative demands. In HancockPA, Desmond PA (eds): Stress, Workload and Fatigue.Hillsdale, NJ, Lawrence Erlbaum, 2001, pp 290304.

38. Heath C, Luff P. Technology in Action. Cambridge, CambridgeUniversity Press, 2000.

39. Sarter NB, Woods DD, Billings CE. Automation surprises. InSalvendy G (ed): Handbook of Human Factors andErgonomics, 2nd ed. New York, Wiley, 1997;192643.

40. Dinklage KC, White SJ, Lenhart JC, Godwin HN. Accuracyand time requirements of a bar-code inventory system for con-trolled substances. Am J Hosp Pharm. 1989;46(11): 2304307.

41. Woods DD, Patterson ES. How unexpected events produce anescalation of cognitive and coordinative demands. In:

Hancock PA, Desmond PA (eds): Stress, Workload andFatigue. Hillsdale, NJ, Lawrence Erlbaum, 2001, pp 290304.

42. Puckett F. Medication-management component of a point-of-care information system. Am J Health-Syst Pharm. 1995;52:130509.

43. Renner SW, Howanitz PJ, Bachner P. Wristband identificationerror reporting in 712 hospitalsa College of AmericanPathologists Q-probes study of quality issues in transfusionpractice. Arch Pathol Lab Med. 1993;117(6):5737.

44. Creasman GL, Dervin D, Widdowson N, Bacher M. Bar-codemedication administration: Quality management and medica-tion error evaluation following implementation. Presented atthe ASHP Midyear Clinical Meeting. Las Vegas, December37, 2000.

45. Ebright PL, Patterson ES, Chalko B, Render ML. Identification

of environmental factors that increase complexity in acute caresettings. Proceedings of the Midwest Nursing ResearchSociety 2002 Annual Research Conference, March 14, 2002,Chicago, p 148.

46. Bates DW, Cohen M, Leape LL, et al. Reducing the frequencyof errors in medicine using information technology. J Am MedInform Assoc. 2001; 8(4):299308.

47. Woods DD, Watts JC. How not to have to navigate through toomany displays. In Helander MG, Landauer TK, Prabhu P(eds): Handbook of Human-Computer Interaction, 2nd ed.Amsterdam, Elsevier Science, 1997, pp 61750.

48. Patterson ES, Watts-Perotti J, Woods DD. Voice loops as coor-dination aids in space shuttle mission control. ComputerSupported Cooperative Work: The Journal of CollaborativeComputing. 1999; 8(4):35371.

49. Murff HJ, Kannry J. Physician satisfaction with two orderentry systems. J Am Med Inform Assoc. 2001;8:499511.

50. Aydin C. Occupational adaptation to computerized medicalinformation systems. J Health Social Behav. 1989;30:16379.

51. Cook RI, Woods DD. Adapting to new technology in the oper-ating room. Hum Fact. 1996;38(4):593613.

52. Woods DD, Sarter NB. Evaluating the impact of new technol-ogy on human-machine cooperation. In Wise JA, Hopkin VD,Stager P (eds): Verification and Validation of ComplexSystems: Human Factors Issues. New York, Springer-Verlag,1993, pp 13358.


Documents

Improving Patient Safety by Identifying Side Effects From Introducing Bar Coding in Medication Administration