summarize 5 articles in your won words article is attached…

The topic is medication error, so make sure u focus on that and plz summarise in ur own words.Thank you

the summary has to be half a page each at least for each article.

ORIGINAL RESEARCH

Evaluation of antiretroviral-related errors and
interventions by the clinical pharmacist in hospitalized
HIV-infected patients

E Carcelero,1 M Tuset,1 M Martin,1 E De Lazzari,2 C Codina,1 J Miró3 and JM Gatell3

1Department of Pharmacy, Hospital Clinic, Barcelona, Spain, 2Biostatistic Unit, Hospital Clinic, Barcelona, Spain and
3Department of Infectious Diseases, Hospital Clinic, Barcelona, Spain

Objectives
The aim of the study was to identify antiretroviral-related errors in the prescribing of medication to
HIV-infected inpatients and to ascertain the degree of acceptance of the pharmacist’s interventions.

Methods

An observational, prospective, 1-year study was conducted in a 750-bed tertiary-care teaching
hospital by a pharmacist trained in HIV pharmacotherapy. Interactions with antiretrovirals were
checked for contraindicated combinations. Inpatient antiretroviral prescriptions were compared with
outpatient dispensing records for reconciliation. Renal and hepatic function was monitored to
determine the need for dose adjustments.

Results

The prescriptions for 247 admissions (189 patients) were reviewed. Sixty antiretroviral-related
problems were identified in 41 patients (21.7%). The most common problem was contraindicated
combinations (n 5 20; 33.3%), followed by incorrect dose (n 5 10; 16.7%), dose omission (n 5 9;
15%), lack of dosage reduction in patients with renal or hepatic impairment (n 5 6; 10% and n 5 1;
1.7%, respectively), omission of an antiretroviral (n 5 6; 10%), addition of an alternative
antiretroviral (n 5 5; 8.3%) and incorrect schedule according to outpatient treatment (n 5 3; 5%).
Fifteen out of 20 errors were made during admission. A multivariate analysis showed that factors
associated with an increased risk of antiretroviral-related problems included renal impairment [odds
ratio (OR) 3.95; 95% confidence interval (CI) 1.39–11.23], treatment with atazanavir (OR 3.53; 95%
CI 1.61–7.76) and admission to a unit other than an infectious diseases unit (OR 2.50; 95% CI 1.28–
4.88). Use of a nonnucleoside reverse transcriptase inhibitor was a protective factor (OR 0.33; 95% CI
0.13–0.81). Ninety-two per cent of the pharmacist’s interventions were accepted.

Conclusion

Antiretroviral-related errors affected more than one-in-five patients. The most common causes of
error were contraindicated or not recommended drug–drug combinations and dose-related errors.
A clinical pharmacist trained in HIV pharmacotherapy could help to detect errors and reduce the
duration of their effect.

Keywords: antiretroviral agents, drug interactions, hospital, medication errors, pharmacy service

Accepted 4 January 2011

Background

Previous studies suggest that patients receiving long-term
medication are at risk of accidental prescription errors on
admission to hospital [1,2]. HIV-infected patients receiving

highly active antiretroviral therapy (HAART) are at
substantial risk of antiretroviral medication errors during
hospitalization, because of the complexity of HAART
regimens and the possibility of drug–drug interactions
(which can place patients at risk of toxicity or drug
resistance) [3]. These errors may not have been resolved
when patients are discharged. When part of the antire-
troviral regimen is missed or antiretroviral levels are low as

Correspondence: Esther Carcelero, Department of Pharmacy, Hospital Clinic,
Villarroel, 170 escalera 8, Sótano 08036, Barcelona, Spain. Tel: 1 34
932275479; fax: 1 34 932275457; e-mail: ecarcelero@yahoo.es,
ecarcele@clinic.ub.es

DOI: 10.1111/j.1468-1293.2011.00915.x
HIV Medicine (2011), 12,

494

–499 r 2011 British HIV Association

494

mailto:ecarcelero@yahoo.es

mailto:ecarcele@clinic.ub.es

a result of interactions or dosage errors, the virus can
replicate and resistance to treatment can appear. In
contrast, toxicity can occur when an interaction leads to
increased antiretroviral concentrations or the patient
receives a higher dose than the correct one. Resistance or
toxicity is more likely to occur when the error is extended
in time or when the error has not been resolved before the
patient’s discharge.

Some authors have confirmed that

HAART-related errors

are common in hospitalized patients and that admission of
an HIV-infected patient by a physician not specialized in
infectious diseases could be a risk factor for drug-related
problems [4].

The aims of this study were to identify and describe
HAART-related errors in medication prescribed to HIV-
infected patients admitted to a tertiary teaching hospital
and to determine the degree of acceptance of the
pharmacist’s interventions.

Methods

We conducted an observational, prospective, 1-year study
(between 1 January and 31 December 2007). Twice a week
(on Tuesday and Thursday), a pharmacy resident trained in
HIV pharmacotherapy and supported by a staff infectious
diseases pharmacist identified patients aged at least 18
years who had been admitted to the Hospital Clinic (a 750-
bed tertiary teaching hospital in Barcelona, Spain) and
prescribed HAART.

A list was made of all inpatients who were prescribed
antiretroviral drugs. Admissions made on Fridays, at
weekends and on Mondays were recorded on Tuesday
afternoon. Admissions made on Tuesdays, Wednesdays and
Thursdays were recorded on Thursday afternoon.

The following data were recorded for all patients: age,
gender, risk factors for HIV infection, admitting service,
serum creatinine level and liver function (serum albumin,
total bilirubin, transaminases, and international normal-
ized ratio). For those patients with an altered creatinine
value (41.2 mg/dL), the glomerular filtration rate was
calculated using the Cockcroft–Gault equation [5]. For
those patients with any abnormal liver function test result,
the admission report was checked to determine whether
they had cirrhosis, in which case the Child–Pugh score [6,7]
was also recorded. Concomitant medication was reviewed
twice weekly to check for drug–drug interactions.

HAART errors were classified as follows: contraindicated
or not recommended drug–drug combinations, incorrect or
incomplete antiretroviral regimen, omitted dose, incorrect
dose (not matching the outpatient prescription), lack of
dose reduction for renal or hepatic impairment and
incorrect schedule [8].

In Spain, HIV-infected patients pick up their antiretro-
viral medication in the outpatient pharmacy unit of the
hospital that they attend for care. Therefore, it was easy for
us to determine the patient’s HAART regimen. In our
hospital, pharmacists dispensing medication to HIV-
infected patients have access to an internal database
containing the medical records of all HIV-infected
outpatients followed at our hospital HIV clinic. Both this
database and pharmacy dispensing records were checked to
identify discrepancies. The inpatient regimen was consid-
ered correct if it matched the outpatient regimen. For those
patients not followed at the hospital HIV clinic, admission
data were also checked to rule out transcription errors.

Drug–drug interactions were checked for contraindicated
or not recommended combinations using national and
international HIV websites [9–11]. If an error or interaction
was detected, the pharmacist phoned the attending
physician or nurse or added a footnote with a recommen-
dation to the computerized prescription, so that the
attending physician could see it the following day. The
acceptance of the pharmacist’s recommendations was also
reviewed during the following days. If the error was not
corrected within 48 h of the recommendation, the prescrip-
tion was classed as not accepted.

Data were entered into an ACCESS 2.0 database (Microsoft
Corp., Redmond, WA, USA).

Statistical analysis

For the descriptive analysis, qualitative variables were
expressed as percentages and frequencies; quantitative
variables were expressed as the mean (standard deviation
[SD]). Fisher’s exact test was used to analyse contingency
tables. Odds ratios (ORs) for risk factors associated with
HAART-related problems were analysed using a general-
ized estimating equation model. This multivariate model
takes into account the correlation between different
admissions belonging to the same patient.

The statistical analysis was performed using STATA
(StataCorp. 2007, Stata Statistical Software, Release 10;
Stata Corporation, College Station, TX, USA).

Results

Over a 1-year period, we reviewed the prescriptions for 247
admissions of 189 HIV-infected patients who received
antiretroviral therapy. Forty-one patients were admitted
more than once during the study period. Table 1
summarizes the demographic characteristics of these
patients. The distribution of admissions by service was as
follows: infectious diseases unit, 135 (54.7%); other
medical units, 58 (23.5%); surgery services, 38 (15.4%);

Evaluation of HAART related errors in hospitalized HIV-patients 495

r 2011 British HIV Association HIV Medicine (2011) 12, 494–499

intensive care units, nine (3.6%); and units with surgical
and nonsurgical patients, seven (2.8%).

A total of 60 antiretroviral drug-related problems were
identified in 41 patients (21.7% of the admitted patients
had at least one antiretroviral problem).

The types of HAART-related errors found are shown in
Table 2. The most common was drug–drug interaction
(33.3%), not only between antiretroviral agents, but also
between antiretrovirals and other drugs. Atazanavir was
the drug most commonly involved in interactions. The
second most common problem was incorrect dose (16.7%),
and the third most common was dose omission (15%),
followed by lack of dosage reduction in patients with renal

or hepatic impairment (11.7%), omission of one or more
antiretroviral medications (10%), addition of an alternative
antiretroviral drug (8.3%) and incorrect schedule according
to outpatient treatment (5%). Table 3 provides the
antiretroviral prescribing errors detected in this study.

Almost all the antiretroviral-related errors occurred at
admission (15; 75%). The error occurred in the HIV clinic in
only five cases and was not resolved on admission (four
cases of lack of dosage reduction in patients with renal
impairment; one case of a contraindicated interaction).

Of 112 admissions to services other than infectious
diseases in which antiretroviral agents had been prescribed,
39 had at least one antiretroviral drug-related error
(34.8%), compared with 21 out of 135 admissions in the
infectious diseases unit (15.6%).

In the multivariate analysis, the factors associated with
an increased risk of HAART-related problems (Table 4)
were renal impairment [OR 3.95; 95% confidence interval
(CI) 1.39–11.23], treatment with atazanavir (OR 3.53; 95%
CI 1.61–7.76) and admission to a unit other than an
infectious diseases unit (OR 2.50; 95% CI 1.28–4.88).
Prescription of a nonnucleoside reverse transcriptase
inhibitor was a protective factor (OR 0.33; 95% CI 0.13–
0.81). No statistical relationship was found between
HAART-related problems and the following factors: age,
sex, risk group, liver impairment, nucleoside reverse
transcriptase inhibitor-based HAART, a protease inhibitor
other than atazanavir, and being treated with an anti-
retroviral with different presentations.

The most common intervention by the pharmacist was a
footnote on the prescription (45 of 60; 75%), followed by a
telephone call to the attending physician (22 of 60; 36.7%)
or nurse (6 of 60; 10%).

The pharmacist made an intervention in all of the 60
errors detected. This was well accepted in most cases (55 of
60; 91.7%), and the error was resolved. Five interventions
were not accepted (8.3%): lack of dosage reduction in
patients with renal impairment (three cases), lack of
efavirenz dosage reduction in a patient with hepatic
impairment (one case), and a contraindicated combination
(atazanavir and omeprazole; one case).

Discussion

There is evidence that antiretroviral errors are common
during hospital admission. Mok et al. [4] prospectively
reviewed the medical records of 83 HIV-infected patients
who received antiretroviral therapy for 20 months and
identified a total of 176 drug-related problems in 71
patients (86% of the patients had at least one problem
associated with their antiretroviral regimen). Over 4
months, Pastakia et al. [12] prospectively evaluated

Table 1 Demographic characteristics of hospitalized patients receiving
antiretroviral therapy (n 5 189)

Variable

HAART-related errors

P-valueYes No

Male [n (%)] 34 (25.2) 101 (74.8) 0.174*
Female [n (%)] 7 (12.9) 47 (87.0)
Age (years) 47 � 11w 45 � 10w 0.230z
Risk group [n (%)]§

Injecting drug use 16 (21.9) 57 (78.1) 0.462*

Heterosexual 11 (24.4) 34 (75.6)
Homosexual 9 (18) 41 (82)
Other 3 (42.9) 4 (57.1)

HAART, highly active antiretroviral therapy.
*Fisher’s exact test.
wMean (standard deviation).
zt-test.
§n 5 175.

Table 2 Types of highly active antiretroviral therapy (HAART)-related
error

Type of error

Number of
patients*
(n 5 41)

% of all
errors
(n 5 60)

Contraindicated or not recommended
drug–drug combinations

20 33.3

Incorrect dose (not matching outpatient
prescription)

10 16.7

Higher dose 5
Lower dose 5

Dose omission (antiretroviral marketed in
different doses; dose at prescription was
omitted)

9 15

Lack of dose reduction in patients with renal
or hepatic impairment

7 11.7

Renal impairment 6
Hepatic impairment 1

Omission of one or more antiretroviral drugs 6 10
Prescription of alternative antiretroviral drugs 5 8.3
Incorrect schedule according to outpatient
treatment and/or current guidelines

3 5

*A patient could have more than one type of HAART-related error.

496 E Carcelero et al.

r 2011 British HIV Association HIV Medicine (2011) 12, 494–499

antiretroviral prescribing errors in 68 hospitalized HIV-
infected patients and found that there was at least one error
in 72% of cases; in 56% of cases, the error had the potential
to cause moderate to severe discomfort or clinical
impairment. In a retrospective study, Purdy et al. [13]
identified 108 clinically significant prescribing errors
involving antiretrovirals during a 34-month study period
in hospitalized HIV-infected patients. Overall, errors
occurred in 5.8% of inpatients prescribed antiretroviral
medication. Rastegar et al. [14] retrospectively identified 61
antiretroviral errors in 54 admissions (percentage of total
admissions, 25.8%) over a 1-year period. In a 6-month
study, Heelon et al. [3] found 73 HAART errors in 41
patients (21% of hospitalized patients with HIV infection),
most of which were the result of incomplete regimens. In
our study, 21.7% of HIV-infected patients admitted and
prescribed antiretroviral therapy had at least one prescrip-
tion-related problem. These results are similar to those of
Rastegar et al. and Heelon et al. The most commonly
observed problems are inappropriate dosage and drug–
drug interactions. Mok et al. [4] found that, among 251
prescriptions for antiretroviral agents, the dosage was

Table 3 Antiretroviral prescribing errors detected in this study

Type of error Description

Contraindicated
combination (n 5 20)

Patients were prescribed omeprazole while receiving
atazanavir (n 5 18)
A patient treated with efavirenz began voriconazole (no
dose adjustment data from pharmacokinetic studies
were available at the time) (n 5 1)
A patient treated with lopinavir/ritonavir was prescribed
rifampicin (n 5 1)

Incorrect dose of an
antiretroviral agent
(n 5 10)

A patient was prescribed nevirapine 200 mg qd when his
current regimen was 400 mg qd (n 5 1)
A patient was prescribed ritonavir 200 mg instead of
100 mg (n 5 1)
Patients were prescribed ritonavir 100 mg bid instead of
100 mg qd (n 5 2)
A patient was prescribed saquinavir 200 mg tid. His
current regimen was 1500 mg qd (n 5 1)
A patient with renal impairment was prescribed abacavir
150 mg/day (one-quarter of a tablet bid). He was
receiving one tablet of abacavir 300 mg bid (600 mg/
day), as the dosage of abacavir does not need to be
adjusted in renal impairment (n 5 1)
A patient with renal impairment was prescribed
lamivudine 50 mg qd. His current regimen was 25 mg
(n 5 1)
A patient was prescribed 1 capsule of atazanavir qd. His
current dose was 2 capsules qd (n 5 1)
A patient was prescribed atazanavir 200 mg bid. His
current dose was 300 mg/day (n 5 1)
A patient was prescribed efavirenz 200 mg qd instead of
600 mg qd (n 5 1)

Lack of dose
reduction in patients
with renal or hepatic
impairment (n 5 7)

A patient with hepatic cirrhosis Child C was prescribed
efavirenz 600 mg qd. Efavirenz is contraindicated in
Child C patients (risk of toxicity). Plasma concentrations
were not determined (n 5 1)
A patient was prescribed didanosine 400 mg qd. His
current dose was 125 mg qd (he was undergoing dialysis)
(n 5 1)
A patient was prescribed lamivudine 300 mg qd. His
current dose was 25 mg qd (he was undergoing dialysis)
(n 5 1)
A patient was prescribed lamivudine 300 mg qd. He had
renal impairment and the correct dose adjusted to renal
function was 150 mg (n 5 1)
A patient was prescribed stavudine 30 mg bid. The
correct dose adjusted to renal function was 20 mg bid
(n 5 1)
A patient was prescribed abacavir 600 mg 1 lamivudine
300 mg. His serum creatinine was 2.6 mg/dL (n 5 1)
A patient was prescribed lamivudine 50 mg qd. His
current dose was 25 mg qd (he was undergoing dialysis)
(n 5 1)

Antiretroviral dose
omitted (n 5 9)

Patients were prescribed atazanavir 2 capsules qd. The
dose was not specified (n 5 2)
Patients receiving didanosine as outpatient treatment
were prescribed didanosine 1 capsule qd and the dose
was not indicated (n 5 2)
Patients were prescribed lamivudine 1 tablet qd (the
dose was not specified) (n 5 3)
Patients were prescribed stavudine 1 capsule bid (the
dose was not specified) (n 5 2)

One or more
antiretrovirals
omitted (n 5 6)

A patient was prescribed ritonavir 100 mg bid but not
tipranavir, which was the current protease inhibitor
(boosted by ritonavir) (n 5 1)

Table 3 (Contd.)

Type of error Description

A patient was prescribed ritonavir 100 mg bid but not
fosamprenavir, which was the current protease inhibitor
(boosted by ritonavir) (n 5 1)
A patient whose current regimen was stavudine,
didanosine, darunavir and ritonavir was not prescribed
stavudine (n 5 1)
A patient whose current regimen was tenofovir,
nevirapine, atazanavir and ritonavir was not prescribed
atazanavir (n 5 1)
Patients were prescribed only part of their current
regimens. Emtricitabine 200 g 1 tenofovir 300 mg was
missing (n 5 2)

Prescription of an
antiretroviral that
was not part of the
patient’s current
regimen (n 5 5)

Patients were prescribed ritonavir 100 mg instead of
lamivudine (n 5 2)
A patient was prescribed his current antiretroviral
regimen and efavirenz, which did not belong to his
chronic treatment schedule (n 5 1)
A patient was prescribed amprenavir instead of
tipranavir (n 5 1)
A patient was prescribed zidovudine bid instead of
zidovudine 1 lamivudine (n 5 1)

Incorrect schedule
(n 5 3)

A patient treated with atazanavir was prescribed one
capsule at breakfast and the other at dinner instead of
both capsules at the same time (n 5 1)
A patient was prescribed 2 tablets of
abacavir 1 lamivudine 1 zidovudine at breakfast instead
of 1 tablet bid (n 5 1)
A patient treated with atazanavir and ritonavir was
prescribed atazanavir at breakfast and ritonavir at dinner
instead of both drugs at the same time (n 5 1)

bid, twice daily; qd, once daily; tid, three times a day.

Evaluation of HAART related errors in hospitalized HIV-patients 497

r 2011 British HIV Association HIV Medicine (2011) 12, 494–499

inappropriate in 57 cases (37 excessive and 20 insufficient),
accounting for 32.4% of all detected problems. The lack of
an adjustment for renal insufficiency was also considered
an excessive dosage; this happened on 19 occasions. Forty-
six drug–drug interactions were identified (26.1% of all
detected problems); 36 of the 83 patients included in the
review (43.4%) had an incomplete antiretroviral regimen
(20.4% of all problems detected). Dosage error was also the
most common type of error detected by Rastegar et al. [14]
(34 admissions; 16.3%); 18 of these errors were inap-
propriate dosage adjustment in patients with renal
insufficiency. The next most common error was contra-
indicated combinations (12 admissions; 5.2%), followed by
receiving two or fewer antiretroviral agents (eight cases;
3.8%). In seven admissions (3.3%) there was an unex-
plained delay in continuing HAART. Gray et al. [15]
analysed the causes of HIV medication errors in MED-
MARX, a voluntary database reporting inpatient medica-
tion errors. They found that the most common causes of
error were inappropriate dosing (38%), followed by
incorrect medication (32%). In our study, interactions
caused by contraindicated or not recommended drug–drug
combinations (33.3%) were slightly higher than in the
study by Mok et al. [4]. We found that, in total, dose-related
problems (incorrect dose, dose omission, and lack of dose
adjustment in patients with renal or hepatic impairment)
accounted for 43.3% of all errors. This result is comparable
to those of Mok et al. [4] and Gray et al. [15]

Risk factors associated with a HAART-related error in
our study were similar to those found by Mok et al. [4]:
renal impairment, an atazanavir-containing regimen,
and admission by a service other than the infectious
diseases service. We also found that receiving a nonnucleo-
side reverse transcriptase inhibitor was a protective
factor.

There is abundant evidence that antiretroviral drug-
related errors on admission are frequent and may be of
clinical relevance. Clinical pharmacists with training in
HIV pharmacotherapy can play an important role in
correcting such errors. They should closely monitor
prescriptions to identify problems and resolve them as
soon as possible in order to prevent toxicity or drug
resistance. Our results show that the pharmacist’s recom-
mendations were frequently accepted.

This study has several limitations. First, hospitalized
patients prescribed an antiretroviral were only followed
twice a week. Admissions made on Fridays, at weekends, and
on Mondays were recorded on Tuesday afternoon, so some
patients could have been missed if they were admitted and
discharged between our monitoring dates. Secondly, the
method used did not allow us to detect errors of complete
HAART omission during hospitalization. Delays in continu-
ing the outpatient regimen were not detected either. Thirdly,
we did not assess dispensing or administration errors, or the
clinical outcomes of our interventions (prevention of drug
toxicity or drug resistance). These limitations mean that it is
difficult to make generalizations based on our results.
Finally, the current recommendations for atazanavir in
combination with proton pump inhibitors differ from those
available when the study was performed: atazanavir can be
used with proton pump inhibitors at present, although only
at low doses in treatment-naı̈ve patients. Most of the
patients admitted during the study period were treatment-
experienced.

Conclusion

Errors in, or problems with, the HAART regimen were
common among HIV-infected hospitalized patients pre-
scribed antiretroviral agents (approximately one-in-five
patients). The most common issues were contraindicated or
not recommended drug–drug combinations and dose-
related errors. Factors associated with an increased risk of
such problems were renal impairment, receiving atazana-
vir, and admission to a unit other than an infectious
diseases unit. Receiving nonnucleoside reverse transcrip-
tase inhibitors was a protective factor.

Clinical pharmacists trained in HIV pharmacotherapy
could help to detect errors and reduce the duration of their
effects, thus improving the quality of prescription in
hospitalized HIV-infected patients.

Acknowledgements

We are grateful to Kenneth Lawrence (Tufts Medical Center,
Boston, MA) for useful suggestions and to Thomas O’Boyle
for editorial assistance.

Table 4 Adjusted odds ratios for risk factors associated with a highly
active antiretroviral therapy (HAART)-related error (245 episodes)

Variable
Adjusted odds
ratio 95% CI P-value

Renal impairment
No 1 0.010
Yes 3.950 (1.390–11.228)

NNRTI
No 1 0.015
Yes 0.329 (0.134–0.805)

Atazanavir
No 1 0.002
Yes 3.532 (1.608–7.756)

Service
Infectious diseases 1 0.007
Other 2.504 (1.285–4.881)

CI, confidence interval; NNRTI, nonnucleoside reverse transcriptase
inhibitor.

498 E Carcelero et al.

r 2011 British HIV Association HIV Medicine (2011) 12, 494–499

References

1 Cornish PL, Knowles SR, Marchesano R et al. Unintended

medication discrepancies at the time of hospital admission.

Arch Intern Med 2005; 165: 424–429.

2 Classen DC, Pestotnik SL, Evans RS et al. Adverse drug events

in hospitalized patients. Excess length of stay, extra costs, and

attributable mortality. JAMA 1997; 277: 301–306.

3 Heelon M, Skiest D, Tereso G et al. Effect of a clinical

pharmacist’s interventions on duration of antiretroviral-related

errors in hospitalized patients. Am J Health Syst Pharm 2007;

64: 2064–2068.

4 Mok S, Minson Q. Drug-related problems in hospitalized

patients with HIV infection. Am J Health Syst Pharm 2008; 65:

55–59.

5 Cockcroft DW, Gault MH. Prediction of creatinine clearance

from serum creatinine. Nephron 1976; 16: 31–41.

6 Child CG, Turcotte JG. Surgery and portal hypertension, In:

Child CG. ed. The Liver and Portal Hypertension. Philadelphia:

Saunders, 1964: 50–64.

7 Pugh RNH, Murray-Lyon IM, Dawson JL, Pietroni MC,

Williams R. Transection of the esophagus in bleeding

oesophageal varices. Br J Surg 1973; 60: 648–652.

8 Panel de expertos de GESIDA; Plan Nacional sobre el Sida.

Recommendations from the GESIDA/Spanish AIDS Plan

regarding antiretroviral treatment in adults with human

immunodeficiency virus infection (update January 2007).

Enferm Infecc Microbiol Clin 2007; 25: 32–53.

9 Tuset M, Miro JM, Codina C et al. Guia de interacciones en VIH.

Available at www.interaccioneshiv.com (accessed 8 August

2008).

10 University of Liverpool HIV drug interactions website.

Available at www.hiv-druginteractions.org (accessed 8

August 2008).

11 Gagnon A, Therrien R. HIV medication guide. Available at

www.hivmedicationguide.com (accessed 8 August 2008).

12 Pastakia SD, Corbett AH, Raasch RH et al. Frequency of

HIV-related medication errors and associated risk factors in

hospitalized patients. Ann Pharmacother 2008; 42: 491–497.

13 Purdy BD, Raymond AM, Lesar TS. Antiretroviral prescribing

errors in hospitalized patients. Ann Pharmacother 2000; 34:

833–838.

14 Rastegar DA, Knight AM, Monolakis JS. Antiretroviral

medication errors among hospitalized patients with HIV

infection. Clin Infect Dis 2006; 43: 933–938.

15 Gray J, Hicks RW, Hutchings C. Antiretroviral medication

errors in a national medication error database. AIDS Patient

Care STDS 2005; 19: 803–812.

Evaluation of HAART related errors in hospitalized HIV-patients 499

r 2011 British HIV Association HIV Medicine (2011) 12, 494–499

www.interaccioneshiv.com

www.hiv-druginteractions.org/

www.hivmedicationguide.com/

multiple sites or posted to a listserv without the copyright holder’s express written permission. However, users

may print, download, or email articles for individual use.

Objective: This study utilized usability testing and
human factors engineering (HFE) principles to create
efficient code cart medication drawer modifications
to improve code blue medical emergency (code)
medication management.

Background: Effective access to medications
during a code is a key component in delivering optimal
care and has been found to be a major problem among
health care organizations; however, little research has
been conducted to improve the efficiency of medication
management during a code.

Method: A total of 26 health care professionals
(13 pharmacists and 13 nurses) were asked to locate
items within a code cart medication drawer during two
independent simulated code scenarios alternately using
either a baseline medication drawer (control; Drawer 1) or
a prototype medication drawer (prototype; Drawer 2),
which was developed using HFE principles and usability
testing. Overall medication retrieval time, wasteful
actions, and survey responses were recorded.

Results: Drawer 2 had significantly faster trial
completion times (p = .005) and fewer wasteful actions
(p < .001) compared to Drawer 1. Participant survey results rated Drawer 2 (prototype) significantly higher (more favorable) for medication drawer visibility (p < .001), usability (p = .011), and organization (p < .001) compared to Drawer 1 (baseline).

Conclusion: The HFE redesign concepts incorpo-
rated into Drawer 2 (consisting of visibility, grouping,
and organization) produced successful, low-cost, and
generalizable modifications that can improve patient
care.

Application: The findings demonstrate that HFE
and usability applied to code cart design are effective,
are customizable, and can affect patient safety by saving
valuable time and reducing wasted motions (including
errors) during code situations.

Keywords: health care, code cart, medical emergency,
usability, human factors engineering

IntroductIon
Standard code carts (aka crash carts or resus-

citation carts) typically have drawers organized
by intervention (i.e., intubation, intravenous
access, and resuscitation medications) to assist
health care personnel during a code blue medical
emergency (code; Hand & Banks, 2004). The
items in each drawer must also be sorted to find
appropriately sized equipment and medications
for patients ranging in age and size. Since codes
are infrequent and chaotic in nature, health care
professionals become unfamiliar with the loca-
tion of specific items in the cart, and this may
cause undue delay in securing the appropriate
equipment during resuscitation (Agarwal et al.,
2005). Inaccurate and incomplete information
transmission during codes has also been found to
add complexity and delay medication manage-
ment (Bogenstatter et al., 2009).

High variability of supplies can lead to poorly
organized code carts and carts without necessary
items; this ultimately adds time to procedures
and affects patient outcomes (Agarwal et al.,
2005; Jankouskas, 2001). Nurses may also
unknowingly look elsewhere for supplies that
are stored in the code cart right next to the col-
lapsed patient during codes (Jankouskas, 2001),
adding a significant delay during a code when
time is critical. A delay in opening the airway or
instituting other life support procedures impairs
the arrest victim’s chances for survival (American
Heart Association [AHA], 2010; Brenner &
Kauffmann, 1996).

Code cart standardization has been found to
affect patient safety (Janda, Coleman, &
Sullivan, 2004; Peterson & Berns, 2006). Nurses
are typically found to “float” between different
patient care units, which can affect both the
nurses’ confidence and performance during a
code because of unfamiliarity with the code cart
layout. Janda et al. (2004) concluded that stan-
dardizing the code cart improved patient care by

Address correspondence to M. Susan Hallbeck, W342
Nebraska Hall, Lincoln, NE 68588-0526; Hallbeck@unl.edu.

HUMAN FACTORS
Vol. 53, No. 6, December 2011, pp. 626-63

DOI:10.1177/0018720811426427
Copyright © 2011, Human Factors and Ergonomics Society.

Improving Medication Management
Through the Redesign of the Hospital
Code Cart Medication Drawer

Justin B. Rousek and M. Susan Hallbeck, University of Nebraska–Lincoln

2011 Human Factors Prize Finalist

ImprovIng medIcatIon management 627

(a) making nurses more comfortable in emer-
gency situations when working outside their
usual areas, (b) adding advanced cardiac life
support (ACLS) drugs to the code cart, and
(c) decreasing pharmacy refill times. Additional
studies have found similar results regarding
medication restocking compliance and the
importance of having all supplies available at all
times (Smith et al., 2008; West, Halls, Coleman,
& Lowe, 2008). It is evident from these studies
that poorly organized code cart medication
drawers may cause the health care personnel in
charge of the code cart to overlook certain medi-
cations because of nonstandardized locations.

McLaughlin (2003) utilized usability testing
to examine code cart organization standardiza-
tion and found that labeling, spacing, and sys-
tematic placement of medications and supplies
decreased overall retrieval time when compar-
ing the baseline medication drawer to the rede-
signed drawer. In addition to labeling improve-
ments, the Joint Commission on Accreditation of
Healthcare Organizations (JCAHO) has set label-
ing requirements on all emergency medications
and supplies (medication management standard
MM 2.30; Kienle, 2006). Agarwal et al. (2005)
conducted a study on pediatric code carts and
found that having medications systematically
organized (i.e., designed to facilitate their iden-
tification, use, and replacement) reduced medi-
cation retrieval times when compared to
baseline drawers. Schultz et al. (2010) collected
feedback from relevant health care personnel
and determined standardization, grouping of
like medications, and organizing based on fre-
quency of use were the most important factors
to consider during code cart redesign; however,
no formal testing was conducted. Last, it has
been concluded that humans’ cognitive archi-
tecture consists of a limited working memory
(Paas, Tuoviinen, Tabbers, & Van Gerven,
2003), which adds to the importance of these
code cart organizational improvements since a
well-organized code cart can reduce load on
working memory.

In addition to code cart organizational issues,
expired medications and wrong drug and dosing
amounts in code carts have been found to be pre-
scribed to patients during medical emergency
procedures (Bussieres et al., 2009). To reduce

the amount of prescription errors and omissions
in dose entries, Bussieres et al. (2009) recom-
mend the introduction of a preprinted admission
order sheet with doses and volumes calculated
for the weights of each person admitted. West et
al. (2008) took an innovative redesign approach
to the code cart as well. The study split the newly
designed code cart (called the Resus:station)
into three connected subcarts, so each code team
member has his or her own designated specialty
equipment laid out for fast access.

The code cart redesign strategies in the pre-
viously mentioned studies, which involve end
users to identify shortfalls in development, are
essential in optimizing code cart usability
(Gosbee, 2004). Gosbee (2004) recommends
interviewing experienced medical staff involved
with codes to identify issues involving design
flaws that are present in emergency care equip-
ment and code carts. Interviewing end users can
identify problems that have not been considered
previously in addition to potential solutions.
This study expands on this concept and utilizes
health care personnel throughout the design and
planning process. The purpose of this study was
to take an innovative approach to code cart
redesign by utilizing usability testing and
human factors engineering (HFE) principles,
such as improved medication visibility, organi-
zation, and grouping. We hypothesize that these
principles will lead to the creation of an effi-
cient code cart medication drawer that will ulti-
mately improve medication management and
patient safety by reducing medication locating
time and wasteful actions during a code. It is
also hypothesized that the participants will pre-
fer a code cart medication drawer based on
usability testing and HFE principles to one that
is not based on end user testing.

Method
Study design

A randomized, controlled, crossover trial
was utilized in this study. Health care profes-
sionals were asked to locate items within a code
cart medication drawer (10 items per drawer)
during two simulated code blue scenarios (tri-
als). The study participants were assigned to
locate the items using either a baseline code
cart medication drawer (control; Drawer 1) or a

628 December 2011 – Human Factors

prototype medication drawer (intervention;
Drawer 2) during their first trial. They were
subsequently assigned to the opposite medica-
tion drawer during the second trial.

Study Population

A total of 26 health care professionals con-
sisting of 13 pharmacists (8 females) and 13
nurses (12 females) participated in this study,
which was conducted at the Nebraska Medical
Center in Omaha. All participants were ACLS
certified and were either registered nurses or
pharmacists. All participants gave informed
consent prior to the study.

Apparatus

Two code cart medication drawers were uti-
lized in this study. The first baseline medication
drawer (Drawer 1) was currently being imple-
mented within code carts during this study and
was being utilized in a medical setting for code
alerts; the contents of Drawer 1 can be seen in
Figure 1.

The second code cart medication drawer used
in this study was a prototype drawer designed
through pharmacist and nurse usability testing.
The experimenters met with several pharma-
cists, nurses, and pharmacy and nursing manag-
ers on several different occasions to discuss and
determine what the usability problems were

with the current medication drawer and how to
improve performance. The pharmacists, nurses,
and managers who provided input during usabil-
ity testing were not included as participants dur-
ing the medication drawer testing. The most
common problems stated by the pharmacists and
nurses were poor organization, visibility, and
grouping of the medications. The experimenters
took all comments and suggestions into consid-
eration and developed several modified drawer
layout designs that the pharmacists and nurses
interacted with until a final prototype drawer
was created (Drawer 2; Figure 2). Drawer 2
included the same medications as Drawer 1 but
was reorganized by grouping like medications,
creating new dividers and vial holders, stacking
multiple medications differently, and making all
of the medication labels visible. Because of the
pharmacists’ and nurses’ experience with the
baseline drawer, general concepts (e.g., vials on
the front left and preloaded syringes on the right
of the drawer) were retained in the prototype
drawer to prevent a steep learning curve.
Epinephrine, vasopressin, amiodarone, and
magnesium sulfate were also moved toward the
front of the drawer because of their frequency of
use (AHA, 2010). In addition to these modifica-
tions, the drawer design had to be customizable
based on varying lengths and types of medica-
tion packages (e.g., different-length preloaded

Figure 1. Baseline code cart medication drawer (Drawer 1) with fluid bags (A) and without fluid bags (B).

ImprovIng medIcatIon management 629

syringes or vials instead of preloaded syringes)
and future additions of new medications.

Questionnaire

One Likert-type scale questionnaire handout
was utilized in this study, which was completed
twice after two similar trials by the participants.
A 7-point Likert-type scale was used to assess
the impression of each medication drawer
based on three given statements concerning
medication drawer content, usability, and orga-
nization (Figure 3).

Procedures

On being informed of the aims of the
research, the voluntary nature of participation,
and Institutional Review Board approval from
the hospital, each participant was given an
informed consent form. After consent was
received, the experimenter explained the study
briefly again, and then the participant opened
the code cart door and pulled out the medica-
tion drawer two thirds of the way. This partially
closed drawer was used to mimic the distance
the drawer can be extended in a crowded room,
and the experimenter made sure the distance
was consistently positioned prior to beginning
the study. The entire code cart can be seen in
Figure 4, with the prototype drawer layout
(Drawer 2).

The study consisted of two successive code
simulation scenarios that were assigned to two
medication drawers, creating two trials within

the study. The scenarios were timed and ran-
domly ordered (Scenario 1 and Scenario 2; see
Table 1), and the 10 medications were chosen
based on typical items used during code events
(McLaughlin, 2003). The two medication draw-
ers consisted of the baseline code cart medication
drawer (Drawer 1) and a code cart medication
drawer prototype (Drawer 2). The number of
pharmacists and nurses using Drawer 1 or
Drawer 2 first was controlled so half of the phar-
macists and nurses used Drawer 1 first and the
other half used Drawer 2 first. The first scenario
used was randomized independently by using a
random number generator. The second scenario
used was determined by assigning the alternate
scenario (i.e., if Scenario 1 was used first, then
Scenario 2 was used second, and vice versa).

The first trial began by having the experi-
menter ask the participant to hand him the first
medication on the scenario list (this is when the
timer started). Once the appropriate medication
was located and handed to the experimenter, the
next medication on the list was read. If the incor-
rect medication was handed to the experimenter,
then the initial time to retrieve the wrong medi-
cation was added to the time it took to retrieve
the correct medication after a second request
from the experimenter and was counted as a
wasteful action. The time stopped when the 10th
medication was located and handed to the
experimenter.

The participants were given a 5-min break
after completion of the first trial. During this
time each participant completed a 7-point
Likert-type scale questionnaire (Figure 3) ask-
ing about his or her experience with the specific
medication drawer. After the break, the second
medication drawer was placed in the code cart
(replacing the other drawer), and once the par-
ticipant was ready the second trial began. The
second trial was identical to the first trial, except
for the design of the medication drawer (Drawer 1
or Drawer 2) and scenario utilized (Scenario 1 or
Scenario 2). The same Likert-type scale ques-
tionnaire was completed after Trial 2, and a final
question was asked of the participant. The ques-
tion was, “Which medication drawer would you
prefer to use during a code blue event (Drawer 1
or Drawer 2)?” The study was complete after the
participant gave his or her response.

Figure 2. Prototype code cart medication drawer
(Drawer 2).

630 December 2011 – Human Factors

Analytical Method

This randomized, controlled, crossover trial
utilized repeated measures analysis of variance
(ANOVA), the paired-sample Wilcoxon signed
rank test, and McNemar’s chi-square test. All
statistical tests were performed using SPSS (V19)
and used a .05 level of significance. Data col-
lected include trial completion time, wasteful

actions (which included errors), medication
drawer and scenario used, drawer order, partici-
pant occupation, preferred medication drawer,
and the three Likert-type scale responses for each
scenario. The wasteful actions variable was
defined as moving an incorrect medication greater
than 2 in. or rotating a vial greater than 90 degrees
(e.g., turning an incorrect vial to find the label,
moving medications to retrieve or view other
medications, etc.). The occurrences of wasteful
actions were determined by the experimenters
after watching retrospective video recordings of
each trial after the completion of the study.

The dependent variables that were collected
for the repeated measures ANOVA consisted of
the scenario completion time and wasteful
actions. The independent variables (IVs) that
were tested are medication drawer (Drawer 1 or
Drawer 2), medication drawer scenario (Scenario
1 or Scenario 2), drawer order (Drawer Order 1
or Drawer Order 2), and occupation (pharmacist
or nurse). The drawer order consisted of which
medication drawer was used in the first trial and
which one was used in the second trial.

The Likert-type scale questions were graphed
by the frequency of answer 1 (strongly disagree)
to 7 (strongly agree). The six ordinal responses
by each participant (2 trials × 3 questions)

Code cart medication drawer questionnaire

Participant # ________

Drawer used ________

Please circle the number that is most appropriate as an answer to the given comment.

Strongly
Disagree

Disagree
Slightly
Disagree

Neutral
Slightly

Agree

Agree
Strongly

Agree

1.
The medications in this
drawer were easily
visible

1 2 3 4 5 6 7

2.
I could easily find all
the needed medications
in this drawer

1 2 3 4 5 6 7

3.
Overall, the drawer
contents are well
organized

1 2 3 4 5 6 7

Figure 3. The 7-point Likert-type scale questionnaire that was used in the study.

Figure 4. The code cart that was used in this study
with the medication drawer (Drawer 2) pulled out.

ImprovIng medIcatIon management 631

regarding the Likert-type scale questions were
then tested using the paired-sample Wilcoxon
signed rank test to verify if the responses were
significantly different depending on the medi-
cation drawer used.

McNemar’s chi-square test was used to ana-
lyze if there was a significant difference in the
proportion of participants who preferred Drawer 1
compared to Drawer 2. The medication drawer
preference of each participant was determined
by a final question that was asked by the experi-
menter after completion of both medication
drawer trials.

reSultS
trial completion time

Participants completed the trial with the pro-
totype drawer (Drawer 2; M = 59.7 s, SD = 9.14)
significantly faster than that with the baseline
drawer (Drawer 1; M = 66.8 s, SD = 8.33), F(1,
50) = 8.72, p = .005. The medication drawer
assigned in Trial 2 (Drawer Order 2; M = 59.7 s,
SD = 8.04) was also found to produce signifi-
cantly faster completion times compared to the
medication drawer assigned in Trial 1 (Drawer
Order 1; M = 67.3 s, SD = 8.97), F(1, 50) =
12.15, p < .001. The differences in trial comple- tion times within the IVs scenario used and occupation were not found to be statistically significant.

The same ANOVA test was also conducted
on only the Drawer Order 2 data since the medi-
cation drawer assigned in Trial 2 was found to
be completed significantly faster than the

drawer assigned in Trial 1. This test accounts
for a possible medication learning curve by
going from Trial 1 to Trial 2 for the participants.
This test found similar results showing that the
participants completed the trial with Drawer 2
(M = 54.6 s, SD = 6.37) significantly faster than
that with Drawer 1 (M = 63.8 s, SD = 6.93), F(1,
24) = 14.16, p < .001.

Wasteful Actions

Participants performed significantly fewer
wasteful actions with Drawer 2 (M = 1.3 occur-
rences/trial, SD = 1.25) compared to Drawer 1
(M = 4.1 occ/trial, SD = 1.74), F(1, 50) = 45.42,
p < .001. The participants also performed sig- nificantly fewer wasteful actions with Drawer Order 2 (M = 1.8 occ/trial, SD = 1.61) com- pared to Drawer Order 1 (M = 3.4 occ/trial, SD = 2.20), F(1, 50) = 8.38, p = .006. The different numbers of occurrences of wasteful actions within the IVs scenario used and occupation were not found to be statistically significant.

The same ANOVA test was also conducted
on only the Drawer Order 2 data since the medi-
cation drawer assigned in Trial 2 was found to
produce significantly fewer wasteful actions
compared to the drawer assigned in Trial 1 (a
similar test was completed for the trial comple-
tion time variable and Drawer Order 2 data).
This test found similar results showing that the
participants performed significantly fewer waste-
ful actions with Drawer 2 (M = 0.6 occurrences/
trial, SD = 0.77) compared to Drawer 1 (M = 3.2
occ/trial, SD = 1.14), F(1, 24) = 38.13, p < .001.

TABLE 1: The Two Medication Scenarios (Scenario 1 and Scenario 2) That Were Used in the Two Trials

Scenario 1 Scenario

1 Dobutamine (500 mg) 1 Dopamine (400 mg)
2 Atropine (1 mg) 2 Magnesium sulfate (1 g)
3 Sodium bicarbonate (50 mEq) 3 Epinephrine (1 mg)
4 NS 0.9% (1 L) 4 Atropine (1 mg)
5 Epinephrine (1 mg) 5 Vasopressin (20 U/ML)
6 Lidocaine (100 mg) 6 Atropine (1 mg)
7 Atropine (1 mg) 7 NS 0.9% (1 L)
8 Magnesium sulfate (1 g) 8 Sodium bicarbonate (50 mEq)
9 Vasopressin (20 U/ML) 9 Dobutamine (500 mg)
10 Dopamine (400 mg) 10 Lidocaine (100 mg)

632 December 2011 – Human Factors

likert-type Scale responses

The 7-point Likert-type scale questionnaire
(see Figure 3 for questionnaire), which was
completed after each of the two trials, produced
statistically significant results for all three
questions (Figure 5). Question 1 on visibility
(p < .001), Question 2 on usability (p = .011), and Question 3 on organization (p < .001) all pro- duced significantly higher ratings directly after using Drawer 2 when compared to Drawer 1.

The same statistical test was also conducted
on only the Drawer Order 2 data because of the
learning effect identified with completion time
and wasteful actions in Drawer Order 2 trials.
This test found similar results showing that
Question 1 on visibility (p < .001), Question 2 on usability (p = .002), and Question 3 on orga- nization (p < .001) all produced significantly higher ratings directly after using Drawer 2 compared to Drawer 1.

drawer Preference

After completion of the two study trials, the
participants were asked which of the two drawers
used in this study they would prefer to use during
a future code event. Of the participants, 96% (25
out of 26) preferred to use Drawer 2 during a
code event compared to Drawer 1 (p < .001).

dIScuSSIon
Despite no prior experience with the prototype

code cart medication drawer (Drawer 2), the
participants in this study found it easier to use
and preferred it over the currently used, baseline
drawer (Drawer 1). The results of this study sup-
port the original hypothesis that usability testing
and HFE principles will lead to the creation of an
efficient medication drawer that will ultimately
improve medication management and patient
safety by reducing medication locating time and
wasteful actions during a code. The results also

A B

1 2 3 4 5 6 7

P
ar

�
ci

p
an

Ra�ng

Ques�on 1 Responses – Visibility

Drawer 1
(Baseline)

Drawer 2
(Prototype)

0
2

4
6
8

10
12

14
1 2 3 4 5 6 7
P
ar
�
ci
p
an
ts
Ra�ng

Ques�on 2 Responses – Usability

Drawer 1
(Baseline)
Drawer 2
(Prototype)
0
2
4
6
8
10
12
14
1 2 3 4 5 6 7
P
ar
�
ci
p
an
ts
Ra�ng

Ques�on 3 Responses – Organiza�on

Drawer 1
(Baseline)
Drawer 2
(Prototype)

1 = Strongly disagree 7 = Strongly agree

Figure 5. Likert-type scale responses for Questions 1 (A), 2 (B), and 3 (C) asked after completion of the two
study trials.

ImprovIng medIcatIon management 633

support the hypothesis that the participants will
prefer a medication drawer based on these same
principles compared to one that is not based on
end user testing.

reduction of Medication
Searching time

It is evident from past research that a delay in
patient care impairs the coding patient’s chance
of survival (AHA, 2010; Brenner & Kauffmann,
1996). There are several factors that contribute
to patient care delays (e.g., response time and
personnel competency) but few that can be as
readily controlled as medication management.
Reducing the delay time from a medication
being requested to actually being in the neces-
sary caregiver’s hands was one of the main
objectives of this study. By simply reorganizing
the medication drawer in the prototype based
on usability testing (Drawer 2), the retrieval
time of 10 medications dropped on average 7.1 s
compared to Drawer 1 (p = .005). Although this
may not seem like a long time, every second
counts when a patient is undergoing cardiac or
respiratory arrest (AHA, 2010). The relative
difference is very important since during a real
code situation (where several additional medi-
cations may be needed) the retrieval time being
saved may increase.

Focusing solely on Trial 2 data of the study
(because of participant medication familiarity
by going from Trial 1 to Trial 2) further strength-
ens the evidence in favor of Drawer 2. When the
participants were more familiar, in general, with
the code cart medications (in Trial 2), they
could retrieve the 10 medications on average
9.2 s faster with Drawer 2 compared to Drawer
1 (p < .001). This shows that as pharmacists and nurses obtain higher competency levels on code cart contents, their medication retrieval times will improve more dramatically with Drawer 2 compared to Drawer 1.

reduction of Wasteful Actions

Wasteful actions (including medication error)
by the pharmacist or nurse in charge of the code
cart during a code can add not only time to the
retrieval of medications but also stress when
everyone involved in a chaotic code needs to
perform at their best. An additional extra

(wasteful) step is introduced when fluid bags
are placed on top of other medications, which is
the case in Drawer 1 (Figure 1a). These fluid
bags must be moved and rearranged in the
medication drawer and consequently cover up
additional medications. In the baseline layout,
the vials are also loosely placed in drawer slots
where they roll around and must be rolled over
or picked up to read the labels (Figure 6a).
Drawer 2 utilized a vial holder that has pre-
defined slots for each vial that prevents rolling
and displays all vial labels (this also helps phar-
macists notice when vials are missing and
restocking is required). Drawer 1 produced 3.2
times as many wasteful actions per trial as
Drawer 2 (p < .001) and 5.3 times as many when only Trial 2 data were analyzed (p < .001). These results show that as pharmacists and nurses (on code teams) obtain higher code medication competency levels, the number of wasteful actions will decrease more drastically with Drawer 2 compared to Drawer 1 (similar to the “medication searching time” analysis).

hFe Improvements of Prototype
Medication drawer

Visibility. One of the major issues with the
Drawer 1 was that not all medications were vis-
ible at once. When Drawer 1 is initially pulled
out of the code cart, there are several fluid bags
lying on top of the drawer (Figure 1a). Since
codes are infrequently called, pharmacists and/
or nurses cannot typically memorize the exact
medication drawer layout. This leads to several
bags and medications being moved to locate the
needed item, causing further viewing obstruc-
tion and stress. Drawer 2 alleviates this problem
by moving the bags to the back of the drawer
(placed on an incline for easy access) and dis-
playing the vials within holders to prevent roll-
ing and ensuring the label is visible at all times.
In addition to making all medication labels vis-
ible, Drawer 2 also made almost all of the labels
horizontal to facilitate label reading. Question 1
from the study survey (which asked about
drawer content visibility) supports these find-
ings as well; the participants rated Drawer 2 sig-
nificantly higher than Drawer 1 (p < .001). Both drawers met the JCAHO medication manage- ment standard (MM 2.30; Kienle, 2006) by

634 December 2011 – Human Factors

having all medications labeled, but (as the
results revealed) this may not be sufficient. We
recommend further expansion of this regulation
to include the use of HFE principles during the
development, design, and modification stages
of medication drawers throughout hospitals.

Grouping. Grouping of medications can aid
pharmacists and nurses when they are asked to
retrieve a specific medication and they are unfa-
miliar with its location. The baseline layout
(Drawer 1) does not group like medications
together (Figure 7a). This problem was stated
repeatedly throughout the usability testing stages
of Drawer 1. Drawer 2 solved this problem by
grouping like medications together and keeping
the general location of these medications consis-
tent with Drawer 1 (Figure 7b). This assists the
pharmacist or nurse (who is in charge of the code
cart) to know where to look for a specific preloaded
syringe or a vial even if he or she does not know
the exact location. The study survey (Question 3)
further supports these findings; the participants
rated Drawer 2 significantly higher than Drawer 1
(p < .001) for well-organized contents.

Generalizability of Medication
drawer Improvements

Many of the code cart medication drawer
modifications (incorporated into Drawer 2) that
stemmed from the usability testing and HFE
principles can be generalized for code carts at
other hospitals. The design and contents of hos-

pital code carts vary by organization, but gen-
eral drawer concepts are common. Typically
code carts are divided into drawers or compart-
ments based on intervention (Hand & Banks,
2004), including drawers for medications. The
low-cost and customizable modifications are
financially viable within almost any budget.
Effective medication access during a code is a
key component in delivering optimal care and
improving patient safety. We recommend that
the medication drawer modifications regarding
visibility, grouping, and organization (that were
successfully implemented into Drawer 2 in this
study) be considered for hospital code carts.
Future research will consist of expanding usabil-
ity testing and HFE principles to additional code
cart drawers and testing in a clinical setting.

limitations

This study used pharmacist and nurse partici-
pants from only one hospital. Also, some hospitals
assign only pharmacists or nurses to gather medi-
cations from the code cart during code situations,
whereas this study used both; however, no sig-
nificant performance differences by occupation
emerged. Because of the crossover study design, a
learning effect was found to occur, as evidenced
by the enhanced performance in the participant’s
second trial resulting from experience gained dur-
ing the participant’s first trial. The influence of the
learning effect in this study was minimized by
randomizing the order of both the scenarios used

Figure 6. Vial containers for baseline Drawer 1 (A) and prototype Drawer 2 (B).

ImprovIng medIcatIon management 635

in each trial and the type of medication drawer
used. The crossover study design, with each par-
ticipant serving as his or her own control, was
chosen to prevent confounding variables within
participants (i.e., clinical experience) and to
increase the statistical power available given the
limited number of participants.

AcknoWledGMentS
The authors of this study would like to acknowledge

(a) the Nebraska Medical Center in Omaha for their
access and assistance in this study, (b) David Gannon
and Patrick Fuller for their guidance, and (c) Vincent
Cao and Sarai Obenland for their assistance in data
collection.

key PoIntS

• This study utilizes concrete evaluation findings
that can be used in policy analysis and develop-
ment within hospitals.

• This study demonstrates how effective communi-
cation and involvement (usability testing) among
various professionals and frontline personnel may
lead to systematic improvements.

• This study demonstrates how human factors
engineering principles and usability testing with
professionals can decrease time for tasks and errors

while increasing organization for easier code cart
stocking.

• The code cart medication drawer redesign concepts
are low in cost and generalizable across hospitals.

reFerenceS
Agarwal, S., Swanson, S., Murphy, A., Yaeger, K., Sharek, P., &

Halamek, L. P. (2005). Comparing the utility of a standard
pediatric resuscitation cart with a pediatric resuscitation cart
based on the Broselow tape. Pediatrics, 116, 326–333.

American Heart Association. (2010). 2010 American Heart
Association guidelines for cardiopulmonary resuscitation
and emergency cardiovascular care science. Circulation, 122,
S639–S933.

Bogenstatter, Y., Tschan, F., Semmer, N. K., Spychiger, M., Breuer,
M., & Marsch, S. (2009). How accurate is information trans-
mitted to medical professionals joining a medical emergency?
A simulator study. Human Factors, 51, 115–125.

Brenner, B. E., & Kauffmann, J. (1996). Response to cardiac
arrests in a hospital setting: Delays in ventilation. Resuscita-
tion, 31, 17–23.

Bussieres, J. F., Scharr, K., Marquis, C., Saindon, S., Toledano, B.,
Diliddo, L., & Charrette, S. (2009). Reevaluation of emergency
drug management in a tertiary care mother-child hospital. Hos-
pital Pharmacy, 44, 584–593.

Gosbee, J. (2004). Human factors engineering series: Who left the
defibrillator on? Joint Commission Journal on Quality and
Safety, 30, 282–285.

Hand, H., & Banks, A. (2004). The contents of the resuscitation
trolley. Nursing Standards, 18(44), 43–52.

Janda, S. M., Coleman, D. S., & Sullivan, B. F. (2004, December).
Standardization of crash carts in a community hospital. Paper

Figure 7. Grouping of preloaded syringes (circles) and vials (squares) for Drawer 1 (A; does not include all
fluid bags on cart, but B does) and Drawer 2 (B).

636 December 2011 – Human Factors

presented at the 39th annual American Society of Health-System
Pharmacists midyear clinical meeting, Orlando, FL.

Jankouskas, T. S. (2001). A continuous curriculum for building
code blue competency. Journal for Nurses in Staff Develop-
ment, 17, 195–198.

Kienle, P. C. (2006). JCAHO med management: Meeting the stan-
dards for emergency medications and labeling. Hospital Phar-
macy, 41, 888–892.

McLaughlin, R. C. (2003). Redesigning the crash cart: Usability
testing improves one facility’s medication drawers. American
Journal of Nursing, 103(4), 64A–64H.

Paas, F., Tuoviinen, J. E., Tabbers, H., & Van Gerven, P. W. (2003).
Cognitive load measurement as a means to advance cognitive
load theory. Educational Psychologist, 38, 63–71.

Peterson, R., & Berns, S. (2006). Trade confusion for order. Nurs-
ing Management, 37(9), 22–25.

Schultz, J., Davies, J. M., Caird, J., Chisholm, S., Ruggles, K., &
Puls, R. (2010). Standardizing anesthesia medication drawers
using human factors and quality assurance methods. Canadian
Journal of Anesthesia, 57, 490–499.

Smith, A., Kinross, J., Bailey, M., Aggarwal, R., Toresen, D., & Vin-
cent, C. (2008). Re-stocking the resuscitation trolley: How good
is compliance with checking procedures? Clinical Risk, 14, 4–7.

West, J., Halls, S., Coleman, R., & Lowe, C. (2008, July).
Resus:station: Breathing life back into resuscitation. Paper
presented at the Ergonomics Society: Improving Patient Safety
conference, Cambridge, UK.

Justin B. Rousek is a biomedical engineering PhD
student from the University of Nebraska–Lincoln
(UNL). He has a MS in industrial engineering from
UNL (2009). He is also currently enrolled in the
Public Health department at the University of
Nebraska Medical Center (UNMC) in Omaha, pur-
suing an MPH degree to go along with his doctoral
degree.

M. Susan Hallbeck is a professor of industrial and
management systems engineering at the University
of Nebraska–Lincoln. She received her PhD in indus-
trial engineering and operations research from
Virginia Polytechnic Institute and State University,
her MS in industrial engineering from Texas Tech
University, and her BS in industrial engineering from
Iowa State University. She is both a certified profes-
sional ergonomist and a professional engineer.

Date received: May 29, 2011
Date accepted: September 4, 2011

emailed to multiple sites or posted to a listserv without the copyright holder’s express written permission.

However, users may print, download, or email articles for individual use.

Feature Article_739 371..379

Identifying the ‘right patient’: Nurse and
consumer perspectives on verifying patient
identity during medication administration

Teresa Kelly,1,2 Cath Roper,2 Stephen Elsom2 and Cadeyrn Gaskin3
1Northern Area Mental Health Service, Melbourne Health, 2Centre for Psychiatric Nursing, The University of
Melbourne and 3The Centre for Quality and Patient Safety Research, Deakin University, Melbourne, Victoria,
Australia

ABSTRACT: Accurate verification of patient identity during medication administration is an impor-
tant component of medication administration practice. In medical and surgical inpatient settings, the
use of identification aids, such as wristbands, is common. In many psychiatric inpatient units in
Victoria, Australia, however, standardized identification aids are not used. The present paper outlines
the findings of a qualitative research project that employed focus groups to examine mental health
nurse and mental health consumer perspectives on the identification of patients during routine
medication administration in psychiatric inpatient units. The study identified a range of different
methods currently employed to verify patient identity, including technical methods, such as wristband
and photographs, and interpersonal methods, such as patient recognition. There were marked simi-
larities in the perspectives of mental health nurses and mental health consumers regarding their
opinions and preferences. Technical aids were seen as important, but not as a replacement for the
therapeutic nurse–patient encounter.

KEY WORDS: consumer perspective, medication administration, mental health, patient identifica-
tion, psychiatric inpatient unit.

INTRODUCTION

Improving patient identification in health-care settings is a
priority for international and national patient safety orga-
nizations (Australian Commission on Safety and Quality
in Health Care 2008; 2009; The Joint Commission 2011;
WHO Collaborating Centre for Patient Safety Solutions
2007). The Joint Commission (2011) and the WHO
Collaborating Centre for Patient Safety Solutions (2007)
recommend that on admission, at least two identifiers be

determined for every patient for the purpose of verifying
that patient’s identity throughout the episode of care. The
Australian Commission on Safety and Quality in Health
Care (ACSQHC) upheld the identification wristband as
an important technical aid in patient identification and
outlined specifications for a standard national identifica-
tion band for use in private and public hospitals across
Australia (Australian Commission for Safety and Quality in
Health Care 2008a; 2008b; 2008c; Australian Commission
on Safety and Quality in Health Care 2008). The specifi-
cations were endorsed by the Australian Health Ministers
that same year (Australian Commission on Safety and
Quality in Health Care 2009).

In Australian non-mental health settings, such as
medical and surgical inpatient units, the use of identi-
fication aids (e.g. wristbands) is common. In Victoria,
Australia, despite the nurse regulatory authority stating
that for the purpose of duty to care and to prevent harm,

Correspondence: Teresa Kelly, Northern Area Mental Health
Service, c/The Northern Hospital, 185 Cooper Street, Epping, Vic.
3076, Australia. Email: teresa.kelly@mh.org.au

Teresa Kelly, RN, MHN, PGradDip(AdvClinNursMH), MGest-
Ther, BHIM.

Cath Roper, BA, DipEd.
Stephen James Elsom, RN, BA, MNurs, PhD.
Cadeyrn James Gaskin, BBS(Hons), MBS, PhD.
Accepted December 2010.

International Journal of Mental Health Nursing (2011) 20, 371–379 doi: 10.1111/j.1447-0349.2010.00739.x

nurses must make sure that ‘all patients have appropriate
identification such as wristbands’ or photographs (Nurses
Board of Victoria 2007, p. 2), in most public psychiatric
inpatient units, identification aids are not used. The
ACSQHC (Australian Commission for Safety and Quality
in Health Care 2008b) stated that when patients refuse or
are unable to wear an identification band, ‘risk-assessed
alternatives’ (p. 2) should be employed. Anecdotal
reports, however, suggest that nurses working in psy-
chiatric inpatient units in Victoria employ a variety of
non-risk-assessed practices to identify patients during
medication administration.

Failure to correctly identify patients during medication
administration results in medication errors (Australian
Commission on Safety and Quality in Health Care 2008;
2009; WHO Collaborating Centre for Patient Safety Solu-
tions 2007). Medication administrations made in error
can reduce the effectiveness of treatments, cause adverse
drug reactions, lead to the deterioration of health, and
threaten the lives of consumers (Grasso et al. 2003; Haw
et al. 2005). There is an urgent need to determine effec-
tive ways of identifying patients to improve medication
safety in psychiatric inpatient units.

In keeping with the ideals of shared leadership
between health-care staff and consumers in the design of
health systems (Bate & Robert 2006), it would make sense
to seek the perspectives of both mental health consumers
and mental health nurses towards the identification of
patients in psychiatric inpatient facilities. In using such
an approach, the chances of developing solutions to the
problem of correctly identifying patients during medica-
tion administration might be maximized.

The purpose of this study was to investigate nurse and
consumer perspectives on verifying patient identity
during medication administration. We examined consum-
ers’ experiences of being identified during medication
administration, their opinions of alternative forms of iden-
tification (e.g. wristbands, photographs), and their prefer-
ences for how they would like to be identified during
medication administration. We also explored nurses’
current practices for identifying patients during medica-
tion administration, their opinions of alternative forms of
identifying patients (e.g. wristbands, photographs), and
their preferred methods of identifying patients during
medication administration.

Throughout this paper, we use the terms ‘mental health
consumer’ and ‘patient’. We use the term ‘mental health
consumer’ when we refer to people who have a lived
experience of accessing mental health care at some point in
their lives. We use the term ‘patient’ when we refer spe-
cifically to people in the context of an inpatient admission.

METHODS

Ethics approval
The study progressed following approval by The Univer-
sity of Melbourne’s Human Research Ethics Committee.

Participants
Flyers advertising our research were used to recruit con-
sumer (n = 9) and nurse (n = 13) participants to this study.
The nurse participants met the following inclusion crite-
ria: (i) registration with the Nurses Board of Victoria; (ii)
current clinical practice in psychiatric inpatient units; and
(iii) roles that included medication administration prac-
tice. The inclusion criteria for consumer participants were
that they had: (i) experienced an episode of inpatient
psychiatric care; and (ii) been the recipients of the nursing
medication administration practice in psychiatric inpa-
tient settings. Consumers who were being treated in inpa-
tient settings, those residing in community care facilities,
and those on community treatment orders were excluded
from this study, because participation during episodes of
care had the potential to cause unnecessary burden and
distress.

Design
We employed focus groups to explore the experiences,
opinions, and preferences of nurses and consumers
towards methods of correctly identifying patients during
medication administration. We used similar questions for
the consumer and nurse focus groups. The main ques-
tions for participants in the nurse focus groups were:

1. What are your current practices for identifying
patients during medication administration?

2. What are your opinions of alternative forms of identi-
fying patients during medication administration (e.g.
wristbands, photographs)?

3. What are your preferred methods of identifying
patients during medication administration?

The main questions for participants in the consumer
focus group were:

1. What were your experiences of being identified during
medication administration?

2. What are your opinions of alternative forms of identi-
fying patients during medication administration (e.g.
wristbands, photographs)?

3. What are your preferences for how you would like to
be identified during medication administration?

The focus groups were audio-taped and transcribed ver-
batim prior to content analysis.

372 T. KELLY ET AL.

Procedures
We advertised our research to nurses through the
Victorian branch of the Australian College of Mental
Health Nurses, and to consumers through the Victorian
Mental Illness Awareness Council. The study was
explained to the research participants (both verbally
and in writing), and informed consent was gained
before the focus groups began. Participants were
reminded that the researchers were interested in learn-
ing about a broad range of experiences and opinions.
Throughout the focus groups, the moderators concen-
trated on engaging participants in interactive discussion,
while at the same time, guiding conversations in a way
that encouraged all participants to contribute to the
discussions.

The focus groups were audio-taped and transcribed
verbatim. The transcripts of the focus groups were used in
the data analysis.

Analysis
We performed content analysis on the focus group tran-
scripts. Content analysis is a ‘data reduction and sense-
making effort that takes a volume of qualitative material
and attempts to identify core consistencies and meanings’
(Patton 2002, p. 453). Three of our research team (TK, a
mental health nurse; CR, a mental health consumer
researcher; CG, a social scientist) synthesized themes
from the data using an inductive framework. We then
discussed our interpretations of the data and came to a
consensus regarding the themes present in the data.
Finally, we compared and contrasted the findings to iden-
tify the similarities and differences between the perspec-
tives of nurses and consumers.

Although yielding rich, qualitative data, the discursive
and interactive nature of the focus groups did present
some challengers for the researchers. Specifically, the
voices of individual participants in the audio-recordings
were unable to be accurately matched to particular quo-
tations. Therefore, the quotations presented within this
paper are representative of several participant voices and
cannot be attributed definitively to individual focus group
participants.

RESULTS

Description of current practices
The study revealed a range of technical and interpersonal
approaches used to verify ‘right patient’ during routine
medication administration.

Technical methods
Technical methods for verifying ‘right patient’ identified
by nurses and consumers included the use of wristbands
and photographs.

Nurses reported that most patients admitted to public
psychiatric inpatient units were not routinely provided
with a wristband. Where wristbands were used, their
use was inconsistent. The exceptions to this were those
patients who were scheduled for electroconvulsive
therapy (ECT) or those who had been issued with a
wristband in another area of the hospital, such as the
emergency department (ED), prior to their transfer
to the psychiatric inpatient unit. Nurses and consumers
described the use of wristbands and/or photographs as
routine practice in psychiatric inpatient units in aged
persons’ mental health, correctional, and private hospital
settings.

Interpersonal approaches
Interpersonal approaches for verifying patient identity
described by nurses and consumers included patient rec-
ognition and knowing the patient, checking with the
patient, checking with another nurse, and conversing with
the patient.

In public adult and adolescent psychiatric inpatient
units where technical identification aids were not used,
nurses described ‘patient recognition’ as an approach
commonly used to verify patient identity. Essentially, this
approach relied on the nurse who was administering the
medication knowing the patient. Nurses explained that
knowing the patient incorporated a range of levels: the
first, simply being able to visually recognize the patient;
the second, remembering the patient in relation to their
medication regime; and the third, knowing the patient in
a deeper and more holistic way informed by the nurse–
patient relationship, the quality of rapport, and knowl-
edge of the patient’s personal narrative.

Consumers reported that staff ‘seemed to know’ who
the patients were and described being individually
sought out by nurses at medication administration times.
Some added that there were often no additional checks
to verify patient identity prior to medication being
administered.

‘Checking with the patient’ involved the nurse asking
the patient for identification details and cross-checking
the information provided by the patient with the identi-
fication information recorded on the medication chart.
Nurse participants reported, however, that in practice,
this method was flawed when nurses failed to have the
medication chart with them at the time of medication
administration.

IDENTIFYING THE ‘RIGHT PATIENT’ 373

‘Checking with another nurse’ was described as an
important part of verifying patient identity, particularly in
those psychiatric inpatient units where technical identifi-
cation aids were not used. This method involved clarifying
the patient’s identity by asking another nurse who knew
the patient.

‘Conversing with the patient’ during medication
administration was identified by nurses and consumers as
an important part of verifying patient identity. Nurses
reported that conversing with the patient incorporated a
variety of approaches, including calling the patient’s
name, using the patient’s name in conversation, and
engaging in therapeutic conversations with the patient.

Several consumers were familiar with nurses calling
out the patient’s name as the sole method used by nurses
to verify patient identity. Some described an active
involvement in verifying patient identity during the medi-
cation administration process. This involved the patient
drawing on their personal knowledge of their medication
regime and checking the medication that nurse adminis-
tered to them to be sure it was correct:

I’d just woken up when I heard my name called, and I
knew what I was taking, and I knew that all the tablets
were spot on.

Similarly, nurses reported that conversations with
patients during medication administration provided an
opportunity for patients to ‘pick up’ potential errors rel-
evant to verifying ‘right patient’.

Opinions about current practices
Opinions on technical methods
In the opinion of nurses working in private psychiatric
inpatient settings, requesting patients to wear a wristband
was appropriate, given the unit was located within the
hospital setting:

Patients are within a hospital setting . . . so it’s probably
quite appropriate to put name tags on their wrists.

Likewise, nurses working in aged persons’ psychiatric
inpatient units were of the opinion that wristbands were
appropriate in that setting. Some nurses attributed the
acceptability of wristbands to the patient’s familiarity with
wristbands from previous admissions to inpatient units in
general health-care settings.

Nurses working in aged persons’ mental health inpa-
tient units described the challenges associated with the
use of wristbands with patients who were distressed or
agitated. These nurses emphasized the importance of pri-

oritizing the patient’s needs and exercising clinical judg-
ment in decisions regarding whether or not a wristband
should be applied:

I just suppose it was about what was going on for her at
the time . . . we were trying to build up a rapport and
trusting relationship and (the wristband) wasn’t a priority
at that point in time.

The nurses’ discussion of the routine use of wristbands
for patients scheduled for ECT highlighted an apprecia-
tion of the contribution of the wristband to patient safety
in the context of ECT, and the existent contradiction in
not extending these same concerns to verification of
patient identity during medication administration:

It’s interesting though . . . we put wristbands on people
when they have ECT, and often their response is . . . ‘At
least they’ll know you have the right person’.

Consumers had mixed views regarding the use of wrist-
bands. Some supported the use of wristbands as an impor-
tant safety measure, but expected that there would always
be some people who would prefer not to wear them:

It’s inevitable with all the patients . . . that go through the
hospital that you’re going to get a percentage (of) wrist-
bands which get ripped off.

In contrast, other consumers viewed wristbands and
other technical aids as potentially dangerous instruments
supporting an impersonal and coercive system:

But the step up from those plastic wrist things is like
completely attached, a prison collar . . . you can just
imagine the process . . . the wristbands get put on, they
get taken off, and they’ll go: ‘Well, we could go a step
further here’.

Nurses working in public adult and adolescent psychi-
atric units described a culture of wristband non-use that
was informed by many nurses’ beliefs that in psychiatric
inpatient units, the nurses know who the patients are. The
strength of the cultural influence upon the non-use of
wristbands in public psychiatric inpatient units was cap-
tured in one nurse’s telling of a conversation between
herself as a nurse new to a psychiatric inpatient unit and
another nurse who had worked in the unit for some time:

I’ve even been told when . . . I’m new to the ward . . . that
they don’t want it to appear like a hospital; they don’t want
people to feel institutionalized. . . . They would prefer not
to have name bands on everybody. . . . I haven’t asked
anybody in a senior position, but that was told to me by
another nurse when I asked her. . . . so it’s not meant to

374 T. KELLY ET AL.

feel like a surgical ward . . . they want to make it more like
a home environment, that sort of thing.

A belief that wristbands contributed to stigma and
patient distress was another characteristic of the culture
where wristbands were not used:

Anyone who . . . was perhaps suffering paranoid schizo-
phrenia is unlikely to want . . . anything which they think
. . . there might be a chip in it or something . . . that
would just add to their distress, suspiciousness.

Nurses described an assumption held by many nurses
that patients do not want to wear wristbands and that
some patients would actively remove the bands. In con-
trast, other nurses felt that often the ‘patients don’t mind
them, (and) most of the time, it’s the nurses who
don’t . . . put them on’.

One consumer, experienced in working with nurses in
the capacity of consumer consultant, reported negative
attitudes expressed by some nurses about consumers in
relation to wristbands and the impact this attitude had on
progressing medication safety initiatives in one mental
health service:

Also the bad attitudes . . . previously as a consumer con-
sultant, it came up . . . ‘Oh, we should introduce wrist-
bands’. I said, ‘Yes, yes, that’s terrific, I have no problems
with wristbands’, you know, as a consumer consultant, but
then it got stopped in one of the committees, because ‘Oh
no, they’ll just rip them off ’ . . . I wouldn’t rip mine
off . . . just the attitude; like I was for it because I read
about medication safety issues, but they just, it was this
attitude in . . . the senior staff . . . ‘No, no, the patients
will just tear them off’. Well, it’s so hard to tear them off,
and I don’t think most people would . . . sort of pre-
empting what we might do based on prejudice.

Nurses working in public acute psychiatric inpatient
units explained that patients who were admitted to the
unit through the ED were often issued with a wristband in
the ED, but that the wristband was often removed by
nurses upon admission to the psychiatric inpatient unit:

My experience is . . . patients only come in with wrist-
bands if they’ve come through ED. . . . If not, there
doesn’t seem to be a policy of putting them on. I’ve even
seen nurses cut them off, saying, ‘Oh look, that’d be
uncomfortable, let’s cut that off ’.

A number of nurses identified the use of wristbands as
a ‘standard of quality’ and advocated for the use of wrist-
bands in psychiatric inpatient units as medication safety
strategy:

I think I’d be most comfortable with wristbands . . . be-
cause that’s what they have in general, and . . . there’s this
standard of quality and of making sure you’ve got the right
patient in general, and that is the standard, the wristband.

Some nurses working in public psychiatric inpatient
units were concerned about the lack of formal processes
for verifying patient identity and the impact of this on
patient safety:

Just the way I often see (nurses) just handing medications
to patients saying, ‘Here’s your medication’. . . . There’s
no verification, and I’ve seen a few near misses happen.

The absence of technical methods for verifying patient
identity contributed to some nurses feeling anxious and
fearful of making mistakes, particularly when working in
new environments:

Giving medication . . . can be quite a nerve-wracking
experience, making sure that I’m on the . . . five rights
. . . identifying a patient is so important . . . and I feel
really nervous sometimes, because the background (I)
come from is . . . medical, where everybody wears a wrist-
band, so it’s quite easy to be sure that you’re giving the
medication to the right person.

Nurses held mixed opinions on the use of patient pho-
tographs. They identified a number of challenges associ-
ated with the use of patient photographs in practice.
These included practical problems associated with taking
the photograph, challenges associated with keeping the
photographs current, the importance of locating the
patient’s photograph with the patient’s medication chart,
and concern that taking the patient’s photograph is an
invasion of the patient’s privacy.

Some consumers experienced in the use of photo-
graphs accepted the taking of the photograph as a neces-
sary part of the information collection associated with
admission to a mental health service:

I found it no trouble at all. . . . By the time I sort of went
into the . . . private system, I was fairly well prepared to
have my photo taken and . . . yeah, well it didn’t occur to
me . . . to be a great intrusion.

Others, however, felt that photographs were an inva-
sion of the patient’s privacy and expressed concerns about
how the photograph might be used in the future:

I just thought forever more that I’d have this photograph
lingering around this space . . . I found it an invasion.

Opinions on interpersonal methods
Nurses differed in their opinions regarding ‘patient rec-
ognition’ as a method for verifying patient identity.

IDENTIFYING THE ‘RIGHT PATIENT’ 375

Nurses who supported this method claimed that remem-
bering patients’ names and faces was not a difficult task.
Others challenged ‘patient recognition’ as a safe method
in inpatient units with a high patient turnover, or when
nurses who were unfamiliar with the patients were allo-
cated to the role of medication administration. These
nurses argued that the reliance on this method in the
absence of other definitive methods of verification of
patient identity compromised patient safety:

I think any type of identification that doesn’t positively
identify the patient is a risk. . . . There is a margin for
error, and then it’s about whether you feel comfortable
with that margin of error or not.

One nurse emphasized the importance of nurses allo-
cated to medication administration roles being familiar
with the patients and with the unit:

If I’ve had the chance to chat with the person and . . . am
able to recall who they are, then that’s generally enough
for me. . . . If I haven’t had that chance . . . I’ll actually ask
not to give the medication. If I’m in a situation where I’m
running the shift and I’m allocating the medication, then
I won’t give it to somebody who doesn’t know the patients
well enough.

Consumers appreciated verification of patient identity
as an important part of medication safety. Most consum-
ers argued that all methods of verifying patient identity
should be underpinned by the nurse’s personal knowl-
edge of the patients:

It shouldn’t become as impersonal as it is. . . . Like may-
be . . . when staff cross paths with a patient on the ward, if
they said . . . ‘Hello James’, you know, acknowledge
you . . . there would be a rapport, and that would build a
familiarity.

Nurses claimed that ‘checking with another nurse’ was
an important part of verifying patient identity when the
nurse administering medication was less familiar with
the patient, the patient had recently been admitted, the
patient was unwilling or unable to provide identification
details, the nurse was in doubt of the patient’s identity, or
when the nurse suspected the patient had provided incor-
rect identification details. Nurse participants, however,
did acknowledge that workplace contextual issues might
prevent nurses from seeking assistance from other nurses:

I think (nurses) are less likely to go find another nurse to
verify the patient, and are more likely to . . . hope that it’s
the right person . . . because there’s a lot of pressure to

get the job done quickly . . . you want to go home, the rest
of the shift wants to go home, the patients want their
meds.

Nurses expressed mixed views about asking patients
for identification details. Some thought that most patients
would cooperate with such requests:

Usually, most consumers are really good if you explain,
‘Look, this is just a routine legal thing I need to do. I know
that I know who you are, but I need to ask you your name
and date of birth’, and everybody’s usually fine and coop-
erative with that.

Other nurses were not convinced that all patients
would be so cooperative:

I’ve done quite a lot of shifts in places where I haven’t
worked before, picking up the occasional shift in a new
environment and having to give medication to people that
(I’ve) never met before . . . and there’s no way to identify
who they are (and) people that don’t want to tell you who
they are.

Some consumers were familiar with being asked their
date of birth and did not experience this request as an
intrusion. One consumer, however, raised concern about
whether asking patients to confirm their date of birth was
a reasonable method for verifying patient identity:

If you’re new on the ward, they’ll ask you for a date of
birth, just to confirm it with their drug sheet. . . . I mean
of course, you still could be anyone, and someone
could’ve overheard . . . a date of birth previously and
doubled . . . you know.

Nurses working in units where technical aids were not
used held strong views that conversations with the patient
were a good way to double-check that they were admin-
istering medication to the right patient:

If . . . I’m not familiar with the person, I might ask them
something about the medication . . . ‘How long have you
been on this one for?’ If they know . . . then generally
their recognizing their medication gives me an indication
I’ve got the right person. . . . I’ve seen (nurses) who’ve
realized they’ve identified the wrong person because that
person’s gone, ‘Oh no, I’m not on this medication’, and
that’s where it’s stopped, and that’s where they’ve realized
they’ve got the wrong person.

Opinions regarding organizational systems and processes
Several nurses stressed the importance of policy or guide-
lines that outline the expected method for identifying

376 T. KELLY ET AL.

patients, and expressed concern that the absence of such
frameworks exposed nurses to the risk of making
mistakes.

Some nurses had experience of primary nursing, where
each nurse was responsible for all aspects of care for a
small group of patients, including medication manage-
ment. These nurses viewed primary nursing as superior to
the common practice in public psychiatric inpatient units
in Victoria, of allocating a single nurse to the task of
administering medication to all patients in the unit, which
according to one nurse, compromised medication safety:

You can have people lined up waiting for medication
. . . standing up at the window, and a couple of people
talking with you, and you know, was it the person on the
left which was supposed to get the medication or the
person on the right? Or people that kind of look similar
and a lot of people in the one area . . . a lot of people
talking to you at the one time. I’ve definitely seen circum-
stances where the medication’s just been handed over to
the wrong person.

Consumers also challenged the appropriateness of the
single medication nurse approach and advocated for
nurses to administer medication to small groups of
patients to promote a more therapeutic and interpersonal
exchange where the nurse knows the patient to whom he
or she administers medication.

Suggestions for best practice
Some nurses held the firm view that nurses should know
the patients to whom they were administering medica-
tion, and that primary nursing models were the best way
to achieve this. Others emphasized that the use of iden-
tification aids, such as wristbands or photographs, should
be routine practice, and that this should be supported by
clear policies and guidelines.

A number of nurses proposed that nurses engage
patients as partners in medication safety, and more spe-
cifically, educate them on the importance of verification
of patient identity during medication administration, and
offer them choice with regard to method of verification.
Some nurses related positive experiences of using indi-
vidualized medication units, such as ‘dosette boxes’ with
photographs attached. They also noted the limitations of
such systems in high-turnover settings, such as acute inpa-
tient units.

Consumers made a number of suggestions to improve
the patient verification process. These included a digital
photograph printed on the patient’s medication chart,
wristbands with a small patient photographs attached, and
a computerized swipe card containing patient identifica-

tion details. Some consumers advocated for the use of
technical aids, such as wristbands and photographs, as
technical backups for the nurses:

You’ve got to have something, a backup as well, so their
photo and wristband, as well as asking a few questions.
You know . . . a couple of different things . . . backup, I
suppose.

Other consumers, however, were concerned that pro-
moting technical aids had the potential to further com-
promise person-centred care:

I just think these impersonal aids and this scanning of the
patient is dangerously impersonal . . . if you initiate
change in that direction, it’s not going to be as pleasant for
the patient . . . on a friendly or on an open sort of, you
know, emotional level with staff, because by doing that
(the nurses will) defer their contact with you and just, you
know, sit in the office. . . .

Consumers agreed that nurses administering medication
to smaller groups of patients would be preferable to the
single medication nurse system commonly used in psychi-
atric inpatient units in Victoria. Like the nurses, consum-
ers cognisant with ‘primary nurse’ and ‘contact nurse’
systems suggested the application of such approaches
to improving patient verification during medication
administration:

A good way to do it would be to have the nurse dispense
the medication for his group, for his five or six (patients),
and then the next nurse would dispense for his group, and
so on.

One consumer advocated for a nurse–patient induc-
tion upon admission, thereby providing an opportunity for
the nurse to develop a comprehensive and holistic knowl-
edge of the patient:

The best thing would be for a friendly induction . . . indi-
vidually or in a group. . . . (The nurses) sit down with you,
they say ‘hello’, you know; full names, they shake your
hand, and wish you well and, you know, that recognition
would last your admission.

DISCUSSION

In this study, nurse participants described the routine use
of wristbands or photographs as technical identification
aids during medication administration in public aged
persons’ mental health, correctional, and private inpatient
psychiatric units. Nurses working in public adult and ado-
lescent psychiatric inpatient units, however, reported that
wristbands were not routinely employed to verify patient

IDENTIFYING THE ‘RIGHT PATIENT’ 377

identity during medication administration. Furthermore,
when wristbands were used, their use was inconsistent,
erratic, and often at odds with a workplace culture that
did not support the use of technical aids.

When the ACSQHC endorsed the identification wrist-
band for use in private and public hospitals across Aus-
tralia, they identified the mental health patient population
as an exception (Australian Commission for Safety and
Quality in Health Care 2008b). Interestingly, in the
present study, consumers emphasized the importance of
not excluding people with mental health problems from
medication safety initiatives by pre-empting what they
might or might not agree to, based on prejudice and
stereotypes of people with mental illness.

Consistent with the findings of this study, contempo-
rary quality and safety health-care literature emphasizes
the importance of systems, processes, and cultures that
promote accurate patient identification in all health-care
settings (Australian Commission on Safety and Quality
in Health Care 2008; WHO Collaborating Centre for
Patient Safety Solutions 2007).

This study revealed that across public adult and ado-
lescent psychiatric inpatient units in Victoria, there is a
nursing culture of wristband non-use that is embedded in
tradition and is informed by a belief that, in psychiatric
inpatient units, the nurses know who the patients are. In
contrast, the WHO Collaborating Centre for Patient
Safety Solutions (2007) maintains that patient identifica-
tion must be verified for each patient on each and every
care administration occasion, even when the health-care
worker feels familiar with the patient. Furthermore, they
assert the importance of organizational systems and
processes that emphasize that the health-care provider
administering the care holds primary responsibility for
verifying a patient’s identity (WHO Collaborating Centre
for Patient Safety Solutions 2007).

Consistent with earlier studies (Duxbury et al. 2010;
Haglund et al. 2004), the researchers in this study found
that in mental health inpatient settings, nurses knowing
patients and patients knowing nurses is an important
part of the medication administration process for nurses
and consumers. The findings of this study indicate that
in many psychiatric inpatient units in Victoria, the task
of medication administration is allocated to a single
medication-administration nurse. Nurse and consumer
participants however, advocated for more patient-centred
approaches that support nurses administering medication
to smaller groups of patients. Furthermore, the nurse and
consumer participants suggested that all methods of veri-
fying patient identity should be underpinned by patient-
centred practices that incorporate engaging with patients

in a holistic way, informed by the nurse–patient relation-
ship, rapport, and knowledge of the patient’s mental,
physical, and social health history. The real-world experi-
ences of consumer participants, however, suggested that
this was not always the case.

Engaging patients as partners in medication safety pro-
vides opportunity for patients to take an active role in
patient identification and in reducing medication errors
(Australian Commission on Safety and Quality in Health
Care 2008; Institute of Medicine 2000; Walrath & Rose
2008; WHO Collaborating Centre for Patient Safety
Solutions 2007). In this study, nurse and consumer par-
ticipants identified nurse–patient interpersonal inter-
actions that invite the patients to check their medication
as an important medication error-reduction strategy.

LIMITATIONS

This modest study was limited in scope, and as such, its
findings should be interpreted with caution. The extent to
which the views expressed by the nurses and consumers in
this study are representative of those of other nurses and
consumers is unknown. Larger-scale research is needed
to ensure that any changes to practice are feasible and
acceptable to the nurses and consumers affected by such
changes.

CONCLUSION

This study highlights important implications and raises a
number of challenges for nurses. Challenging cultures
that do not value standardized methods of verifying
patient identity is fundamental to nurses progressing
medication-safety agendas in mental health settings.
Further, nurses must take lead roles in progressing sys-
temic, systematic, and multifaceted practice improve-
ment initiatives specifically relevant to accurate
verification of patient identity during medication admin-
istration. At an organizational level, policies and guide-
lines must clearly outline the nurse’s legal, ethical, and
organizational responsibilities (WHO Collaborating
Centre for Patient Safety Solutions 2007). At a patient
care level, nurses, patients, and carers need to be
informed of the importance of accurate patient identifi-
cation to medication safety. Practices that engage patients
as active partners in medication administration, particu-
larly in regard to verifying patient identity, should be
promoted (Australian Commission on Safety and Quality
in Health Care 2008; Institute of Medicine 2000; Walrath
& Rose 2008; WHO Collaborating Centre for Patient
Safety Solutions 2007).

378 T. KELLY ET AL.

Educating patients and their carers (WHO Collaborat-
ing Centre for Patient Safety Solutions 2007) about the
importance of accurate patient identification, explaining
policies, and offering patients practical and realistic
choices about how they would prefer to be identified
during the medication administration is a vital component
of progressing patient-centred, safety focused, and prag-
matic patient-identification practices.

ACKNOWLEDGEMENTS

The authors thank the Victorian Mental Illness Awareness
Council and the Australian College of Mental Health
Nurses (ACMHN), Victorian branch, for assistance in
recruiting participants, and Finbar Hopkins for her
helpful comments on an earlier draft. This study was
supported by the 2007 ACMHN and Bristol Myers
Squibb Research Grant.

REFERENCES
Australian Commission for Safety and Quality in Health Care

(2008a). Fact sheet: Specifications for a standard patient
identification band. [Cited 29 Mar 2010]. Available from:
URL: http://www.health.gov.au/internet/safety/publishing.
nsf/Content/EAC2DBC0F54777B5CA2574DE00111B73/
$File/FactSheet-PatID-Band

Australian Commission for Safety and Quality in Health Care
(2008b). FAQ: Specification for a standard patient identifi-
cation band. [Cited 29 Mar 2010]. Available from: URL:
http://www.health.gov.au/internet/safety/publishing.nsf/
Content/EAC2DBC0F54777B5CA2574DE00111B73/
$File/FAQ-PatID-Band

Australian Commission for Safety and Quality in Health Care
(2008c). Specifications: Specification for a standard patient
identification band. [Cited 29 Mar 2010]. Available from:
URL: http://www.safetyandquality.gov.au/internet/safety/
publishing.nsf/Content/EAC2DBC0F54777B5CA2574DE
00111B73/$File/Specs-PatID-Band

Australian Commission on Safety and Quality in Health Care
(2008). Windows into Safety and Quality in Health Care
2008. Sydney: Australian Commission on Safety and Quality
in Health Care.

Australian Commission on Safety and Quality in Health Care
(2009). Windows into Safety and Quality in Health Care
2009. Sydney: Australian Commission on Safety and Quality
in Health Care.

Bate, P. & Robert, G. (2006). Experience-based design: From
redesigning the system around the patient to co-designing
services with the patient. Quality and Safety in Health Care,
15, 307–310.

Duxbury, J. A., Wright, K., Bradley, D. & Barnes, P. (2010).
Administration of medication in the acute mental health
ward: Perspectives of nurses and patients. International
Journal of Mental Health Nursing, 19, 53–61.

Grasso, B. C., Rothschild, J. M., Genest, R. & Bates, D. W.
(2003). What do we know about medication errors in inpa-
tient psychiatry? Joint Commission Journal on Quality and
Safety, 29, 391–400.

Haglund, K., von Essen, L., von Knorring, L. & Sidenvall, B.
(2004). Medication administration in inpatient psychiatric
care: Get control and leave control. Journal of Psychiatric
and Mental Health Nursing, 11, 229–234.

Haw, C. M., Dickens, G. & Stubbs, J. (2005). A review of
medication administration errors reported in a large psychi-
atric hospital in the United Kingdom. Psychiatric Services,
56, 1610–1613.

Institute of Medicine (2000). To Err is Human: Building a Safer
Health System. Washington DC: National Academy Press.

Nurses Board of Victoria (2007). Rising complaints of poor
documentation concerns Board. Nexus, 14 (1), 2–3.

Patton, M. Q. (2002). Qualitative Research and Evaluation
Methods, 3rd edn. Thousand Oaks, CA: Sage Publications.

The Joint Commission (2011). National patient safety goals.
[Cited 14 Jan 2011]. Available from: URL: http://www.
jointcommission.org/assets/1/6/2011_NPSGs_AHC

Walrath, J. M. & Rose, L. E. (2008). The medication adminis-
tration process: Patients’ perspectives. Journal of Nursing
Care Quality, 23, 345–352.

WHO Collaborating Centre for Patient Safety Solutions
(2007). Patient identification. Patient safety solutions.
[Cited 29 Mar 2010]. Available from: URL: http://www.
ccforpatientsafety.org/common/pdfs/fpdf/presskit/PS-
Solution2

IDENTIFYING THE ‘RIGHT PATIENT’ 379

may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder’s express

written permission. However, users may print, download, or email articles for individual use.

Practice rePorts Direct refill program

1659Am J Health-Syst Pharm—Vol 69 Oct 1, 2012

Effects of a direct refill program for automated
dispensing cabinets on medication-refill errors

Pieter J. Helmons, AsHley J. DAlton, AnD CHArles e. DAniels

Purpose. The effects of a direct refill pro-
gram for automated dispensing cabinets
(ADCs) on medication-refill errors were
studied.
Methods. This study was conducted in
designated acute care areas of a 386-bed
academic medical center. A wholesaler-
to-ADC direct refill program, consisting
of prepackaged delivery of medications
and bar-code-assisted ADC refilling, was
implemented in the inpatient pharmacy
of the medical center in September 2009.
Medication-refill errors in 26 ADCs from the
general medicine units, the infant special
care unit, the surgical and burn intensive
care units, and intermediate units were
assessed before and after the implemen-
tation of this program. Medication-refill
errors were defined as an ADC pocket con-
taining the wrong drug, wrong strength, or
wrong dosage form.
Results. ADC refill errors decreased by
77%, from 62 errors per 6829 refilled

Pieter J. Helmons, PHarm.D., M.A.S., is Hospital Pharmacist,
Department of Pharmacy, St. Jansdal Hospital, Harderwijk, The
Netherlands; at the time of writing he was Pharmacist Specialist—
Pharmacoeconomics, University of California San Diego (UCSD)
Health System, San Diego. asHley J. Dalton, PHarm.D., is Inpa-
tient Pharmacy Manager, UCSD Health System La Jolla, San Diego.
CHarles e. Daniels, PH.D., B.s.PHarm., is Professor of Clinical
Pharmacy and Associate Dean for Clinical Affairs, Skaggs School of
Pharmacy and Pharmaceutical Sciences, University of California,
La Jolla.

Address correspondence to Dr. Helmons at the Department of
Pharmacy, St. Jansdal Hospital, Postbus 138, 3840 AC Harderwijk,
The Netherlands (pieter.helmons@gmail.com).

Supported by an unrestricted Cardinal Health Patient Safety
Grant.

The authors have declared no potential conflicts of interest.

cists, Inc. All rights reserved. 1079-2082/12/1001-1659$06.00.
DOI 10.2146/ajhp110503

pockets (0.91%) to 8 errors per 3855 re-
filled pockets (0.21%) (p < 0.0001). The predominant error type detected before the intervention was the incorrect medi- cation (wrong drug, wrong strength, or wrong dosage form) in the ADC pocket. Of the 54 incorrect medications found before the intervention, 38 (70%) were loaded in a multiple-drug drawer. After the imple- mentation of the new refill process, 3 of the 5 incorrect medications were loaded in a multiple-drug drawer. There were 3 instances of expired medications before and only 1 expired medication after imple- mentation of the program. Conclusion. A redesign of the ADC refill proc

ess using a wholesaler-to-ADC direct
refill program that included delivery

of
prepackaged medication and bar-code-
assisted refill significantly decreased the
occurrence of ADC refill errors.
Am J Health-Syst Pharm. 2012; 69:1659-
64

M
ost acute care hospitals in the
United States use automated
dispensing cabinets (ADCs)

as the core of their medication dis-
tribution system. In 2008, an ASHP
survey of 527 hospitals found that
82.9% used ADCs.1 If hospitals with
fewer than 100 staffed beds are ex-
cluded, the percentage of hospitals
using ADCs increases to 95–98.7%.1
ADCs offer a variety of benefits to
the organization and the user, such
as secure and timely access to the
most commonly used medications
in a specific patient care area and
more accurate tracking and capture
of charge data for the medications
used.

However, the impact of ADCs on
medication safety is less well defined,
and several reports have indicated
that the incorrect use or poor design
of ADCs results in medication er-
rors.2,3 ADCs have been the source of
almost 15% of all medication-error
reports received by the Pennsylvania
Patient Safety Reporting System since
its inception in 2004.3 In addition,
123 ADC-related medication errors
have been reported to the National
Medication Errors Reporting Pro-

gram, operated by the Institute for
Safe Medication Practices (ISMP),
since 1971.3

In 2008, ISMP identified 12 core
processes to ensure the safe use of
ADCs4:

1. Provide ideal environmental condi-
tions for the use of ADCs,

2. Ensure ADC system security,
3. Use pharmacy-profiled ADCs,
4. Identify information that should

appear on the ADC screen,

Practice rePorts Direct refill program

1660 Am J Health-Syst Pharm—Vol 69 Oct 1, 2012

5. Select and maintain proper ADC
inventory,

6. Select appropriate ADC
configuration,

7. Define safe ADC restocking
processes,

8. Develop procedures to ensure the
accurate withdrawal of medications
from the ADC,

9. Establish criteria for ADC system
overrides,

10. Standardize processes for transport-
ing medications from the ADC to
the patient’s bedside,

11. Eliminate the process for returning
medications directly to their origi-
nal ADC location, and

12. Provide staff education and compe-
tency validation.

At the beginning of this study, the
processes used at the University of
California San Diego (UCSD) Health
System were aligned with some of
these core processes. At this institu-
tion, almost all ADCs in the acute
care setting required pharmacist
review and approval before dispens-
ing medications from an ADC and
subsequent administration to the pa-
tient (core process #3). UCSD Health
System also predominantly used
single-drug pockets. These pockets
contain only one specific medication
(core process #6), thereby decreasing
the opportunity for fill errors. In ad-
dition, standard safeguards were in
place to ensure appropriate stocking
of the ADC (core process #7), such as
mandatory checks of any drug prod-
uct to be refilled before it leaves the
pharmacy and an additional phar-
macist check after refilling the prod-
uct in the ADC. Despite these efforts,
ADC refill errors continued to occur.

A prospective before-and-after
study was conducted to determine
the impact of a new ADC refill proc-
ess on medication-refill errors.

Methods
Background. This study was con-

ducted in designated acute care areas
of UCSD Health System, a 386-bed

academic medical center. A total of
2 7 A D C s ( P y x i s Me d S t a t i o n ,
CareFusion, San Diego, CA) from
the general medicine units, the in-
fant special care unit, the surgical
and burn intensive care units, and
intermediate units were included
in this study. These areas predomi-
nantly rely on ADCs for medication
distribution, with more than 90%
of medications billed to the patient
originating in these ADCs.

The typical configuration of the
ADC in the acute care areas com-
prises a cabinet containing predomi-
nantly single-drug pockets and some
multiple-drug pocket drawers, a
refrigerator unit, and an ADC tower
containing bins to store large items
such as large-volume i.v. bags.

At the time of the study, orders
were entered into the computerized
prescriber-order-entry system, which
was interfaced with the pharmacy in-
formation system. Medications could
be administered only after the or-
ders were reviewed by a pharmacist.
Because the pharmacy information
system was interfaced with the ADC,
nurses could view only the medica-
tions on the ADC that had been veri-
fied by a pharmacist.

Pharmacy technicians manu-
ally restocked ADCs twice daily
(morning and evening) by manually
retrieving (“picking”) the medica-
tions to be restocked from the phar-
macy inventory. Pharmacists visually
checked the contents of the retrieved
medications before the products left
the pharmacy and again after the
pockets were restocked. However, the
time period between the technician’s
refilling of the ADC and the second
pharmacist check was variable, de-
pending on the availability of the
pharmacist to perform the double
check. As a result, the ADC restock-
ing process was suboptimal. Manual
retrieval of medications from phar-
macy inventory is time-consuming
and allows for human error. In ad-
dition, the lag between ADC refill
and the pharmacist’s check of the

ADC is a potential vulnerability, as
unchecked (and potentially incor-
rect) medications remain available
for retrieval.

Intervention. In September 2009,
the inpatient pharmacy implemented
a wholesaler-to-ADC direct refill
program. Only unit-of-use packaged
medications are available through
this program. Figure 1 illustrates the
ADC refill process before and after
implementation of this program. The
wholesaler-to-ADC refill program is
offered to hospital pharmacies at an
additional charge. In the redesigned
process, pharmacy technicians no
longer have to manually select most
of the ADC refill orders from the
central pharmacy supply, and the
pharmacist no longer has to check
the selected products before refilling
the ADC. In addition, the software
from the wholesaler-to-ADC direct
refill program automatically creates a
recommendation when the inventory
of a pocket containing medication in
the program falls below the prespeci-
fied level. The wholesaler prepack-
ages and delivers medications in an
ADC pocket-specific bag containing
sufficient medication to fit the pocket
and a bar code with the identity
of the contents. When refilling the
ADC, the pharmacy technician scans
the bar code on the ADC pocket-
specific bag, and the corresponding
pocket automatically opens. This
eliminates the error-prone step of
manually browsing for the product
from an alphabetized list in the ADC.
The double check of the identity and
condition of the refilled medication
at the ADC by a pharmacist is still
required.

Statistical analysis. The pharma-
cists performing the ADC refill checks
collected data on medication-refill
errors before and after implementa-
tion of the new program. However,
data collection for this study was
voluntary. Medication-refill errors
were defined as an ADC pocket
containing the wrong medication,
wrong strength, or wrong dosage

Practice rePorts Direct refill program

1661Am J Health-Syst Pharm—Vol 69 Oct 1, 2012

Figure 1. Redesign of the automated dispensing cabinet (ADC) refill process before and
after implementation of a wholesaler-to-ADC direct refill program.

6. Technician

Manually selects the
drug record of the

item to be refilled and
pocket opens

Before Intervention

1. Technician

Prints ADC refill list

2. Technician

Manually picks items

3. Technician

Sorts items per ADC
for delivery

4. Pharmacist

Checks items before
ADC refill

5. Technician

Goes to the ADC to be
refilled

7. Technician

Refills the pocket with
the picked medication

8. Pharmacist

Checks if the right
drug is filled in the
appropriate ADC

location

After Intervention

1. ADC

Automatically places
order if ADC pocket
falls below par level

4. Technician

Goes to the ADC to be
refilled
5. Technician

Scans the bar code
on the prepackaged
bag containing the
medication to be

refilled. Pocket opens
automatically.

6. Technician
Refills the pocket with
the picked medication

7. Pharmacist

Checks if the right
drug is filled in the
appropriate ADC
location

2. Wholesaler

Picks, checks,
and delivers items
prepackaged per

individual ADC pocket
and containing product

specific bar code

3. Wholesaler

Sorts items per ADC
for delivery

form. A check of the expiration date
of the medications was included, as
the prepackaging step by the whole-
saler could result in the acquisi-
tion of shorter dated medications.
After each ADC refill check, the
pharmacist filled out a data collec-
tion form capturing the date, ADC
location, duration of the ADC refill
check, and details of any fill errors
(Figure 2). Electronic reports from
the ADC were used to capture the
number of pockets checked by each
pharmacist. Lastly, we used elec-
tronic reports to document the type
of pockets associated with an error.
We based our sample-size calcula-
tion on a previous study by Klibanov
and Eckel5 in a similar-sized hospital
that used a similar ADC system and
refill process. Of the 2858 pockets
inspected, this study found a misfill
rate of 2.3%. Based on a baseline
misfill rate of 2.3% and a power of
80%, we calculated that 6600 pock-
ets would need to be inspected to
detect a misfill error reduction of
30%. An interim analysis during the
postimplementation period showed
an error reduction of more than
70%, larger than was expected. It
was then decided that sufficient data
had been collected for the study to
be adequately powered. Data collec-
tion postimplementation was sub-
sequently halted after 3855 refilled
pockets had been checked.

Data were entered into spread-
sheets (Microsoft Excel, Redmond,
WA) for initial analysis and sum-
mary statistics. Stata 10 (StataCorp
LP, College Station, TX) was used for
the power calculation and additional
statistical tests. Chi-square analysis
was used to compare error rates be-
fore and after the intervention. Con-
tinuous data were analyzed using the
unpaired t test. The a priori level of
significance was set at 0.05.

Results
Totals of 6829 pockets in 26 ADCs

and 3855 pockets in 24 ADCs were
inventoried 5 months before and 18

months after implementation of the
new program, respectively. Since we
relied on voluntary data collection by
the pharmacists assigned to the unit
during a fixed data collection period,
refill data during the preimplemen-
tation and postimplementation pe-

riods were not collected from 1 and
3 ADCs, respectively. Data collected
during both periods were mostly
similar (Table 1), except that medica-
tions were more frequently stored in
a single-drug pocket during the post-
implementation period (73% versus

Practice rePorts Direct refill program

1662 Am J Health-Syst Pharm—Vol 69 Oct 1, 2012

51%, p < 0.0001). ADC refill errors decreased by 77%, from 62 errors per 6829 refilled pockets (0.91%) to 8 er- rors per 3855 refilled pockets (0.21%) (p < 0.0001). The predominant error type detected before the intervention was the incorrect medication (wrong drug, strength, or dosage form) in the ADC pocket (Table 2). Of the 54 in- correct medications found before the intervention, 38 (70%) were loaded in a multiple-drug drawer.

After the implementation of the
new refill process, 3 of the 5 incor-
rect medications were loaded in a
multiple-drug drawer. There were

Figure 2. Data collection form completed by the pharmacist (RPh) while checking the refilled medications in the automated dispensing
cabinet (ADC).

Error details (Medication Involved in Error—Name,
Strength, Form)

A
D

C
E

rr
o

r
D

a
ta

C
o

ll
e
c
ti

o
n

F
o

—

P
le

a
s
e
r

e
tu

rn
t

o
J

o
h

n
D

o
e
o

r
J
a
n

e
D

o
e
w

e
n

c
o

m
p

le
te

Date:

Pyxis Machine Name:

Start time:

Stop time:

RPh:

How many
times were
you
interrupted?Error type

W
ro

d
ru

g
W
ro

n
g

s
tr

e
n

g
th

W
ro
n
g

f
o

E
x
p

ir
e
d

m
e
d

3 instances of expired medications
before and only 1 expired medica-
tion after implementation of the
program.

Discussion
ADC refill errors decreased by

77% after implementation of a
wholesaler-to-ADC direct refill pro-
gram without increasing the fre-
quency of expired medication. How-
ever, these results should be viewed
in light of the study’s limitations.
First, this study required extensive
data collection, because medication-
refill errors are rare. Twenty-nine

pharmacists collected data during
the preimplementation period, com-
pared with 16 pharmacists during the
postimplementation period. Eleven
pharmacists collected data during
both time periods. Data collection
by different pharmacists could have
led to differences in the consistency
of the data collected. However, the
electronic reports used to capture
ADC refill data are identical for every
ADC. This should result in minor
variance only and would not account
for the large decrease in ADC refill
errors. In addition, the baseline ADC
refill error rate in this study (0.91%)

Practice rePorts Direct refill program

1663Am J Health-Syst Pharm—Vol 69 Oct 1, 2012

is similar to the rate reported by
Klibanov and Eckel5 (2.3%), further
strengthening the validity of this
study’s results.

Second, a separate effort to de-
crease ADC refill errors was focused
on increasing the use of single-drug
pockets when storing medications
in ADCs. As a result of these efforts,
medications were more frequently
stored in single-drug pockets in the
postimplementation period. Scan-
ning the wholesaler prepackaged
medication bar code at the ADC
automatically opens the correct
single-drug pocket, making it al-
most impossible to refill the incor-
rect pocket. Multiple-drug pocket
drawers, however, are more prone
to errors, as these pockets do not
contain a lid. The new process re-
quires an additional scan of the bar
code in the specified pocket inside
the multiple-drug pocket drawer as
an added safety feature. It is pos-
sible to misplace a medication in the
compartment without performing a
second scan. During the time of the
study, it was not possible to measure
scanning compliance when refilling
the ADC, which would have quanti-
fied this limitation. However, it is
unlikely that this potential “work-
around” influenced the results of
this study: omitting the second scan
requires the user to cancel the entire
bar-code-assisted refill process and
resume the refill using a much more
labor-intensive manual process.

Third, the redesigned ADC refill
process eliminated two error-prone
steps: (1) medications are no longer
manually collected by the pharmacy
technician in the inpatient phar-
macy but are delivered to the ADC
prepackaged per pocket and (2)
pharmacy technicians no longer have
to browse through an alphabetized
list on the screen of an ADC for the
appropriate pocket. Scanning the bar
code on the prepackaged bag auto-
matically opens the appropriate ADC
pocket. Both error-prone steps were
eliminated at the same time; there-

Table 1.
Comparison of Data Collected Before and After Implementation
of the Wholesaler-to-ADC Direct Refill Programa

Variable
Before

Implementation

After

Implementation

No. pockets checked
Type of pocket, no. (%)
Single-drug pocketb

Multiple-drug drawers
No. (%) ADCsc

Median no. (range) pockets per ADC
Median no. (range) pockets per

medication check
Median duration (range) of medication

check, min

6829

3500 (51)
3329 (49)

26 (96)
169 (3–773)

6 (0–47)

3 (0–23)

3855

2821 (73)
1034 (27)

24 (89)
109 (1–537)

6 (0–50)

2 (0–38)
aUnless otherwise stated, differences were not significant. ADC = automated dispensing cabinet.
bp < 0.0001. cSince data collection was voluntary, not all 27 ADCs were represented in the analysis.

Table 2.
Comparison of Error Rates Before and After Implementation of
Wholesaler-to-ADC Direct Refill Programa

Error
Before

Implementation
After

Implementation
Wrong drug 30 (48) 1 (13)
Wrong strength 16 (26) 4 (50)
Wrong dosage form 8 (13) 0 (0)
Expired medication 3 (5) 1 (13)
Otherb 5 (11) 2 (25)
Total 62 (100) 8 (100)

aADC = automatic dispensing cabinet.
bExamples include nonmedication items such as broken glass found in the drawer, loose dividers in the

matrix drawer, and technical issues.

No. (%) Errors

fore, it cannot be concluded whether
wholesaler-to-ADC prepackaging
or the use of bar-code-assisted ADC
refilling prevented the most errors.

Not all medications are available
through the wholesaler-to-ADC
program. Although the percentage of
incorrect medication (wrong drug,
wrong strength, and wrong dos-
age form) errors decreased, only 28
(47%) of the medications involved
in incorrect medication errors were
obtained through the new pro-
gram. At the time of the study, only
medications obtained through the
wholesaler-to-ADC program were
available for bar-code-assisted ADC
refilling, as only these products con-

tained a bar code scannable at the
ADC. The decrease in the percentage
of errors related to medications not
obtained through the wholesaler-to-
ADC refill program could potentially
be attributed to other changes insti-
tuted during program implementa-
tion. Nevertheless, there are plans
to expand bar-code-assisted ADC
refilling to all medications stocked in
the ADC to reap the full benefit from
the error-reduction potential of bar-
coding technology.

Conclusion
A redesign of the ADC refill proc-

ess using a wholesaler-to-ADC direct
refill program that included delivery

Practice rePorts Direct refill program

1664 Am J Health-Syst Pharm—Vol 69 Oct 1, 2012

of prepackaged medication and
bar-code-assisted refill significantly
decreased the occurrence of ADC
refill errors.

References
1. Pedersen CA, Schneider PJ, Scheckelhoff

DJ. ASHP national survey of pharmacy
practice in hospital settings: dispensing
and administration—2008. Am J Health-
Syst Pharm. 2009; 66:926-46.

2. Gaunt MJ, Johnston J, Davis MM. Auto-
mated dispensing cabinets. Don’t assume
they’re safe; correct design and use are
crucial. Am J Nurs. 2007; 8:27-8.

3. Paparella S. Automated medication dis-
pensing systems: not error free. J Emerg
Nurs. 2006; 32:71-4.

4. Institute for Safe Medication Practices.
Guidance on the interdisciplinary safe
use of automated dispensing cabinets.
www.ismp.org/Tools/guidelines/ADC_
Guidelines_Final (accessed 2011 Jun
30).

5. Klibanov OM, Eckel SF. Effects of auto-
mated dispensing on inventory control,
billing, workload, and potential for medi-
cation errors. Am J Health-Syst Pharm.
2003; 60:569-72.

System Pharmacists and its content may not be copied or emailed to multiple sites or posted to a listserv without

the copyright holder’s express written permission. However, users may print, download, or email articles for

individual use.

JONA
Volume 42, Number 12, pp 562-566
Copyright B 2012 Wolters Kluwer Health | Lippincott Williams & Wilkins

T H E J O U R N A L O F N U R S I N G A D M I N I S T R A T I O N

Bar-code Verification: Reducing but not
Eliminating Medication Errors

Philip L. Henneman, MD

Jenna L. Marquard, PhD

Donald L. Fisher, PhD

Justin Bleil, BS

Brendan Walsh, BS

Justin P. Henneman, MS

Fidela S. Blank, MS, RN

Ann Marie Higgins, RN

Brian H. Nathanson, PhD

Elizabeth A. Henneman, PhD, RN

Using observation, eye tracking, and clinical simu-
lation with embedded errors, we studied the impact
of bar-code verification on error identification and
recovery during medication administration. Data sup-
ported that bar-code verification may reduce but does
not eliminate patient identification (ID) and medica-
tion errors during clinical simulation of medication
administration.

Serious medication errors, many related to medica-
tion administration, are common in hospitals.1,2 The
medication administration process is complex and
is subject to numerous potential errors. Two critical
subprocesses of the medication administration pro-
cess are necessary for safe medication administration.
They are verifying that the patient’s identity (VPtID)

matches the identity (identify) on the medication and
orders and then verifying that the medication name
and dose match the medication name and dose on
the patient’s medication orders (VMed).

Nurses play the central role in medication admin-
istration. Studies suggest that nurses recover (ie, iden-
tify, interrupt, and correct) the majority of medication
errors.3-7 It is crucial that efficient and effective tech-
nologies be implemented to assist the nurse in per-
forming the recovery process.

The VPtID process is among the most common
safety process performed by healthcare workers. It
should be completed prior to performing most patient-
specific tasks. The 1st Joint Commission’s National
Patient Safety Goal calls for identifying patients with
at least 2 patient identifiers when providing care, treat-
ment, and services.8 The VPtID process requires match-
ing at least 2 unique patient identifiers, such as name,
date of birth (DOB), or medical record number (MRN)
on the task artifact (eg, patient identity label on a phar-
macy prepared medication) directly to the patient or
indirectly to another artifact (eg, patient’s ID band).

The VMed process involves matching the med-
ication name, dose, route, and scheduled time to the
patient’s orders. It includes confirming that the pa-
tient is not allergic to the ordered medication.

Bar-code verification technology in conjunction
with computer provider order entry and an electronic
medication administration system (eMAS) has been
introduced to reduce certain medication administra-
tion errors, in particular those related to incorrect pa-
tient ID and/or medication. In the absence of bar-code
eMAS, the nurse must (1) visually or verbally match the
patient’s orders to the patient or their ID band (VPtID),

562 JONA � Vol. 42, No. 12 � December 2012

Author Affiliations: Professor of Emergency Medicine (Dr P. L.
Henneman) and Assistant Professor of Emergency Medicine (Ms
Blank), Tufts University School of Medicine, Baystate Medical Cen-
ter, Springfield; Assistant Professor of Engineering (Dr Marquard),
Professor of Engineering (Dr Fisher), and Undergraduate Students
(Mr Bleil and Ms Walsh), University of Massachusetts, Amherst;
Research Assistant (Mr Henneman) and Registered Nurse (Ms
Higgins), Baystate Medical Center, Springfield; and Chief Ex-
ecutive Officer (Dr Nathanson), OptiStatim, LLC, Longmeadow,
Massachusetts; and Associate Professor of Nursing (Dr E. A.
Henneman), University of Massachusetts, Amherst.

This study was funded in part by National Science Foundation
(awards 0829901 [to P.L.H.], 1032574 [to J.L.M.], and 0820198
[to E.A.H.]).

The authors declare no conflicts of interest.
Correspondence: Dr P. L. Henneman, 109 Lake Ave, Sunapee,

NH 03782 (philip.henneman@bhs.org).
Supplemental digital content is available for this article. Direct

URL citations appear in the printed text and are provided in the
HTML and PDF versions of this article on the journal’s Web site
(www.jonajournal.com).

DOI: 10.1097/NNA.0b013e318274b545

http://www.jonajournal.com

(2) match the patient ID on the medication to the
patient or their ID band, and (3) match the informa-
tion on the medication itself to the medication infor-
mation listed on the patient’s order sheet (VMed).

To start the process of bar-code eMAS, the nurse
scans the bar code on the patient’s ID band (VPtID).
The bar code in the patient’s ID band contains a unique
identifier signifying patient identity and date of the
present visit. The nurse’s action results in a patient-
specific, documentation-ready medication administration
record appearing on the computer screen (ie, matches
the patient to the patient’s computer record/order).
Next, the nurse scans the bar code on the medication
container itself (package, intravenous bag, etc). The
bar code on the medication contains the drug’s na-
tional drug code, medication name, and dose. Other
information may be contained in the bar code (eg lot
number, expiration date, etc) but usually does not con-
tain the patient-specific ID. Scanning the medication
bar-code results in computer verification of the name
and dose of the medication compared with the name
and dose on the medication order (VMed). The lack of
a match between medication and order (name or dose)
will result in an error message being displayed on the
computer screen. If the medication name and dose
match, the nurse verifies such, and the system will doc-
ument the medication as administered.

Bar-code eMAS will successfully complete VPtID
and VMed when stock medications are administered.
When a medication is mixed by the pharmacy for a
specific patient (eg, unique dose or medication), pa-
tient ID information is placed on the medication bag
often as a label but not in the bar code. The nurse is
expected to match the patient’s ID information on the
medication to the patient or ID band.

In a previous study, using observation, eye track-
ing, and clinical simulation of medication administra-
tion without bar-code eMAS, we found that nurses
gave a medication to the wrong patient 39% of the
time when presented with an unexpected patient iden-
tity error similar to that used in this study.9 The pres-
ent study evaluated the impact of bar-code eMAS on
the incidence of a subset of medication errors com-
mitted by nurses during the medication administra-
tion process in a clinical simulation setting.

The research questions were as follows:

1. Does bar-code eMAS reduce patient ID (VPtID)
and medication (VMed) errors?

2. Is there a difference in the numbers of patient
ID (VPtID) and medication (VMed) errors with
and without bar-code eMAS?

3. What specific patient identifiers are viewed
by nurses during the administration of the
medication?

Methods

This was a prospective, observational study of emer-
gency department (ED) nurses administering a med-
ication to a patient (actor) in a simulated setting. The
study was approved by the institutional review board
at Baystate Medical Center, and all subjects gave their
signed, informed consent before participation. Nurses
volunteered to participate during one of their day or
evening shifts. Student volunteers were trained as pa-
tient actors and were given instructions on what to
say during the simulation scenario.

All nurses worked in a busy, urban ED with more
than 100 000 annual visits. The nurses were told that
the purpose of the study was to evaluate how experts
used visual cues during medication administration
using bar-code eMAS. Nurses were not told that there
were embedded patient ID and medication errors in
the simulation scenarios. All nurses were trained and
experienced in using the Cerner millennium bar-code
eMAS that was used in the simulation. All nurses
wore an eye-tracking device (see Figure, Supplemental
Digital Content 1, http://links.lww.com/JONA/A139)
that would video the field in front of them and place
crosshairs on the recorded video where the nurse was
looking at each moment during the simulation. This
was used to determine which specific patient identi-
fiers the nurses examined during the process of medi-
cation administration.

After placing and calibrating the eye-tracking de-
vice, the subject was given 2 labeled intravenous med-
ication bags to be administered to 2 separate patients.
Subjects were asked the complete this task using the
same bar-code eMAS process used in their clinical
practice. The embedded error in the simulation was
that the medication bag intended for patient 2 had
the same patient name but different DOB than the
DOB reported by the patient. In addition, the MRN
on the medication label did not match the MRN on
the patient’s ID band or on the patient’s computer
screen. The medication bag also listed the correct med-
ication name but a different dose than had been or-
dered for the patient (ie, gentamicin 400 mg instead of
60 mg).

An observer followed the nurse to each simulated
patient and completed a standardized data collection
sheet. If the nurse did not give the medication, the
observer asked the reason why. Identifying the medi-
cation error was defined as the nurse not administer-
ing the medication and voicing that there was a dose
discrepancy. Identifying the patient ID error was de-
fined as not giving the medication and voicing that the
medication was intended for a different patient.

The ASL (Applied Science Laboratory) mobile
eye is a tetherless eye-tracking system, which can

JONA � Vol. 42, No. 12 � December 2012 563

http://links.lww.com/JONA/A139

be worn by an active person in a free-moving envi-
ronment (see Figure, Supplemental Digital Content,
http://links.lww.com/JONA/A139). The eye tracker in-
cludes a scene camera, optics, and reflecting mirror
all mounted on safety glasses. Pupil corneal reflec-
tions are used to measure the position of the eye. The
eye-tracking device is 1st calibrated to each user. The
calibration process has the subject look at multiple
specific reference points in both the area where the
patient’s ID band would be and the area where the
computer screen would be. A mark will appear on
the video near each reference point. The marks are
adjusted to each of the specific reference points. The
output is stored on tape. The tape is analyzed with
the mobile eye software program, which after cali-
bration is able to overlay crosshairs at the approx-
imate 1-cm2 location in a scene where the individual
was looking.

Following the experiment, all videos were re-
viewed by 2 independent observers who recorded
whether the nurse looked or did not look at each of
the patient identifiers on the medication label, the
patient’s ID band, or the computer screen. Disagree-
ments between the 2 observers were resolved by a 3rd
observer. A nurse was assumed to have looked at a
specific patient identifier if 2 observers agreed that the
specific identifier was in the imaginary 1-cm2 box out-
lined by the crosshairs during a 0.4-second interval.

Eye-tracking or video failures were documented.
Eye-tracking failures occurred when cross hairs were
absent or only intermittently observed on the video.
Video failures occurred when patient ID information
could not be discerned because the image was washed
out from excessive glare.

Verifying patient ID on the medication bag to the
patient was defined as looking at 2 specific identifiers
on the medication label and matching them to the 2
same identifiers with the patient (verbal report), their
ID band (looking), or the ID information on the com-
puter screen (looking). Verifying patient to their ID
band required looking at 2 identifiers on the ID band
and verbally matching them to the patient’s self-report.

Comparison of study results were made to histori-
cal controls in a similar study previously reported.9

In this previous study, again using observation, eye
tracking, and clinical simulation, 28 nurses gave in-
travenous medications to 3 simulated patients with-
out using bar-code eMAS. The medication bag for the
3rd patients had the same name but different DOB
and MRN than reported by the patient, present on
their ID band, or present on the patient’s order sheet.

For statistical analysis, we used Stata/SE 11.1
(StataCorp, College Station, Texas). We calculated
95% binomial exact confidence intervals (CIs) for all
percentages. Chi square tests were used for all cat-

egorical inferences, except when the cell counts were
less than 4, in which case Fisher exact test was used.
P G 0.05 (ie, ” = .05) was considered significant.
Percentage agreement and 0 statistics were calcu-
lated for interrater agreement.

Results

Twenty-five nurses participated in 50 patient scenarios
(2 per nurse). Eighty-four percent of nurses (21/25;
95% CI, 64%-96%) determined that the medication
dose was incorrect for patient 2 before starting the
medication and did not give the medication. Nineteen
percent of the nurses (4/21; 95% CI, 5%-42%) who
identified the medication error also identified the pa-
tient ID error.

Sixteen percent of nurses (4/25; 95% CI, 5%-36%)
failed to identify the medication or ID error and
administered the medication to the wrong patient.
Two of the 4 nurses who started to administer the
medication to the wrong patient promptly recovered
the medication portion of the error when they noted
the error message on the computer and stopped the
medication.

Eye-tracking data could not be used in 8% (4/50)
of the patient scenarios because of intermittent or
absent crosshairs on the videos. Glare from the com-
puter screen prevented eye-tracking data from being
recovered when participants were looking at the com-
puter screen for an additional 28% (14/50) of the
patient scenarios. There was an 85% agreement be-
tween the initial 2 independent observers (0 = 0.71).

Twenty percent of nurses (9/46; 95% CI, 9%-34%)
verified the patient’s ID band to the patient, using 2
identifiers, yet all used the ID band for bar coding.
Forty-six percent (21/46; 95% CI, 31%-61%) verified
the patient’s name (asked name or looked at name
on ID band) to the patient name on the medication
bag; 4 of these nurses also checked the DOB or MRN
on patient 2 and noted the discrepancy between the
medication bag and the patient, their ID band, or the
computer screen (ie, detected patient ID error).

Table 1 outlines the comparison of medication
administration with bar-code eMAS in this study with
medication administration without bar-code eMAS in
historical controls. Eighty percent or more of nurses in
either experiment did not verify the patient to the ID
band (they assumed that they matched). The percent-
age of RNs verifying patient ID on the medication bag
to the patient, their ID band, or the patient’s verified
computer screen was 16% (5/32; 95% CI, 5%-33%)
with bar-code eMAS and 67% (45/67; 95% CI,
55%-78%) without bar-code eMAS.

Verification using bar-code eMAS in clinical
simulation reduced the percentage of nurses giving

564 JONA � Vol. 42, No. 12 � December 2012

http://links.lww.com/JONA/A139

a medication to the wrong patient from 39% (11/28)
without bar-code eMAS to 16% (4/25; 95% CI,
5%-36%) with bar-code eMAS (P = .06). If we in-
clude nurses who promptly recovered the error by
noting the error message on the computer screen and
stopping the medication, only 8% of nurses (2/25;
95% CI, 1%-26%) using bar-code eMAS did not
identify and recover the error and give the complete
medication to the wrong patient (P = .01).

Discussion

Verification using bar-code eMAS reduced the inci-
dence of patient ID and medication errors in clinical
simulation. This has also been observed in a before-and-
after clinical study where implementation of bar-code
eMAS reduced the incidence of errors in medication
administration.1 Our study demonstrates ways that
errors continue to occur.

Few nurses verified that the information provided
by patient matched the information on their ID band;
most assumed that the ID band was accurate. In a
review of 2.4 million patient ID bands at 712 hospi-
tals, 8.6% had erroneous information, and 0.5% of
patients were wearing an ID band with another pa-
tient’s information.10 In clinical simulation, we found
that clerks, when presented with an unexpected pa-
tient ID error similar to that used in this study, would
place an incorrect wrist band on as many as 71% of
patients.9 Perhaps hospitals need to develop a verifi-
cation process of the patient’s ID band by someone
other than the individual who applies it.

Few nurses identified the patient ID error in our
simulation experiments; only 4 of the 25 noted the
different DOB or MRN. Only half of the nurses in
our trial verified the patient’s name, but we have
found that verifying the name alone is insufficient in
that 11% of patients will have another patient in the
ED with the same last name at the same time.9 Most
nurses did not give the medication in our simulation

because they noted the dosing error; however, few
nurses noted the patient ID error. Adding patient ID
information (eg, account number) to the medication
bar code would allow the software to add an error
message for the ID error and therefore might reduce
or prevent prepared medications being given to the
wrong patient even with bar-code eMAS system.

The bar-code eMAS should routinely pick up a
medication name and dose error and create an error
message. In our study, a small percentage of nurses
either did not look at the computer prior to starting
the medication or ignored the error message. Multi-
ple different types and causes for workarounds by
nurses with bar-code eMAS have been identified.
One study found that nurses overrode bar-code
eMAS alerts for 4% of patients and 10% of medica-
tions charted.11 Perhaps adding audible alarms that
cannot be silenced for specific patient safety alerts
might reduce delays in error recognition. Improving
the process to alleviate the need for workarounds
and raising awareness of the impact of workarounds
during patient safety processes might help reduce
errors.

We observed inefficient and interrupted visual
scanning patterns by nurses during medication
administration in the historical control study with-
out bar-code eMAS. Many nurses looked at multi-
ple identifiers in between key VPtID steps, such as
looking at the patient’s name and MRN in between
looking at the DOB on 2 artifacts.12 Although this
could be considered an appropriate match between
the 2 artifacts, the nurse likely could not keep the
DOB in working memory so could not remember
the DOB correctly. We also observed random visual
scanning patterns by nurses who did not identify
patient ID errors. These findings suggest that train-
ing nurses to use specific visual surveillance techni-
ques might improve the effectiveness and shorten
the time required to perform VPtID and VMed dur-
ing medication administration. We are continuing to

Table 1. Comparison of Simulated Medication Administration With and Without Bar-code eMAS

With Bar-code eMAS
Without Bar-code eMAS

(Historical Controls)9 P

Verify patient to ID band 20% (9/46) (95% CI, 9%-34%) 13% (9/67) (95% CI, 6%-24%) .38
Verify patient ID on medication to

patient/ID band/computer
16% (5/32) (95% CI, 5%-33%) 67% (45/67) (95% CI, 55%-78%) G.001

Did not recover errorVgave
medication to wrong patient

8% (2/25) (95% CI, (0%-26%) 39% (11/28) (95% CI, 22%-59%) .01

Recovered errorVdid not give
entire medicationa

92% (23/25) 61% (17/28) .01

aIncludes 2 nurses who started the medication but promptly stopped the infusion when they noted the error message on the computer screen
(ie recovered the error).

JONA � Vol. 42, No. 12 � December 2012 565

look at the visual scanning patterns of nurses during
bar-code eMAS.

Limitations

The main limitation in this study was that it was
performed in a simulated setting without any of the
usual stressors found in clinical practice. However,
only in simulations could specific errors be embedded
to observe nurse behavior. Our study was performed
using nurses from a single institution with 1 kind of
bar-code eMAS. We do not know if the behavior we
observed is similar to that found in other settings. The
eye-tracking device failed in 8%, and glare on the
computer screen prevented complete data collection
during another 28% of patient scenarios. Improving
the lighting from the computer screen will be nec-

essary in future work. The comparison of medica-
tion administration with and without bar-code eMAS
used historical controls instead of a direct comparison,
but the historical control was derived from the same
hospital. Finally, looking at 2 patient identifiers for
0.4 seconds does not necessarily mean that patient
identity was verified; we know that inattention
occurs.9

Conclusion

Bar-code eMAS reduces but does not eliminate pa-
tient ID and medication verification errors during
medication administration in a simulated setting. This
study demonstrates that further human and techno-
logical improvements are needed to ensure that the
right patient receives the right medication.

References

1. Poon EG, Keohane CA, Yoon CS, et al. Effect of bar-code

technology on the safety of medication administration.

N Engl J Med. 2010;362:1698-1707.
2. Henneman PL, Blank FS, Smithline HA, et al. Voluntarily

reported emergency department errors. J Patient Saf. 2005;1:
126-132.

3. Henneman EA, Gawlinski A. A ‘‘near-miss’’ model for

describing the nurse’s role in the recovery of medical errors.
J Prof Nurs. 2004;20:196-201.

4. Henneman EA, Blank FS, Gawlinski A, et al. Strategies used

by nurses to recover medical errors in an academic emergency
department setting. Appl Nurs Res. 2006;19:70-77.

5. Henneman EA, Roche JP, Fisher DL, Cunningham H, et al.

Error identification and recovery by student nurses using hu-

man patient simulation: opportunity for improving patient
safety. Appl Nurs Res. 2010;23:11-21.

6. Rothschild JM, Hurley AC, Landrigan CP, et al. Recovery

from medical errors: the critical care nursing safety net.

J Qual Patient Saf. 2005;32:63-72.

7. Dykes PC, Rothchild JM, Hurley AC. Medical errors

recovered by critical care nurses. J Nurs Adm. 2010;40:
241-246.

8. JCAHO National Safety Goals. http://www.jointcommission
.org/assets/1/6/NPSG_EPs_Scoring_HAP_20110706 .

Accessed October 25, 2011.

9. Henneman PL, Fisher DL, Henneman EA, et al. Patient

identification errors are common in a simulated setting. Ann
Emerg Med. 2010;55:503-509.

10. Renner SW, Howanitz PJ, Bachner P. Wristband identifica-

tion error reporting in 712 hospitals. Arch Pathol Lab Med.
1993;117:573-577.

11. Koppel R, Wetterneck T, Telles JL, et al. Workarounds to

barcode administration systems: their occurrences, causes,

and threats to patient safety. J Am Med Inform Assoc. 2008;
15:408-423.

12. Marquard J, Henneman P, Jo J, et al. Nurses’ behaviors and

visual scanning patterns may reduce patient identification

errors. J Exp Psychol Appl. In press.

566 JONA � Vol. 42, No. 12 � December 2012

http://www.jointcommission.org/assets/1/6/NPSG_EPs_Scoring_HAP_20110706

Turn in your highest-quality paper
Get a qualified writer to help you with

“ summarize 5 articles in your won words article is attached… ”

Get high-quality paper

NEW! AI matching with writer

Still stressed with your coursework?

Get quality coursework help from an expert!

Order now