ACAD EMERG MED

March 2004, Vol. 11, No. 3

289

www.aemj.org

SPECIAL CONTRIBUTIONS
Profiles in Patient Safety: Medication Errors in the
Emergency Department
Pat Croskerry, MD, PhD, Marc Shapiro, MD, Sam Campbell, MB, ChB,
Connie LeBlanc, MD, Douglas Sinclair, MD, Patty Wren, MD, Michael Marcoux, RPh
Abstract
Medication errors are frequent in the emergency department (ED). The unique operating characteristics of the ED
may exacerbate their rate and severity. They are associated
with variable clinical outcomes that range from inconsequential to death. Fifteen adult and pediatric cases are
described here to illustrate a variety of errors. They may
occur at any of the previously described five stages, from
ordering a medication to its delivery. A sixth stage has been
added to emphasize the final part of the medication
administration process in the ED, drawing attention to
considerations that should be made for patients being

discharged home. The capability for dispensing medication,

without surveillance by a pharmacist, provides an errorproducing condition to which physicians and nurses should
be especially vigilant. Except in very limited and defined
situations, physicians should not administer medications.
Adherence to defined roles would reduce the team
communication errors that are a common theme in the
cases described here. Key words: medication errors;
emergency medicine. ACADEMIC EMERGENCY MEDICINE 2004; 11:289299.

Medication errors are common throughout the health

care system. In the ambulatory setting, the annual cost
of drug-related morbidity and mortality in the United
States was estimated to be in excess of $76 billion in
1997, a significant part of which is associated with
hospitalizations.1 The unique milieu of the emergency
department (ED) provides a number of factors that
may exacerbate medication error rate and severity
(Table 1). Most of these are self-explanatory. Tight
coupling2 refers to medication administration that is

time-dependent, has a rapid onset of action, is

delivered in an invariant sequence, and is relatively
inflexible; examples are rapid sequence intubation
and thrombolysis. ED dispensing refers to the ability
of most EDs to function as a pharmacy and dispense
drugs. This is a necessary capability, but it obviates
input from pharmacists who might otherwise detect
prescription and other errors.
There are five steps in the sequence of ordering
a drug to its delivery: prescription ! transcription !
dispensing ! administration ! monitoring.3 A
comprehensive review of medication errors in the
ED has recently been provided.4 The following 15
cases have been selected to illustrate a variety of
medication errors in the ED.

From the Department of Emergency Medicine, Dartmouth General

Hospital, Dartmouth, Nova Scotia, Canada (PC); Department of
Emergency Medicine (MS) and Pharmacy Department (MM),
Rhode Island Hospital, Brown Medical School, Providence, RI;
Department of Emergency Medicine, Queen Elizabeth II Health
Sciences Centre, Dalhousie University, Halifax, Nova Scotia, Canada
(SC, CL); and Department of Emergency Medicine, Isaac Walter
Killam Health Centre, Dalhousie University, Halifax, Nova Scotia,
Canada (DS, PW).
Received May 15, 2003; accepted May 17, 2003.
Series editors: Pat Croskerry, MD, PhD, Department of Emergency
Medicine, Dartmouth General Hospital Site, Dalhousie University,
Halifax, Nova Scotia, Canada; and Marc J. Shapiro, MD, Department of Emergency Medicine, Rhode Island Hospital, Brown
University School of Medicine, Providence, RI.
Supported by a Senior Clinical Research Fellowship from the
Faculty of Medicine, Dalhousie University, Halifax, Nova Scotia,
Canada, to Dr. Croskerry. Drs. Croskerry and Shapiro were
supported in part by Grant P20HS11592-02 to the Center for Safety
in Emergency Care from the Agency for Healthcare Research and
Quality.
Address for correspondence and reprints: Pat Croskerry, MD, PhD,
Department of Emergency Medicine, Dartmouth General Hospital,
325 Pleasant Street, Dartmouth, Nova Scotia, Canada, B2Y 4G8. Fax:
902-460-4148; e-mail: xkerry@accesscable.net
doi:10.1197/j.aem.2003.05.013

CASE 1
A 65-year-old man presented to the ED of a teaching
hospital at 2218 hours with left-sided weakness, which
started at 2040 hours that day. He also complained of
mild pain (a cranial headache) and nausea, both of
which started at the same time as the weakness. He
had no other neurological symptoms and had never
had similar symptoms in the past. The patient and his
family were well known to the emergency physician.
The patients history included lung cancer for which
he was currently undergoing chemotherapy, adultonset diabetes, and hypertension. He had no allergies.
Medications included dexamethasone, ondansetron,
metoprolol, and ranitidine.
Physical Examination. Vital signs at triage were
Glasgow Coma Scale score (GCS) 15; heart rate (HR)

290
TABLE 1. Factors Contributing to Medication
Errors in the Emergency Department (ED)*
Patients are usually strangers
Multiple patients being treated concurrently
Frequent reliance on verbal orders
Wide range of drugs in use
Variety of administration routes
Wide variety of dangerous drugs
Time pressures
Interruptions/distractions
Tight coupling
ED dispensing
Physician administration of medication
Team communication problems
Laboratory errors

Croskerry et al.

MEDICATION ERRORS IN THE ED

is unlikely because she believed she was giving

Maxeran, and not midazolam, and the order was
clearly written. The vials of the two medications were
color-coded differently and easily distinguishable
from each other, but they were similarly sized and
adjacent to each other in the same drawer. Therefore,
a spatial discrimination error in selection of the
correct vial might have occurred. Another possibility
is a simple slip of action, or execution failure, due
to fatigue, distraction, or attentional capture by
something other than the task at hand.5 The proximity
and similarity of the two medications seem to provide
the most plausible explanation.

*Based on Croskerry and Sinclair14 and Peth.4

130 beats/min; blood pressure (BP): 170/96 mm Hg;

respiratory rate 16 breaths/min; and oxygen (O2)
saturation 95% on room air. His glucose check was
9.6 mmol/L. His speech was normal, and he had
no cranial nerve abnormalities. Chest, heart, and
abdomen were unremarkable. Marked weakness
and increased reflexes were evident on the left side.
Sensory examination was normal. A computed
tomography (CT) scan was completed and appeared
normal. A consult was made to neurology, describing
the stable condition of the patient and the diagnosis of
a nonhemorrhagic cerebrovascular accident. The
physician then gave a verbal and written order for
10 mg Maxeran IV for nausea and headache.
The physician was called to the patients bedside 10
minutes later because the patient had become unresponsive with a respiratory rate of 8 breaths/min. A
discussion followed with his family, in which the
gravity of his apparently deteriorating condition was
discussed. A decision was made to intubate him, repeat the CT scan, and reassess him at that time. However, a second nurse noticed shortly afterward that the
medication vial attached to his intravenous ( IV) bag
was not Maxeran (metoclopramide), but midazolam.
He was immediately given the benzodiazepine
antagonist flumazenil IV, after which he awakened
and appeared to recover to his previous state.
The physician immediately disclosed the error to
the family, with a full explanation of events. They
were reassured that the patient had not suffered any
apparent ill effects. One family member became
extremely angry but eventually settled with reassurances from the other family members. The neurology
resident was forewarned of the error before meeting
with the family.
Comment. This is an example of a transcription error.
In the investigation that followed, no obvious explanation emerged to explain why the error occurred. At
first sight, it might appear this was a sound-alike error
such that the nurse misheard the order as midazolam, confusing it with Maxeran. However, this

CASE 2
An 18-year-old female student presented to the ED of
a regional hospital in the early afternoon with
a complaint of urinary frequency and back pain. Her
last menstrual period was normal. She was taking no
medications other than birth control pills. She had no
allergies. She had no prior history of urinary tract
infection. Her medical history was significant for
glucose-6-phosphate dehydrogenase (G6PD) deficiency.
She was a healthy-looking woman, in no distress,
with normal vital signs. Her physical examination
was unremarkable other than some mild suprapubic
tenderness on palpation. A pregnancy test was negative. Her urinalysis showed moderate leukocytes,
nitrites, and bacteria.
She was informed that she had a urinary tract
infection and that it would be treated with an antibiotic. A starting dose of trimethoprim-sulfamethoxazole was ordered in the ED. She indicated she was
aware of problems of some drugs with G6PD deficiency and asked if the antibiotic was safe. The
emergency physician said that it was and she took the
medication. A prescription was given for a course of
the antibiotic.
Approximately one hour later the emergency physician received a call from the patients pharmacy
where she had presented her prescription. The pharmacist was aware of the patients G6PD deficiency and,
on routinely checking the antibiotic, noted a precaution. Sulfonamides may cause hemolysis in patients
with G6PD deficiency, an effect that is frequently
dose-related. The physician ordered a substitution. He
later called the patient at home and explained the
error. She was advised that it was unlikely she would
develop a problem with one dose. Nevertheless, she
was cautioned to seek immediate medical attention if
she appeared to be suffering any adverse effects.
Comment. This is a prescribing error attributable
largely to a knowledge deficiency. In Reasons Generic
Error-Modeling System (GEMS), it is classified as

ACAD EMERG MED

March 2004, Vol. 11, No. 3

291

www.aemj.org

a knowledge-based mistake.5 In one study, knowledge

deficits accounted for about 30% of medication
errors.6 This physician had treated several patients
in the past with G6PD deficiency and, other than salicylate, could not recall any specific issues around
medication precautions. He also thought that he was
guilty of an overconfidence error7,8 in assuring the
patient that the medication was safe, leading to an
error of commission.7,8 The case illustrates clearly
how the dispensing process provides an opportunity
to correct a medication error.

CASE 3
A 45-year-old woman was brought to the ED of
a regional hospital by ambulance. She had been found
unresponsive by her boyfriend. It appeared that she
had taken a mixed overdose, including a significant
amount of a tricyclic antidepressant. Her pulse was
130 beats/min and her BP was 105/50 mm Hg. Her
GCS was 7.
She was prepared for rapid sequence intubation.
The emergency physician called for 5 of midazolam,
IV for induction, administering it himself by slow IV
push over 2 minutes, and the patient was successfully
intubated. Shortly afterward, as the record was being
completed, the recording nurse confirmed with the
physician the medication that had been given, specifically that 25 mg of midazolam had been administered.
The physician corrected her, saying that he had given 5
mL of a solution containing 1 mg/mL of midazolam,
for a total of 5 mg. The nurse informed the physician
that, in fact, he had been given a solution containing
5 mg/mL of midazolam and that the patient
had, therefore, received a total of 25 mg. The patient
appeared to suffer no adverse consequences from
the medication overdose. Specifically, there were no
cardiac arrhythmias. She subsequently underwent
stomach lavage, was further stabilized, and was
transferred to the intensive care unit (ICU) where she
made an uneventful and full recovery.
Comment. This is a form of transcription error. It was
an emergent situation and, therefore, a verbal order
was acceptable. The physician assumed that the
concentration of midazolam was 1 mg/mL, believing
that was the only concentration available in the ED.
He was not aware that the alternative higher
concentration was also in stock. The nurse, who was
relatively unfamiliar with the drug and new to the
ED, was also not aware that more than one concentration was available. Nevertheless, the error would
probably have been avoided if the nurse had stated
the concentration of the solution, or if the physician
had verbally confirmed what he understood the
solution strength to be. Subsequently, the higher dose
was removed from ED stock.

CASE 4
An elder woman with type 1 diabetes presented to the
ED of a general hospital feeling weak and dizzy. She
had recently experienced symptoms of a gastrointestinal infection and had not taken her insulin for 24
hours. She had been vomiting and felt dehydrated.
Her glucometer reading in the ED was 24 mmol/L.
Routine blood work was ordered, an IV line was
started for hydration, and the attending physician
wrote an order for 10U insulin to be given intravenously. The nurse read the order as 100 insulin
and proceeded to give it. The patient subsequently
became obtunded. The error was discovered almost
immediately, and a bolus of IV dextrose was given.
The patient was further stabilized in the ED over
several hours, was transferred to the ICU, and made
an uneventful recovery.
Comment. This is a frequently reported transcription
error. The abbreviation U for units is read as a zero
and the dose subsequently magnified by a factor of 10.
To avoid the error, units should be written out in
full and never abbreviated. Furthermore, any unusual
dose should be questioned by the nurse before
administration. Failure to challenge may be due to
authority gradient effects referred to in the Institute of
Medicine (IOM) report9 and discussed by Turnbull.10,11

CASE 5
A 52-year-old woman presented to the ED of a general
hospital with progressive dyspnea over several days.
She had symptoms of an upper respiratory infection.
She had a history of bronchitis, but her medical
history was otherwise unremarkable. She had a mild
fever of 37.58C, and the remainder of her vital signs
were stable. On examination, she had mild inspiratory
and expiratory wheezes and an exudative pharyngitis, with some tender anterior cervical nodes.
She was given a series of bronchodilators by mask,
resulting in an eventual clearing of her chest. A throat
swab was done, and the patient was informed she
would be given a prescription for amoxicillin. The ED
was extremely busy at the time of her presentation,
and the physician had been interrupted several times
during his care of the patient.
The patient returned to the ED three hours later
with facial swelling, nausea, and tightness in her
chest. She said she had filled the prescription on
leaving the ED and had developed the symptoms
after taking the first dose. She was seen by the same
physician who had first seen her, and was immediately treated with subcutaneous epinephrine, bronchodilators by mask, intramuscular diphenhydramine
hydrochloride, and IV methylprednisolone. Her
symptoms resolved over several hours.

292

Croskerry et al.

She was very angry with the physician and asked

him why he had prescribed penicillin for her when
she was known to be allergic to penicillin. The
physician checked her chart and noted that she was
clearly flagged with a penicillin allergy. He apologized to the patient and said that he had failed to
check the chart, probably because he had been
distracted by the high level of activity and acuity in
the ED at the time of her first visit.
Later, when she had settled down somewhat, he
asked her why she had taken a penicillin knowing
that she was allergic to it. She became angry again,
asking how she was expected to know that amoxicillin was a penicillin.
Comment. This is a prescription error. Before any
prescription is given to a patient, physicians should
always apply a simple forcing function, asking the
question: Is there any reason why this patient should
not receive this medication? In the case presented
here, the error was a rule-based mistake5 (i.e., the
physician clearly knew the rule of checking for an
allergy before prescribing the medication but failed to
apply it). He attributed the failure to the extreme
busyness of the ED at the time. High rates of
interruption and distraction have been described in
EDs12 and clearly contribute to error. Attentional
capture associated with distractions and preoccupations is a necessary condition for the occurrence of
some errors.5 Although patients themselves may be
the last opportunity to stop the administration of
a drug to which they are allergic, clearly there should
be no expectation that they will have any understanding of drug classes.

CASE 6
In a busy ED of a tertiary care hospital, an emergency
physician had several patients in various stages of
workup. He had seen and assessed a female patient in
a monitored bed (bed 16), made a diagnosis of migraine
headache, and prescribed prochlorperazine 10 mg/IV.
Another female patient who presented with chest pain
was in an unmonitored bed (bed 12) and being
investigated for a possible acute coronary syndrome.
When the physician reassessed the patient with
headache, he found the response to the initial therapy
to be inadequate. A few minutes later, at the nursing
station, he asked one of the two nurses to Give the
lady in bed 16 one milligram of dihydroergotamine
(DHE). The nurse promptly carried out the verbal
order and administered the drug to the patient in bed
16. However, before he had spoken to the nurse, the
other nurse working in the area had switched the
chest pain patient to bed 16 so that she could be on
a cardiac monitor. The patient with migraine was
moved to the now vacated bed 12. Some 20 minutes
later, when the first nurse asked the physician about

MEDICATION ERRORS IN THE ED

the new treatment for unstable angina, the potentially catastrophic mix-up was discovered.
Fortunately, the chest pain of the patient given DHE
was found to be of a noncardiac origin and she
suffered no ill effects. When informed of the mistake,
she took it cheerfully without complaint. The administration of a vasoconstrictor to a patient with
vaso-occlusive cardiac disease could have caused
significant exacerbation of her condition.
Comment. This case illustrates several errors. First,
there is a dispensing error. The process of dispensing
a medication is to provide it to the person who will
give the drug, a task usually performed by a pharmacist. In this case, however, the nurse dispenses the
medication from the ED herself, ultimately for the
wrong patient. If a pharmacist had been involved in
the transaction, a bed number would not have
sufficed, and a verification process would have
occurred, matching the drug to a particular patient.
Second, there is an administration error in that the
nurse physically administered the drug to the wrong
patient, without verifying the patients identity. Recent Joint Commission on Accreditation of Healthcare
Organizations (JCAHO) recommendations are that
the identity of the patient be doubly confirmed (e.g.,
by asking the patient his or her name as well as
a numeric check [e.g., date of birth]) prior to
administration. Third, this case raises issues around
the use of verbal orders in the ED. Protocols for the
use of verbal orders vary between EDs, some allowing
them only in urgent or emergent situations. In certain
circumstances, however, protocols may be difficult to
apply because physicians may be involved in urgent
or emergent situations with other patients and unable
to leave to write nonurgent orders. Often nurses will
take verbal orders in these nonurgent situations to
maintain patient flow through the department. In this
particular case, the ED did not have a protocol in
place and, therefore, a safety violation did not
technically occur. However, bed switching of patients
is not uncommon in the ED, and the error might have
been avoided if the order had been written on the
patients chart instead of being given verbally.

CASE 7
A 35-year-old woman presented to the ED of a general
hospital with severe flank pain, accompanied by
nausea and vomiting, which had started early in the
morning. A physician was interrupted in his evaluation of another patient to come to assess the patient
because of her level of distress. After a rapid
assessment, the physician determined that the patient
probably had ureteral colic and asked the nurse to
give fentanyl 100 micrograms intravenously and 100
milligrams of indomethacin rectally. He then returned
to his previous patient.

ACAD EMERG MED

March 2004, Vol. 11, No. 3

293

www.aemj.org

Shortly afterward, hurriedly walking past the

patient with flank pain, he noticed she was still in
distress. He asked the nurse if the fentanyl was ready.
She replied that she had first given the rectal
medication and that she was about to draw up the
fentanyl. Several minutes later, the physician returned
and saw the medication on the patients bedside table.
The nurse was not present. He immediately gave the
fentanyl IV push into a port in the IV line.
About 15 minutes later, passing by the patient
again, he observed that she was still in some distress
and thought this unusual given the rapid onset of
relief usually associated with fentanyl. He went to the
nursing station and asked the nurse how the patient
was doing. The nurse replied that the patient was
still in some pain but that the bed felt much better!
She explained that she had set up the IV bag on a pole
with an IV line but had been called away to help with
a resuscitation. As she left the patient, she had pushed
the end of the IV line under the patients bedclothes
and under the mattress to secure it. When the
physician came to administer the medication he had
followed the IV line going from the IV bag until he
found a port. The remainder of the IV line disappeared under the bedclothes and he assumed it was
connected to the patient. When the nurse returned to
complete the IV setup, the patient informed her that
the doctor had come in briefly and injected the
medication into the IV line, telling her she would
feel better shortly.
Comment. This is an unusual medication administration error arising largely out of interruptions and
discontinuities in both nursing and physician care.
When patient care is limited by availability of resources
of the ED, conditions are created for a potential tradeoff with quality, referred to as RACQITO (resource
availability continuous quality improvement tradeoff).13 The RACQITO state can be exacerbated by errorproducing conditions (EPCs) in the ED.14,15 The
physician was ultimately the victim of a visual illusion
in presuming that an IV line going from an IV bag in
close proximity to the patient and disappearing under
a patients bedclothes was connected to the patient. In
addition, there were rule violations in that the patency
of IV access should always be checked before administering medication, and the physician administered
the drug in a nonemergent situation.

CASE 8
A 23-year-old woman presented to the ED of a tertiary
care hospital with a complaint of abdominal cramps.
After assessing the patient, the emergency physician
ordered 20 mg hyoscine butylbromide (Buscopan) IV,
saying Ill give her Buscopan. The department was
busy. He asked a nurse to start an IV line, intending to
administer the drug himself to save her from the time-

consuming procedure of hanging it in a bag and

giving it over 20 minutes as directed by nursing
protocol. The nurse started the IV line while the
physician left to draw up the medication himself. On
his return, and seeing the IV line in place, the
physician administered the drug and charted it in
the nursing note. While he was doing this, the nurse,
misunderstanding his order to believe that she should
just do an IV push without a bag, administered the
same dose to the patient. The double dose was
discovered when the nurse went to document the
dose she had just given. The patient was told of the
mistake and suffered no ill effects.
Comment. This is a form of transcription error,
a breakdown in communication occurring between
the prescriber of the drug and the person responsible
for dispensing and administering it. It arises out of an
ambiguous statement by the physician declaring his
intent to give the medication himself, which the nurse
interprets as verbalizing a general plan, seeing herself
as the intended administrator of the medication. The
physician is responsible for the error through his
effort in trying to save time and resources in a busy
ED. Again, this breakdown in teamwork communication is an example of a RACQITO-type error13 (see
case 7), compounding the error of the physician
administering the drug himself.

CASE 9
A 36-year-old man was brought to a teaching hospital
ED by friends. He had collided with another player
and fallen in a rugby game. He complained of an
injury to his right shoulder. He was placed in a bed in
the minor trauma area and an IV line was started. His
vital signs were stable. He was healthy with no
significant medical history. He was seen by a firstpostgraduate-year (PGY1) resident, who suspected
a shoulder dislocation and ordered an x-ray, which
confirmed his diagnosis. The PGY1 resident asked the
nurse to draw up 10 mg diazepam intravenously and
later administered it to the patient. He then went to
discuss the case with the attending physician.
The attending was already discussing a complex
case with another PGY1 resident that took some time
to complete. It was about 20 minutes before the first
PGY1 was able to present his shoulder dislocation
case. The attending discussed the particular case and
then went on to a general discussion of both anterior
and posterior dislocations and their treatment. He
then asked the PGY1 what he thought should be done
in this particular case. The PGY1 responded that the
patient should be sedated and proposed a maneuver
to reduce the shoulder. The attending concurred with
the general plan and asked where the patient was. The
PGY1 gave the location and then said he had already

294

Croskerry et al.

initiated the sedation to speed things along. The

attending hurried to the minor treatment room.
The patient had stopped breathing. The nurse in the
room was unaware the medication had been given,
and the patient was not on a monitor or pulse
oximeter. The patient was immediately bagged and
given 0.2 mg flumazenil intravenously. The attending
quickly reduced the shoulder before the patient woke
up. There appeared to be no ill effects of what was
probably a relatively brief period of anoxia.
Comment. The first error is a rule violation in that the
resident should not have initiated treatment without
conferring with the attending. This was the residents
first day on the service, and he claimed that he was
unaware of the rule, having missed part of the
orientation. The second is one of lack of knowledge
combined with anchoring. The resident had just
completed a rotation on a psychiatric service where
he had been involved in detoxification of patients
with alcoholism. He had been accustomed to patients
receiving fairly high doses of oral diazepam (up to 120
mg/day) and had anchored7 on this as an acceptable
dose range. In the discussion that followed, he said he
had thought that a big rugby player should have
been able to handle 10 mg Valium. In fact, 10 mg is
the upper limit for IV administration, reflecting
a knowledge error. The third error is another rule
violation in that a conscious sedation procedure was
initiated without following the department protocol.
The fourth is a team communication error. Had the
nurse been aware that the medication had been given
for a conscious sedation procedure, she would have
placed the patient on appropriate monitoring according to the ED protocol.

CASE 10
A 70-year-old man at a shopping mall experienced
acute onset of shortness of breath and was brought to
a community hospital ED by ambulance. He was
placed in a resuscitation bed. The monitor showed
a wide-complex ventricular tachycardia with a rate of
140 beats/min. The emergency physician was called
and immediately ordered an intravenous bolus of 100
mg lidocaine. Shortly after, an internist assumed the
care of the patient and proceeded to order a bolus
dose of procainamide according to the ED protocol.
The Kardex, opened at the patients bedside, specified
the dose at 100 mg in 10 mL D5W or NaCl.
Procainamide was supplied to the ED in a 10-mL vial
containing 1,000 mg.
There were three nurses in the resuscitation room.
Nurse A, who was unfamiliar with the protocol, drew
up the full 10 mL into a 10-mL syringe. She asked the
internist if he wanted the whole thing. He assumed
he was getting the dose prescribed in the protocol and
said Yes. The patient, therefore, received 1,000 mg

MEDICATION ERRORS IN THE ED

of procainamide. There was no effect on the rhythm

and the internist ordered a second dose of procainamide. He was given the same dose as the first time
(1,000 mg in 10 mL) and again gave the full 10 mL,
assuming it contained 100 mg. The nurses were
unable to recall who prepared this second dose.
However, the IV line was observed to be nonfunctional and had infiltrated. A second IV access was
obtained by nurse B, and the physician ordered a third
bolus of the drug. He was again handed 1,000 mg, this
time by nurse C, and he administered it. The patient
became hemodynamically unstable, requiring intubation and was transferred to a regional hospital for
further management. He died several days later. The
errors were discovered immediately after the attempted resuscitation, when the notes were being written
up. Throughout the resuscitation, the ED was short
staffed, and the three nurses involved in the case
periodically left and returned to the room, attending
to other patients in the ED.
Comment. The immediate error is one of communication in transcription. As in case 3, the physician
believes he is receiving a particular dose while getting
another. A critical feature of dilution errors is that they
are usually not small deviations from the intended
dose, but often differ by tenfold. Decimal point errors
are particularly common in pediatrics.16 Transmission
of information in emergency situations is most often in
the verbal mode and usually expedient and safe.
However, occasional error is inevitable, especially in
high-stress, tightly coupled situations.2
A rule violation error also occurs when the protocol
for dilution and administration of the drug is not
followed. Error probabilities are known to increase
under a variety of task situations and conditions.17 In
this case, unfamiliarity with the drug, time pressures,
impoverished quality of information transfer, unclear
allocation of responsibility, and other EPCs of the
ED14 are all contributory. Another critical aspect of
this case is disrupted team function. There was no
team leader directing the nursing resources in the ED,
and there were significant discontinuities in the
immediate care of this patient. Teamwork failures in
the ED are associated with significant error.18 Finally,
it is of note that the same error is repeated three times
in this case. Clearly, this is not coincidental. Once the
error trajectory has been established, and there is no
immediate negative feedback, it is relatively easy for
copycat17 errors to occur.

CASE 11
A 41-year-old man presented to the ED of a general
hospital in the early evening complaining of dizziness
and weakness. He stated that he had two seizures the
previous day. When he tried to walk earlier in the day,
he fell onto his face and had a nosebleed. He crawled

ACAD EMERG MED

March 2004, Vol. 11, No. 3

www.aemj.org

back to his room and later called a cab to bring him to

the ED because he was afraid he would have more
seizures. At triage, his vital signs were normal other
than a mild tachycardia at 103 beats/min. There was
an abrasion across his nose but no active bleeding or
deformity. The patient was well known to the ED,
having had numerous previous visits for a variety of
complaints. His medical history included personality
disorder, alcoholism, depression with several overdoses, and gastroesophageal reflux disease. He had
been started on phenytoin seven months earlier by his
family physician following a seizure, which may have
been related to alcohol withdrawal. His compliance
on the medication was unclear. Initially, at triage he
said he had stopped taking his phenytoin two days
earlier and had flushed the remainder of the prescription down the toilet that day. He then changed
his story and said he had taken three 100-mg tablets
on the day of presentation to the ED.
When he was examined by the emergency physician, he stated again that he had taken phenytoin
within the last 24 hours. He described a seizure the
day before during which he lost consciousness for
20 minutes and had urinary incontinence. He had also
experienced some visual symptoms, reporting scintillations with red and black dots, headache, leg
numbness, abdominal discomfort, and chest pain.
He appeared apprehensive and in some distress. He
was fully oriented, with a normal neurological
examination, and there were no significant findings
on a complete physical examination. Routine blood
work, a drug screen, and phenytoin level were
ordered. An IV line was established.
The phenytoin level was reported to the ED as less
than 12 mmol/L or none detected (therapeutic range:
4080 mmol/L), and the physician ordered a loading
dose of 1 g IV over one hour, which was started at 2150
hours. At 2240 hours, on nursing assessment, the
patient was found to have vomited and appeared
disoriented. It was thought he might have had an
unwitnessed seizure and was now postictal. He was
reassessed by the emergency physician, who ordered
pantoprazole 40 mg IV and 10 mg diazepam IV. A
second infusion of phenytoin was started, and a second
phenytoin level was ordered. According to the nurses
notes, at 0030 hours he had a witnessed complex/
partial seizure, followed by vomiting. At 0040 hours,
the laboratory reported the phenytoin level at 228
mmol/L, and the second infusion of phenytoin was
immediately stopped. He was placed on a cardiac
monitor and an oxygen saturation monitor, and neurovitals were done hourly. For the remainder of the night
he had disorientation, with intermittent vomiting. His
cardiac and respiratory status remained stable.
At 0700 hours, a CT scan of the head was done and
reported as normal. He was reassessed by an
oncoming physician. His vital signs were normal.
The patient was lethargic but rousable. He had

295
slurred speech, ataxia, nystagmus, and dysarthria.
Later, he was seen to exhibit some choreoathetoid
movements. He was transferred to the ICU. His
phenytoin level peaked at 259 mmol/L at 1310 hours.
The intensivist ordered a repeat phenytoin level on
the first sample that had been taken in the ED, to be
done instead at the laboratory of the local tertiary care
hospital, and this came back at 185 mmol/L. The
patient appeared to suffer no further deterioration.
His neurological status normalized over several days,
and he made an uneventful recovery.
Comment. This case illustrates a variety of errors. The
first is anchoring.7 Both the triage nurse and the
physician appear to accept the patients account of
seizures. Self-reports of seizures and periods of
unwitnessed loss of consciousness are notoriously
unreliable. In fact, it appears clear from the patients
account that the fall was not associated with a seizure
and was probably due to ataxia from phenytoin
toxicity. However, once the physician and nurses had
anchored to the idea that the patient had been having
seizures, then the none detected phenytoin result
would be readily accepted and incorporated into
a prematurely closed diagnosis,7 an example of
confirmation bias.7
The next error occurs in the laboratory. The three
phases of laboratory error are preanalytic, analytic, and
postanalytic (for a recent review, see Bonini et al.19).
The present case is an example of an analytic error (i.e.,
it occurs during the stage of specimen analysis). This is
the phase during which errors are least likely.19
Following the discovery of the error, the laboratory
technician recalled getting a code of ** NR when the
sample was run. The NR refers to No Result and
indicates that the level is outside the dynamic range,
either too low or too high. The procedure then is to
dilute the sample and repeat the test. However, the
code was misunderstood by the technician, read as
not detectable, and reported as such.
The laboratory error is further compounded by
assumptions and subsequent events in the ED. The
patient had told both the triage nurse and the
physician that he had taken his usual dose of
phenytoin (3 3 100 mg) that day. Thus, when the
phenytoin result was reported as none detected, it
should have been questioned immediately. (In fact,
the patient had actually overdosed on phenytoin
before presenting to the ED). If another patient simply
had presented with a seizure disorder despite taking
antiseizure medication, the laboratory value would
not have gone unchallenged. However, this patients
psychiatric diagnosis of personality disorder, coupled
with the inconsistent history, resulted in the laboratory values being given more credence, and he
subsequently received a loading dose of phenytoin
on top of an already toxic phenytoin state. Credibility
of the patient is one of a number of errors uniquely

296

Croskerry et al.

associated with the management of psychiatric

patients in the ED.15 Overall, whenever the laboratory
value is unexpected, inconsistent with the history
provided, or inconsistent with other data, a forcing
function should be applied and the value challenged.
Finally, the continued seizure-like activity in this
case was unlikely due to true seizures but, instead, to
some manifestation of phenytoin-induced dyskinesias
(tremor, chorea, asterixis, and dystonia). The order to
give a second bolus of phenytoin in these circumstances would push the patient toward the lower end
of the lethal range (25 g), and, therefore, is a prescription error due to a knowledge deficiency. In the
face of presumed continuing seizure activity, an
adjunctive antiseizure medication such as phenobarbital would have been indicated.

CASE 12
A 62-year-old man was brought to a community
hospital by ambulance. About an hour earlier he had
developed severe midline chest pain radiating between his shoulder blades. It was associated with
numbness in both arms, dyspnea, diaphoresis, and
vomiting. He was taken immediately to the cardiac
room where vital signs, blood work, and a 12-lead
electrocardiogram (ECG) were done. A second IV line
was started. His blood pressure was 180/100 mm Hg,
pulse 110 beats/min, and respiratory rate 22 breaths/
min. The ECG showed significant ST-segment elevation in several leads. The cardiac nurse took it to the
physician, a family practitioner who did regular shifts
in the ED. The nurse asked whether she should start
mixing the thrombolytic. The physician agreed and
went to assess the patient.
The patient was in significant distress. He was pale
and sweaty. He was taking no medications other than
antihypertensives and had no cardiac history. The
nurses had prepared the patient for thrombolysis and
asked for permission to proceed. The physican had
thrombolysed a patient only twice before and asked
for a call to be placed to the cardiologist on call. He
ordered morphine for the patient. The cardiologist,
who was out of hospital, responded promptly and
discussed the case with the physician. On the basis of
the history he was given, and a description of the ECG
findings, he advised that the patient was likely
experiencing an acute myocardial infarct and the ED
could go ahead with thrombolysis or wait until the
cardiologist arrived in about 30 minutes. The physician decided to go ahead.
The cardiologist arrived in about 25 minutes, by
which time a bolus of thrombolytic had been given,
and an infusion started. A portable chest x-ray was
done but the results were not yet available. While he
was reviewing the ECG, the patient experienced
cardiac arrest. All efforts at cardiopulmonary resuscitation failed and he was declared dead. The

MEDICATION ERRORS IN THE ED

chest x-ray was reviewed and showed some dilation

of the aortic shadow. A clinical autopsy revealed
a proximal dissection of his thoracic aorta.
Comment. This is a prescription error. The wrong
medication was given to the patient because the
wrong diagnosis was made. It is also a knowledgebased error. Several factors contributed to the wrong
diagnosis. First, the physician is unfamiliar with the
uncommon, but not rare, presentation of acute
thoracic dissection. There is a clear imperative to
exclude it on the differential for chest pain.20 Second,
he is inexperienced with a medication that is tightly
coupled2 (see case 9). Thrombolytic agents are timedependent, their administration is relatively inflexible, and they have the potential for rapid and
devastating action. He admitted later that he had felt
pressured to act by the nurses, who, in turn had been
pressured at a recent in-service to reduce their doorto-needle time. These team-coercive pressures17
and the unintended consequences of achieving
quality indicators21 are known to lead to error. More
thorough assessment of the patients arm numbness, comparing blood pressures in both arms, and
reviewing the chest x-ray might have offset the early
search satisficing7 and anchoring7 to the incorrect
diagnosis of acute myocardial infarct. In tightly
coupled situations that carry the potential for a catastrophic outcome, cognitive forcing strategies should
be used to improve patient safety.20

CASE 13
A 6-month-old female with no significant medical or
surgical history was brought to the ED with a fever
(maximum body temperature 1028F), overnight irritability for two nights, and having vomited once that
day. On examination, a bulging fontanel was noted, as
well as a new flat, blanching rash. In the ED, the
patients temperature was 101.48F (rectal), pulse was
160 beats/min, and respiratory rate was 40 breaths/
min. A lumbar puncture was ordered to rule out
meningitis. The patients weight on the ED record was
listed as 15.6 kg, so the patient was given acetaminophen 240 mg rectally and ceftriaxone 1.5 g intravenously.
It was subsequently discovered that the patients
weight of 15.6 that was entered in the space marked
for kilograms was actually pounds. The patients
actual weight was 7.1 kg. This error resulted in
a twofold dosing error of both acetaminophen and
ceftriaxone. The patient appeared to suffer no adverse
effects from these two overdoses.
Comment. The pediatric chart in this ED had a box
for the childs weight in kilograms only. However, the
weighing scales had the option for weighing in either
pounds or kilograms. In this case, the nurse weighed

ACAD EMERG MED

March 2004, Vol. 11, No. 3

www.aemj.org

the patient in pounds, a unit of measurement with

which she was more familiar, and entered this weight
in the kilogram box. The weight of a child is critically
important in pediatric dosing, and strategies have
been developed to reduce cognitive errors associated
with weighing.22 Where direct weighing is necessary,
it is clearly important to adopt one unit of weight only
in the ED to remove any ambiguity.

CASE 14
A 40-year-old man with no significant medical history
and taking no medications came to the ED complaining of right-sided chest wall pain intermittently
for one week. The pain was exacerbated by deep
inspirations and sometimes by light activity. No
shortness of breath, retractions, or nasal flaring was
noted, and the monitor showed normal sinus rhythm.
Blood pressure was 140/80 mm Hg, pulse 84 beats/
min, respiratory rate 16 breaths/min, and pulse
oximetry 99%. Tests included complete blood count
(CBC), Chem 7, liver function tests (LFTs), chest x-ray,
ECG, and ventilation-perfusion (V-Q).
The laboratory later called to report a critical result
of a blood glucose of 39 mg/dL. The patient was
immediately given 1 ampule of 50% dextrose (25 g in
50 mL). Approximately one hour later, the laboratory
notified the ED that the result previously reported
was for the wrong patient. This patients correct
glucose value was 94 mg/dL. No adverse effects
resulted from the error.
Comment. This is an example of a postanalytic
laboratory error, involving patient identification. Postanalytic errors constitute a relatively small proportion
of all laboratory errors,19 but ED personnel should be
vigilant for them. In this case, the erroneous laboratory
value is accepted uncritically by the ED and acted on.
Whenever an abnormal laboratory value is received
that is inconsistent with the patients clinical assessment, it should always be challenged and, if possible,
repeated before action is taken (e.g., potassium values
that are aberrant due to hemolysis or sampling procedures are not uncommon in EDs).

CASE 15
A 5-week-old female was brought to the ED at 1754
hours with a history of fever since the previous
evening. She had been irritable during the day and
had refused to feed. Her 2-year-old sibling had been
unwell a few days earlier with a presumed viral
respiratory and gastrointestinal illness. The babys
medical history was unremarkable. She had been born
at term by caesarian section because of maternal hip
dislocation. Her birth weight was 3.4 kg.
At triage, she appeared mottled with a full fontanel.
HR was 190 beats/min, respiratory rate 52 breaths/

297
min, O2 saturation 99%, and rectal temperature
40.38C. Her weight was 4.7 kg. She was brought into
the department at 1800 hours and seen by a pediatric
PGY2 resident at 1810 hours. On examination she was
irritable. The head and neck examination was normal,
chest clear, cardiovascular examination normal, and
abdomen soft and nontender. She had a fine maculopapular rash over her abdomen. A full septic
workup was performed, including a catheter urine
specimen, blood culture, and lumbar puncture.
Ampicillin 230 mg and gentamicin 35 mg IV were
ordered by the resident. There was no documentation
on the ED record from the staff physician. The
gentamicin was given in the ED at 2025 hours. This
dose was three times the usual dosage. Ampicillin
was not given until the baby arrived on the ward at
2250 hours. The admitting orders were written by
a second PGY4 resident who transcribed the same
doses of antibiotics. The nurse on the ward identified
the error, and a verbal order was received to hold
gentamicin for two doses and to reduce the dose to 12
mg every eight hours. No further gentamicin was
given, and cefotaxime and acyclovir were added to
the treatment.
The medication error was disclosed to the family.
The father, who worked in quality control in the
airline industry, remained very angry throughout the
hospital stay. Hospital quality resources were involved in the case. A final diagnosis of enteroviral
meningitis was made, and the baby made an uneventful recovery. Arrangements were made to follow
her for a potential hearing loss, but it was anticipated
that causality with the medication error might prove
difficult to establish because viral meningitis is also
associated with hearing deficits.

Comment. The primary error is one of dose calculation, which is common in pediatrics, and classified as
a rule-based mistake. In this case, a threefold error
was made because the first resident calculated the
total dose for 24 hours but then wrote it as
gentamicin 35 mg/IV/q8h. For this hospitals
formulary, the recommended total daily dose for this
weight is 7.5 mg/kg/day. The error would have been
less likely if a forcing function had been applied,
requiring this order to be written as total daily dose
of 35 mg, based on 7.5 mg/kg/day, to be given as 12
mg IV q8h. The obvious discrepancy with the childs
weight would then more likely have been detected.
The transcription of the same dose by the second and
more senior resident is a form of copycat error. It may
have been due, in part, to fatigue, which is known to
contribute to error.23 She had been working since 0800
hours and had already been at work for 15 hours when
she wrote the admitting orders. There is also a timedelay error in that with a working diagnosis of meningitis, the first dose of antibiotic was administered

298

Croskerry et al.

MEDICATION ERRORS IN THE ED

TABLE 2. The Medication Process, Error-producing Conditions, and Prevention Strategies in the
Emergency Department (ED)*
Stage
Prescribing

.
Transcribing

.
Dispensing

.
Administration

.
Monitoring

.
Discharge drugs

Error-producing Conditions

Prevention Strategies

Incomplete knowledge of drug

Incomplete knowledge of patient

Readily available drug reference systems

Pharmacist availability in the ED
Take thorough medication/medical/allergy history
Physician order entry
Computerized decision support
Pediatric patients
Determine accurate weight in kilograms
Be alert for calculation/decimal point errors
Caution with off-label prescribing
Geriatric patients
Consider comorbidities and drugdrug interactions in particular
Consider possibility of falls with new medications
Consider renal and hepatic function
Pregnant patients
Rule out pregnancy if a possibility
Careful evaluation of riskbenefit in pregnant patients
Avoid verbal orders except for emergencies
Write legibly; print if necessary
Electronic order transcription
Attend carefully to drugs of like-sounding name
Avoid acronyms or abbreviations
Indicate decimal point clearly
No trailing zeros
Avoid apothecary terms
Include physician phone no. and pager no. for patients leaving ED
Nurses dispense to another nurse and not themselves
No dispensing by physicians
Automated dispensing
Check with emergency physician for any ambiguity in order
Check correct placement of decimal points
Check that weight is correct in kilograms
Always check for allergies to drug class
Doubly confirm patient identification; bar-coding
Double-check arithmetic
Be prepared to challenge orders
Clarify if any ambiguity or doubt concerning medication order
Consistent consultation with reference materials
Consultation with hospital pharmacist if available
Call back verbal orders
Implement systematic safety checks
Avoid physician administration of drug wherever possible
Ensure adequate monitoring technology
Ensure adequate monitoring personnel
Clear ED protocols for conscious sedation, rapid sequence
intubation, etc.
Ensure adequate monitoring time following drug administration
No verbal discharge instructions to patients given amnestic
Provide written information on drug to patient if possible
Advise patients of any necessary follow-up after ED visit
Consider effect of drug if patient driving home from ED

Verbal orders
Poor penmanship
Team communication errors

Dispensing by nurses
Dispensing by physicians
Patient ID

Multiplicity of drugs used

Potency of drugs
Multiple patients in ED
Parenteral administration
Drug incompatibilities
Physician administration
Potent drugs
Parenteral administration
Emergent procedures
Complex procedures
Medicated patients leaving ED

*Based on Bates3 and Peth.4

more than two hours after the patient was first seen
by the resident, and the second antibiotic dose was
not given until more than four hours after. This error
proved inconsequential because the final diagnosis
was viral meningitis.

CONCLUSIONS
This series illustrates a variety of adult and pediatric
medication errors, occurring in each of the five stages

described by Bates.3 That they were gathered fairly

effortlessly by simply asking colleagues if they could
recall a recent medication error indicates they are
probably abundant in the ED. However, the magnitude of the problem is presently unknown because,
for a variety of reasons, many would not be recorded
in incident reports. The principal error-producing
conditions and strategies to avoid them are summarized in Table 2. A sixth stage has been added to the
schema of Bates to emphasize the final part of the

ACAD EMERG MED

March 2004, Vol. 11, No. 3

www.aemj.org

medication administration process in the ED, drawing

attention to considerations that should be made for
patients being discharged home.
The unique and often prevailing conditions in many
EDs predispose to these errors, and ED personnel
should be aware of their possible antecedents. A
particular operating feature of EDs, the capability for
dispensing medication without surveillance by a pharmacist, provides an error-producing condition to
which physicians and nurses should be constantly
vigilant. A further problem is that in some of these
cases the error arose because a physician was involved
in administration of the medication. In general,
physicians should administer medication only in
conscious sedation procedures, similar to the role of
an anesthesiologist. Wherever possible, the responsibility for medication administration should be restricted to nurses. Adherence to defined roles would
reduce the team communication errors that are
a common theme here. Finally, in several of the cases
described, recovery patterns are evident (i.e., nurses
and physicians react to and may often correct the error
and its consequences). This would be expected, of
course, because EDs are uniquely equipped to deal
with the aftermath of error. However, it should be
emphasized that although many errors provide an
opportunity for recovery, they all carry the potential for
learning, provided that error management is judicious
and applied in an appropriate educational milieu.
The administrative assistance of Sherri Lamont at the Emergency
Department at Dartmouth General Hospital is appreciated.

References
1. Johnson JA, Bootman JL. Drug-related morbidity and mortality
and the economic impact of pharmaceutical care. Am J Health
Syst Pharm. 1997; 54:5548.
2. Perrow C. Normal Accidents. Princeton, NJ: Princeton University Press, 1999.
3. Bates DW. Using information technology to reduce rates of
medication errors in hospitals. BMJ. 2000; 320:78891.
4. Peth HA. Medication errors in the emergency department:
a systems approach to minimizing risk. Emerg Med Clin North
Am. 2003; 21:14158.

299
5. Reason J. Human Error. New York, NY: Cambridge University
Press, 1990.
6. Lesar TS, Briceland L, Stein DS. Factors related to errors in
medication prescribing. JAMA. 1997; 277:3127.
7. Croskerry P. Achieving quality in clinical decision making:
cognitive strategies and detection of bias. Acad Emerg Med.
2002; 9:1184204.
8. Croskerry P. The importance of cognitive errors in diagnosis
and strategies to minimize them. Acad Med. 2003; 78:
16.
9. Kohn LT, Corrigan JM, Donaldson MS (eds). To Err Is Human.
Building a Safer Health System. Washington, DC: National
Academy Press, 1999.
10. Turnbull JE. A systems approach to error reduction in health
care. Jpn Med Assoc J. 2001; 44:392403.
11. Morath JM, Turnbull JE. To Do No Harm: Ensuring Patient
Safety in Health Care Organizations. San Francisco, CA: JosseyBass, 2004.
12. Chisholm CD, Collison EK, Nelson DR, Cordell WH. Emergency department workplace interruption: are emergency
physicians interrupt-driven and multitasking? Acad
Emerg Med. 2000; 7:123943.
13. Croskerry PG. The cognitive imperative: thinking about how
we think. Acad Emerg Med. 2000; 7:122331.
14. Croskerry P, Sinclair D. Emergency medicine: a practice prone
to error? Can J Emerg Med. 2001; 3:2716.
15. Croskerry P, Wears RL. Safety errors in emergency medicine.
In: Markovchick VJ, Pons PT (eds). Emergency Medicine
Secrets. 3rd ed. Philadelphia, PA: Hanley and Belfus, 2003.
16. Koren G, Haslam RH. Pediatric medication errors: predicting
and preventing tenfold disasters. J Clin Pharmacol. 1994;
34:10435.
17. Williams JC. Assessing & Reducing the Likelihood of Violation
BehaviourA Preliminary Investigation. Proceedings of an
International Conference on the Commercial & Operational
Benefits of Probabilistic Safety Assessments. Edinburgh,
Scotland: Institute of Nuclear Engineers, October 1997.
18. Risser DT, Rice MM, Salisbury ML, et al. The potential for
improved teamwork to reduce medical errors in the emergency
department. Ann Emerg Med. 1999; 34:37383.
19. Bonini P, Plebani M, Ceriotti F, Rubboli F. Errors in laboratory
medicine. Clin Chem. 2002; 48:6918.
20. Croskerry P. Cognitive forcing strategies in clinical decision
making. Ann Emerg Med. 2003; 41:11020.
21. Croskerry P, Chisholm C, Vinen J, Perina D. Quality and
education. Acad Emerg Med. 2002; 9:110815.
22. Luten R, Wears R, Broselow J, Croskerry P, Joseph M, Frush K.
Managing the unique size related issues of pediatric
resuscitation: reducing cognitive load with resuscitation aids.
Acad Emerg Med. 2002; 9:8408.
23. Gaba DM, Howard SK. Fatigue among clinicians and the safety
of patients. N Engl J Med. 2002; 347:124955.