You are on page 1of 7

ORIGINAL INVESTIGATION

Medication Errors Observed


in 36 Health Care Facilities
Kenneth N. Barker, PhD; Elizabeth A. Flynn, PhD; Ginette A. Pepper, PhD;
David W. Bates, MD, MSc; Robert L. Mikeal, PhD

Background: Medication errors are a national con- Clinical significance was judged by an expert panel of
cern. physicians.

Objective: To identify the prevalence of medication er- Main Outcome Measure: Medication errors reach-
rors (doses administered differently than ordered). ing patients.

Design: A prospective cohort study. Results: In the 36 institutions, 19% of the doses (605/
3216) were in error. The most frequent errors by cat-
Setting: Hospitals accredited by the Joint Commission egory were wrong time (43%), omission (30%), wrong
on Accreditation of Healthcare Organizations, nonac- dose (17%), and unauthorized drug (4%). Seven per-
credited hospitals, and skilled nursing facilities in Geor- cent of the errors were judged potential adverse drug
gia and Colorado. events. There was no significant difference between er-
ror rates in the 3 settings (P=.82) or by size (P=.39). Er-
Participants: A stratified random sample of 36 insti- ror rates were higher in Colorado than in Georgia (P=.04).
tutions. Twenty-six declined, with random replace-
ment. Medication doses given (or omitted) during at least Conclusions: Medication errors were common (nearly
1 medication pass during a 1- to 4-day period by nurses 1 of every 5 doses in the typical hospital and skilled nurs-
on high medication–volume nursing units. The target ing facility). The percentage of errors rated potentially
sample was 50 day-shift doses per nursing unit or until harmful was 7%, or more than 40 per day in a typical 300-
all doses for that medication pass were administered. patient facility. The problem of defective medication ad-
ministration systems, although varied, is widespread.
Methods: Medication errors were witnessed by obser-
vation, and verified by a research pharmacist (E.A.F.). Arch Intern Med. 2002;162:1897-1903

T
HE 1999 Institute of Medi- searchers. They found that 56% of the
cine report1 on the quality events they detected were due to prescrib-
of care, entitled To Err Is ing errors and 44% involved administra-
Human: Building a Safer tion. Obviously, drug therapy cannot be
Health System, has drawn successful unless prescribing and deliv-
From the Center for Pharmacy
national attention to the occurrence, clini- ery and administration are performed cor-
Operations and Designs, School
of Pharmacy, Auburn cal consequences, and cost of adverse drug rectly.
University, Auburn, Ala events (ADEs) in hospitals. The report calls A key variable in assessing the medi-
(Drs Barker and Flynn); the for more systematic approaches to the pre- cation system in health care facilities is
School of Nursing, University of vention of injuries due to medical care. whether the patient receives the pre-
Colorado Health Sciences Many of these ADEs are viewed as origi- scribed medication. A medication error was
Center, Denver (Dr Pepper); nating from systems problems (ie, prob- defined for this study as a discrepancy be-
the Division of General Internal lems with the processes of the medica- tween the dose ordered and the dose re-
Medicine and Primary Care, tion use system). We divide those ceived. This definition takes a systems view
Brigham and Women’s processes into (1) prescribing and (2) de- of medication error, because the focus is
Hospital, and the Center for
livery and administration. The focus of this on the system outcome rather than on the
Applied Medical Information
Systems, Partners Healthcare article is on the latter. actions of individual health care work-
and Harvard Medical School, Leape and associates2 studied ADEs ers. Medication error is operationalized as
Boston, Mass (Dr Bates); and involving medications using methods that an easily understood rate that is simply cal-
DACE Co, West Monroe, La included solicited self-report and daily culated: (doses in error/total doses given
(Dr Mikeal). medical record review by clinical nurse re- or omitted) ⫻ 100. This measure of medi-

(REPRINTED) ARCH INTERN MED/ VOL 162, SEP 9, 2002 WWW.ARCHINTERNMED.COM


1897

©2002 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a New York University User on 05/21/2015


cation error rate has been extensively used to test hy- counted for 21% of all acute-care hospitals in the United States
potheses about system improvements. For example, it was in 1998, with a mean bed size of 67 (median, 44).10 In the sample,
used to evaluate the impact of the unit dose system that the mean bed size was 48, compared with 268 for the accredited
was ultimately adopted by 90% of US hospitals.3-6 hospitals. A consequence was that the research team (E.A.F. and
This report is part of a study to seek the best method G.A.P.) arriving on the previously negotiated day sometimes en-
countered fewer than 50 doses for study, and sometimes none
for detecting and counting the frequency of medication for several days, due to unanticipated changes in the census. In
errors in US hospitals and skilled nursing facilities, com- contrast, in the larger accredited hospitals and skilled nursing fa-
paring validity with cost-effectiveness. This article re- cilities, the workload of doses offered many more than 50 doses
ports the errors verified by a research pharmacist (E.A.F.) per day for study, at minimum incremental cost. These addi-
using the observation method data as a gold standard in a tional data were collected to achieve a better description of the
study comparing different methods and data collectors. Ob- error rate in that facility type.
servation was superior to medical record review, and to
the examination of incident reports.7 RECRUITMENT AND TRAINING
Observation uses as the primary outcome measure OF DATA COLLECTORS
the percentage of doses ordered that are in error when
administered to the patient (or omitted). Uses of this mea- Two registered nurses, 2 licensed practical nurses, and 2 phar-
macy technicians per state were sought from the general popu-
sure have included benchmarking to help hospitals test lation by placing advertisements in the newspapers and on the
and evaluate new systems (eg, unit dose) in 40 studies, Internet in Denver and Atlanta. Only 1 pharmacy technician
comparing with “best practice” hospitals, evaluating ex- was hired in Colorado because of a lack of qualified appli-
pensive interventions (eg, automated pharmacy sys- cants. Applicants took a qualifying test to determine their base
tems) before and after installation, and enforcing gov- knowledge of medication and administration techniques.
ernmental standards and regulations.8,9 Training in the observation technique required 20 hours
Because this measure may vary by accreditation sta- and included classroom lectures, an interactive videotape pro-
tus, whether the site is an acute-care or skilled nursing gram, practice observations on a nursing unit, and 2 examina-
facility, or geographic location, we performed a study to tions. Additional practice observations were performed after
assess the medication error rate in various hospitals and training. One registered nurse in Georgia withdrew after train-
skilled nursing facilities in 2 states. ing for personal reasons.
The final examination included a paper test of the observ-
er’s ability to detect errors when provided with a typed list of
PARTICIPANTS AND METHODS drugs administered and a typed set of drug orders for the pa-
tients involved. A test set of 49 doses, which included 27 er-
SAMPLE rors and 22 nonerrors, was constructed. The frequency of each
error type was proportional to the occurrence of error catego-
The areas from which the samples of each type of facility were ries in 12 previous observation-based studies (wrong time, 16;
drawn were the Atlanta, Ga, metropolitan statistical area and omissions, 8; wrong dose, 2; and unauthorized drug, 1). The
the Denver-Boulder-Greeley, Colo, consolidated metropoli- scores on the examination served as the basis for interrater and
tan statistical area, using lists provided by the Health Care Fi- intrarater reliability assessments. The percentage agreement on
nancing Administration. Data provided for each facility in- each question was used to calculate the interrater reliability score.
cluded address, telephone number, accreditation status, and bed A repeated-measures analysis of variance was performed on the
size. From these lists, 18 facilities were randomly selected for split-halves test scores to determine intrarater reliability.
each of 3 facility types in each state: 6 accredited hospitals, 6
nonaccredited hospitals, and 6 skilled nursing facilities, for a DATA COLLECTION PROCEDURES
total of 36 sites. Facilities were invited to participate via letter
and telephone. When a facility declined, the hospital or skilled Direct observation was used to detect medication errors, based
nursing facility in the same positional order in the next ran- on the method of Barker and McConnell.11 An orientation to
dom sample was contacted in turn until enough facilities of that each site was provided by facility personnel before observa-
category agreed to participate. Facilities were required to have tions started. On each day of observation, the observer arrived
an incident report system in place (and all approached did). A on the nursing unit in time to attend the change-of-shift re-
minimum bed size requirement of 24 was established after it port, to meet the staff and allow nurses to ask questions about
was found that nonaccredited hospitals with fewer beds often the study. An information sheet approved by the Auburn Uni-
had too few patients; 6 such hospitals had to be excluded. versity Institutional Review Board was provided to the nurse
Based on previous experience, a sample size of 50 doses subjects. The observer witnessed the preparation and admin-
per nursing unit was chosen as large enough to obtain an ad- istration of 50 doses by the first nurse encountered plus a sec-
equate measure of an observation-based error rate for each of ond nurse if necessary. The period for the observation was 2
the 36 facilities. The doses were those occurring during a medi- hours, or until all doses due were administered. The observer
cation pass on a nursing unit identified as high volume by an wrote down exactly what the subject did, including all details
official of the facility. Up to 4 different nursing units were in- about the medication, and witnessed the administration to the
cluded if available at each site, so that 200 doses per facility patient. Data recorded included patient names (which were later
could have been observed. coded), drug product, amount of drug, dose form, route of ad-
The nonaccredited hospitals presented special sampling ministration, time of administration, and medication-related pro-
problems. Five achieved accreditation status during the study cedures (such as measuring the patient’s heart rate or giving
period (7 months). The judgment was made that the data from with food). After the medication pass, the observer and re-
these hospitals should be analyzed as nonaccredited and then search pharmacist made their own independent copies of the
as accredited. original medication orders for patients involved in the obser-
Another problem was the small number of doses per day in vation. Each dose observed was compared with what the pre-
some of the nonaccredited facilities. Nonaccredited hospitals ac- scriber ordered. If there was a difference, the error was de-

(REPRINTED) ARCH INTERN MED/ VOL 162, SEP 9, 2002 WWW.ARCHINTERNMED.COM


1898

©2002 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a New York University User on 05/21/2015


scribed and categorized. After comparing all doses witnessed, 7. Wrong technique: exclusion, or incorrect perfor-
the observer determined if any other drugs should have been mance, of a procedure ordered by the prescriber immediately
given at the time of the observation based on what the pre- before administration of each dose of medication. Examples in-
scriber ordered. If any were identified, they were recorded as clude lack of heart rate or blood pressure measurement before
omission errors unless a valid reason was discovered. Doses given giving a dose.
based on orders judged difficult to interpret were excluded from 8. Wrong time: administration of a dose more than 60
the study (0.2% of the orders were deemed uninterpretable). minutes before or after the scheduled administration time. A
The medication error rate was calculated as follows: [(number 30-minute window was used for medications that were or-
of errors, with no more than 1 error per dose)/(number of doses dered before, with, or after a meal. Routine administration
given+number of omissions)]⫻100. times were obtained from each site, and times assigned on the
After the observer finished the error determination, all data medication administration record were used when no other
were turned over to a research pharmacist. The researcher made policy was available.
a blinded independent determination of errors by comparing Each dose observed to be given or omitted was operation-
each dose on the observer’s drug pass worksheet with the phar- ally defined to be a dose (ie, opportunity for error), and is the
macist’s copy of the prescriber’s orders (correcting 210 false basic unit of data. Any dose could be only in error or not in er-
negatives and 87 false positives). The research pharmacist for ror. Doses included only those for which the preparation and
the Colorado area (E.A.F.) reviewed the data collected at the administration of the medication were witnessed by an ob-
Georgia sites to address inconsistencies. Only doses con- server or that the observer was certain were not administered
firmed as in error or not in error by the research pharmacist (ie, omitted). Doses labeled by the pharmaceutical manufactur-
are reported herein. ers were assumed to be correct.

DEFINITIONS STATISTICAL METHODS

A medication error was defined in general as a dose adminis- The overall medication error rates (with and without wrong
tered differently than as ordered on the patient’s medical rec- time errors) for each site were compared between states, facil-
ord. Such medication errors were viewed as system defects (ie, ity types, accreditation status, and facility size categories us-
outcomes different from those the system was designed to de- ing an analysis of variance. The Tukey test was used to deter-
liver and administer to the patients). Categories of medication mine the means between which significant differences existed
errors were defined as follows. in the comparison of facility types. Computer software (SAS
1. Unauthorized drug: the administration of a dose of medi- statistical software for Windows, version 6.12; SAS Institute Inc,
cation that had never been ordered for that patient. Cary, NC) was used. The ␣ level was set at .05.
2. Extra dose: any dose given in excess of the total number
of times ordered by the physician, such as a dose given based on ASSESSMENT OF POTENTIAL FOR HARM
the expired order, after a drug had been discontinued, or after a
drug had been put on hold. A potential ADE was defined in general as a medication error
3. Wrong dose: any dose of preformed dosage units (such that had the potential to cause a patient discomfort or jeopar-
as tablets) that contained the wrong strength or number; if an dize the patient’s health and safety. In this study, the opera-
injectable product, then any dose that was ±10% or more dif- tional definition was the expert judgment (and majority deci-
ferent from the correct dosage; if any other dosage form, then sion) of a 3-physician advisory panel, each experienced in
any dose that was ±17% or more of the correct dose in the judg- making such judgments, who evaluated the same descriptive
ment of the observer. In judging dosage, measuring devices and information for each medication error detected. Health Care
graduations were those provided for routine use by the insti- Financing Administration guidelines (available from the au-
tution: graduations on the syringe for injections, graduations thors) for judging significance were provided to the panel.
on medicine cups for oral liquids, and drops for the dropper The information sent to the physicians’ panel (at
provided. Wrong dose errors were counted for ointments, topi- Brigham and Women’s Hospital) included a description of
cal solutions, and similar medications only when the dose was each individual error, the drug involved, and the error cat-
specified quantitatively by the prescriber (eg, in inches of egory. The information excluded the data collection method
ointment). used and the data collector type so as to blind the panel to
4. Omission: failure to give an ordered dose. If no at- these factors. Information sent about each patient’s condition
tempt was made to administer the dose, an omission error was included sex, age, allergies, disease states, selected laboratory
counted. If the patient refused the medication, an opportunity data if associated with a medication, red flag drugs ordered,
for error was not counted provided the nurse responsible for and physician or nurse progress notes when deemed notewor-
administering the dose tried to give it. Doses withheld accord- thy by the research pharmacist reviewing the patient’s medi-
ing to policies calling for the withholding of medication doses, cal record.
such as nothing by mouth before surgery, were not counted as An institutional review board application was submitted
errors or opportunities for errors. Omissions were detected by and approved by Auburn University and the Colorado Mul-
comparing the medications administered at a given time with tiple Institutional Review Board.
doses that should have been given at that time based on the
physician’s written order and protocols. RESULTS
5. Wrong route: medication administered to a patient us-
ing a different route than ordered (eg, oral administration of a Overall percentage agreement on the Drug Pass Exami-
drug ordered intramuscularly). Included in this category were
doses given in the wrong site, such as the right eye instead of
nation was 96%, with the range of agreement on each in-
the left eye. dividual question between 89% and 100%, indicating in-
6. Wrong form: the administration of a dose in a differ- terrater reliability. The result of the repeated-measures
ent form than ordered by the physician. If enteric-coated as- analysis of variance test found that the split-halves test
pirin was ordered, but plain aspirin was administered, a wrong scores were not significantly different within subjects, in-
form error was counted. dicating intrarater reliability (F1,8 =2.26, P = .0541).

(REPRINTED) ARCH INTERN MED/ VOL 162, SEP 9, 2002 WWW.ARCHINTERNMED.COM


1899

©2002 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a New York University User on 05/21/2015


Table 1. Error Rates by Error Category and Facility Type*

Facility Type

Error Category Overall Accredited Hospitals Nonaccredited Hospitals Skilled Nursing Facilities
Omission 183 (6) 73 (5) 8 (3) 102 (7)
Wrong dose 103 (3) 46 (3) 12 (4) 45 (3)
Unauthorized drug 22 (1) 12 (1) 2 (1) 8 (1)
Wrong form 20 (1) 10 (1) 2 (1) 8 (1)
Extra dose 10 (0) 7 (0) 0 (0) 3 (0)
Wrong route 6 (0) 1 (0) 1 (0) 4 (0)
Wrong technique 2 (0) 0 (0) 0 (0) 2 (0)
Wrong time 259 (8) 85 (6) 31 (11) 143 (10)
Total errors 605 (19) 234 (16) 56 (20) 315 (22)
No error 2611 (81) 1247 (84) 228 (80) 1136 (78)
Total Doses 3216 (100) 1481 (100) 284 (100) 1451 (100)

*Data are given as number (percentage) of doses for each facility type. Percentages are rounded to the nearest whole number.

The mean error rate detected in the 36 sites in At- The 3-physician panel rated 7% of the errors de-
lanta and Denver was 19% (605 of 3216 doses). Exclud- tected (48 of 675 errors assessed) as potential ADEs. When
ing wrong time errors, the error rate was 10%. The range wrong time errors are excluded, 10% of the errors were
was 0% to 67%, with a 95% confidence interval of ±4.5%. considered potential ADEs (45 of 448 errors). Table 4
All data were collected during 81 observation days from lists examples of those errors rated as potential ADEs.
May 4 to November 11, 1999. The error rates by cat- Table 5 shows the potential clinical significance for each
egory (Table 1) demonstrate that the most frequent er- error type.
rors were wrong time (8%), omission (6%), and wrong
dose (3%); as a percentage of all errors, the results in-
cluded wrong time (43%), omission (30%), wrong dose COMMENT
(17%), and unauthorized drug (4%).
The distribution of error rates by error category was The results show that medication errors were common,
similar between accredited and nonaccredited hospitals and occurring in 19% or nearly 1 of every 5 doses in the typi-
skilled nursing facilities (Table 1). When rate by site was cal site. Assuming 10 doses per patient day, this would
compared (Table 2), however, substantial variation be- mean the typical patient was subject to about 2 errors
tween sites was found, with error rates ranging from 0% every day. There was substantial variation by site and re-
at one site to 66.7% at another. To help maintain ano- gion, however; therefore, the results can only be de-
nymity of the sites, size was categorized as large (⬎100) scribed for the sample observed.
and small (ⱕ100) based on number of certified beds.
There was no significant difference in error rates by POTENTIAL FOR HARM
type or size of facility. The statistics comparing the 3 types
of sites were as follows: F2,33 = 0.20, P = .82; and exclud- A panel of 3 physicians, experienced with such judg-
ing wrong time errors, F2,33 = 0.39, P= .68. The 17 large ments, rated 7% of these errors as potential ADEs (10%
sites had a mean ± 95% confidence interval error rate of if wrong time errors are excluded). This is comparable
16.5%±5.8% (excluding wrong time errors, 9.7%±2.9%), to the 8% of all errors found in the teaching hospital
and the 19 small sites had a mean ± 95% confidence in- study by Bates et al.12 (The methods and definitions
terval error rate of 20.5% ± 6.9% (excluding wrong time used, although not identical, were similar.) For 300 in-
errors, 10.2%±3.0%), resulting in F1,34 =0.75, P=.39 (ex- patients, assuming 10 doses per patient on 1 day, this
cluding wrong time errors: F1,34 = 0.06, P = .80). would be almost 40 potential ADEs per day in that fa-
The one teaching hospital had a low error rate of cility. Many drugs could have the potential for harm in
4.7% (8.2% including wrong time errors). The error rates some patients, but were judged safe herein because
for the accredited and nonaccredited facilities are shown these particular patients were not susceptible. For ex-
by error category in Table 1. There was no significant dif- ample, enteric coated aspirin, 325 mg, was adminis-
ference in error rates by accreditation status, with tered to a patient without an order. This unauthorized
(F1,22 =0.30, P= .59) or without (F1,22 =0, P = .95) wrong drug error was rated as not significant by the 3-physi-
time errors. This also proved true when those 5 in tran- cian panel. However, if the patient were also receiving
sition were treated as nonaccredited (F1,22 = 0.05, P =.82) warfarin sodium therapy, this could have been a clini-
(excluding wrong time errors: F1,22 =0, P = .96). cally significant error. When pharmacists in other stud-
The mean ± 95% confidence intervals error rates in ies were asked to judge the potential for harm from
the Colorado sites (Table 3), 23.4% ± 7.6% (excluding drugs involved in errors based on their pharmacological
wrong time errors, 13.0% ± 3.1%) (F1,34 = 4.75, P = .04), class alone, they found 67% of the doses threatening
were significantly greater than in the Georgia sites, harm.13-15 Systems should be designed to eliminate
13.8%±4.0% (excluding wrong time errors, 7.0%±1.9%) threats to patients for the full range of clinical condi-
(F1,34 =10.25, P = .003). tions that might be encountered.

(REPRINTED) ARCH INTERN MED/ VOL 162, SEP 9, 2002 WWW.ARCHINTERNMED.COM


1900

©2002 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a New York University User on 05/21/2015


Table 2. Error Rates by Site, Facility Type, and Region

No. of Wrong Error Rate


Site No. Site Size* No. of Errors No. of Doses Error Rate, % Time Errors (Excluding Wrong Time Errors), %
Accredited Hospitals in Colorado
1 Small 12 41 29.3 7 12.2
2 Large 41 174 23.6 4 21.3
3 Large 14 97 14.4 8 6.2
4 Large 18 247 7.3 3 6.1
5 Large 7 85 8.2 3 4.7
6 Large 27 102 26.5 14 12.7
Total or average ... 119 746 16.0 39 10.7
Accredited Hospitals in Georgia
7 Small 12 97 12.4 0 12.4
8 Small 21 159 13.2 1 12.6
9 Small 17 76 22.4 13 5.3
10 Small 7 24 29.2 5 8.3
11 Large 9 63 14.3 6 4.8
12 Large 7 80 8.8 2 6.2
Total or average ... 73 499 14.6 27 9.2
Nonaccredited Hospitals in Colorado
13 Small† 23 58 39.7 13 17.2
14 Small† 13 42 31.0 2 26.2
15 Small 9 66 13.6 4 7.6
16 Small 22 33 66.7 18 12.1
17 Small 8 37 21.6 0 21.6
18 Small 1 22 4.5 0 4.5
Total or average ... 76 258 29.5 37 15.1
Nonaccredited Hospitals in Georgia
19 Small† 6 59 10.2 4 3.4
20 Small 4 60 6.7 0 6.7
21 Small 2 21 9.5 0 9.5
22 Small† 0 30 0.0 0 0.0
23 Small 10 45 22.2 9 2.2
24 Large† 0 47 0.0 0 0.0
Total or average ... 22 262 8.4 13 3.4
Skilled Nursing Facilities in Colorado
25 Small 9 80 11.2 1 10.0
26 Small 17 104 16.3 1 15.4
27 Large 15 97 15.5 1 14.4
28 Large 6 88 6.8 0 6.8
29 Large 90 182 49.5 47 23.6
30 Large 85 242 35.1 59 10.7
Total or average ... 222 793 28.0 109 14.2
Skilled Nursing Facilities in Georgia
31 Small 8 27 29.6 6 7.4
32 Large 23 217 10.6 9 6.5
33 Large 19 84 22.6 9 11.9
34 Large 5 88 5.7 0 5.7
35 Large 20 126 15.9 8 9.5
36 Large 18 116 15.5 2 13.8
Total or average ... 93 658 14.1 34 9.0
Overall ... 605 3216 18.8 259 10.8

*Large sites had more than 100 licensed beds; small sites, 100 or fewer licensed beds. Ellipses indicate data not applicable.
†These hospitals were not accredited at the start of the study, but achieved accreditation status before data collection.

ACCREDITATION BY THE JOINT COMMISSION riod) to 26.2% (also an accredited hospital). The Joint
ON ACCREDITATION OF Commission on Accreditation of Healthcare Organiza-
HEALTHCARE ORGANIZATIONS tions has identified medication errors as one of the most
frequent sentinel events.16
Statistically, accreditation by the Joint Commission on
Accreditation of Healthcare Organizations was irrel- GEOGRAPHIC LOCATION
evant for differentiating the hospitals by error rate. The
error rates, excluding wrong time errors, ranged from 0% It is unclear why the error rates for the 3 types of sites
(in 2 hospitals gaining accreditation during the study pe- were significantly higher in Colorado than in Georgia.

(REPRINTED) ARCH INTERN MED/ VOL 162, SEP 9, 2002 WWW.ARCHINTERNMED.COM


1901

©2002 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a New York University User on 05/21/2015


Table 3. Error Rates by Error Category, Facility Type, and Region*

Facility Type

Accredited Hospitals Nonaccredited Hospitals Skilled Nursing Facilities

Error Category Colorado Georgia Colorado Georgia Colorado Georgia


Omission 54 (6) 19 (3) 6 (4) 2 (2) 64 (8) 38 (6)
Unauthorized drug 8 (1) 4 (1) 2 (1) 0 8 (1) 0
Wrong dose 27 (3) 19 (3) 7 (4) 5 (4) 29 (4) 16 (2)
Extra dose 5 (1) 2 (0) 0 0 3 (0) 0
Wrong route 0 1 (0) 1 (1) 0 2 (0) 2 (0)
Wrong form 7 (1) 3 (0) 2 (1) 0 6 (1) 2 (0)
Wrong technique 0 0 0 0 1 (0) 1 (0)
Subtotal 101 (12) 48 (7) 18 (11) 7 (6) 113 (14) 59 (9)†
Wrong time 54 (6) 31 (5) 22 (14) 9 (7) 109 (14) 34 (5)
Total errors 155 (18) 79 (12) 40 (25) 16 (13) 222 (28) 93 (14)
No error 691 (82) 556 (88) 118 (75) 110 (87) 571 (72) 565 (86)
Total Doses 846 (100) 635 (100) 158 (100) 126 (100) 793 (100) 658 (100)

*Data are given as number (percentage) of doses for each facility type in each region. Percentages are rounded to the nearest whole number.
†This subtotal does not equal the sum of the individual percentages because of rounding.

Table 4. Examples of Errors Rated by the Physician Panel Table 5. Potential Clinical Significance
as Potential Adverse Drug Events* for Each Error Category

Error Description Error Category Significant % Significant


Within Each
Glyburide, 2.5 mg, given, but serum glucose Extra dose
Error Category No Yes Total Error Category
level at 6.4 mmol/L (115 mg/dL) and ordered
to hold if level is ⬍6.7 mmol/L (⬍120 mg/dL) Extra dose 8 1 9 11
Warfarin sodium, 10 mg/d, ordered, Omission Omission 271 17 288 6
but not given Unauthorized drug 24 4 28 14
Phytonadione, 10 mg, not given intramuscularly Omission Wrong dose 84 15 99 15
at 9 AM Wrong form 12 6 18 33
Valproic acid delayed-release tablet, Unauthorized Wrong route 4 1 5 20
250 mg, given, but no order found drug Wrong technique 0 1 1 100
Insulin human, isophane, ordered, Unauthorized Wrong time 224 3 227 1
but isophane/regular 70/30 combination given drug Total 627 48 675 7
Verapamil, 120 mg, ordered, but 240 mg given Wrong dose
Lisinopril, 20 mg PO, given for 10 mg Wrong dose
Metoprolol tartrate ordered, but metoprolol Wrong form
succinate sustained-release tablet given LIMITATIONS
Isosorbide dinitrate SA tablet, 40 mg PO BID, Wrong form
crushed
Morphine sulfate (MS Contin), 15 mg PO BID, Wrong route
The 36 institutions studied were selected at random (or
ordered, but given rectally via random replacement) from 2 metropolitan statistical
Insulin human, isophane/regular, 25 U once Wrong time areas and were limited to those agreeing to be studied. Re-
every day subcutaneously due at 7:30 AM, but marks by those 26 institutions declining were to the effect
given at 10:10 AM that they might have poor scores and wanted to improve
their performance first. Two institutions were prevented
*PO indicates orally; SA, sustained action form; BID, twice daily; and IV,
intravenously. from participating as a matter of corporate policy. Two were
planning to close. Most did not give reasons. Thus, the er-
ror rates reported likely represent a lower bound.
The possibility that the difference was in part due to a The doses selected for examination were a conve-
difference in the skills of the observers was investigated, nience sample of a medication pass from a nursing unit
but no evidence of this was found. (All observers were identified as high volume. The typical medication pass
checked by the same pharmacist.) does not include contrast media, respiratory therapy, or
most chemotherapy. The number of doses examined was
ERRORS BY TYPE less for the nonaccredited hospitals, because of the dif-
ficulty in anticipating medication workloads in these typi-
In general, the prescribers in the typical facility faced cally small hospitals.
the reality that almost 1 in every 5 doses they ordered The possibility of an effect of the presence of an ob-
(605/3216) would be given in error, 30% of which server on the subjects observed is always a concern, but
would be omissions—the most common error type after it is not a severe problem when the subjects are ob-
wrong time errors. However, the rates across facilities served doing an activity familiar to them, such as their
differed widely. regular jobs, and when the observer is trained to be un-

(REPRINTED) ARCH INTERN MED/ VOL 162, SEP 9, 2002 WWW.ARCHINTERNMED.COM


1902

©2002 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a New York University User on 05/21/2015


obtrusive and nonjudgmental.13,17-20 It is possible that some Corresponding author and reprints: Kenneth N. Barker,
errors represented intercepted prescribing errors de- PhD, Center for Research on Pharmacy Operations and De-
tected and, therefore, not followed by pharmacists or signs, School of Pharmacy, Auburn University, 128 Miller Hall,
nurses. However, there was no evidence to support this. Auburn, AL 36849-5506 (e-mail: barkekn@auburn.edu).

REFERENCES
CONCLUSIONS
1. Institute of Medicine. To Err Is Human: Building a Safer Health System. Wash-
Medication errors were frequent, occurring at a rate of ington, DC: National Academy Press; 1999.
nearly 1 of every 5 doses in the typical hospital and skilled 2. Leape LL, Bates DW, Cullen DJ, et al. Systems analysis of adverse drug events.
JAMA. 1995;274:35-43.
nursing facility. The percentage of errors rated poten- 3. Allan EL, Barker KN. Fundamentals of medication error research. Am J Hosp Pharm.
tially harmful was 7%, or more than 40 per day per 300 1990;47:555-571.
inpatients, on the average. Accreditation by the Joint Com- 4. Barker KN, Heller WM, Brennan JJ, et al. The development of a centralized unit
mission on Accreditation of Healthcare Organizations was dose dispensing system, part VI: the pilot study—medication errors and drug
not associated with significantly lower error rates. Error losses. Am J Hosp Pharm. 1964;21:609-625.
5. Barker KN, Harris JA, Webster DB, et al. Consultant evaluation of a hospital medi-
rates were higher in Colorado than in Georgia. Substan- cation system: analysis of the existing system. Am J Hosp Pharm. 1984;41:
tial variations in error rates by facility were identified. If 2009-2016.
the rates detected are durable over time, it should be pos- 6. Barker KN, Pearson RE. Medication distribution systems. In: Brown TR, Smith
sible to identify organizations that deserve closer study. MC, eds. Handbook of Institutional Pharmacy Practice. 2nd ed. Baltimore, Md:
Williams & Wilkins; 1986:341-342.
The error rates are likely to be understated because 7. Flynn EA, Barker KN, Pepper GA, Bates DW, Mikeal RL. Comparison of methods
of the large proportion of facilities that declined to par- for detecting medication errors in 36 hospitals and skilled-nursing facilities. Am
ticipate. This evidence of a high rate of medication er- J Health Syst Pharm. 2002;59:436-446.
rors in many of the institutions in the sample supports 8. Flynn EA, Barker KN. Medication error research. In: Cohen MR, ed. Medication
the implications of the Institute of Medicine report that Errors. Washington, DC: American Pharmaceutical Association; 1999:6.1-6.30.
9. Barker KN, Felkey BG, Flynn EA, Carper JL. White paper on automation in phar-
the medication delivery and administration systems of macy. Consult Pharm. 1998;13:256-293.
the nation’s hospitals and skilled nursing facilities have 10. Healthcare QuickDisc CD-ROM Database Ver. 2.0. Chicago, Ill: Health Forum, Ameri-
major systems problems. These results are especially valu- can Hospital Association; 2000.
able because they provide data from primarily nonteach- 11. Barker KN, McConnell WE. The problems of detecting medication errors in hos-
pitals. Am J Hosp Pharm. 1962;19:360-369.
ing sites, complementing data from large teaching hos- 12. Bates DW, Boyle DL, Vander Vliet MB, et al. Relationship between medication
pitals, and examine the association of accreditation with errors and adverse drug events. J Gen Intern Med. 1995;10:199-205.
error rates. 13. Barker KN, Kimbrough WW, Heller WM. A Study of Medication Errors in a Hos-
pital. Fayetteville: University of Arkansas; 1966.
14. Hynniman CE, Conrad WF, Urch WA, Rudnick BR, Parker PF. A comparison of
Accepted for publication February 13, 2002. medication errors under the University of Kentucky unit dose system and tradi-
This study was supported by grant 500-96-P605 from tional drug distribution systems in four hospitals. Am J Hosp Pharm. 1970;27:
802-814.
the Alabama Quality Assurance Foundation, Birmingham. 15. Hall KW, Ebbeling P, Brown B, Shwortz I. A retrospective-prospective study of
We thank Robert M. Cisneros, RPh, MS, for his valu- medication errors: basis for an ongoing monitoring program. Can J Hosp Pharm.
able assistance at 16 sites in Georgia. We appreciate the in- 1985;38:141-143, 146.
put and advice of Samuel W. Kidder, PharmD, MPH, phar- 16. Nadzam DM. A systems approach to medication use. In: Cousins DM, ed. Medi-
cation Use. Oakbrook Terrace, Ill: Joint Commission on Accreditation of Health-
macy consultant at Health Care Financing Administration.
care Organizations; 1998:5.
We thank Linda A. Pfaff, RN, MS, coordinator for opera- 17. Kerlinger FN, Lee HR. Foundations of Behavioral Research. 4th ed. New York,
tions in Georgia, for her valuable assistance. We also thank NY: Harcourt College; 2000:728-729.
Helen Deere-Powell, RPh; Lucian L. Leape, MD; Loriann E. 18. Barker KN. Data collection techniques: observation. Am J Hosp Pharm. 1980;
DeMartini, PharmD; G. Neil Libby, PhD, RPh; Richard Shan- 37:1235-1243.
19. Barker KN, Mikeal RL, Pearson RE, Illig NA, Morse ML. Medication errors in nurs-
non, RPh; Robert E. Pearson, RPh, MS; Tejal Gandhi, MD; ing homes and small hospitals. Am J Hosp Pharm. 1982;39:987-991.
Rainu Kaushal, MD; and Jeffrey Rothschild, MD, for the vari- 20. Dean B, Barber N. Validity and reliability of observational methods for studying
ous roles they played in the preparation of the manuscript. medication administration errors. Am J Health Syst Pharm. 2001;58:54-59.

(REPRINTED) ARCH INTERN MED/ VOL 162, SEP 9, 2002 WWW.ARCHINTERNMED.COM


1903

©2002 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a New York University User on 05/21/2015

You might also like