RESEARCH REPORTS

Adverse Drug Reactions

Clinical and Economic Impact of

Adverse Drug Reactions in Hospitalized Patients

Dong-Churl Suh, Betsy S Woodall, Soung-Kook Shin, and Evelyn R Hermes-De Santis

OBJECTIVE:To identify the classes of drugs that most commonly cause adverse drug reactions (ADRs) and the characteristics of
these ADRs and to determine the economic impact of ADRs on patients’ length of stay and hospitalization costs.
METHODS: Data on ADRs from patients admitted to a hospital in New Jersey were collected, studied, and analyzed over a five-month
period. To determine the economic impact of ADRs, patients who experienced ADRs during hospitalization were matched to
controls. Each ADR was rated with regard to its severity, the patients’ outcomes were determined, and specific classes of
medications were identified as particularly causative of ADRs.
RESULTS: A total of 196 patients experienced ADRs; 131 of these individuals were matched with 1338 patients who did not
experience an ADR, based on their diagnosis-related group code. The leading causal drugs according to therapeutic class were
antiinfective (17%), cardiovascular (17%), antineoplastic (15%), and analgesics/antiinflammatory agents (15%). The organ systems
most often affected were gastrointestinal (24%), dermatologic (19%), and immune systems (15%). The mean length of stay per
patient differed significantly between the ADR case group and matched control group (10.6 vs. 6.8 d; p = 0.003), as did the total
hospitalization cost ($22 775 vs. $17 292; p = 0.025).
CONCLUSIONS: Length of hospital stay and total hospitalization costs were significantly higher for patients experiencing ADRs than
those who did not experience ADRs. ADR reporting systems in hospitals need to be changed and strengthened to decrease the
incidence of avoidable reactions.
KEY WORDS: adverse drug reactions, costs.
Ann Pharmacother 2000;34:1373-9.

dverse drug reactions (ADRs) have been implicated as
A the fourth to sixth leading cause of death in the US,
1
amounting to 106 000 deaths annually. The national aver-
ages and previous studies predict that 2.4 –30% of hospi-
talized patients may experience an ADR during their hos-
pital stay.1-5 The economic burden of drug-related morbidi-
ty and mortality costs was estimated to range anywhere
from $30 billion to $130 billion annually.6,7
ADRs result in transient or permanently debilitating ef-
fects, including death, as well as financial costs not only to
patients and healthcare institutions, but also to society. Be-
cause of the high prevalence of ADRs, a number of initia-
tives in patient safety have been undertaken at both state
Author information provided at the end of the text.
and national levels, and many hospitals have intensified
their efforts to prevent ADRs and medication errors. The
Joint Commission on Accreditation of Healthcare Organi-
zations (JCAHO) mandates hospitals and other healthcare
organizations to have a system in place for reporting the
occurrence of ADRs. JCAHO specifically recommends
that institutions establish programs to monitor, track, and
prevent ADRs.
A few studies2,8-10 involving patients who were admitted
to the hospital due to ADRs or who experienced ADRs
during their hospital stay found that ADRs prolong hospi-
talization and/or increase hospitalization costs. However,
other studies1,5 have revealed differences in the overall im-
pact of ADRs. This could be explained by variations in the
patient populations, as well as the methodology used in
previous studies. Study institutions often vary significantly

www.theannals.com The Annals of Pharmacotherapy ■ 2000 December, Volume 34 ■ 1373

Downloaded from aop.sagepub.com by guest on October 11, 2013
from one another in terms of patient population character-
istics and the services they provide. In addition, some stud-
ies11,12 failed to use a validated algorithm or method to es-
tablish the likelihood of a cause-and-effect relationship be-
tween the alleged agents and the ADRs. Because previous
studies present various frequencies and resource utilization
associated with ADRs, further studies are needed to deter-
mine the overall impact of ADRs.
This study was conducted to analyze the characteristics
of all ADRs occurring in a hospital, including the classes
of drugs that commonly caused the ADRs, the probability
that the adverse events were caused by the agents, and the
impact of the ADRs on patients’ hospitalization costs and
length of stay at a healthcare institution during a specified
period, when compared with a control group.
Methods
SAMPLE PATIENTS
Patients who experienced ADRs while admitted at The University of
Medicine and Dentistry of New Jersey–Robert Wood Johnson Universi-
ty Hospital, an institution with 453 beds, over a five-month period from
August 1, 1998, through December 31, 1998, were identified and includ-
ed in the study. Information on ADRs was obtained by either of two re-
porting systems: (1) healthcare practitioners (i.e., pharmacists, nurses,
physicians) who reported ADRs to the pharmacy as they occurred or (2)
the medical records department, which identified ADRs that occurred
during the study period through chart review as requested and informed
the pharmacy department. All cases reported were reviewed to ensure
that no ADR was reported in duplicate via the two reporting systems.
To measure the impact of ADRs on hospitalization costs and length
of stay, patients experiencing an ADR during their hospital stay (i.e., the
case group) were matched with hospitalized patients not experiencing
ADRs (i.e., the control group) during the study period. Criteria for selec-
tion as matched control included classification by diagnosis-related
group (DRG), hospitalization during the study period, age (±2 y), and
gender.2 The DRGs used in this study were modified from the original
DRGs by the New Jersey Department of Health in order to classify all
diseases. The process of selection of matched controls was as follows:
patients admitted during the study period who belonged to the same
DRG at discharge as patients in the case group were selected and
grouped as cohort. The cohort was matched with the case group based
on DRG, then age, and finally gender. Only cohort patients whose char-
acteristics matched those of the case patients served as matched control
patients. For every ADR patient, there was a matched control group with
varying numbers of patients. Patients who were admitted to the hospital
during the study period as a result of an ADR were not included in the
match.
CLASSIFICATION OF ADVERSE DRUG REACTIONS
ADRs were defined in accordance with the World Health Organiza-
tion’s definition13 of an ADR as “any response to a drug which is nox-
ious, unintended, and which occurs at doses normally used in man for
the prophylaxis, diagnosis, or therapy of disease.” An ADR reporting
form was created to aid in the reporting and collection of ADRs. These
forms were distributed to healthcare professionals throughout the institu-
tion. Each patient’s demographic data (nursing unit, name, medical
record number, account number, date of birth), causative medication, re-
action, intervention taken, severity, and outcome were documented.
The causality relationship between the ADR and the suspected drug
therapy was assessed using the Naranjo algorithm,11 according to which
each ADR was given a probability category based on the Naranjo scale.
To avoid variations in the assessment of the ADRs, two investigators
were used to evaluate the severities and outcomes of the ADRs.
For the assessment of the outcome of each ADR, patients were moni-
tored for their entire length of stay or until the ADR subsided, whichever
1374 ■ The Annals of Pharmacotherapy ■ 2000 December, Volume 34
occurred first. Otherwise, if the ADR was identified retrospectively by
the medical records department, the outcome was determined by docu-
mented information in the chart. Mild ADRs were defined as self-limit-
ing and able to resolve over time without treatment. Moderate ADRs
were defined as those that required therapeutic intervention and hospital-
ization prolonged by one day. Severe ADRs were those that were life-
threatening, carcinogenic, or permanently disabling, and those that pro-
longed hospitalization (by >1 d) or caused death. Patient outcomes were
reported as fully recovered (i.e., patient fully recovered during hospital-
ization), not yet recovered (i.e., patient recovering, but not fully recov-
ered during hospitalization), unknown (i.e., not documented after initial
report in chart), or death.
DATA ANALYSIS
The data observed were analyzed in order to study the characteristics
of the ADRs and to determine differences in hospital length of stay and
costs between patients who experienced ADRs and patients who did not.
The characteristics of the ADRs included the severity, the type of reac-
tion (e.g., rash, acute respiratory failure), the number of ADRs in a given
patient, the class of drugs that caused the reaction, the final outcome, and
the probable cause of the ADR based on the Naranjo probability scale.11
A Student’s t-test was performed to test for differences in age be-
tween the case and matched control groups. Differences in the propor-
tion of men and women in the case and control groups were tested using
the χ2 test. Hospital charges for each patient were obtained from the
study institution. The charges were converted to estimate costs by using
the hospital–specific cost–charge ratio. Outliers were identified by the
extreme studentized deviate method and were excluded from the analy-
ses.14,15
The distribution of the length of stay and hospitalization costs was
positively skewed, due to wide variations among patients both in the
case and control groups. Therefore, the numbers representing length of
stay and hospitalization costs were transformed to their corresponding
logarithm. A paired Student’s t-test was performed to assess statistically
significant differences in transformed logarithm of hospital costs and
length of stay between the cases and the controls. A Student’s t-test was
applied to determine whether the hospitalization costs and length of stay
differed between the age groups of the case patients (i.e., ≥65 vs. <65 y).
Owing to the nature of the data, median values are presented in addition
to mean values.
Multivariate analyses were conducted to obtain adjusted means and
to estimate differences in length of stay and hospitalization costs be-
tween cases and matched controls, while controlling for age. Age was
controlled for because ADR patients were matched with non-ADR pa-
tients within the age range of ±2 years. All other variables were exactly
matched between the cases and the controls. All statistical analyses were
conducted using an SAS statistical software package.16
Results
A total of 196 patients were reported to have experi-
enced ADRs out of 9311 admissions during the study peri-
od. One hundred forty-six ADRs occurred while the pa-
tients were in the hospital, and 50 patients were admitted
as a result of ADRs. The patients admitted as a result of an
ADR were not included in the study. Nine ADR patients
were identified as outliers, and six ADR patients were un-
able to be matched with a non-ADR patient in the same
DRG; they were therefore excluded from the analysis. The
remaining 131 ADRs were matched to a total of 1338
matched control patients who were also admitted during
the five-month study period. The demographic characteris-
tics of the study patients are provided in Table 1.
Age did not differ significantly between the case and
control groups (mean ± SD 56.6 ± 20.3 vs. 55.0 ± 20.0 y,
respectively). The groups were also similar in gender, with
www.theannals.com
 Research Reports

women constituting 49.6% of the case group and 49.0% of (93.8%), followed by ADRs affecting the immune systems
the control group. (91.4%). ADRs with the lowest recovery rates were hema-
Table 2 lists the top therapeutic classes of medications tology related (66.7%).
that caused the ADRs during the study period, and the organ Table 3 presents the causality, severity, and outcomes of
systems affected by ADRs, according to the intensity and the most commonly isolated reactions. Although there were
outcome. All causative medications were classified in ac- 67 different reactions, only the seven most common reac-
cordance with the American Hospital Formulary Service.17 tions are presented in the table. The category titled “Others”
ADRs that affected the central nervous system had the included reactions such as hypotension, chills, fever, flush-
highest rate of severe intensity (21.8%). However, such ing, acute renal failure, gastrointestinal bleeding, hives,
ADRs were also associated with the highest recovery rates shortness of breath, tachycardia, rigors, seizures, alopecia,
neutropenia, anaphylaxis, encephalopathy, and headache.
Generally, reactions that were more likely to be associ-
Table 1. Characteristics of Sample Patientsa ated with the suspected medication, according to the
Naranjo probability scale,11 were undifferentiated in rela-
Matched Controls
Case Group Groupb Cohortc tion to the organ system they affected. The reaction most
(n = 131) (n = 1338) (n = 3219) frequently ranked with a definite causality was hyper-
Parameter n, % n, % n, %
glycemia as a result of corticosteroids.
Age (mean ± SD) 56.6 ± 20.3 55.0 ± 20.0 58.6 ± 20.3
The most frequently occurring reaction was a rash, with
<15 18, 13.7 45, 3.4 203, 6.3
15–29 14, 10.7 65, 4.9 147, 4.6
most cases classified as mild. Most patients recovered fully
30–44 15, 11.5 142, 10.6 312, 9.7 from the rash. The second most common ADR was nau-
45–64 39, 29.8 334, 25.0 965, 30.0 sea; 21% of patients who experienced nausea had end
≥65 45, 34.4 752, 56.2 1592, 49.5 points documented as “not yet recovered” or “death.” It
Gender can be assumed that these outcomes were associated with
male 66, 50.4 683, 51.0 1837, 57.1
female 65, 49.6 655, 49.0 1382, 42.9
an underlying disease state and corresponding therapeutic
management rather than solely due to nausea. Itching was
a
No significant differences between case and matched controls groups the third most common reaction; all patients experiencing
at the significance level of 0.05.
b
Includes patients who did not have adverse drug reactions (ADRs), this reaction recovered fully. Patients least likely to recover
but matched study criteria for ADR patients. from their ADR were those who experienced vomiting, di-
c
Includes patients who had identical diagnosis-related group codes arrhea, hyperglycemia, and/or thrombocytopenia. Fortu-
with the ADR patients, but did not have an ADR.
nately, none of these reactions resulted in death.

Table 2. Therapeutic Class of Causative Agents and Affected Organ Systems by

Intensity of Adverse Drug Reactions and Outcomea
Intensity Outcomeb
Frequency
Parameter n, % Mild Moderate Severe Fully Recovered Not Fully Recovered Death Unknown

Therapeutic class
antiinfective 28, 17.1 9 16 3 22 1 0 3
cardiovascular agents 27, 16.5 9 13 5 20 5 0 2
antineoplastic agents 24, 14.6 2 13 9 19 5 0 0
ANA/AIG 24, 14.6 6 16 2 13 4 1 6
psychotropic agents 9, 5.5 6 2 1 7 2 0 0
others 52, 31.7 14 32 6 46 2 2 2
TOTAL 164, 100 46 92 26 127 19 3 13

Organ system
gastrointestinal 59, 24.4 6 43 10 50 6 2 1
dermatology 45, 18.6 26 17 2 38 2 0 2
immunology 35, 14.5 9 19 7 32 1 0 1
CNS 32, 13.2 10 15 7 30 1 1 0
hematology 24, 9.9 6 14 4 16 6 0 2
others 47, 19.4 8 28 11 32 8 0 7
TOTAL 242, 100 65 136 41 198 24 3 13

ANA/AIG = analgesics/antiinflammatory agents; CNS = central nervous system.

a
Intensity and outcome may not add up to frequency because of missing observations.
b
Outcomes following adverse drug reaction (i.e., patient fully recovered during hospitalization, or patient recovering, but not fully recovered during
hospitalization.

www.theannals.com The Annals of Pharmacotherapy ■ 2000 December, Volume 34 ■ 1375

 Table 4 describes the length of stay and hospital costs
associated with the number of ADRs per patient. The
lengths of stay and hospital costs were divided into those
associated with patients who had one to three ADRs and
those who had four or more ADRs. Most patients (n =
119) experienced one to three ADRs, and 12 patients had
four or more ADRs. Patients who had four or more ADRs
had a higher average length of stay compared with patients
experiencing one to three ADRs. Patients who had four or
more reactions were hospitalized longer than patients with
one to three reactions, and patients <65 years old with one
to three ADRs averaged a shorter length of stay than those
≥65 years. However, among patients with four or more
ADRs, those <65 years old were hospitalized for more days
than their older counterparts. These differences were not
statistically significant.
The average hospitalization costs also varied among the
subgroups. As with the differences in hospital stay, these
differences were not statistically significant.
The length of stay and hospitalization costs for the case
and control groups are presented in Table 5. There was a
statistically significant difference between the average
length of stay of the case patients and that of the control
patients. The total hospitalization costs also differed signif-
icantly between the groups. Length of stay and hospitaliza-
tion costs between the groups were signifcantly different in
patients <65 years old, but these differences were not sig-
nificant in patients >65 years old.

Table 3. Frequency of Reactions by Causality, Severity, and Outcomes

n, %
Rash Nausea Itching
Parameter n, % n, % n, % Hyperglycemia Thrombocytopenia Vomiting Diarrhea Others Total

Causality
definite 0 1, 7 1, 8 4, 36 0 1, 14 0 8, 7 15, 8
probable 14, 67 10, 72 9, 69 5, 46 5, 56 6, 86 3, 50 80, 73 132, 69
possible 7, 33 2, 14 3, 23 2, 18 4, 44 0 2, 33 19, 18 39, 21
doubtful 0 1, 7 0 0 0 0 1, 17 2, 2 4, 2
TOTALa 21 14 13 11 9 7 6 109 190
Severity
mild 13, 62 3, 21 6, 46 5, 45 3, 33 0 1, 17 27, 24 58, 30
moderate 7, 33 10, 71 7, 54 6, 54 6, 67 7, 100 5, 83 55, 48 103, 53
severe 1, 5 1, 7 0 0 0 0 0 32, 28 34, 17
TOTALa 21 14 13 11 9 7 6 114 195
Outcomesb
recovered 18, 86 11, 79 13,100 4, 36 6, 67 5, 71 3, 50 93, 84 153, 80
not recovered 2, 10 2, 14 0 3, 27 2, 22 2, 29 2, 33 11, 10 24, 13
death 0 1, 7 0 0 0 0 0 2, 2 3, 2
unknown 1, 4 0 0 4, 36 1, 11 0 1, 17 4, 4 11, 5
TOTALa 21 14 13 11 9 7 6 110 191
a
Total frequencies of probability, severity, and outcomes may vary due to missing observations.
b
Outcomes following adverse drug reaction (i.e., patient fully recovered during hospitalization, or patient recovering, but not fully recovered during
hospitalization).

Table 4. Length of Stay and Hospitalization Costs by the Number of ADRs per Case Patient During Hospitalization
Length of Stay in Days per Pt. Age Hospitalization Costs per Pt. Age ($)
Parameter ≥65 <65 Total ≥65 <65 Total

No. of ADRs
1–3
mean ± SD 11.7 ± 13.1 9.1 ± 8.3 10.3 ± 10.7 23 638 ± 24 368 18 385 ± 15 617 20 745 ± 20 040
median 6 7 7 17 983 12 097 14 816
25th and 75th percentiles 4, 13 3, 11 4, 12 5613, 27 631 6369, 28 776 6207, 27 631
number of patients 37 82 119 37 82 119
≥4
mean ± SD 12.5 ± 9.4 13.0 ± 6.0 12.8 ± 6.8 25 075 ± 16 313 39 131 ± 26 791 34 445 ± 24 025
median 12 14 12 28 402 38 393 36 507
25th and 75th percentiles 6, 18 8, 17 7, 17 12 689, 37 461 19 368, 49 971 16 190, 41 898
number of patients 8 4 12 8 4 12

ADRs = adverse drug reactions.

1376 ■ The Annals of Pharmacotherapy ■ 2000 December, Volume 34 www.theannals.com

 Research Reports

Table 6 provides the adjusted mean differences between of hospital stay was increased by 2.2–3.2 days and hospital
the case and matched control groups, with ADR patients costs were increased by $3244 – 4655 in patients with
staying significantly longer than non-ADR patients. The ADRs compared with their non-ADR counterparts.
average hospitalization cost per ADR patient was signifi- Bates et al.10 reported that hospitalization costs were
cantly higher than that of a patient in the matched control $3244 higher for patients who experienced adverse drug
group. Although the mean length of stay and hospitaliza- events. They calculated the costs after the occurrence of
tion costs calculated change for the case and matched con- adverse drug events; our study calculated the total cost of
trol groups when age was controlled for, the mean differ- hospitalization. This difference in their approach when
ences in length of stay and hospitalization costs were con- compared with ours is most likely responsible for the larg-
sistent. er differences in cost between the case and control groups
determined in our study. A study by Classen et al.2 showed
Discussion that adverse drug events resulted in a mean difference in
hospitalization costs of $4655 between the cases and
This study found significant differences in the average matched controls. These smaller differences in mean hos-
length of stay and average total hospitalization costs be- pitalization cost in previous studies when compared with
tween the case group and the matched control group; the our study can also be explained by differences in the crite-
adjusted length of hospital stay and hospitalization costs ria for patient selection (e.g., patients who experienced ad-
were higher for ADR patients. These findings support verse drug events vs. ADRs).
those of previous studies,2,10 which reported that the length The length of stay and hospitalization costs for patients
<65 years of age experiencing ADRs were sig-
nificantly longer and higher than those in the
matched control group. However, differences
Table 5. Differences in Length of Stay and Hospitalization Costs in the length of stay and total hospitalization
Between ADR and Non-ADR Patientsa costs between the case and control groups were
not seen with patients ≥65 years old. Suggest-
Case Group Matched Control Group
Parameter (n = 131) (n = 1338) p Value ed reasoning for this is that, because patients
Length of stay
≥65 years often have multiple comorbidities
all 10.6 ± 10.2 6.8 ± 4.5 0.0029
and weakened immune systems, which makes
percentileb 4, 7, 13 3, 5, 9 them vulnerable to experiencing certain ADRs,
≥65 11.8 ± 12.7 8.6 ± 4.2 0.2972 they are often treated prophylactically when
percentile 4, 7, 13 4, 7, 9 admitted to the hospital. Another explanation
<65 9.8 ± 8.0 6.7 ± 5.9 0.0007 for these differences is that elderly patients are
percentile 4, 8, 13 3, 5, 9
more likely to have had previous experience
Total hospitalization costs ($)
with an ADR and therefore are generally treat-
all 22 775 ± 21 088 17 292 ± 13 323 0.0251
percentile 6463, 17 983, 31 965 5722, 12 612, 28 091
ed based on their medical history, and thus
≥65 23 802 ± 23 402 18 800 ± 13 965 0.5792
have fewer major ADRs.
percentile 5613, 18 152, 31 965 6885, 15 175, 31 817 Length of stay and hospitalization costs
<65 21 993 ± 19 375 16 144 ± 12 849 0.0053 were not significantly different between ADR
percentile 6807, 15 908, 32 485 4726, 11 482, 24 700 patients in both age groups when classified by
ADR = adverse drug reaction.
the number of reactions. These results corre-
a
Mean ± SD. spond with previous studies,18,19 which re-
b
The 25th percentile, median, and 75th percentile. vealed that the occurrence of ADRs does not
increase with the age of the population. How-
ever, some researchers20,21 still argue that elder-
ly patients present with more severe disease
Table 6. Adjusted Mean Differences Between and thus more complicating factors. The re-
ADR and Non-ADR Patientsa sults of our study suggest that age is not a fac-
Case Matched Control tor that influences differences in length of stay
Group Group
Parameter (n = 131) (n = 1338) Difference R 2
p Value
and hospitalization costs among patients expe-
riencing ADRs.
Length of stay (d)
Since all drugs have the potential to cause
unadjusted 10.6 ± 0.88 6.8 ± 0.88a 3.8 0.56 0.0001
a a
ADRs, the drugs associated with ADRs in par-
adjusted for age 7.7 ± 1.96 3.9 ± 1.96 3.8 0.57 0.0001
ticular studies vary, depending on patients’
Hospitalization costs ($)
unadjusted a
22 775 ± 1959 17 292 ± 1959 a
5483 0.57 0.0001
characteristics and clinical settings. Many re-
a
adjusted for age 15 650 ± 4358 10 194 ± 4345 a
5456 0.58 0.0001
ports2,20-23 have attempted to identify which
drugs are commonly implicated in ADRs. We
ADR = adverse drug reaction. found the leading causal agents of ADRs to be
a
Mean ± SEM.
antiinfective, cardiovascular, antineoplastic,

www.theannals.com The Annals of Pharmacotherapy ■ 2000 December, Volume 34 ■ 1377

and analgesics/antiinflammatory agents. Other studies9,18
also documented that antiinfective and cardiovascular
agents caused a considerably high percentage of adverse
drug events.
ADRs can have a detrimental effect on a patient’s well-
being during the hospital stay and impact the overall health-
care system.1,2,10 Unfortunately, many healthcare practition-
ers in the US fail to monitor and track the occurrence of
ADRs; fear of liability may contribute to the underreport-
ing. By documenting such occurrences, reliable informa-
tion on the patient population in question would be made
available and would aid in preventing recurrences. By
recording and studying the results of interventions that
were made as part of the management of an ADR, we can
better target our efforts to the prescribing medications to
prevent or treat an ADR. Identifying trends in the occur-
rences of ADRs will help us to better direct our efforts to-
ward prevention through the development of clinical pro-
tocols and improved patient screening.
It has been suggested7,24 that efforts be made to reduce
the incidence of ADRs that are considered preventable and
that a monitoring program be implemented in all health-
care institutions. Another article5 suggested that sponta-
neous ADR reporting programs are the most appropriate
way of monitoring the safety of a drug. The establishment
of a nationwide, mandatory public reporting system25 was
also recommended as a way to learn about medical treat-
ments that lead to serious injury or death and to prevent fu-
ture occurrences and avoidable costs.
The matching of controls allowed us to closely approxi-
mate how much the hospital stay was extended, as well as
the excessive costs incurred once a hospitalized patient ex-
perienced an ADR, but some limitations warrant discus-
sion. Although we were able to show a statistically signifi-
cant difference between the case and control groups with
regard to length of stay and hospital charges, there are
many confounding variables that could have influenced
these outcomes. One factor that could have explained a
large proportion of the detected differences was the uncon-
trolled differences in severity of patient disease. Future
studies of ADRs need to incorporate more variables to de-
termine economic costs directly associated with an ADR.
It remains unknown whether the longer hospital stay was a
cause or a consequence of the ADRs. Patients with more
serious conditions would probably require more medica-
tions and theoretically would be more prone to the devel-
opment of ADRs. In addition, the short data collection pe-
riod for this study, and the voluntary reporting system
used, most likely resulted in a relatively low rate of docu-
mentation. We also recognize that this study does not be-
gin to assess the impact of ADRs on the quality of life of
the patients.
Summary
Length of hospital stay and hospitalization costs were
significantly higher for patients who experienced ADRs
than those who did not experience ADRs. The leading
1378 ■ The Annals of Pharmacotherapy ■ 2000 December, Volume 34
types of ADRs reported were rash, nausea, itching, throm-
bocytopenia, vomiting, hyperglycemia, and diarrhea. The
leading causal drugs classes were antiinfective, cardiovas-
cular, antineoplastic, and analgesics/antiinflammatory
agents. The organ systems most often affected were gas-
trointestinal, dermatologic, and immune systems.
Changing and strengthening the ADR reporting system
are needed to improve patient outcomes and save health-
care dollars by reducing the occurrence of such devastating
and costly events.
Dong-Churl Suh MBA PhD, Assistant Professor, College of Phar-
macy, Rutgers–The State University of New Jersey, Piscataway, NJ
Betsy S Woodall PharmD, at time of study, Drug Information Res-
ident, UMDNJ–Robert Wood Johnson University Hospital, New
Brunswick, NJ; now, Clinical Assistant Professor, Arnold & Marie
Schwartz College of Pharmacy and Health Sciences, Long Island
University, Brooklyn, NY; Coordinator, Drug Information Services,
Brookdale University Hospital and Medical Center, Brooklyn
Soung-Kook Shin PhD, Pharmacoeconomics Fellow, College of
Pharmacy, Rutgers–The State University of New Jersey
Evelyn R Hermes-De Santis PharmD, Clinical Assistant Profes-
sor, College of Pharmacy, Rutgers–The State University of New Jer-
sey; Associate Director, Drug Information Service, Department of
Pharmacy, UMDNJ–Robert Wood Johnson University Hospital
Reprints: Dong-Churl Suh MBA PhD, College of Pharmacy, Rut-
gers–The State University of New Jersey, 160 Frelinghuysen Rd., Pis-
cataway, NJ 08854-8020, FAX 732/445-2533, E-mail dsuh@rci.rut-
gers.edu
We thank John P Burke MD, University of Utah; Kitaw Demissie MD PhD, Univer-
sity of Medicine and Dentistry of New Jersey; and Steve Selvin PhD, University of
California, Berkeley, for helpful comments on statistical analysis.
References
1. Lazarou J, Pomeranz BH, Corey PN. Incidence of adverse drug reactions
in hospitalized patients: a meta-analysis of prospective studies. JAMA
1998;279:1200-5.
2. Classen DC, Pestotnik SL, Evans RS, Lloyd JF, Burke JP. Adverse drug
events in hospitalized patients. JAMA 1997;277:301-6.
3. Davies D, ed. Textbook of adverse drug reactions. New York: Oxford
University Press, 1991.
4. Brennan TA, Leape LL, Laird NM, Hebert L, Localio AR, Lawthers
AG, et al. Incidence of adverse events and negligence in hospitalized pa-
tients. N Engl J Med 1991;324:370-6.
5. Pirmohamed M, Breckenridge AM, Kitteringham NR, Park BK. Ad-
verse drug reactions. BMJ 1998;316:1295-8.
6. Johnson JA, Bootman JL. Drug-related morbidity and mortality, Arch
Intern Med 1995;155:1949-56.
7. White TJ, Arakelian A, Rho JP. Counting the costs of drug-related ad-
verse events. Pharmacoeconomics 1999;15:445-58.
8. Schneider PJ, Gift MG, Lee Y-P, Rothermich EA, Sill BE. Cost of medi-
cation-related problems at a university hospital. Am J Health Syst Pharm
1995;52:2415-8.
9. Pearson TF, Pittman DG, Longley JM, Grapes ZT, Vigliotti DJ, Mulls
SR. Factors associated with preventable adverse drug reactions. Am J
Hosp Pharm 1994;51:2268-72.
10. Bates DW, Spell N, Cullen DJ, Burdick E, Laird N, Petersen LA, et al.
The costs of adverse drug events in hospitalized patients. JAMA 1997;
277:307-11.
11. Naranjo CA, Busto U, Sellers EM, Sandor P, Ruiz I, Roberts EA, et al.
A method for estimating the probability of adverse drug reactions. Clin
Pharmacol Ther 1981;30:239-45.
12. Nelson KM, Talbert RL. Drug-related hospital admission. Pharma-
cotherapy 1996;16:701-7.
13. Internal drug monitoring: the role of the hospital. Geneva: World Health
Organization, 1966.
14. Rosner B. On the detection of many outliers. Technometrics 1975;17:
221-7.
www.theannals.com

15. Grubbs FE. Procedures for detecting outlying observations in samples.
Technometrics 1969;11:1-21.
16. SAS/STAT user’s guide, version 6. Cary, NC: SAS Institute, 1989.
17. American Hospital Formulary Service drug information. Bethesda, MD:
American Society of Health-System Pharmacists, 1999.
18. Gray SL, Sager M, Lestico MR, Jalauddin M. Adverse drug events in
hospitalized elderly. J Gerontol 1998;53A:M59-63.
19. Bates DW, Miller EB, Cullen DJ, Burdick L, Williams L, Laird N, et al.
Patient risk factors for adverse drug events in hospitalized patients. Arch
Intern Med 1999;22:2553-60.
20. Moore N, Lecointre D, Noblet C, Mabille M. Frequency and costs of se-
rious adverse drug reactions in a department of general medicine. Clin
Pharmacol 1998;45:301-8.
21. Seeger JD, Kong SX, Schumock GT. Characteristics associated with
ability to prevent adverse drug reactions in hospitalized patients. Phar-
macotherapy 1998;18:1284-9.
22. Vargas E, Simon J, Martin JC, Puerro M, Gonzalez-Callejo MA, Jaime
M, et al. Effect of adverse drug reactions on length of stay in intensive
care units. Clin Drug Invest 1998;15:353-60.
23. Bates DW, Cullen DJ, Laird N, Petersen LA, Small SD, Servi D, et al.
Incidents of adverse drug events and potential adverse drug events.
JAMA 1995;274:29-34.
24. Moore TJ, Psaty BM, Furberg CD. Time to act on drug safety. JAMA
1998;279:1571-3.
25. Corrigan J, Donaldson MS, Kohn LK, eds. To err is human: building a
safer health system. Washington, DC: National Academy Press, 2000.
EXTRACTO
OBJETIVO: Identificar las clasificaciones de medicamentos que más
comunmente causan Reacciones Adversas a Medicamento (RsAM) y
las características de éstas reacciones; determinar el impacto económico
de las RsAM en el tiempo de estadía y los costos de hospitalización.
MÉTODO: Los datos de las RsAM fueron tomados, estudiados, y
analizados por un período de cinco meses entre los pacientes admitidos
en un hospital en New Jersey. Los pacientes que experimentaron RsAM
durante la hospitalización fueron comparados con grupos de control
(pacientes que no experimentaron RAM durante su estadía en el
hospital), de manera que se pudiera determinar el impacto económico de
las RsAM. Cada RAM fue categorizada en cuanto a su severidad. Los
efectos en los pacientes fueron determinados y las clasificaciones
específica de medicamentos fueron identificadas como causal de las
RsAM.
RESULTADOS: Un total de 196 RsAM fueron documentadas. De estos
pacientes, 131 de ellos fueron comparados con 1338 pacientes los cuales
no experimentaron una RAM basado en su código DRG. Las
clasificaciones terapeúticas de los medicamentos que causaron mayor
número de RsAM fueron antiinfectivo (17%), cardiovascular (17%),
author has requested enhancement of the downloadedThe
Thewww.theannals.com file. Annals of references
Pharmacotherapy
All in-text
Research Reports
antineoplásico (15%), y agentes análgesico/anti-inflamatorio (15%). Los
sistemas de órganos más afectados por RsAM fueron gastrointestinal
(24%), dermatológico (19%), y sistema inmunológico (15%). El
promedio en el tiempo de estadía por paciente se diferenciaba
significamente entre los casos con RsAM y los grupos de control (10.6
vs 6.8 días; p = 0.003); como también en los costos de hospitalización
($22 775 vs $17 292; p = 0.025).
CONCLUSIONES: El tiempo de estadía y el total en los costos de
hospitalización fueron significativamente más altos para los pacientes
que experimentaron RsAM que aquellos que no experimentaron dichas
reacciones. Un cambio y fortalecimiento en el sistema de reporte de
RAM en los hospitales es necesario para disminuir la incidencia de las
RsAM prevenibles.
Wilma M Guzmán
RÉSUMÉ
OBJECTIF: Les objectifs de cette étude étaient d’identifier les classes de
médicaments causant le plus souvent des effets indésirables (EI) et
caractériser ceux-ci et de déterminer l’impact économique des EI en
particulier sur la durée du séjour hospitalier et les coûts d’hospitalisation.
DEVIS EXPÉRIMENTAL: Les données sur les EI ont été colligées chez les
patients séjournant dans un hôpital de l’état du New Jersey au cours
d’une période de cinq mois (août à décembre 1998). L’information
provenait soit des professionnels de la santé qui rapportaient les EI au
département de pharmacie au fur et à mesure, soit des archives
médicales. Afin d’évaluer l’influence des EI sur les coûts
d’hospitalisation et la durée du séjour, des patients qui n’ont pas
développés de EI durant leur séjour hospitalier ont été identifiés (groupe
témoin) et appariés au groupe de patients ayant développés des EI selon
leur groupe de diagnostique (diagnosis-related group – DRG), leur âge
et leur genre. Chaque EI a aussi été examiné pour sa sévérité et son
évolution et des fréquences par classe de médicaments ont été calculées.
RÉSULTATS: Un total de 196 EI ont été documentées chez 131 patients.
Le groupe témoins, auquel ces 131 patients ont été comparés,
comprenait 1338 patients. Les antibiotiques sont les médicaments le plus
souvent impliqués (17%), suivi des médicaments du système cardio-
vasculaire (17%), des anticancéreux (15%), et des médicaments
analgésiques/anti-inflammatoires (15%). Les systèmes les plus souvent
touchés par ces EI sont le système gastro-intestinal (24%), la peau
(19%), et le système immunitaire (15%). La durée moyenne du séjour
étaient plus longue chez les patients ayant développés des EI (10.6 vs.
6.8 jours; p = 0.003) et conséquemment, les coûts d’hospitalisation
étaient plus élevés ($22 775 vs. $17 292; p = 0.025).
CONCLUSIONS: La durée du séjour hospitalier et les coûts
d’hospitalisation sont plus élevés chez les patients développant des EI
que chez ceux qui n’en développent pas.
Suzanne Laplante
2000areDecember, Volume 34on ResearchGate.
1379
■ ■
underlined in blue linked to publications