Healthy.org.il

Incidence of Adverse Drug Reactionsin Hospitalized Patients Jason Lazarou, MSc; Bruce H. Pomeranz, MD, PhD; Paul N. Corey, PhD Objective.—To estimate the incidence of serious and fatal adverse drug reac-
Definitions
Data Sources.—Four electronic databases were searched from 1966 to 1996.
Study Selection.—Of 153, we selected 39 prospective studies from US
not use the following specific definitions: Data Extraction.—Data extracted independently by 2 investigators were ana-
Adverse Drug Reaction (ADR).—Ac-
lyzed by a random-effects model. To obtain the overall incidence of ADRs in hos- pitalized patients, we combined the incidence of ADRs occurring while in the hos- tion definition,8 this is any noxious, un- pital plus the incidence of ADRs causing admission to hospital. We excluded errors intended, and undesired effect of a drug, in drug administration, noncompliance, overdose, drug abuse, therapeutic failures, and possible ADRs. Serious ADRs were defined as those that required hospital- ization, were permanently disabling, or resulted in death.
definition excludes therapeutic failures,intentional and accidental poisoning (ie, Data Synthesis.—The overall incidence of serious ADRs was 6.7% (95% con-
fidence interval [CI], 5.2%-8.2%) and of fatal ADRs was 0.32% (95% CI, 0.23%-0.41%) of hospitalized patients. We estimated that in 1994 overall 2 216 000 (1 721 000-2 711 000) hospitalized patients had serious ADRs and 106 000 compliance (taking more or less of a drug (76 000-137 000) had fatal ADRs, making these reactions between the fourth and Conclusions.—The incidence of serious and fatal ADRs in US hospitals was
found to be extremely high. While our results must be viewed with circumspectionbecause of heterogeneity among studies and small biases in the samples, these For editorial comment see p 1216.
data nevertheless suggest that ADRs represent an important clinical issue.
adverse drug event (ADE), which is an in- drug. In contrast to the World Health Or- ganization definition of ADR, the defini- tion of ADE includes errors in administra- tients; those admitted to the hospital due Health Organization definition for ADR be- cause of its frequent use in the studies that we analyzed, and because of our goal to es- timate injuries incurred by drugs that were those articles that did not use the World From the Departments of Zoology (Mr Lazarou and administration. However, this was not al- Dr Pomeranz), Physiology (Dr Pomeranz), and Public ways feasible since a few articles may have Health Sciences (Dr Corey), University of Toronto,Toronto, Ontario.
included errors in administration but did Reprints: Bruce H. Pomeranz, MD, PhD, Depart- not report them separately. Therefore, un- ments of Physiology and Zoology, University of Toronto, pare benefits of drugs to the side effects fortunately, these latter articles added to 25 Harbord St, Toronto, Ontario, Canada M5S 3G5(e-mail: pomeranz@zoo.utoronto.ca).
Adverse Drug Reactions in Hospitalized Patients—Lazarou et al 1998 American Medical Association. All rights reserved.
Possible ADR.—This is an ADR that
ies). Second, ADRs classified as “possible” were excluded. Attributing causality is al- the number of false positives in the data.
tient’s clinical state.10 Possible ADRs Heterogeneity
Serious ADR.—This is an ADR that
requires hospitalization, prolongs hospi- confidence intervals (CIs) to draw atten- Prospective Studies.—Patients were
heterogeneity of the various studies,13,14 patient’s discharge from the hospital.
abuse, or therapeutic failures, (4) for ad- Retrospective Studies.—
had left the hospital. These studies were ample, the mean estimate for the overallnumber of serious ADRs per year (see Literature Search
Data Extraction
adverse drug or adverse reaction or number of hospital patients in each study drug-related or drug-induced and hos- pital. Three MeSH (Medical Subject ing this value by the total number of hos- appropriate (ie, hospitalization, drugs, drug therapy/adverse effects) in combi- test.17 All statistical analyses were per- hospital type in which the study was per- Number of Patients With ADRs
sent letters to researchers in the field to en in each study. To test for reliability of Selection Criteria
selected subset of the data was extracted tiplying this value by the number of hos- 1. The patients studied were not se-
tics.18 In 1994, there were 33 125 492 hos- serious, fatal, and all severities (intra- pital admissions in the United States.
2. Sufficient information was re-
3. English translations of the papers
Analysis of ADR Incidence
pitals in 1994 (63 000)=Incidence of Fatal 4. Prospective monitoring was used
5. Definitions used in the studies co-
incided with ours (see “Definitions” sub- tion that our sample is representative of that were serious (which included fatal), Quality of the Data
ity of each study,11 we chose instead to im- prove the quality of our database. First, Using our 5 selection criteria, 39 of the quality studies (ie, the retrospective stud- Adverse Drug Reactions in Hospitalized Patients—Lazarou et al 1998 American Medical Association. All rights reserved.
Table 1.—Studies on ADRs in Patients While in the Hospital (ADRIn)* countries from our meta-analysis since aproper analysis for representativeness Incidence of ADRs, %‡
Source, y
Studied†
Severities
Incidence of ADRs
2.3%) of hospital patients, while the inci- of hospital patients and the incidence of tients and the overall incidence of fatal ADRs was 0.32% (95% CI, 0.23%-0.41%).
10.9% (95% CI, 7.9%-13.9%) of hospitalpatients. The overall incidence of ADRIn *ADR indicates adverse drug reaction; ADRIn, an ADR occurring in patients while in the hospital; and ellipses, †Wards studied: 1, medical; 2, surgical; 3, geriatric; 4, pediatric; 5, psychiatric; 6, internal medicine; 7, intensive ‡Incidence of ADRs = (number of patients with ADR/total patients studied) ϫ 100.
§This study performed by the Boston Collaborative Drug Surveillance Program was categorized as United States in our analysis since only 1787 of the 11 526 patients were from hospitals outside the United States.
proportion of type A44 (dose-dependentADRs) and type B44 (idiosyncratic and/ Table 2.—Studies on Patients Admitted to the Hospital Due to an ADR*† or allergic ADRs). Of the “all severities”ADRIn, 76.2% (95% CI, 71.0%-81.4%) Incidence of ADRs, %§
Source, y
Studied‡
CI, 18.6%-29.0%) were type B reactions.
Number of Hospital
Patients With ADRs
hospital patients in the United States ex- 1 033 000-2 060 000) hospital patients ex- 1 721 000-2 711 000) hospital patients ex- *ADR indicates adverse drug reaction; ADRAd, an ADR causing admission to the hospital; and ellipses, data not †Unlike Table 1, the column “All Severities” is missing from Table 2 because all ADRAds are classified as serious ‡Wards studied: 1, medical; 2, surgical; 3, geriatric; 4, pediatric; 5, psychiatric; 6, internal medicine; 7, intensive §Incidence of ADRs = (number of patients with ADR/total patients studied) ϫ 100.
4.6% (95% CI, 3.3%-6.0%) of the 2 286 000 of these 39 studies are given in Tables 1 and 2.4-7,9,19-43 Fifty-seven studies were States to allow us to perform these tasks, Adverse Drug Reactions in Hospitalized Patients—Lazarou et al 1998 American Medical Association. All rights reserved.
Table 3.—ADR Incidence According to ADR Severity* Table 5.—Is Our Sample Representative of USHospitals? Total Patients
Incidence of
ADR Group
ADRs in Patients While in the Hospital (ADRIn)
Hospitals* Sample† Studies‡
Patients Admitted to the Hospital Due to an ADR (ADRAd)
*Statistics in this column were derived from data by the National Hospital Discharge Survey.50 Overall ADR Incidence (ADRIn + ADRAd)‡
†Values in this column were derived from combining our ADRIn (adverse drug reaction [ADR] occurring in patients while in the hospital) and ADRAd (ADR causingadmission to the hospital) studies to increase the sample size, except for average drug exposure, for which datawere unavailable for the ADRAd group.
*ADR indicates adverse drug reaction; ADRIn, an ADR occurring in patients while in the hospital; CI, confidence ‡The number of studies among the 39 US articles that interval; and ADRAd, an ADR causing admission to the hospital.
†By definition, all ADRAds are serious, hence there is no “All Severities” category for ADRAd.
‡Overall incidence is adjusted to avoid double counting (see “Methods” section).
¶No statistic could be obtained for the average drug exposure in US hospital patients; ellipses indicate data Table 4.—Estimated Number of Hospital Patients in 1994 With ADRs, in Thousands (95% CI)*† sample, there may not be a major bias.
*ADR indicates adverse drug reaction; CI, confidence interval; ADRIn, an ADR occurring in patients while in the hospital; and ADRAd, an ADR causing admission to the hospital.
†Based on 33 125 492 US admissions18 in 1994: estimates use values from Table 3 (eg, for all severities ADRIn: 33 125 492 ϫ 0.1089 = 3 607 000 patients with an ADR).
‡By definition all ADRAds are serious, hence there are no data for nonserious ADRs in this category.
§From these numbers, we estimated that ADRs were the fourth to sixth leading cause of death in the United States.
Representativeness of Our Sample
ber of drugs per day may be rising to com- pensate. Therefore, while the actual inci- last 32 years, the pattern of their occur- of stay,45,46 age,45,47 gender,48,49 and drug fect on ADR rate: renal function, hepatic function, alcoholism, drug abuse, and se- sentative of the US hospital population50 verity of illness.44,52 Unfortunately, these vis-a`-vis these 4 factors. We determined that the differences were significant for sample of studies and, thus, could not be length of stay and gender but not for age.
used to determine representativeness.
have an effect on ADR incidence.9,40,53,54 were classified as “possible,” (5) we ex- in length of stay or gender are estimated individually. Without these data, we can- cidence and year of study using a random- tration.9,19,20 One of the goals of ADE re- effects linear regression model and found (r=0.27, P=.14, n=18) or for ADRAd may also introduce bias into our results.
(r=0.23, P=.34, n=21). The Figure shows It is thought that teaching hospitals con- cludes medication errors, had a different tain more seriously ill patients than non- objective: to show that there are a large proven.35,55 Teaching hospitals are over- US hospitals that should have affected the of hospital stay is decreasing,51 the num- Adverse Drug Reactions in Hospitalized Patients—Lazarou et al 1998 American Medical Association. All rights reserved.
76 000-137 000) to 93 000 (95% CI, 67 000- variation such as this is a limitation of altogether, there probably is a small net Representativeness of Our Sample
Incidence of adverse drug reactions (ADRs) in 39 In the “Results” section, we found that studies distributed over 32 years. All 39 points are for the 5 factors examined 3 were possible not visible as several are superimposed on each sources of bias: length of stay, gender, and other. Linear regression, using a random-effectsmodel, showed no significant correlation for either those experiencing an ADR while in the hospital estimate the size of the sampling bias due (ADRIn) (r=0.27, P=.14) or those admitted to the to these 3 factors as follows. As seen in hospital due to an ADR (ADRAd) (r=0.23, P=.34).
of hospital stay than the US national av- erage (10.6 days vs 7.6 days).18 While the hospitals leads to a reduction of many of and length of stay,21,45,46 there are no quan- CONCLUSIONS
formed a linear regression analysis on our obtain a slope of 0.007 (P=.008) and de- from the fourth to sixth leading cause of tal costs directly attributable to an ADR pital stay from 7.6 to 10.6 days would pos- death. Even if the lower confidence limit sibly cause the incidence of ADRIn of all ditional $1.56 to $4 billion in direct hos- severities to rise from the adjusted value cause of death in the United States, after tion of female patients in our sample was heart disease (743 460), cancer (529 904), Heterogeneity
As outlined in the “Methods” section, scenario, the adjusted value for the over- all incidence of ADRs of all severities in erogeneity in our data using a linear re- value of 15.1% (95% CI, 12.0%-18.1%).
the variance (r=0.65, P=.009, n=18). For This work was supported by a grant (Dr Pomeranz) and a scholarship (Mr Lazarou) from the National reduced by 27% (r=0.52, P=.04, n=14).
Science Engineering Research Council, Ottawa, ance. Thus, a great deal of the heteroge- J. L. Lazarou did this work in partial fulfillment of neity could be attributed to factors well obstetrical patients, and, since there are his MSc degree at the University of Toronto, On-tario; B. H. Pomeranz, MD, PhD, was the principal about 4 million obstetrical ward patients investigator; and P. N. Corey, PhD, was the statis- tician who contributed to the conception, design, hospital stay, and the age of patients.
million total hospital admissions,18 then analysis and interpretation of the data, and also particpated in writing the manuscript.
A complete list of the 104 papers excluded from our meta-analysis is available on request from the Adverse Drug Reactions in Hospitalized Patients—Lazarou et al 1998 American Medical Association. All rights reserved.
References
1. Gray J. Bill would force drug makers to give cus-
of adverse drug reactions in adult medical inpa- problems upon patient admission to small hospitals.
tomers data on risks. New York Times. July 25, 1996: tients. Can J Hosp Pharm. 1994;47:209-216.
Am J Hosp Pharm. 1977;34:738-742.
22. Steel K, Gertman PM, Crescenzi C, Anderson J.
41. McKenney JM, Harrison WL. Drug-related hos-
2. D’Arcy PF, Griffin JP. Thalidomide revisited.
Iatrogenic illness on a general medical service at a pital admissions. Am J Hosp Pharm. 1976;33:792- Adverse Drug React Toxicol Rev. 1994;13:65-76.
university hospital. N Engl J Med. 1981;304:638-642.
3. Cluff LE, Thornton CF, Seidl LG. Studies on the
23. Mitchell AA, Goldman P, Shapiro S, Slone S.
42. McKenzie MW, Marchall GL, Netzloff ML, Cluff
epidemiology of adverse drug reactions, part 1: Drug utilization and reported adverse reactions in LE. Adverse drug reactions leading to hospitaliza- methods of surveillance. JAMA. 1964;188:976-983.
hospitalized children. Am J Epidemiol. 1979;110: tion in children. J Pediatr. 1976;89:487-490.
4. Seidl LG, Thornton GF, Smith JW, Cluff LE.
43. Caranasos GJ, Stewart RB, Cluff LE. Drug-
Studies on the epidemiology of adverse drug reac- 24. Bennett BS, Lipman AG. Comparative study of
induced illness leading to hospitalization. JAMA.
tions, III: reactions in patients on a general medical prospective surveillance and voluntary reporting in service. Johns Hopkins Hosp Bull. 1966;119:299- determining the incidence of adverse drug reactions.
44. Rawlins M, Thompson J. Mechanisms of adverse
Am J Hosp Pharm. 1977;34:931-936.
drug reactions. In: Davies D, ed. Textbook of Ad- 5. Borda IT, Slone D, Jick H. Assessment of ad-
25. May FE, Fuller S, Stewart RB. Drug use
verse Drug Reactions. 4th ed. Oxford, England: Ox- verse reactions within a drug surveillance program.
and adverse drug reactions prior to and during hospitalization. J Am Pharm Assoc. 1977;17:560-598.
45. Carbonin P, Pahor M, Bernabei R, Sgadari A. Is
6. Miller RR. Hospital admissions due to adverse
26. Miller RR. Drug surveillance utilizing epidemio-
age an independent risk factor of adverse drug re- drug reactions: a report from the Boston Collabo- logical methods: a report from the Boston Collabo- actions in hospitalized medical patients? J Am Geri- rative Drug Surveillance Program. Arch Intern rative Drug Surveillance Program. Am J Hosp 46. Hurwitz N, Wade OL. Intensive hospital moni-
7. Smith JW, Seidl LG, Cluff LE. Studies on the
27. McKenzie MW, Stewart RB, Weiss CF, Cluff
toring of adverse reactions to drugs. BMJ. 1969;1: epidemiology of adverse drug reactions, V: clinical LE. A pharmacist-based study of the epidemiology factors influencing susceptibility. Ann Intern Med.
of adverse drug reactions in pediatric medicine pa- 47. Ogilvie RI, Ruedy J. Adverse drug reactions
tients. Am J Hosp Pharm. 1973;30:898-903.
during hospitalization. Can Med Assoc J. 1967;97: 8. World Health Organization. International Drug
28. Wang R, Terry LC. Adverse drug reactions in a
Monitoring: The Role of the Hospital. Geneva, Swit- Veterans Administration hospital. J Clin Pharma- 48. Domecq C, Naranjo CA, Ruiz I, Busto U. Sex-
zerland: World Health Organization; 1966. Techni- col New Drugs. 1971;11:14-18.
related variations in the frequency and character- 29. Gardner P, Watson LJ. Adverse drug reactions:
istics of adverse drug reactions. Int J Clin Phar- 9. Bates DB, Leape LL, Petrycki S. Incidence and
a pharmacist-based monitoring system. J Clin macol Ther Toxicol. 1980;18:362-366.
preventability of adverse drug events in hospital- 49. Zilleruelo I, Espinoza E, Ruiz I. Influence of the
ized adults. J Gen Intern Med. 1993;8:289-294.
30. Sidel VW, Koch-Weser J, Barnett GO, Eaton A.
assessment of the severity on the frequency of ad- 10. Karch FE, Lasagna L. Adverse drug reactions:
Drug utilization and adverse reactions in a general verse drug reactions. Int J Clin Pharmacol Ther a critical review. JAMA. 1975;234:1236-1241.
hospital. Hospitals. 1967;41:80-88.
11. Chalmers TC, Smith H, Blackburn B, et al. A
31. Reichel W. Complications in the care of five hun-
50. National Center for Health Statistics. National
method for assessing the quality of a randomized dred elderly hospitalized patients. J Am Geriatr Soc.
Hospital Discharge Survey: Annual Summary, control trial. Control Clin Trials. 1981;2:31-49.
1992. Hyattsville, Md: US Dept of Health and Hu- 12. Karch FE, Smith CL, Kerzner B, Mazzullo JM,
32. Schimmel EM. The hazards of hospitalization.
man Services; 1994. Publication 94-1779.
Weintraub M, Lasagna L. Adverse drug reac- Ann Intern Med. 1964;60:100-110.
51. Kahn L, Rubenstein L, Draper D, et al. The ef-
tions—a matter of opinion. Int J Clin Pharmacol 33. Nelson KM, Talbert RL. Drug-related hospital
fects of the DRG-based prospective payment sys- admissions. Pharmacotherapy. 1996;16:701-707.
tem on quality of care for hospitalized medicare pa- 13. Colditz GA, Burdick E, Mosteller F. Heteroge-
34. Col N, Fanale JE, Kronholm P. The role of medi-
tients. JAMA. 1990;264:1953-1955.
neity in meta-analysis of data from epidemiologic cation noncompliance and adverse drug reactions in 52. Hurwitz N. Predisposing factors in adverse
studies: a commentary. Am J Epidemiol. 1995;142: hospitalizations of the elderly. Arch Intern Med.
drug reactions to drugs. BMJ. 1969;1:536-539.
53. Smidt NA, McQueen EG. Adverse reactions to
14. DerSimonian L, Laird N. Meta-analysis in clini-
35. Mitchell AA, Lacouture PG, Sheehan JE, Kauff-
drugs: a comprehensive hospital inpatient survey.
cal trials. Control Clin Trials. 1986;7:177-188.
man RE, Shapiro S. Adverse drug reactions in chil- 15. Berkey CS, Hoaglin DC, Mosteller F, Colditz
dren leading to hospital admission. Pediatrics. 1988; 54. SmidtNA,McQueenEG.Adversedrugreactions
GA. A random-effects regression model for meta- in a general hospital. N Z Med J. 1973;78:39.
analysis. Stat Med. 1995;14:395-411.
36. Bigby J, Dunn J, Goldman L, et al. Assessing the
55. O’Hara D, Carson N. Reporting of adverse
16. Einarson TR. Drug-related hospital admissions.
preventability of emergency hospital admissions: a events in hospitals in Victoria, 1994-1995. Med J Ann Pharmacother. 1993;27:832-840.
method for evaluating the quality of medical care in 17. Rosner B. Fundamentals of Biostatistics. 4th
a primary care facility. Am J Med. 1987;83:1031- 56. Evans RC, Classen DC, Horn SD, Bass SB,
ed. New York, NY: Duxbury Press; 1995.
Burke JP. Preventing adverse drug events in hos- 18. Morgan K, Morgan S, Quitno N. Health Care
37. Lakshmanan MC, Hershey CO, Breslau D. Hos-
pitalized patients. Ann Pharmacother. 1994;28:523- State Rankings 1996. 4th ed. Lawrence, Kan: Mor- pital admissions caused by iatrogenic disease. Arch Intern Med. 1986;146:1931-1934.
57. Classen DC, Pestonik SL, Evans RS, Lloyd JF,
19. Bates DW, Boyle DL, Vander Vliet MB, Sc-
38. Salem RB, Keane TM, Williams JG. Drug-
Burke JP. Adverse drug events in hospitalized pa- heider J, Leape LL. Relationship between medica- related admissions to a Veterans’ Administration tients: excess length of stay, extra costs, and attrib- tion errors and adverse drug reactions. J Gen Intern psychiatric unit. Drug Intell Clin Pharm. 1984;18: utable mortality. JAMA. 1997;277:301-306.
58. Bates DW, Spell N, Cullen DJ, et al. The costs of
20. Bates DW, Cullen DJ, Laird N, et al. Incidence
39. Stewart RB, Springer PK, Adams JE. Drug-
adverse drug events in hospitalized patients.
of adverse drug events and potential adverse drug related admissions to an inpatient psychiatric unit.
events: implications for prevention. JAMA. 1995; Am J Psychiatry. 1980;137:1093-1095.
59. American Hospital Association. Hospital Sta-
40. Frisk PA, Cooper JW, Campbell NA. Commu-
tistics, 1993-1994. Chicago, Ill: American Hospital 21. Bowman L, Carlstedt BC, Black CD. Incidence
nity-hospital pharmacist detection of drug-related Adverse Drug Reactions in Hospitalized Patients—Lazarou et al 1998 American Medical Association. All rights reserved.

Source: http://www.healthy.org.il/files/jama.pdf