related to vaccines (which go to the Labora- tory Centre for Disease ... the national Drug Adverse Reporting Pro- gram is of very ... c Adverse drug reaction definite, probable, or possible. Table 2. .... scribed. Central nervous system (CNS) drugs ...
Adverse Drug Re actions Review of the Canadian literature JOEL LEXCHIN, MID, CCFP(EM\t
A combined manual and computer search of the literature on the Canadian experience with adverse drug reactions (ADRs) to prescribed drugs yielded a total of 23 articles. The risk of developing an ADR, factors influencing that risk, the incidence of ADRs, ADRs as a cause of hospitalization, case fatality ratios, whether ADRs increase hospitalization time, and the drugs that are most commonly implicated in ADRs were analyzed. Une recherche informatisee et manuelle de la litterature concernant l'experience canadienne avec les medicaments prescrits et leurs reactions indesirables (RMI) a fourni un total de 23 articles. l'auteur analyse le risque de developper une RMI, les facteurs qui influencent ce risque, I'incidence des RMI, les RMI comme cause d'hospitalisation, les taux de deces, la question de savoir si les RMI augmentent le sejour hospitalier et les medicaments qui sont le plus souvent impliques dans les RMI. Con Fhm Physkian 1991;37:109-118.
DVERSE
DRUG
REACTIONS
(ADRs) involving prescribed drugs have generated much attention over the years. Some researchers claim that ADRs increase hospitalization costs by more than $300 million yearly' and that more than 20% of seniors admitted to acute care hospitals in Ontario are diagnosed as having prescription drug-related complications.2 Aside from one recent partial review of this topic," however, there has been no attempt to collect and analyze the Canadian experience. Karch and Lasagna,4 in 1975, defined an ADR as "any response to a drug that is noxious and unintended and that occurs at doses used in man for prophylaxis, diagnosis or therapy, excluding failure to accomplish the intended purpose." While that definition seems relatively clear-cut, a number of complexities interfere with the recognition and reporting of ADRs. ' 1. In nearly all cases the patient is already ill, and the specific event could be the result of the underlying disease rather than the medication. 2. Many patients, especially those hospitalized or in other institutions, are taking multiple medications. 3. There is a lack of a practical definition as to what information to collect. Dr Lexchin s on the Active Staff of the Department of Eme?geng Medicine, Toronto Hospital, Ontario.
4. Many reporters lack a standard approach to reporting. Even clinical pharmacologists with an interest in ADRs often cannot agree on whether a specific event was drug related or which drug was the causative agent. i67 In an attempt to circumvent some of these difficulties, various investigators have developed algorithms for deciding whether a given patient has experienced an ADR and for categorizing the reported reactions as definite, probable, possible, or unlikely." Even the use of algorithms may not yield definitive answers. WVhen the same clinical cases were analyzed by four different published algorithms, there was only 25% concurrence. 5 Currently, in Canada, the Product-Related Disease (PRD) Division of the Health Protection Branch has the responsibility for receiving ADR reports, except for ADRs related to vaccines (which go to the Laboratory Centre for Disease Control). Physicians who suspect an ADR has occurred are asked to complete an ADR form and mail it to the PRD. (A copy of this form can be found at the back of the Compendium of Pharmaceuticals and Specialties.9) According to a recent report, however, the national Drug Adverse Reporting Program is of very limited value.'0 The PRD does not have regular, ongoing solicitation of medical, hospital, or pharmaceutical associations for their reporting of ADRs. In Canadian Family Physician V'OL 37: Januay 1991
109
Table 1. ADVERSE DRUG REACTIONS IN AMBULATORY AND NURSING HOME PATIENTS: Researchers have found conflicting results. STUDY
RESULTS NO. OF PATIENTS OBSERVED
HUTCHINSON ET AL24
1026
REYNOLDSII
2411
REYNOLDS'2
--
SOON'3
826
YEAR OF STUDY
POPULATION STUDIED Outpatients attending medical chnic
1981-1982
Patients of 10 family physicians
--
Patients of 232 family physicians
--
Nursing home
1978-1979
222 (26.9)
260C
826
1979-1980
165 (20.0)
189c
Nursing home
49 (4.8)
58a
REACTIONS PER DRUG EXPOSURE N (% 58/3330 (1.7)
DRUGS AND GROUPS OF DRUGS MOST COMMONLY IMPLICATED CVS, CNS, diuretics/ electrolytes
29 (1.2)
29b
Antibiotics, CVS, misc
patients SOON14
PATIENTS WITH NO. OF REACTIONS N TOTAL N H REACTIONS
patients
314
314b
Antibiotics, CNS, respiratory/ bronchodilator 260/4014 (6.5) CNS, digoxin, diuretics
189/3420 (5.5) Digoxin, CNS, diuretics
Adverse drug reaction definite or probable Degree of certaint_ of an ADR not stated c Adverse drug reaction definite, probable, or possible a
b
Table 2. ADVERSE DRUG REACTIONS AS A CAUSE OF HOSPITAL ADMISSION: Adverse drug reactions accounted for 3% to 20% of hosp~ital admissions. STUDY RESULTS
NO. OF PATIENTS OBSERVED
ASTHANA,
170
BOSTON COLLABORATIVE"6
700
CROMARTY26
142
SOODIS
POPULATION STUDIED
YEAR OF STUDY
ADMISSIONS FROM ADR N (%)
DRUGS AND GROUPS OF DRUGS MOST
IOMPNLYTE
Patients admitted to geriatric ward
1984-1985
33 (19.4)a,b
CNS, CVS
Patients admitted
1969-1972
24 (3.4)a,b
Digoxin, ASA, prednisonec
1974-1975
7 (4.93)a
1983
83 (11.6)e
1965-1966
48 (6.6)a
to medical ward
Patients admitted to medical ward
GRYMONPRE ET AL25
7 18d
Patients aged 50 or over admitted to medical service
731
Patients admitted
to medical serice
Systemic steroids, CVS, NSAIDs
Degree of certainty of an ADR not stated Admission partially or solely due to an ADR c Implicated in admission adverse reactions in 7 different hospitals, only one of which was in Canada d A total of 863 patients were admitted, but only 718 were using prescribed drugs. Out of 193 drug-related adverse patient events (DRAPEs), 93 were ADRs (some patients experiencing more than one ADR); of the 193 DRAPEs, 52% were definite or probable. a
b
110
Canadian Family
Physician VOL 37: January 1991
addition there is minimal feedback to or even acknowledgment of physicians who are participating and reporting to the PRD on suspected reactions. There is no follow up on individual reports, and the PRD does not currently carry out regular detailed analysis of the database. In addition to this national program, since 1981 the Ontario Medical Association has organized and run a voluntary ADR reporting program. Summaries of the reports received, as well as short articles on related topics, are published approximately every 4 to 6 months in their newsletter, The DrugReport. Adverse drug reaction registries are also operated by the Saskatchewan Medical Association, the Canadian Ocular Disease Program, and the Medical Examiners of Canada. In the absence of a well-organized country-wide reporting scheme for ADRs, reviews are the best method of drawing together the acquired knowledge in this area. Individual articles are simply not comprehensive enough to deal with such a large topic. It is unlikely that the ADR situation in Canada will be significantly different from that in other industrialized countries, but I know of no recent complete review on this subject, at least in any of the English-speaking countries; therefore, there is no database from which to extrapolate. My objective here is to summarize and interpret previous Canadian work on ADRs and to outline directions for future research.
French (owing to my limited ability in that language), case reports, papers commenting only on the potential for adverse reactions, articles dealing with intentional drug overdoses, and papers describing adverse effects noted during clinical evaluations of drugs. A total of 23 papers were ultimately included in this review. When studies included information about reactions to drugs and chemicals other than prescribed drugs, the data were reanalyzed, if possible, to reflect only the findings for prescribed drugs. Similarly, where possible, only data on definite and probable ADRs, as classified by the original author(s), were used.
RESULTS AND DISCUSSION
Methodology employed Many of the studies reviewed employed suboptimal methodology, which renders their findings suspect. The family physicians who reported ADRs in their ambulatory patients '12 followed their "usual practice of eliciting ADRs," but gave no details on what those practices were or on what kind of training the doctors had in this area. No mention was made of the use of an algorithm, and the strength of the association was not graded. Physicians were not expected to actively survey their patients for ADRs. Adverse drug reactions in nursing home patients were actively sought,'3"14 and reactions were graded as to their likelihood. However, the only stated ground for deciMETHOD sion making was that "all the relevant inforPapers dealing with ADRs caused by pre- mation to evaluate the adverse effect was scribed drugs were located through a man- collected." ual search of the following journals from The literature originating from in-hos1965 to the start of 1990: Annals of the Royal pital monitoring is also not immune from College ofPhysicians and Surgeons of Canada, Ca- methodological weaknesses. Asthana and nadian Family Physician, Canadian Journal of Sood'5 give almost no details to back up Hospital Pharmay, Canadian Journal ofPsychia- their contention that ADRs were partially tr, Canadian Journal ofPublic Health, Canadian or solely responsible for 19.4% of admisMedical Association Journal, and Canadian sions that they studied. Two of the older Pharmaceutical Journal. I also performed a hospital studies originated from the Boston MEDLINE search from 1966 to the start Collaborative Drug Surveillance Proof 1990 by computer. References in any gram,'6"7 and two others'8"9 used similar relevant articles were traced by referring to methodology. While specially trained nurse the relevant book or journal or by directly monitors were used, decisions about causacontacting the author(s). tion did not use an algorithm. In the last Five different groups of papers were not two studies, judgments about ADRs were included in this analysis: references in based on the consensus of a clinical Canadian
Family Physician VOL 37: january 1991 111
Table 3. IN-HOSPITAL ADVERSE DRUG REACTIONS: Percentages ofpatients diagnosed with ADRs varied. STUDY
RESULTS
NO. OF NO. OF POPULATION STUDIED
YEAR OF STUDY
200
1977
PATIENTSSTUIE OBSERVED ARANDA ET AL2' BORDA ET AL17
Neonates in NICU
936
PATIENTS WITH REACTIONS (N %) 60
Patients admittcd
1969-1972
CROMARTY26
142
Patients
HODDINOTT ET AL27
104
CNS, digoxin,
antacids
18
(12.5)
medical ward
350
liquid
535
1974-1975
admitted to
FARKAS3
diuretics, TPN:
(57.2)
wards
Patients admitted to
--
23
(6.6)
1966
16
(15.4)
investigative
and treatment ward KOWALYK3
200
LOU ET AL28
100
1938
MIKHAEL, KACEW23
57 881
OGILVIE, RUEDY20
731
OGILVIE,
534
RUEDY29
Autopsies
578
Autopsies
--
34
1970)
26
(26)| Sedative hypnotics, cytotoxic agents,
1977-1982
145
(75)a,b
(l.8)dh
1977-1982
Patients admitted to medical service
1965-1966
All patints discharged
1967-1968
120 (21.4)
1969-1970
70 (12.1)
rececived
analgesics, antineoplastic
(17.0)
Surgical pathology specimens
from or who died in medical wards who
Contrast media agents
1972
MIKHAEL, KACEW22
~~~~~~~~~~~~~~~~~IMPLICATED
Antibiotics,
(30.0)
to 2 medical
DRUGS AND GROUPS OF DRUGS MOST COMMONLY
1015
132 (18.1)
digoxin Antineoplastics,
antibiotics,
Estrogenic
preparation
CVS, antibiotics,
insulin
Digoxin
-
digoxin SMITH3 SPINO ET
AL'8
(16.8)
--
204
Patients in
1974
100
medical and surgical units who received
(49.0)c
--
Digoxin, ampicillin, furosemide
furosemide
112
Adverse drug reaction definite or probable. bAuthors did not distinguish between ADRs due to therapeutic exposure, medication error,
c
Patients with clinical
events; biochemical events occurred in 144 patients.
Canadian Family Physician VOL 37: Januagy1991
or
steroids
intentional or accidental overdose.
pharmacist and a clinical pharmacologist. Other investigators relied heavily on decisions from house staff and nursing staff,20 Only the more recent hospital investigations based the decisions on published algorithms.2 1-25
Ambulatory and nursing home patients Several Canadian studies have reported adverse drug reactions in ambulatory and nursing home patients (Table 11 1i-424). There have been only two studies of adverse reaction reporting by family physicians."1,I2 The first of these" found 29 ADRs (1.2% of cases) during a 30-day period in 2411 patients receiving prescriptions. The authors reported that there was no relationship between type of practice or prescribing habits and reporting of adverse drug reactions. In the second study,'2 232 doctors observed 314 suspected reactions over a 5-month period. However, the total number of patients who were prescribed medications during the study period was unknown. Twenty-four percent of reactions resulted in absence from work, hospitalization, or death. The most commonly implicated drug group in both studies was
antibiotics. Hutchinson and colleagues24 used an algorithm in their prospective telephone surveillance of outpatients visiting the Internal Medicine Unit at the Royal Victoria Hospital in Montreal. The rate of probable or definite reactions was 49/1026 (5%) per patient and 58/3330 (2%) per drug course. Patients may have had more than one reaction, and some patients had more than one drug course. In general, suspected reactions tended to cause discomfort or worry rather than disability, and none of the reported reactions was fatal. Patients who took their medication less frequently than prescribed had more frequent ADRs than patients who took medication at the prescribed frequency. There were no adverse reactions in the nine patients who took the medication more frequently than prescribed. Central nervous system (CNS) drugs, cardiovascular drugs, and electrolytes or diuretics had the highest rate of adverse reactions. Soon'3"4 monitored adverse reactions in 10 nursing homes with personal or inter-
mediate care geriatric patients during two 1-year periods in the late 1970s and early 1980. A consulting pharmacist critically assessed medication profiles every 3 months and used a set of criteria to identify those at risk for an ADR. Additional information was gathered through frequent communication with nursing staff. Only definite, probable, and possible reactions of clinical significance were documented. In the first period of monitoring, there was a 26.9% incidence of ADRs among 826 patients, with 24 of 222 patients affected requiring admission to hospital. In the second period the incidence of ADRs dropped to 20.0% (165/826), and only six needed hospitalization. In both periods the largest number of reactions were attributed to digoxin, CNS drugs, and diuretics, while digoxin and non-steroidal anti-inflammatory drugs (NSAIDs) were responsible for most hospitalizations.
How ADRs cause hospitalization Adverse drug reactions account for 3% to 20% of hospital admissions (Table 215,16202526), but only one study25 used a published algorithm and stated the degree of certainty of the ADR. These reports indicate that ADRs account for 1 0% to 20% of hospital admissions for people over 50 years of age and about 5% of all admissions to medical wards. Cardiovascular drugs and systemic steroids were the main causes of these admissions.
Hospitalized patients In 13 studies (14 surveys), researchers monitored the development of ADRs in hospitalized patients (Table 3317,18,20-23,26-29). The percentages of patients diagnosed with an ADR show enormous variation, from 1.8% of 57 881 who had surgical pathology specimens analyzed for ADRs23 to 57.2% of 936 patients admitted to two medical wards.'7 Likewise the severity of the reactions and the number of reactions per drug exposure also vary (Table 417182021262729) The reactions in the former range from a low of 0.6% major and fatal reactions'7 to a high of 85.9%.29 Reactions in the latter range from 2.3% for a group of patients on an "investigative and treatment" medical ward27 to 21.4% of medical patients who were receiving digoxin.29 Canadian Family Physician VOL 37: Januagy 1991
113
Table 4. SEVERITY OF IN-HOSPITAL ADVERSE DRUG REACTIONS AND REACTIONS PER DRUG EXPOSURE: Varied results could reflect differences in populations, degree of certainty, and monitoring methods.
STUDY
RESULTS TOTAL NO. OF REACTIONS
MINOR REACTIONS (%)
MODERATE REACTIONS (%)
MAJOR AND FATAL REACTIONS (%)
ARANDA ET AL21
71a
63.4
19.7
16.9
BORDA ET AL'7
497b
..
..
0.6
CROMARTY26
24c
12.5
87.5
87.5
HODDINOTT ET AL27
1 7d
27
38
35
OGILVIE, RUEDY20
193c
26.9
38.3
34.7
..
OGILVIE, RUEDY29
120e
..
.
71.7
120/562 (21.4)
71e
..
..
85.9
7 1/578 (12.3)
186f
..
..
..
186/2162 (8.6)
SPINO ET AL'8
REACTIONS PER DRUG EXPOSURE N (%) ..
497/6565 (7.6) ..
17/749 (2.3)
a Adverse drug reaction definite or probable drug reaction definite or probable as determined by clinical investigator (7he prescribing physician believed 490 of the ADRs to be definite or probable.) Degree of certainty of an ADR unstated d Adverse drug reaction probable Degree of certainty of an ADR not directly stated but definite or probable f Clinical ADR definite or probable b Adverse
The variability in all the categories tabulated can probably be attributed to the differences in the populations studied, the degree of certainty of ADRs, and the methods used to monitor for the occurrence of ADRs. In addition, these three factors make it difficult to discern any historical trends in hospital-based ADRs. The other limitation in trying to define any trends is that most of the studies were done during the same 12-year period, roughly 1965 to 1977. However, from studies where the methodology and patients observed appear to be similar, 2026,27,29 it seems reasonable to assume that about 15% of patients on general medical wards will experience an ADR. The case fatality rate can be calculated for only four of the surveys and ranged from 1.2%0 to 12 .9%211 (Table 518202,1130). Mikhael's figure of 1. 2% agrees closely with information that comes from the Ontario Medical Association (OMA) Adverse Drug Reaction Reporting Program. From the in114
Canadian Family Physician
VOL 37:
Januagy 1991
ception of this program in 1981 until the end of 1988, a total of 15 823 reports of suspected ADRs were received, with 289 deaths (1.8%) considered likely to be related to ADRs.i' Most reports to the OMA's program come from hospitals,3 and Mikhael's observations"" are derived from clinicopathological examination of surgical specimens from most of the hospitalized patients (57 881 of 72 653) during a 5-year period, thus accounting for the similarity of the results. The OMA data3' implicated NSAIDs in more deaths than any other group of drugs, but because proof of a direct cause-effect relation is not available and because the ratio of fatalities to number of prescriptions is unknown, the authors caution against drawing any inferences about the comparative safety of various classes of drugs. It is difficult to draw any conclusions about the drugs, other than digoxin and antibiotics, that most commonly cause ADRs
in hospitalized patients. Mikhael and Kacew23 found that estrogenic preparations caused the largest number of suspected ADRs, an association that probably was not recognized earlier because endometrial tissue is so rarely examined. The OMA data32 list antimicrobials, radiologic dyes, and psychotropics as the drugs causing the largest numbers of suspected ADRs. These figures, once again, reflect the hospital source of most of the reports to the OMA program. Most authors3"8'20'26 agree that ADRs prolong the length of hospitalization, although Hoddinott and colleagues27 claim that prolonged hospitalization resulted in more ADRs being detected, but was not the result of an ADR. The study by Spino and co-workers'9 was structured to help resolve this question. The researchers could not find any correlation between duration of stay and ADRs during the first 2 weeks of hospitalization, but for the third through fifth weeks of hospital stay, there was a significant correlation. Even when ADRs did prolong hospital stay, however, additional factors had a greater influence on the duration of stay. Ogilvie and Ruedy20'29 found a significant increase in the mortality rate for patients suspected of having an ADR compared with those without one. These authors20 also noted that the incidence of second ADRs both in patients admitted to hospital with an ADR and in patients who had a first ADR in hospital was higher than the incidence of a single ADR. Similarly, Soon'4 observed that nursing home patients with a history of a previous reaction were at higher risk of experiencing another ADR.
Effect of gender, age, and number of drugs In general, investigators seem to agree that women are more likely to experience an ADR than men,'213"7'23'25 with only one dissenting observation20; however, women receive more prescriptions than men. In the study by Borda and colleagues,'7 58% of the observed ADRs occurred in women, but they accounted for 57% of all drug exposures. Similarly, although Soon'3 reported that 30% of women had ADRs | compared with 19% of men, women had
72% of all drug exposures and 77% of the ADRs. In Soon's follow-up survey,'4 women had 7 1/% of drug exposures and 7O0% of the ADRs.
Tabl 5. IN-HOSPITAL CASE FATALITY RATE: Rates ranged
frnm 1.2% to 12.9%. STUDY
CAFATAITYR^E
%
MIYjHAEL30
1.2a
1015
SPINO ET ALI"
4.0
100
a Author did not distinguish betzeenADRs due to therapeutic exposure, medication error, or intentional or accidental overdose. __j
I
Grymonpre and associates25 are the only ones whose data show that women suffer a proportionately greater number of ADRs than would be expected based on their percentage of drug exposures; women made up 63% of admissions because of ADRs, while they had only 50% of drug exposures. Confounding the picture even further is the fact that some drugs to which women are exposed may be particularly likely to cause ADRs. Mikhael and Kacew23 noted that estrogenic preparations were the most common cause of ADRs in their clinicopathological examination of surgical specimens. The elderly seem to be the group most likely to suffer an adverse drug reaction,3"11 2,26,29,3132 although Ogilvie and Ruedy20 found that the age of patients with reactions was not significantly different from the age ofother patients. Mikhael and Kacew23 also noted that the highest incidence of definite tissue adverse reactions occurred in the 41 to 60 age group and concluded that younger individuals are equally as susceptible to adverse reactions as the elderly, except that the causative agent may vary between age groups. The increased risk of suffering an adverse reaction could reflect the greater number of medications that the elderly are prescribed rather than age per se.33 Hoddinott and associates27 and Cromarty26 showed a relationship between the number of drugs administered and the incidence of ADRs. Grymonpre and co-workers25 demCanadian Family Physician VOL 37: january 1991
115
onstrated a significant correlation between the incidence of a drug-related adverse patient event (DRAPE) and the number of drugs consumed by the patient before hospital admission, but did not find any correlation between age and the risk of a DRAPE. In her first study Soon'3 could not demonstrate a relationship between age and the risk of an ADR for either men or women; however, in her follow-up work,'4 women did show a significant increase in ADRs by age. Conversely, the relationship between ADRs and the number of drugs that were present initially for both men and women'3 disappeared for men when Soon repeated her work. 14 In both instances'3'14 women over the age of 70 had an increased susceptibility to ADRs as the number of drugs increased. Finally, Hutchinson and others24 found that neither the age of their patients nor the total number of drugs that they took affected the risk of an adverse reaction to an individual drug course. They explained that the patient was used as the unit of analysis in studies that found age and number of drugs to be risk factors and that the findings could be the result of the simple additive effect of reactions to individual drugs. Their study, on the other hand, used the drug course as the unit of analysis. A second possible confounding factor is that Hutchinson and colleagues monitored a group of outpatients, while most of the other work in the area has been on institutionalized patients. None of the authors studying this question appear to have considered the possibility that the correlation of the incidence of ADRs with the number of drugs prescribed could reflect primarily increasing severity of disease rather than number of drugs prescribed,34 although both Hutchinson and colleagues24 and Grymonpre and associates25 did control for the number of illnesses.
Under-reporting of ADRs Adverse drug reactions are probably greatly under-reported. When Reynolds'2 extrapolated his 5-month study to produce figures for a 12-month period, he calculated that the 232 doctors involved would have reported about 720 suspected adverse 116
Canadian
Famiy Physician VOL 37: 3anuary 1991
drug reactions. The total number of suspected reactions reported to the Health Protection Branch by all Canadian physicians in nonhospital settings is about 800 annually. Although Borda and associates'7 found 535 reactions among 936 monitored patients during a 3-year period, only 350 adverse reactions were reported for the other 75 373 nonmonitored patients admitted to the rest of the hospital over the same 3-year period. Under-reporting is probably partly a consequence of the low priority that the Health Protection Branch gives to the national Drug Adverse Reporting Program. "' Primary care physicians may not report reactions they do not recognize.15 The completeness of in-hospital reporting is directly related to the degree of certainty of the ADR and the severity of the reaction. Physicians also more frequently report ADRs that cause hospital admission or that prolong hospital stay. Finally, allergic or idiosyncratic reactions are reported more often than those that are dose-dependent and related to the basic pharmacologic actions of the drug."t' Some or all of these factors could also be operating in the ambulatory care setting, but no research has been conducted into this question. In-hospital ADR reporting has been markedly improved by attention to three factors.'37 The definition of an ADR was easily applicable, was operationally meaningful, and excluded trivial side effects of drug therapy if the lack of real significance was immediately apparent to physicians. Feedback reports made it clear to physicians that their reports were being carefully observed and evaluated. Finally, there was continuing emphasis on the existence of ADRs.
CONCLUSION The Canadian ADR literature suggests the following: 1. Patients in nursing homes have a 20% to 25% chance of developing an ADR. 2. The risk of developing an ADR increases with age or the number of drugs taken for institutionalized patients, but these have not been shown to be risk factors in the very few outpatient studies where they were examined.
3. Women account for more ADRs than men, but this phenomenon probably reflects their greater number of drug exposures. 4. Adverse drug reactions account for 10% to 20% of hospital admissions for people older than 50 years of age. 5. Adverse drug reactions account for about 5% of all admissions to medical wards. 6. About 15% of patients on medical wards will have an ADR. 7. Adverse drug reactions increase hospitalization time, but are not the most important factor in length of stay. 8. The case fatality rate for hospitalized patients suffering an ADR is 1% to 2%. 9. Digoxin and antibiotic interactions appear to be the single most important cause of ADRs in hospitalized patients, but this conclusion is based on studies 10 to 20 years old and may no longer be accurate. 10. Adverse drug reactions both in the hospital and in ambulatory care settings are
under-reported. There are also large gaps in our knowledge about ADRs: 1. We have very little information about ADRs in outpatients. 2. There have not been any recent studies of hospital-based ADRs, and none of the older ones, except the studies by Mikhael and Kacew,22,23,30 looked at the entire hospital population. 3. The drugs that most frequently cause ADRs are not known with any degree of certainty. 4. The relative safety of different classes of drugs and different drugs in the same class is not accurately known. 5. Reasons for under-reporting, especially in ambulatory care, have not been investigated.
Further research into ADRs needs to incorporate some basic methodological criteria: 1. The number of prescriptions for each drug, especially for outpatients, must be recorded in order to be able to calculate the true incidence of ADRs for specific
drugs and determine their relative safety. Provincial drug programs should be able to supply this information for the population that they cover- usually the elderly
and those on social assistance. (In Saskatchewan coverage is universal.) For drugs that may be at high risk of causing ADRs, or for new drugs where not enough is known about their potential for ADRs, a form of triplicate prescribing could be instituted. Under this program one copy of the prescription would be sent to the Drug Adverse Reporting Program in Ottawa. 2. Adverse drug reactions should be categorized based on a tested algorithm. 3. People need to be trained to look for ADRs; otherwise these reactions will be missed. 17 4. Studies on the frequency of ADRs should ideally be controlled to account for the fact that healthy individuals and patients frequently report symptoms that could be interpreted as ADRs if the person were taking a drug.4 Use of placebos containing excipients but not active ingredients could also help to differentiate true reactions." 5. Attempts to correlate the frequency of ADRs with the number of drugs being consumed need to control for the number and severity of the underlying illnesses. 6. Both patient-based and drug course-based reporting and analysis are needed. If the research question revolves around the risks of suffering an ADR for a patient or group of patients, patient-based reporting would be appropriate. Here such issues as the impact of interactions of multiple drugs can be examined. If the research question is primarily: "What is the risk of an ADR per course of a particular drug (and, by extension, how safe is a given class of drugs)?" appropriate "denominator data" indicating the number of drug courses given is essential for the interpretation of frequencies of ADRs. Finally, Canada needs a vigorous national reporting system for ADRs. There is some indication that we may be moving in that direction (E. Napke, personal communication, May 1990). A * 0 -0 0 - 0 0 0 - le
a * 0 0 e 0 0 0 - - 0 0 0 0 0 0 0 0 - - - 0 0 0 le 0 0 -
Acknowledgment I thank Dr Margaret E. Millson, who made many useful criticisms while reviewing the manuscript, especialy her suggestions for the use of the technique of triplicate prescribing and the addition of the conclusion Canadian Family Physician VOL 37: Janua?y 1991
117
about patient-based and drug course-based reporting and analysis. Most of the wording on the latter topic was borrowedfiom her review.
Requests for reprtints to: Joel Lexchin, MD, CCFP(EM), 121 Walmer Rd, Toronto, ON M5R 2X8 References 1. McLean WM. Some practical aspects of monitoring drug adverse reactions. Can 7 Hosp Phann 1973;26:247-5 1. 2. Nova Scotia Royal Commission on Health Care. Towards a new strategy. Halifax, NS: Nova Scotia Royal Commission on Health Care, 1989:116. 3. Taras-Zasowski KM, Einarson TR. Review of Canadian pharmacist involvement in adverse drug reaction reporting. Can J Hosp Pharm 1989;
42:105-8,133. 4. Karch FE, Lasagna L. Adverse drug reactions: a critical review. JAA4A 1975;234:1236-41. 5. Napke E. Adverse reactions: some pitfalls and postulates. In: Dukes MNG, ed. Side effects ofdrugs annual 7. Amsterdam, The Netherlands: Excerpta Medica, 1983:xvi-xxvi. 6. Karch FE, Smith CL, Kerzner B, MazzulloJM, Weintraub M, Lasagna L. Adverse drug reactions - a matter of opinion. Clin Pharmacol 7her 1976; 19:489-92. 7. Koch-WeserJ, Sellers EM, Zacest R. The ambiguity of adverse drug reactions. Eur3 Clin Pharmacol 1977;11:75-8. 8. Kramer MS, Leventhal JM, Hutchinson TA, Feinstein AR. An algorithm for the operational assessment of adverse drug reactions. 1. Background, description and instructions for use. JAA/JA 1979;242:623-32. 9. Krogh CME, Gillis MC, Bisson R, et al, eds. Compendium ofpharmaceuticals and specialties. 24th ed. Ottawa, Ont: Canadian Pharmaceutical Association, 1989. 10. Turriff C, Parison D. Reporting and monitoring drug adverse reactions. Ottawa, Ont: Health and Welfare Canada, May 1989; technical report no. 6. (Program evaluation study of the drug safety, quality
and efficacy program.) 11. Reynolds JL. Monitoring adverse drug reactions. A preliminary study. Can Fam Physician 1981;27:1235-8. 12. Reynolds JL. A survey of adverse drug reactions in family-practice. Can Fam Physician 1984; 30:81-4. 13. Soon JA. Developing an adverse drug reaction program in nursing homes. Can J Hosp Pharm 1980;33: 158-61,164. 14. Soon JA. Assessment of an adverse drug reaction monitoring program in nursing homes. Can J Hosp Pharm 1985;38:120-5. 15. Asthana S, Sood VP. Prescribing for the elderly: one hospital's experience. Geriatr Med 1987; 3:113-7. 16. Boston Collaborative Drug Surveillance Program. Hospital admissions due to adverse drug reactions. Arch Intern Med 1974; 134:219-23. 17. Borda IT, Napke E, Stapleton C. Drug surveillance data in a Canadian hospital. Can Med Assoc J 1976;114:517-22.
118
Canadian Family Physician VOL 37: January 1991
18. Spino M, Sellers EM, Kaplan HL, Stapleton C, MacLeod SM. Adverse biochemical and clinical consequences of furosemide administration. Can Med Assoc j 1978; 118:1513-8. 19. Spino M, Sellers EM, Kaplan HL. Effect of adverse drug reactions on the length of hospitalization. Am 7 Hosp Pharm 1978;35:1060-4. 20. Ogilvie RI, RuedyJ. Adverse drug reactions during hospitalization. Can Med Assoc j 1967; 97:1450-7. 21. Aranda JV, Portuguez-Malavasi A, Collinge JM, Germanson T, Outerbridge EW. Epidemiology of adverse drug reactions in the newborn. Dev Phannacol Ther 1982;5:173-84. 22. Mikhael NZ, Kacew S. Monitoring of adverse drug reactions by means of autopsy tissue examination. Hum Toxicol 1984;3:133-40. 23. Mikhael NZ, Kacew S. Adverse tissue reactions to drugs: a five-year surveillance study. Hum Toxicol 1985;4:583-90. 24. Hutchinson TA, Flegel KM, Kramer MS, Leduc DG, Ho Ping Kong H. Frequency, severity and risk factors for adverse drug reactions in adult outpatients: a prospective study. J Chronic Dis 1986;39:533-42. 25. Grymonpre RE, Mitenko PA, Sitar DS, Aoki FY, Montgomery PR. Drug-associated hospital admissions in older medical patients. J Am Geriatr Soc 1988;36: 1092-8. 26. Cromarty AJ. Initiation of a drug adverse reaction monitoring and reporting system [Abstract]. Can J Hosp Pharm 1977;30:84. 27. Hoddinott BC, Gowdey CW, Coulter WK, ParkerJM. Drug reactions and errors in administration on a medical ward. Can Med Assoc j 1967;97:1001-6. 28. Lou P, CampbellJS, Henderson IWD, Napke E, Mikhael NZ. Drug adverse reactions in autopsy cases. A preliminary report. Rx Bull 1974;5:2-3. 29. Ogilvie RI, RuedyJ. An educational program in digitalis therapy. JAA'IA 1972;222:50-5. 30. Mikhael NZ. Adverse reactions to drugs in the hospital milieu. Ann R Coll Phys Surg Can 1985; 18:477-83. 31. Brennan M, Gowdey CW. Adverse drug reactions: a review of fatalities reported in Ontario. Ont Med Rev 1989;56(Aug):23-6. 32. Gowdey CW, Brennan M. Adverse drug reaction reporting program of the Ontario Medical Association: the first 3 years. Can Med Assoc] 1985; 132:19-23. 33. McKim WA, Mishara BL. Drugs and aging. Toronto, Ont: Butterworths, 1987. 34. Klein LE, German PS, Levine DM. Adverse drug reactions among the elderly: a reassessment. JAm Geriatr Soc 198 1;29:525-30. 35. Petrie JC, Howie JGR, Durno D. Awareness and experience of general practitioners of selected drug interactions. Br Medjt 1974;2:262-4. 36. Koch-WeserJ, Sidel VW, Sweet RH, Kanarek P, Eaton AE. Factors determining physician reporting of adverse drug reactions. N Engl J Med 1969;280:20-6. 37. Napke E, Stevens DGH. Excipients and additives: hidden hazards in drug products and in product substitution. Can Med Assoc J 1984; 13 1: 1449-52.
