Incident-Event Rate of Preventable Drug-Related Morbidity in Older Adults in Nova Scotia

Abstract

Background: Preventable drug-related morbidity (PDRM) is an issue of special concern in geriatric medicine. The objective of this study was to determine the incident-event rate of PDRM in older adults in one area of Nova Scotia, Canada. Methods: The study population consisted of seniors enrolled in the Nova Scotia Pharmacare program in the western part of the Halifax Regional Municipality. Claims information for all inpatient admissions, emergency room and physician office visits, ambulatory prescription medication use, and laboratory results was contained within a database. PDRM indicators were developed in a previous stage of the study. The incident-event rate was determined by (1) identifying the number of occurrences for each outcome associated with the indicators and (2) identifying the number of outcomes that followed a specified pattern of care. Results: The study population consisted of 22,453 older adults. The computerized search identified 29,796 outcomes, 3,277 of which matched the associated pattern of care and were thus considered occurrences of PDRM. The incident-event rate of PDRM in seniors in Nova Scotia was 110 instances of PDRM per 1,000 occurrences. Conclusions: This study has helped quantify the magnitude of the problem of PDRM in older adults in this region of Nova Scotia.

Full text

Canadian Journal of Geriatrics Volume 9,Issue 5,2006 159
Incident-Event Rate of Preventable Drug-Related Morbidity in Older Adults
in Nova Scotia
Neil J. MacKinnon, PhD, RPh, FCSHP; Nicole R. Hartnell, MSc(Pharm); Susan K. Bowles, PharmD; Susan A. Kirkland, PhD;
Erika J.M. Jones, BSc(Pharm)
ABSTRACT
Background
Preventable drug-related morbidity (PDRM) is an issue of special concern in geriatric medicine. The objective of this study
was to determine the incident-event rate of PDRM in older adults in one area of Nova Scotia, Canada.
Methods
The study population consisted of seniors enrolled in the Nova Scotia Pharmacare program in the western part of the
Halifax Regional Municipality. Claims information for all inpatient admissions, emergency room and physician office visits,
ambulatory prescription medication use, and laboratory results was contained within a database. PDRM indicators were
developed in a previous stage of the study. The incident-event rate was determined by (1) identifying the number of
occurrences for each outcome associated with the indicators and (2) identifying the number of outcomes that followed a
specified pattern of care.
Results
The study population consisted of 22,453 older adults.The computerized search identified 29,796 outcomes, 3,277 of which
matched the associated pattern of care and were thus considered occurrences of PDRM. The incident-event rate of PDRM
in seniors in Nova Scotia was 110 instances of PDRM per 1,000 occurrences.
Conclusions
This study has helped quantify the magnitude of the problem of PDRM in older adults in this region of Nova Scotia.
Key words: preventable drug-related morbidity, medication use, prescribing
SOMMAIRE
Généralités
La mortalité évitable liée à l’usage de drogues est un sujet d’intérêt spécial en gériatrie. L’objectif de cette étude était de
déterminer le taux de cette manifestation/incident chez les adultes plus âgés dans une région de la Nouvelle-Écosse au
Canada.
Méthodes
La population à l’étude était composée de personnes âgées inscrites au programme Pharmacare de la Nouvelle-Écosse
dans la partie ouest de la municipalité d’Halifax. L’information sur les réclamations pour toutes les hospitalisations, les
visites à l’urgence ou au cabinet du médecin, l’emploi de médicaments à l’extérieur de l’hôpital et les résultats des tests
de laboratoire était contenue dans une base de données. Les indicateurs de la morbidité évitable liée à l’usage de drogues
ont été déterminés à la phase précédente. Le taux de manifestation/incident a été déterminé (1) en identifiant la fréquence
de chaque résultat associée aux indicateurs et (2) en identifiant le nombre des résultats qui a suivi un type de soins spécifié.
Résultats
La population à l’étude était composée de 22 453 adultes plus âgés. La recherche par ordinateur a identifié 29 796
résultats, dont 3 277 correspondaient au type de soins et étaient considérés comme la mortalité évitable liée à l’usage de
drogues. La fréquence de la mortalité évitable liée à l’usage de drogues chez les personnes âgées de la Nouvelle-Écosse a
été de 110 sur 1 000 occurrences.
Conclusions
Cette étude a permis de quantifier l’ampleur du problème de mortalité évitable liée à l’usage de drogues chez les adultes
âgés dans cette région de la Nouvelle-Écosse.
Neil J. MacKinnon, PhD, RPh, FCSHP
,and
Nicole R. Hartnell, MSc(Pharm)
:College of Pharmacy,Dalhousie University;
Susan K. Bowles, PharmD
:College of Pharmacy and Division of Geriatric
Medicine, Dalhousie University, and Centre for Health Care of the Elderly, Capital District Health Authority;
Susan A. Kirkland, PhD
: Department of Community Health and Epidemiology,Faculty of
Medicine; Division of Geriatric Medicine, Faculty of Medicine, Dalhousie University;
Erika J.M. Jones, BSc(Pharm)
: College of Pharmacy,Dalhousie University, Halifax, Nova Scotia
Address for correspondence: Dalhousie University College of Pharmacy,5968 College St, . Halifax, NS,B3H 3J5
Conflict of interest: None declared
Can J Geriatr
2006;9:159-163

160 Volume 9,Issue 5,2006 Canadian Journal of Geriatrics
Problems with the medication use system have been well
documented in Canada. In the sentinel Canadian Adverse Events
Study, 7.5% of patients admitted to the studied acute care hospitals
were estimated to have experienced an adverse event.1Drug- or fluid-
related events accounted for 23.6% of all events, making it the second
single largest category of adverse events. In a recent article, 134 patients
(average age = 67.3 years) were found to have an average of 3.59 drug-
therapy problems at the time of discharge from hospital, with 72.1%
deemed to be of significant or very significant clinical impact.2An
editorial in this journal reviewed issues regarding the management of
medication use in older adults and proposed strategies for
improvement.3
Several terms are used in the literature to describe the various types
of adverse drug-related complications that patients experience. Some
of the more commonly used terms include adverse drug events,
adverse drug reactions, drug-related morbidities, drug-related
problems, medication errors, and medication misadventures. One of
the broadest of these descriptors is drug-related morbidity (DRM),
defined as the failure of a therapeutic agent to produce the intended
therapeutic outcome or the clinical or biosocial manifestation of
unresolved drug-related problems.4As Hepler argues, “A DRM is an
unintended patient injury with a scientifically plausible relationship
either to (a) drug therapy or (b) an untreated indication for drug
therapy. Plausible means a valid theoretical relationship and
chronology.”5It has been estimated that approximately 50% of DRMs
are preventable.6Admittedly, the term preventability can be a source of
misunderstanding. As described by Hepler and Strand, a preventable
drug-related morbidity (PDRM) has four unique characteristics.
Given an adverse clinical outcome, a pre-existing DRP must have been
recognizable and the adverse outcome or treatment failure must have
been foreseeable. In addition, the cause or causes ofthe DRP and the
outcome must have been both identifiable and controllable.4
As the demand for accountability in health care has increased in
recent years, so has research related to the measurement of adverse
drug-related complications, including the creation of indicators or
performance measures. Typically, such indicators are developed
through consensus-building techniques using a panel of experts, and
the resulting indicators can then be used to measure the magnitude of
the problem and can help in the design of targeted interventions. To
date, clinical indicators of PDRM have been developed and tested in
the United States,7,8 United Kingdom,9and Canada.10 The objective of
this study was to use the Canadian consensus-approved clinical
indicators for PDRM to determine the incident-event rate of PDRM in
older adults in one area of Nova Scotia, Canada.
Methods
This study was approved by the Dalhousie University Health Sciences
Research Ethics Board.
Study Design and Data Sources
This was a retrospective, population-based study using administrative
data from the Population Health Research Unit (PHRU) at Dalhousie
University. The PHRU maintains three principal databases in support
of health services research: the Medical Services Insurance Physician
Billings Database, the Canadian Institute for Health Information
Hospital Separations Database, and the Nova Scotia Pharmacare
Database.
The study population was drawn from the Nova Scotia Pharmacare
Registration File using postal code fields. The western area of Halifax
Regional Municipality was chosen to include those who would use the
Queen Elizabeth II Health Sciences Centre (QEII) as their primary
hospital. There are approximately 200,000 people in this area, with
about 15% of the population enrolled in the Pharmacare program.
Seniors (age 65 years and over) enrolled in Pharmacare for the entire
duration of the study period (January 1998 through December 1999)
were included in the study.
Through the use of encrypted health card numbers, the individual
records from each of the databases of the study population were linked
for the study period. The resulting database contained the relevant
claims data for each study subject, which included age and gender; all
hospital admissions, including admission dates; lengths of stay;
diagnostic codes and procedural codes; all physician visits, including
date; diagnosis and fee code; all medication prescriptions filled under
the Pharmacare Program (including date, drug identification number
[DIN], dosage, and days’ supply). To incorporate clinical laboratory
information, the records from the QEII Laboratory Database were
linked to the PHRU databases through a process that has been
established for the linkage of external databases.
Indicator Development
Detailed methods of the development of the PDRM indicators have
been published.10In short, using a combination of the modified Delphi
Technique and the nominal group technique with 24 experts in
geriatric medicine, clinical pharmacologists, and general practitioners,
52 consensus-based clinical indicators of PDRM in older adults in
Canada were generated. Others who have developed indicators of
PDRM have followed a similar methodology.7–9 Each indicator had a
similar structure: an inappropriate pattern of care that preceeded an
adverse outcome, such as an emergency room visit or hospitalization.
Study Outcomes
The outcomes of interest for this study were (1) the number of
occurrences for each particular outcome of the 52 indicators and
(2) the number of outcomes that followed a specified pattern of care
and thus were considered PRDMs.
The study database was first searched for the outcome(s) related to
each specific indicator. This was done by searching for the diagnosis
codes (International Classification of Diseases [ICD]-9-CM) related to
these outcomes. Once the outcomes (morbidities) were identified,
each of these cases was searched to determine whether they were
preceded by the associated pattern of care. This was accomplished by
searching for particular ICD-9-CM codes and/or particular DINs. A
PDRM was judged to have occurred if both the outcome and the
pattern of care matched the specific operational definition of the
PDRM. For the purposes of this article, cases judged to have been a
PDRM are called a “hit.”
Incident-event rates were determined by dividing the total number
of hits by the total number of occurrences found in the database for
the study period. Incidence rate is defined as the occurrence of new
events among the total population at risk. In this case, we wanted to
include all of the events because it was felt that each PDRM was unique
(new) and was subsequently resolved, and therefore independent of
the next PDRM. For the denominator, it was similarly important to
take the same factors into account. Thus, to exclude events in the
numerator just because they occurred after the first event did not seem
Incident-Event Rate of Preventable Drug-Related Morbidity in Nova Scotia

Canadian Journal of Geriatrics Volume 9,Issue 5,2006 161
to be a defensible position, and to limit the denominator to the
number of people rather than the number of events would be
incorrect as well. In addition to calculating the overall incident-event
rate for the study population, the authors also calculated a process-to-
outcome value (PTOV) for each indicator.8As the name suggests, the
PTOV helps quantify the relationship of the occurrence of
inappropriate patterns of care (processes) to outcomes. This is
important as one would likely be more interested in indicators for
which the PTOV was high (i.e., close to 1) as a high PTOV value
indicates a strong association between the pattern of care and its
associated outcome. Conversely, if the pattern of care that is deemed to
be inappropriate is infrequently associated with adverse outcomes, one
could question how much effort should be directed toward changing
that inappropriate pattern of care.
Results
The study population consisted of 22,453 older adults, of whom 8,809
(39.2%) were men and 13,644 (60.8%) were women. The age and
gender of the study population can be seen in Table 1. Individuals
represented in the study population were predominantly between the
ages of65 and 79 years (72%). Only a very small percentage of the
study population was over the age of 90 years (4.6%).
When the 52 clinical indicators of PDRM were applied to the study
database, 29,796 occurrences associated with outcomes consistent with
one of the 52 indicators were found in 22,453 patients (some patients
experienced more than one occurrence). Of these 29,796 occurrences,
3,277 matched the associated pattern of care for at least one of the
indicators and were thus considered to constitute a PDRM. The overall
incident-event rate of PDRM in the population was 110 per 1,000
occurrences. Table 2 shows the five clinical indicators for PDRM that
accounted for the largest percentage of all instances of PDRM. These
five indicators were responsible for 76% of the total instances of
PDRM in this population of older adults. Twenty-six of the 52
indicators had at least 10 hits, whereas seven indicators did not have a
single hit.
MacKinnon et al.
Table 1. Age and Gender of the Study Population
(N= 22,453)
Age (yr)
n
(%)
Men
8,809 (39.2)
65–69 2,795 (12.4)
70–74 2,214 (9.9)
75–79 1,923 (8.6)
80–84 1,122 (5.0)
85–89 545 (2.4)
90 + 210 (0.9)
Women 13,644 (60.8)
65–69 3,150 (14.0)
70–74 3,078 (13.7)
75–79 3,003 (13.4)
80–84 2,167 (9.7)
85–89 1,422 (6.3)
90 + 824 (3.7)
Table 2. Five Indicators Accounting for the Largest
Percentage of All Instances of Preventable Drug-
Related Morbidity
PDRM
Indicator
1. This outcome has occurred after
the pattern of care below:
Physician office visit/ER
visit/hospitalization owing to
pneumonia during the influenza
season
This is the pattern of care:
1. Over 65 years of age
2. No contraindications to influenza
vaccine (anaphylactic allergy to
eggs, Guillain-Barré syndrome
following previous influenza
vaccine)
3. Has not received annual
influenza vaccine
1,501 45.8
2. This outcome has occurred after
the pattern of care below:
ER visit/hospitalization owing to
symptoms of hyperthyroidism
This is the pattern of care:
1. Use of a thyroid or antithyroid
agent (e.g., levothyroxine,
propylthiouracil)
2. T4/TSH not done within 6 weeks
after initiation of therapy and at
least every 12 months thereafter
519 15.8
3. This outcome has occurred after
the pattern of care below:
ER visit/hospitalization owing to
respiratory distress
This is the pattern of care:
1. Diagnosis of asthma/COPD
2. Use of a β-blocker
236 7.2
5. This outcome has occurred after
the pattern of care below:
Hospitalization owing to fracture in a
patient with osteoporosis
This is the pattern of care:
1. Patient (male or female) has
diagnosis of osteoporosis
2. Is at least 65 years of age
3. Is not receiving adequate levels
of calcium and vitamin D (either
dietary or supplements)
90 2.7
4. This outcome has occurred after
the pattern of care below:
Physician visit/ER visit/hospitalization
owing to hypo-/hyperthyroidism
This is the pattern of care:
1. Use of a thyroid replacement
agent
2. No monitoring of T4/TSH at 3 to 6
months following initiation of
therapy or every 12 months
thereafter
139 4.2
COPD = chronic obstructive pulmonary disease; ER = emergency room;
PDRM = preventable drug-related morbidity; T4= thyroxine;
TSH = thyroid-stimulating hormone.
*For an event to be considered a hit, the outcome had to have occurred, and this
must have been proceeded by the matching pattern of care as specified in the
indicator.If the pattern of care occurred without the following matching outcome, this
would not be considered a hit.
Number of
PDRM “Hits”*
Found in the Study
Population
% of All PDRM
“Hits” Found in the
Study Population
(
n
= 3,277)

Table 3. Five Preventable Drug-Related Morbidity
Indicators with the Greatest Process-to-Outcome Value
PDRM
Indicator
1. This outcome has occurred after
the pattern of care below:
ER visit/hospitalization owing to
symptoms of hyperthyroidism
This is the pattern of care:
1. Use of a thyroid or antithyroid
agent (e.g., levothyroxine,
propylthiouracil)
2.T
4/TSH not done within 6 weeks
after initiation of therapy at
least every 12 months thereafter
836 1:1.6519
2. This outcome has occurred after
the pattern of care below:
Physician office visit/ER
visit/hospitalization owing to
pneumococcal pneumonia or sepsis
This is the pattern of care:
1. Over 65 years of age
2. No contraindications to
pneumococcal vaccine
3. Has not received pneumococcal
vaccine within last 6 years
152 1:2.268
3. This outcome has occurred after
the pattern of care below:
ER visit/hospitalization owing to
hypoglycemia
This is the pattern of care:
1. Use of a long-acting
antihyperglycemic agent
(e.g., chlorpropamide)
2. Hemoglobin A1clevel not
measured at least every
6 months
86 1:2.729
4. This outcome has occurred after
the pattern of care below:
Physician office visit/ER
visit/hospitalization owing to
influenza-related pneumonia
This is the pattern of care:
1. Over 65 years of age
2. No contraindications to
influenza vaccine (anaphylactic
allergy to eggs, Guillain-Barré
syndrome following previous
influenza vaccine)
3. Has not received annual
influenza vaccine
4323 1:2.91501
5. This outcome has occurred after
the pattern of care below:
Physician visit/ER
visit/hospitalization owing to hypo-
/hyperthyroidism
This is the pattern of care:
1. Use of a thyroid replacement
agent
2. No monitoring of T4/TSH at 3 to
6 months following initiation of
therapy or every 12 months
thereafter
836 1:6.0139
Number
of
Outcomes
Number of
PDRM
“Hits”
PTOV
ER = emergency room; PDRM = preventable drug-related morbidity; PTOV = process-
to-outcome value; T4= thyroxine; TSH = thyroid-stimulating hormone.
162 Volume 9,Issue 5,2006 Canadian Journal of Geriatrics
The five PDRM indicators with the greatest PTOVs are shown in
Table 3. The five indicators’ PTOVs range from 1:1.6 to 1:6.0. Three of
the five indicators with the greatest PTOVs are also among the five
indicators that account for the largest percentages of all instances of
PDRM.
Discussion
This study has shown that a significant number of PDRMs occur in
Nova Scotia seniors and that a small number of indicators are
responsible for a majority of the specific types of PDRMs observed.
Indeed, the top two indicators were responsible for over 60% of all
hits. This suggests that simply by ensuring that seniors receive the
annual influenza vaccine if they have no contraindication and that
regular monitoring of thyroxine (T4)/thyroid-stimulating hormone
(TSH) levels is performed for seniors on a thyroid or antithyroid
agent, a significant amount of patient morbidity can be reduced and
economic savings can be realized. Nonetheless, a systematic approach
that helps promote the uptake of these two important activities can be
difficult to implement.
Although information on the magnitude of PDRM is limited to
date, this study does compare favourably with the existing research in
certain aspects but differs considerably in other areas. In the two
previous studies that measured incidence rate, a small number of
PDRM indicators were responsible for a large number of hits, as in our
study.In a study of older adults in a Florida hospital-based health care
plan, the top five indicators were responsible for 46.8% of all hits
found,11 whereas in a study of nine primary care sites in the United
Kingdom, the top four indicators were responsible for 60.0% of all
hits.12 As one would expect, the indicators in the three countries differ
considerably owing to differences in clinical practice13;however, the
indicator with the second greatest number of hits in our study
(hyperthyroidism following improper T4/TSH monitoring) had the
third greatest number of hits in the Florida study.A direct comparison
of the incidence of PDRM is not possible as the previous studies used
an incidence rate, whereas we used an incident-event rate for the
reasons described previously.Yet it is still possible to conclude that we
observed more hits than in the other studies. Part of the difference may
be explained by the fact that neither of these studies included the
PDRM indicator that was responsible for the most hits we observed
(influenza-related pneumonia following no influenza vaccine).
Indeed, this indicator was responsible for nearly half of all of our hits.
Even after excluding this indicator from our analysis, we appear to
have observed a greater frequency of PDRM hits in our population. In
a meta-analysis of 15 studies, a median preventable drug-related
hospital admissions prevalence rate of 4.3% was observed.14 A
difference from this rate may be expected, at least in part, as we
included morbidities in addition to only hospital admissions and
studied a population (older adults) in which one would expect to see
more drug-related morbidities.
What are the practice implications of the findings of this study? As
argued by Easton-Carter and colleagues, studies that investigate drug-
related morbidity and mortality can provide important information
about the quality use of medicines.15 A better picture of the adverse
medication-related issues in this geriatric population has been
realized, although, ideally, these indicators would be used on an
ongoing basis as part of the continuous quality improvement activities
Incident-Event Rate of Preventable Drug-Related Morbidity in Nova Scotia

Canadian Journal of Geriatrics Volume 9,Issue 5,2006 163
of a provincial department of health, regional health authority, or
family practice. For example, in one general practice in the United
Kingdom, PDRM indicators have been used in conjunction with
electronic patient records and multidisciplinary discussion forums to
create changes in medicine management.16 The real value appears to be
in the use of these indicators to better target future interventions. As
Hepler suggests,“The process components of PDRM indicators can be
used prospectively to identify patients who are receiving potentially
inappropriate processes of care, before a DRM has time to occur.”17
Additionally,use of the PTOV can help identify adverse outcomes that
have a strong relationship to medication use.
Our study does have some important limitations that should be
noted. We included only the 52 PDRM indicators that were developed
in a previous stage of this study. Thus,we acknowledge that other types
of PDRMs exist in older adults and our observed incident-event rate is
a lower-bond estimate of the magnitude of PDRM. Others may
question the validity of using a definition of PDRM that mandates a
clear link between a pattern of care judged to be inappropriate by a
panel of experts and the resulting outcome. Moreover, our
observations were made using primarily administration claims data.
In a previous stage of this study, 4 of the 52 indicators were assessed for
validity and reliability and were found to have an overall sensitivity of
81% and a specificity of 56%.18 Thus, we are fairly confident that the
hits found using our computerized search process were, indeed, real
clinical PDRMs.
However, there are some limitations specific to the indicator for
influenza vaccine. First, this study evaluated a time period in which
there was a poor match between the viral strain of influenza
circulating in the community and that contained in the annual
influenza vaccine. Furthermore, although admission rates for
pneumonia do increase significantly during influenza season,19 it is
important to note that not all admissions for pneumonia during
influenza season will be related to influenza. Thus, indicator number
one may represent an overestimate of the morbidity associated with
not receiving an annual influenza vaccine.
Conclusions
This study employed the use of clinical indicators of PDRM to identify
the incident-event rate of PDRM in a population of older adults in
Nova Scotia. Additionally, the results of this study will allow the Nova
Scotia Department of Health, family physicians, and geriatricians to
better design interventions to reduce PDRM in Nova Scotian seniors.
Acknowledgements and Disclaimer
Funding for this study was provided by the Canadian Institutes of
Health Research. The authors would like to acknowledge the
contributions of Priti Flanagan, Jocelyn LeClerc, Heather Robertson,
Megan Rogers, Chris Skedgel, and Mark Smith.
Although this research is based on data obtained from the
Population Health Research Unit, the observations and opinions
expressed are those of the authors and do not represent those of the
Population Health Research Unit.
References
1. Baker GR, Norton PG, Flintoft V, et al. The Canadian Adverse
Events Study: the incidence of adverse events among hospital
patients in Canada. CMAJ 2004;170:1678-86.
2. Nickerson A, MacKinnon NJ, Roberts N, et al. Drug-therapy
problems, inconsistencies and omissions identified during a
medication reconciliation and seamless care service. Healthcare
Quarterly 2005;8:65-72.
3. MacKinnon NJ. Improving the management of medication use in
older adults. Geriatr Today J Can Geriatr Soc 2002;5:63-5.
4. Hepler CD, Strand LM. Opportunities and responsibilities in
pharmaceutical care. Am J Hosp Pharm 1990;47:533-43.
5. Hepler CD. Understanding adverse drug therapy outcomes. In:
Hepler CD, Segal R, eds. Preventing Medication Errors and
Improving Drug Therapy Outcomes. Boca Raton, FL: CRC Press;
29-71; 2003.
6. Hepler CD, Grainger-Rousseau TJ. Pharmaceutical care versus
traditional treatment: is there a difference? Drugs 1995;49:1-10.
7. MacKinnon NJ, Hepler CD. Preventable drug-related morbidity in
older adults: indicator development. J Managed Care Pharm
2002;8:365-71.
8. Faris RJ. Explicit Definitions to Identify Preventable Drug-Related
Morbidity in an Elderly Population and Their Use as an Indicator
of Quality in the Medication Use System [dissertation]. Gainesville,
FL: University of Florida; 2001.
9. Morris CJ, Cantrill JA. Preventing drug-related morbidity—the
development of quality indicators. J Clin Pharm Ther 2003;28:295-
305.
10. Robertson HA, MacKinnon NJ. Development of a list of
consensus-approved clinical indicators of preventable drug-related
morbidity in older adults. Clin Ther 2002;24:1595-604.
11. MacKinnon NJ, Hepler CD. Indicators of preventable drug related
morbidity in older adults: use within a managed care organization.
J Managed Care Pharm 2003;9:134-41.
12. Morris CJ, Rodgers S, Hammersley VS, et al. Indicators for
preventable drug-related morbidity: application in primary care.
Qual Saf Health Care 2004;13:181-5.
13. Morris CJ, Cantrill JA. Preventable drug-related morbidity
indicators in the US and UK. J Managed Care Pharm 2002;8:372-7.
14. Winterstein AG, Sauer BC, Hepler CD, et al. Preventable drug-
related hospital admissions. Ann Pharmacother 2002;36:1238-48.
15. Easton-Carter KL, Chapman CB, Brian JE. Issues in the
interpretation and design of studies investigating drug-related
morbidity and mortality. J Pharm Pract Res 2002;32:235-9.
16. Morris CJ, Cantrill JA, Bate JR. How the use of preventable drug-
related morbidity indicators can improve medicines management
in primary care. The Pharmaceutical Journal 2003;271:682-6.
17. Hepler CD. A promising way forward to the goal of systematic
medicines management. The Pharmaceutical Journal 2003;271:674.
18. Flanagan PS, MacKinnon NJ, Bowles SK, et al. Validation of four
clinical indicators of preventable drug-related morbidity. Ann
Pharmacother 2004;38:20-4.
19. Upshur RE, Goel V. Measuring the impact of influenza on hospital
admission rates of he elderly in Ontario, a 5-year admission rate
analysis, 1988-1993. Can J Public Health 2000;91:144-7.
MacKinnon et al.