Osteoarthritis (OA) is a joint disease characterized
by degenerative changes in the articular
cartilage and progressive joint deterioration. It
is the most common type of joint disease and is a major
cause of disability.1 In Western countries, radiographic
evidence of OA is present in most individuals by 65
years of age and in about 80% of those older than 75
years.2 In Canada, an estimated 4 million individuals 15
years and older experienced arthritic disorders in 2000,
and the direct and indirect costs of these diseases in
1998 were estimated to be at least Can$4.4 billion.3
Osteoarthritis is expected to account for about half of
these costs.4
Osteoarthritis typically presents as pain involving 1
or several joints. The primary goal of treatment is to
relieve pain, thereby improving function and quality of
life. Treatment guidelines in Canada5 and the United
States6 recommend nonsteroidal anti-inflammatory
drugs (NSAIDs) as second-line treatment (after acetaminophen)
for mild OA and as first-line treatment for
moderate-to-severe OA.
Canadian guidelines further recommend that, in
patients at high risk for upper gastrointestinal (GI)
adverse events, the use of traditional NSAIDs should be
accompanied by prophylaxis with misoprostol or a proton
pump inhibitor.5,7 Risk factors include a history of
peptic ulcer disease, GI bleeding, cardiovascular disease,
and aged 65 years or older.8,9
Alternatively, the use of coxib NSAIDs (cyclooxygenase-2–specific inhibitors) as first-line treatment for
high-risk patients is also recommended by Canadian
guidelines.5,7 Large clinical trials have demonstrated
that these drugs have efficacy similar to that of traditional
(non-coxib) NSAIDs10-12 and produce fewer GI
adverse effects.13-16 Although coxibs are not medically
contraindicated for low-risk patients, economic models
suggest that they may not be cost effective in this
group.17,18
The Canadian Osteoarthritis Rx (CANOAR) study
was designed to characterize OA patients in Ontario for
whom primary care physicians prescribed NSAIDs. This
article examines the use of NSAIDs in this study cohort
in relation to recommendations for their appropriate
use as published in the Second Canadian Consensus
Conference arthritis treatment guidelines.5
METHODS
The CANOAR study was an observational study of a
cohort of primary care physicians with high-volume
NSAID prescribing practices. The objective was to
determine the clinical circumstances under which
NSAIDs were prescribed for OA patients in routine clinical
practice. The study took place in Ontario, a
province of 11 million people under a single healthcare
system.
Before conducting the study, a steering committee
was formed with 2 primary care physicians, 3 rheumatologists,
a biostatistician, a health economist, the chief
executive officer of The Arthritis Society, and a patient
with OA. The committee met to review and modify an
initial draft protocol and data collection form. A review
of the modified protocol by 4 rheumatologists, 4 epidemiologists,
and 2 health economists further focused
the study design.
One-page clinical encounter data forms using anonymous
sequential patient identifiers were designed and
pilot-tested in collaboration with 7 primary care physicians.
The form recorded patient demographics, location
of principal OA joint involvement, global physician
and patient assessments, concomitant medical problems,
history of clinically significant GI events, current
drugs, and previous use of and current recommendations
for OA therapies and gastroprotective agents
(GPAs). The form allowed identification of the following
GI risk factors listed in the Second Canadian Consensus
Conference guidelines: history of a clinically significant
GI event, advanced age (aged 65-74 years was scored as
1 risk factor, while age ≥ 75 years was scored as 2 risk
factors), current and continuing use of warfarin sodium,
current and continuing use of corticosteroids, hypertension,
congestive heart failure, renal insufficiency,
and hepatic insufficiency. Renal risk factors recorded
were renal insufficiency, hypertension, congestive heart
failure, hepatic insufficiency, and concomitant antihypertensive
drugs.
The forms were designed to be usable as part of the
routines of normal clinical practice. Seven pilot physicians
tested the protocol by enrolling 10 patients and
provided final assessments and recommendations about
office procedures for patient enrollment and consent,
data form completion, and data transmission to the
data center by toll-free, never-busy facsimile. The pilot
physicians also provided estimates of required per
patient time, for inclusion in the physician recruitment
letter.
After receiving ethics review board approval, the
CANOAR study proceeded with physician recruitment.
Of the 12 000 primary care physicians in Ontario, 1400
of the top NSAID prescribers, based on the number of
NSAID prescriptions filled in 2000, were identified from a
physician list sourced from the IMS Health Canada aggregated
prescriber-level database. These physicians were
invited by letter to participate. Each participating physician
was asked to record up to 130 office visits, including
follow-up visits, of successive patients for whom they
decided to prescribe an NSAID or a coxib (new or renewal)
for OA. Participants were not provided with the
Second Canadian Consensus Conference guidelines as
part of the study protocol and were not told that their
results would be compared with guidelines. Physicians
were reimbursed Can$20 per completed form.
Although a diagnosis of OA is a determinant of eligibility
for coxib coverage by the Ontario Drug Benefits
Program, which provides basic drug coverage to all
Ontario citizens 65 years and older, the program does
not specify OA diagnostic criteria.19 The phrase "working
diagnosis of OA" was therefore used on the data
form to avoid telling participating physicians how to
diagnose OA. Patients were included in the study if they
had a working diagnosis of OA by the treating primary
care physician, received a prescription for an NSAID or
a coxib as part of routine care on the first study visit,
and gave consent for the use of their anonymous data
for aggregate analyses.
Completed data forms were sent by toll-free facsimile
to the data center. Standardized data management
protocols for paperless and semiautomated processes
adapted to the specific needs of the study were used to
aggregate and process the data into a relational database.
The Goodman-Kruskal γ statistic was used as a
measure of ordinal association. All statistical comparisons
were 2-tailed, with P < .05 considered statistically
significant.
RESULTS
Recruitment
Of the 1400 physicians invited, 185 agreed to participate
and were enrolled. It was not possible to examine
whether characteristics of participants differed from
those of nonparticipants, because the aggregated prescriber
list was not ranked and did not preserve prescribing
data or other individual characteristics. Data
collection continued during 14 months, from November
2000 to December 2001, with 119 (64% of those
enrolled) of the physicians registering 8846 office visits
of 5947 OA patients in the study (the number of
patients with 2, 3, and >3 visits were 1034, 431, and
280, respectively). Data were analyzed for the first visits
of the 5459 patients for whom a prescribed NSAID
was identified.
Patient Characteristics
Of the 5459 OA patients for whom data were analyzed,
60% were female and 46% were older than 65
years. All residents of Ontario 65 years and older have
limited public drug coverage through the Ontario Drug
Benefits Program. In this study, 57% of patients younger
than 65 years and 19% of patients 65 years or older had
private drug coverage.
A large number of patients had other concomitant
conditions. The most common comorbidities were
hypertension (34%) and coronary
artery disease (12%). A clinically
significant GI event had
been experienced by 8% of
patients. More than a third (34%)
of patients were receiving concomitant
antihypertensive therapy,
and 17% were receiving
low-dose aspirin (the study protocol
did not specify a threshold
dose because physicians may
have differed in their operational
definitions of low-dose aspirin
for cardiovascular protection).
At the time of the first study
visit, 20% of patients were
already receiving a GPA.
Acetaminophen had previously
been tried by 61% of patients for
their OA symptoms, and 49%
had tried 1 or more NSAIDs
(Figure 1). Five percent of
patients had tried 3 or more NSAIDs. Of the 5459
patients, 1145 (21%) were receiving NSAIDs for the
first time at their first CANOAR study visit.
Nonsteroidal Anti-inflammatory
Drug Prescribing Patterns
Coxibs were prescribed at 56% of visits in this study
cohort, evenly divided between rofecoxib and celecoxib
(Figure 2). Diclofenac, meloxicam, naproxen, and
ibuprofen were less commonly recommended. Forty
patients (0.7%) received aspirin prescriptions
at the study visit, and
because it was unclear whether they
were for OA therapy, these prescriptions
are not included in Figure 2.
Coxib use was higher in certain
high-risk patient subgroups. Among
the 26 patients who had experienced
clinically significant GI events in the
previous 60 days, 85% were prescribed
coxibs (Table 1), while 74%
of the 214 patients whose GI events
had occurred within the last 10 years
received coxibs. Most (79%) of the 72
patients receiving warfarin therapy
and most (68%) of the 163 patients
with congestive heart failure were
also prescribed coxibs. However, the
percentage (57%) of patients with
hypertension who were prescribed
coxibs was similar to that (56%)
among all patients, and coxib use did not vary with
patient age (Goodman-Kruskal γ = 0.07, SE = 0.02, P =
.60; data not shown). Although it appeared that coxibs
were prescribed more commonly as the number of GI
and renal risk factors increased (Table 2), these trends
were not statistically significant (P = .93 and P = .21,
respectively). However, the results suggested a threshold
effect, with higher coxib use for patients with 4 or
more risk factors than for those with 3 or fewer. A post
hoc analysis in which risk factor categories were pooled
confirmed that this difference was significant (GI and
renal risk factors, P < .001 for both; Fisher exact test).
Coxib use increased significantly with increasing global
pain assessments (physician and patient assessments,
P < .001 for both; Table 3).
Gastroprotective Agent Prescribing Patterns
Gastroprotective agents were coprescribed with an
NSAID at 31% of visits. The most common reason given
for prescribing a GPA was NSAID GI prophylaxis (39%).
Other reasons included dyspepsia (19%), gastroesophageal
reflux disease (25%), and a history of ulcer
(8%). Misoprostol was the most commonly prescribed
GPA, usually in the form of a combination pill of
diclofenac and misoprostol, while H2 antagonists (eg,
ranitidine) and proton pump inhibitors (eg, omeprazole)
were also used, although less often (Table 4). Patients
receiving non-coxib NSAIDs were more likely to
receive GPAs than were coxib patients (45% vs 21%).
Gastroprotective agent coprescriptions were more
commonly given to patients with a higher number of
GI risk factors (Goodman-Kruskal γ = −0.20, SE =
0.04; P < .001; Table 5). The percentage
of patients who were
coprescribed a GPA with their
NSAID was 25% for patients younger
than 65 years, 34% for those aged 65
to 69 years, 39% for those aged 70 to
74 years, 39% for those aged 75 to 79
years, and 45% for those older than
79 years.
Acetaminophen Use
Acetaminophen had been tried
previously by 61% of study patients.
The prevalence of its previous use
increased with increasing patient
age and the number of GI and renal
risk factors (P < .001 for all). In addition,
previous acetaminophen use
was more common among patients
with higher patient and physician
global assessments of OA severity (P
< .001).
Appropriateness of Prescribing
Most coxib prescriptions were for patients with 1 or
more GI risk factors, but a large percentage (39%) were
for patients with no identified risk factors and could be
considered to be less cost effective (Table 6). Most
(56%) prescriptions for a traditional NSAID alone were
for patients with at least 1 GI risk factor, for whom a
GPA coprescription or a coxib would be recommended.
Of 1081 prescriptions for traditional NSAIDs plus
GPAs, 31% went to patients with no identified risk factors.
Physicians did not indicate a reason for prescribing
a GPA in 40% of prescriptions for traditional
NSAIDs plus GPAs, so these cases were excluded from
the analysis of appropriate prescribing. In 98% of these
excluded cases, the prescription was
for the NSAID and GPA combination
pill Arthrotec (diclofenac sodium
and misoprostol). Some
physicians who prescribed
Arthrotec may not have considered
that they were actively prescribing
a GPA, which may explain the high
rate of nonresponse to this question.
This problem was addressed
by also excluding from the analysis
of appropriate prescribing all 236
Arthrotec prescriptions for which a
reason for prescribing a GPA was
given. Of the remaining 413 prescriptions
for traditional NSAIDs
plus GPAs, 73% went to
patients with 1 or more
identified risk factors and
were counted as appropriate.
The other 27% went to
patients with no identified
risk factors. Of these prescriptions,
63% were prescribed
for NSAID GI
prophylaxis only and were
counted as inappropriate,
while the remainder were
prescribed for other reasons
(30%) or for NSAID GI prophylaxis
and other reasons
(6%) and were counted as
appropriate. Therefore, 83%
of prescriptions for traditional
NSAIDs plus GPAs
were considered appropriate.
For patients with no identified GI risk factors, 33% of
nonexcluded prescriptions were considered appropriate,
whereas for those with 1 or more risk factors, 74%
were considered appropriate. Overall, 58% of nonexcluded
prescriptions were considered to be appropriate
given patient GI risk factors. If the analysis was restricted
further to exclude prescriptions for traditional
NSAIDs plus GPAs in which the GPA was prescribed for
reasons other than GI prophylaxis alone, 32% of prescriptions
for patients with no identified GI risk factors,
73% of those for patients with 1 or more risk factors, and
56% overall were considered appropriate.
DISCUSSION
Treatment guidelines in the United States,6 Europe,20
and Canada5 stress the importance of tailoring OA
treatment to individual patient needs. Patient age and
the presence of comorbidities should be important factors
in treatment selection. The 2000 Second Canadian
Consensus Conference guidelines5 that were available
when this study was conducted are a relevant benchmark
for prescribing practice at the time. Although subsequently
published guidelines7,21 have refined the
criteria for appropriate coxib, NSAID, and GPA
prescription, they identify the same risk factors as did the
Second Canadian Consensus Conference and agree on
selection of a coxib or an NSAID plus GPA for at-risk
patients. On the basis of the Second Canadian Consensus
Conference guidelines, 42% of prescriptions in
the CANOAR study were inappropriate given patient GI
risk factors. There was underuse of coxibs and of
NSAIDs plus GPAs for patients with 1 or more risk factors
and overuse of these options for patients with no
identified risk factors. The percentage of prescriptions
deemed appropriate was higher for coxibs than for traditional
NSAIDs alone (61% vs 44%) but lower than that
(83%) for traditional NSAIDs plus GPAs. Greater appropriate
use of traditional NSAIDs in combination with
GPAs than of coxibs may reflect the fact that coxibs
were a new prescribing option when this study was
conducted.
All guidelines recommend acetaminophen as a firstline
agent, primarily because of its GI safety. In accord
with guidelines, most patients in this Ontario cohort
had previously tried acetaminophen, particularly older
patients, those with multiple risk factors, and those
with severe OA.
Bleeding, upper GI ulcers, and perforation occur in
approximately 1% of patients treated with non-coxib
NSAIDs for 3 to 6 months.5 Among patients treated for
1 year, 2% to 4% are affected. These events are costly
and can lead to reduced compliance and decreased
quality of life.22 Recent clinical trial evidence found
that, compared with acetaminophen alone, NSAIDs in
combination with a GPA produced significantly greater
improvements in pain scores for patients with moderate
OA.23 In our study, GPAs were prescribed at 31% of
visits, and 39% of these prescriptions were for NSAID
GI prophylaxis. In addition, GPA prescriptions
increased with the number of GI risk factors. For
patients with the same number of GI risk factors, those
receiving coxibs had a somewhat lower rate of GPA
coprescription compared with those receiving traditional
NSAIDs, often for reasons other than NSAID GI
prophylaxis. In general practice, GPAs are commonly
and appropriately prescribed to treat conditions such
as gastroesophageal reflux disease and dyspepsia.24,25
Comparisons of non-coxib NSAIDs and coxibs suggest
similar efficacy,10-12 and a recent clinical trial suggests
that coxibs are similar to or better than
acetaminophen.26 Because they produce fewer GI
adverse effects than traditional NSAIDs,13-16 coxibs are
now recommended in the United States and Canada
as first-line treatment for patients at risk for GI perforation,
ulcer, and bleeding;5,6 recent updates of
European guidelines are in agreement.27 In this study,
85% of patients who had experienced recent (within
60 days) clinically significant GI events were prescribed
a coxib NSAID, while 74% of those whose GI
event occurred up to 10 years earlier received a coxib
NSAID. Also in keeping with Canadian guidelines, a
coxib NSAID was prescribed to most (79%) OA
patients receiving warfarin therapy, who are at higher
risk for GI bleeding.
Osteoarthritis is the most common chronic condition
among older persons,28 and non-coxib NSAIDs are
prescribed frequently in this group.29 A study30 of noncoxib
NSAID prescribing patterns in Canada reported
that older OA patients, who have 3 times the risk for
serious GI adverse effects relative to patients younger
than 50 years, were frequently given unnecessary prescriptions
for non-coxib NSAIDs. In addition, GI complications
stemming from non-coxib NSAID use are a
leading cause of hospitalization for older persons.29
Although Canadian guidelines recommend the use of
coxibs in patients with GI risk factors, including older
patients,5 coxib prescribing in our cohort was not related
to age. As a result, some older patients who could
have benefited from a coxib received non-coxib NSAID
prescriptions instead, whereas some younger patients
for whom traditional NSAIDs may have been appropriate5
were prescribed more costly coxibs. One possible
explanation is that age alone may not have been considered
a risk factor by the physicians participating in
this study. Alternatively, NSAID prescribing practice
may have been correlated with barriers against coxib
use, such as access. The finding that older patients had
higher rates of GPA coprescription with NSAIDs contradicts
the suggestion that physicians failed to recognize
age as a GI risk factor, suggesting instead the
existence of fewer barriers to coxib prescribing among
younger patients than among seniors. Future analyses
of the CANOAR study data will examine whether prescribing
varied by drug coverage.
Although an association has been found between
NSAID use and impaired renal function,31 NSAIDs are
not contraindicated for patients with renal risk factors,
including older patients. Rather, renal function should
be monitored in all high-risk patients taking NSAIDs.
There does not appear to be a difference between noncoxib
and coxib NSAIDs in this regard, and in our
study there was no relationship between coxib prescribing
and elevated creatinine levels. This practice
pattern is in concert with treatment guidelines.
More than a third of patients in this study were
hypertensive. Meta-analyses studying the effect of noncoxib
NSAIDs on blood pressure found that they elevate
blood pressure and antagonize the blood
pressure–lowering effect of antihypertensive medication.32,33 The Second Canadian Consensus Conference
guidelines noted that coxibs can also adversely affect
blood pressure control, but did not mention possible
differences in the hypertensive or prothrombotic effects
of rofecoxib and celecoxib.5 Data collection for this
study was largely complete before publication of an
analysis that suggested differences may exist among
coxibs in the risk of cardiovascular events.34 It is therefore
unlikely that participating physicians prescribed
coxibs differentially in the presence of cardiovascular
risk factors. More recent meta-analyses have not found
increased cardiovascular thrombotic event rates relative
to placebo for either coxib.35,36
This study has limitations. Prescribing patterns were
measured via physician questionnaire responses, without
external verification. Self-reported physician practice
may not accurately reflect actual practice,
introducing the potential for biasing self-reports toward
higher quality levels.37 However, the CANOAR study
participants were not told in advance that their responses
would be evaluated against guidelines. Furthermore,
forms were to be completed during routine clinical practice,
minimizing the potential for self-recall bias.
Participation in the study was voluntary, and the 185
physicians who agreed to participate—and the 119 who
contributed data—may not represent a random sample
of the 1400 high NSAID prescribers who were invited.
Although nonresponding physicians were not followed
up to determine reasons for their nonparticipation, it is
possible that some may have thought that study participation
would be onerous, given the need to fill in
detailed data forms, obtain patient consent, and perform
medical chart review. Some nonparticipants may
also have found the financial incentive to be inadequate.
This could bias results if, for instance, physicians
who are more likely to enroll in such a study and who
actually do participate are also more likely to follow prescribing
guidelines.
High NSAID prescribers were targeted to capture the
highest number of prescriptions with the fewest physicians.
Although the prescribing choices of these physicians
may not reflect those of all primary care
physicians, their disproportionately large share of
NSAID prescriptions is expected to be representative of
prescriptions for OA in Ontario. These results are not
necessarily generalizable to other jurisdictions, however,
because prescribing may differ between Canada and
the United States or other regions. For example, whereas
39% of coxib prescriptions in the CANOAR study
went to patients with no GI risk factors, a recent analysis
of claims data from a large preferred provider organization
in the US Midwest found that as many as 73% of
patients given new coxib prescriptions had no evidence
of GI risk factors.38
To avoid influencing prescribing decisions, this
observational study did not provide physicians with
guidelines for OA diagnosis or therapy. This may have
increased variability in physician ratings of OA severity,
but questions on the data form about other risk factors
required yes or no responses that would not be predicted
to vary with exposure to guidelines.
Previous use of acetaminophen as reported in this
study may be somewhat underestimated, as this commonly
used drug is available by prescription and over
the counter. Its previous use in either form, perhaps
many years earlier, may have been forgotten by physicians
or their patients at the time of completing the
study data forms.
In conclusion, large, simple observational studies
such as the CANOAR study can contribute important
practical assessments of the appropriateness of real-world
prescribing and guide the development of future
interventions designed to enhance the application of
evidence-based care in actual clinical practice. Most
prescriptions observed in this cohort of OA patients
treated by primary care physicians in Ontario were consistent
with guidelines from the Second Canadian
Consensus Conference,5 but some could be considered
inappropriate. Areas in which improved recognition of
risk factors might lead to further increases in the
appropriate use of NSAIDs, coxibs, and GPAs in OA
patients include the relationship between patient age
and GI risk, as well as the benefit of GPAs or coxibs for
patients with multiple GI risk factors.
Acknowledgments
We thank the 119 Ontario primary care physicians who contributed to
the data collection effort. This article was prepared with the assistance of
BioMedCom Consultants Inc, Montreal, Quebec.
|
Previous Article
