Drugs Aging (2013) 30:2330
DOI 10.1007/s40266-012-0037-9
ORIGINAL RESEARCH ARTICLE
COX-2 Inhibitor and Non-Selective NSAID Use in Those
at Increased Risk of NSAID-Related Adverse Events
A Retrospective Database Study
Svetla Gadzhanova Jenni Ilomaki
Elizabeth E. Roughead
Published online: 23 November 2012
Springer International Publishing Switzerland 2012
Abstract
Background Adverse events related to analgesic use
represent a challenge for optimizing treatment of pain in
older people.
Objective The aim of this study was to determine
whether non-selective non-steroidal anti-inflammatory
drug (NS-NSAID) and cyclo-oxygenase (COX)-2 inhibitor
use is appropriately targeted in those with a prior history of
gastrointestinal (GI) events, myocardial infarction (MI) or
stroke.
Methods A retrospective study of pharmacy claims data
from the Australian Government Department of Veterans
Affairs was conducted, involving 288,912 veterans aged
55 years and over. Analgesic utilization from 2007 to 2009
was assessed. Three risk cohorts (veterans with prior hos-
pitalization for GI bleed, MI or stroke) and a low-risk
cohort were identified. Poisson regression was applied to
test for a linear trend over the study period.
Results The prevalence of analgesics dispensed in the
overall study population was approximately 34 % between
2007 and 2009. COX-2 inhibitors were more widely dis-
pensed than NS-NSAIDs in all those at risk of NSAID-
related adverse events. At the end of 2009, the ratio was
5.1 % to 2.5 % in the GI cohort, 3.6 % to 3.2 % in the MI
cohort and 3.6 % to 2.6 % in the stroke cohort.
Conclusions Although COX-2 inhibitors appeared to be
preferred over NS-NSAIDs in those with a prior history of
GI events, 2.5 % of patients were still using an NS-NSAID
S. Gadzhanova
J. Ilomaki (&)
E. E. Roughead
Quality Use of Medicines and Pharmacy Research Centre,
Sansom Institute, School of Pharmacy and Medical Sciences,
University of South Australia, GPO Box 2471,
Adelaide, SA, Australia
e-mail: jenni.ilomaki@unisa.edu.au
at the end of the study period. Consistent with treatment
guidelines, in most of these cases, these drugs were
co-dispensed with proton pump inhibitors. COX-2 inhibitors
were used at slightly higher rates than NS-NSAIDs in those
with a prior history of MI or stroke, which is not consistent
with guidelines recommending NS-NSAID use.
1 Introduction
The recommendations for pharmacological management of
pain have evolved in recent years because of evidence of
adverse events from cyclo-oxygenase (COX)-2 inhibitors
[14]. In the mid-2000s, mounting evidence of an
increased risk of cardiovascular disease (CVD) related to
COX-2 inhibitor use [510] and, to a smaller extent, related
to non-selective non-steroidal anti-inflammatory drug
(NS-NSAID) use caused changes in labelling in several
countries, including Australia [24]. A meta-analysis of
observational studies reported a 35 % increased risk of
cardiovascular ischaemia with the COX-2 inhibitor,
rofecoxib [9]. There was also a 25 % increased risk reported
for meloxicam and a 40 % increased risk for diclofenac.
Meloxicam and diclofenac are traditionally classified as
NS-NSAIDs [11]; however, both have significant COX-2
inhibition [12]. A 30 % increased risk with indomethacin,
an NS-NSAID, was also reported [9]. Although no increase
in risk was found with the use of the COX-2 inhibitor
celecoxib, another large cohort study found a dose-
dependent relationship between celecoxib and myocardial
infarction (MI), where a 61 % increase with doses greater
than 200 mg was observed compared with non-users [13].
Evidence also suggests that individuals aged 65 years and
older are at increased risk of serious cardiovascular-related
events due to COX-2 inhibitor use [14].
24
The risk of gastrointestinal (GI) events due to the use of
NS-NSAIDs is well known. Pooled results from 18 obser-
vational studies showed a fourfold overall increased risk of
GI bleeding or perforation among users of NS-NSAIDs
compared with non-users [15]. The study also reported that
advanced age increased the risk of GI events. The risk of GI
bleeding was 4.5 times higher among those aged
7080 years and 9.2 times higher among those aged over
80 years than among those aged 2549 years. COX-2
inhibitors are reported to be safer in terms of GI events [16].
The CLASS (Celecoxib Long-term Arthritis Safety Study)
trial showed that the risk of symptomatic ulcers and ulcer
complications among celecoxib users was 40 % lower than
among NS-NSAID users [17]. However, a meta-analysis of
clinical trials concluded that the risk of any GI event was
20 % higher with celecoxib than with placebo [16].
Current Australian guidelines recommend paracetamol
(acetaminophen) as the first-line pharmacological treatment
for pain in osteoarthritis and rheumatoid arthritis [18, 19].
NSAIDs are recommended if pain relief with paracetamol is
inadequate, with recommendations to consider the patients
existing GI and CVD risk when selecting NSAIDs [1821].
In patients at high risk of GI events, COX-2 inhibitors are
recommended over NS-NSAIDs. NS-NSAIDs can be used
if adequate GI protection with proton pump inhibitors
(PPIs) is added to the regimen. For those with CVD, NS-
NSAIDs are preferred to COX-2 inhibitors.
Adverse events related to analgesic use present a chal-
lenge for optimizing the pharmacological treatment of
chronic pain, especially in older people, among whom
arthritis is one of the most common chronic conditions
[22]. One in four Australians aged 6574 years reported
having osteoarthritis in 20072008, and this proportion
rose to 32 % amongst those aged 75 years and older [22].
Older people are also most susceptible to NSAID-related
adverse effects [23].
We have previously reported the rapid uptake of all
NSAIDs after celecoxib was first subsidized in Australia in
August 2000 and a subsequent overall decrease in NSAID
use after rofecoxib withdrawal in October 2004 [24].
Recognizing the new recommendations and knowledge
about adverse events, the aim of this study was to deter-
mine whether NS-NSAID and COX-2 inhibitor use is
appropriately targeted in those with a prior history of GI
events, MI or stroke.
2 Methods
2.1 Data Source and Study Design
We conducted a retrospective study using the Australian
Government Department of Veterans Affairs (DVA) [25]
S. Gadzhanova et al.
administrative health claims database. These data include
all veterans who have served in the Australian Defence
Force, their spouses and their dependants. According to the
DVAs most recent statistics, more than half of the veterans
served in the Second World War and the majority are aged
between 80 and 85 years [26]. Overall, 60 % of the DVA
populations are men.
The pharmacy claims database provides de-identified
patient-level information on all prescription medicines
dispensed and subsidized to DVA clients under the Phar-
maceutical Benefits Scheme or Repatriation Pharmaceuti-
cal Benefits Scheme [27]. Dispensing data include the
medicine dispensed, date of dispensing, quantity supplied,
dosage form and strength. Client data include patient age,
gender, date of death and residential status. Hospitalization
data are also available and provide information on all
hospital admissions of DVA clients to public and private
hospitals. Diagnoses associated with the hospitalization are
coded according to the International Classification of Dis-
ease 10th edition, Australian modification (ICD-10-AM)
[28]. Analgesic utilization from January 2007 to December
2009 was investigated in three risk cohorts (GI cohort, MI
cohort and stroke cohort) and one low-risk cohort.
2.2 Definition of Cohort Populations
The study population included all veterans and their
dependants who were aged 55 years and older in January
2006. They contributed to the following cohorts:
1. GI cohortveterans who have had at least one
hospitalization with a primary diagnosis of GI events
(ICD-10-AM codes I21.0I22.9) between January
2006 and December 2009. Veterans entered the cohort
and were considered to be at increased risk of NSAID-
related adverse events from the date of their earliest
(index) hospital admission for a GI event in the period.
2. MI cohortveterans who have had at least one hospital-
ization with primary diagnosis of MI (ICD-10-AM codes
K25.0K27.9) between January 2006 and December
2009. They entered the MI cohort and were considered to
be at increased risk of NSAID-related adverse events
from their index MI admission.
3. Stroke cohortveterans who have had at least one
hospitalization with primary diagnosis of stroke
(ICD-10-AM codes I60.0I69.9) between January
2006 and December 2009. They entered the stroke
cohort and were considered to be at increased risk of
NSAID-related adverse events from their index stroke
admission.
4. Low-risk cohortincluded (a) veterans who had no
hospitalizations for GI events, MI or stroke between
January 2006 and December 2009 (they were
NSAID Use in Those at Risk of Adverse Events
considered at low risk for the whole period); and
(b) veterans who were considered to be at low risk prior
to their first hospital admission due to a GI event, MI or
stroke between January 2006 and December 2009.
Veterans were considered to be part of the relevant risk
cohort after their first hospital admission for any of these
events. How participants moved from the low-risk
cohort to one of the risk cohorts is illustrated in Fig. 1.
Participants who had consecutive hospitalizations for
the same event in the period contributed person-months in
the corresponding risk cohort from the date of the earliest
admission. Participants who had two or more hospital
admissions for different events contributed person-months
in each of the corresponding risk cohorts from the date of
the earliest admission for the given event.
2.3 Definition of Analgesics
Analgesic use each month was assessed for each of the
following medicine classes:
1. Paracetamol only
2. COX-2 inhibitors (including concurrent use with para-
cetamol but excluding use with NS-NSAIDs or selected
opioids)
3. NS-NSAIDs (including concurrent use with paraceta-
mol but excluding use with COX-2 inhibitors or
selected opioids)
4. Selected opioids: tramadol, codeine with paracetamol,
and oxycodone (including concurrent use with para-
cetamol, COX-2 inhibitors and NS-NSAIDs).
The rationale for the definitions of the medicine classes
was to take into account individuals who step up from first-
line to second-line and stronger analgesia. Participants
were excluded from the analyses if they were using COX-2
inhibitors and NS-NSAIDs concomitantly.
First hospitalization
GI event
Veteran population No events Low-risk
MI
aged 55 years cohort
Stroke
No hospitalizations
Low-risk
cohort
Fig. 1 Illustration of how participants of the veteran population could move from a low-risk cohort to a risk cohort after hospitalization for a
gastrointestinal (GI) event, myocardial infarction (MI) or stroke during the study period (20062009)
25
Analgesics were defined according to the World Health
Organizations (WHOs) Anatomical Therapeutic Chemical
(ATC) classification [11]. Paracetamol use was defined as
ATC code N02BE01. All NS-NSAIDs and COX-2 inhibitors
that were on the market in Australia and subsidized to vet-
erans during the study period were included in the analyses.
The NS-NSAIDs that were included were indomethacin
(M01AB01), diclofenac (M01AB05), sulindac (M01AB02),
piroxicam (M01AC01), ibuprofen (M01AE01), naproxen
(M01AE02), ketoprofen (M01AE03), tiaprofenic acid
(M01AE11), mefenamic acid (M01AG01) and diflunisal
(N02BA11). The COX-2 inhibitors that were included were
celecoxib (M01AH01), lumiracoxib (M01AH06; until
August 2007) and meloxicam (M01AC06). Although not
classified a coxib by the WHO, meloxicam was included in
the COX-2 inhibitors because of its high COX-2 selectivity
[12]. The National Prescribing Service (NPS) [29] and
Australian Therapeutic Goods Administration [30] both
classify meloxicam as a COX-2 inhibitor in their recom-
mendations for NSAIDs. The selected opioids included
codeine and paracetamol combinations (N02AA59), oxy-
codone (N02AA05) and tramadol (N02AX02). We did not
include stronger opioids such as morphine, because they are
mainly used for cancer pain and in palliative care, where pain
management may differ from usual management of acute or
chronic non-cancer pain [31].
Monthly utilization was calculated as the proportion of
people dispensed the medicine of interest in each month
among the cohort population in that month. To calculate
the population using medicines each month, prescription
durations were applied. The prescription duration was
calculated from the data for each individual medicine and
reflected the time period within which 75 % of those who
were prescribed that medicine had a repeat prescription
dispensed, thus most likely representing periods of actual
use. For example, 75 % of all prescriptions for meloxicam
were dispensed every 43 days; therefore, for the purposes
Hospitalization for a different event
GI MI or stroke GI and MI/stroke
cohort cohorts
MI GI event or stroke MI and GI/stroke
cohort cohorts
Stroke GI event or MI Stroke and GI/MI
cohort cohorts
26 S. Gadzhanova et al.

of these analyses, it was assumed that the person was using Table 1 Cohort characteristics
the medicine from the date of dispensing until 43 days Cohort No. of participants Age Sex
later. In Australia, subsidized medicines can be dispensed [years; mean [women (%)]
only for the equivalent of a 1-month supply. Jan Dec (SD)] (Jan 2007)
2007 2009 (Jan 2007)
2.4 Statistical Analysis GI cohorta 514 1,277 82.0 (7.78) 43
a
MI cohort 2,025 4,969 83.2 (7.48) 35
Participants were followed up until death or the end of the Stroke 2,481 5,356 84.1 (6.38) 41
study period. They were excluded from a cohort when they cohorta
entered palliative care (defined as the date 3 months before Low-risk 279,361 199,534 79.7 (9.85) 41
the first claim for palliative care item) or when they were cohort
dispensed a medicine for neuropathic pain (if any). This GI gastrointestinal, MI myocardial infarction, SD standard deviation
was done because pain management among these groups a
3.6 % of participants in the risk cohorts were included in more than
differs from that for those with acute or chronic non-cancer one risk cohort
pain [31, 32].
Cohort characteristics are presented as means and pro- COX-2 inhibitor use was more prevalent than NS-
portions. Medicine utilization rates were age-standardized NSAID use in all cohorts (Table 2). At the end of 2009, the
using the veteran population in January 2007 as the stan- use of COX-2 inhibitors was twice as high as the use of
dard population in 5-year categories. Poisson regression NS-NSAIDs in the GI and low-risk cohorts (5.1 % vs.
models were used to calculate age-standardized rate ratios 2.5 % and 5.6 % vs. 3.3 %, respectively). In December
(SRRs), and the rate in 1 year was compared with that for 2009, COX-2 inhibitors were used at slightly higher rates
the previous year to test for linear trends over the period than NS-NSAIDs in the MI cohort (3.6 % vs. 3.2 %), and
20072009. All analyses were done for all analgesic groups the rates were 3.6 % vs. 2.6 % in the stroke cohort.
in each cohort. Analyses were performed using an SAS 9.3 The use of both COX-2 inhibitors and NS-NSAIDs
statistical package (SAS Institute, Cary, NC, USA). declined significantly throughout the study years in the
DVA Ethics Committee and the Human Research Ethics stroke and the low-risk cohorts (Table 2). The rates of
Committee of the University of South Australia approved COX-2 inhibitor and NS-NSAID use also declined in the
this study. GI cohort, but the decrease was not statistically significant
for NS-NSAIDs. A significant decline was observed in NS-
NSAID use in the MI cohort.
3 Results

The study included 288,912 subjects. The mean (SD) age
was 79.8 (9.82) years and 40 % were women. Between
2006 and 2009, 19,544 subjects had at least one hospital-
ization for a GI event, MI or stroke, which accounted for
6.8 % of all subjects. Of the 19,544 hospitalized subjects,
3.6 % experienced hospitalization for more than one type
of event. Characteristics of the cohorts are presented in
Table 1. The age-standardized prevalence of overall anal-
gesic use was approximately 34 % among the overall study
population (Table 2). Analgesic use was higher in the risk
cohorts than in the low-risk cohort. In December 2009, the
prevalence of analgesic use was 47 % in the GI cohort,
40 % in the MI and stroke cohorts and 34 % in the low-risk
cohort.
Paracetamol was the most commonly used analgesic
within each cohort, accounting for more than half of any
analgesic use (Table 2; Fig. 2). The prevalence of para-
cetamol use was higher in the risk cohorts, ranging from
21 % to 23 %, compared with 16 % in the low-risk cohort.
Selected opioids were the next most widely used analgesic
after paracetamol.
4 Discussion
We found that one-third of the overall study population
used analgesics. There was a significant decline in COX-2
inhibitor use among the GI and stroke cohorts, and a
decline in NS-NSAID use among the MI and stroke
cohorts.
Paracetamol was the most commonly dispensed anal-
gesic in all our cohorts, accounting for half of all analgesic
use. This supports guideline recommendations to use par-
acetamol as a first-line therapy [18, 19]. Our results also
demonstrated practice consistent with guidelines for the GI
cohort, where COX-2 inhibitors were the preferred NSAID.
There was a low prevalence of NS-NSAID use among
those with a prior history of a GI event, with only 2.5 %
using NS-NSAIDs at the end of the study period. Further
analysis showed that 82 % of NS-NSAID users in the GI
cohort were using PPIs concomitantly, consistent with the
guidelines [18, 21]. A positive finding was also seen in the
decrease in COX-2 inhibitor use among those with a prior
NSAID Use in Those at Risk of Adverse Events 27

Table 2 Prevalence and changes in analgesics use from 2007 to 2009

Cohort Drug class January 2007 December 2009 Average SRR 95 % CI p valuea
annual
n CR SR n CR SR change
(%) (%) (%) (%) (%)

GI event Any analgesic 251 48.83 47.57 604 47.30 47.10 -0.5 0.995 0.985, 1.004 0.2500
groupb
Paracetamolc 115 22.37 21.31 326 25.53 23.11 ?0.3 1.003 0.985, 1.021 0.7695
COX-2 40 7.78 7.89 64 5.01 5.06 -6.0 0.940 0.904, 0.978 0.0021
inhibitord
NS-NSAIDe 18 3.50 3.76 29 2.27 2.47 -4.7 0.953 0.895, 1.016 0.1400
Selected opioidsf 77 14.98 14.20 184 14.41 16.36 ?1.9 1.019 0.984, 1.055 0.2835
MI Any analgesic 777 38.37 37.20 1,953 39.30 39.66 ?0.0 1.000 0.991, 1.010 0.9750
groupb
Paracetamolc 418 20.64 19.03 1,113 22.40 20.51 ?1.6 1.016 0.998, 1.034 0.0790
COX-2 70 3.46 3.79 154 3.10 3.57 ?0.6 1.006 0.977, 1.037 0.6851
inhibitord
NS-NSAIDe 53 2.62 2.64 134 2.70 3.21 -5.8 0.942 0.910, 0.975 0.0007
Selected opioidsf 234 11.56 11.65 547 11.01 12.27 -1.3 0.987 0.973, 1.000 0.0573
Stroke Any analgesic 1,007 40.59 39.81 2,161 40.35 40.31 -1.6 0.984 0.977, 0.992 0.0001
groupb
Paracetamolc 580 23.38 19.93 1,343 25.07 22.26 ?0.6 1.006 0.993, 1.018 0.3745
COX-2 101 4.07 4.09 176 3.29 3.57 -3.4 0.966 0.949, 0.983 \0.0001
inhibitord
NS-NSAIDe 65 2.62 2.92 110 2.05 2.59 -11.5 0.885 0.854, 0.917 \0.0001
Selected opioidsf 258 10.40 12.75 531 9.91 11.87 -1.6 0.984 0.967, 1.001 0.0692
Low risk Any analgesic 95,425 34.16 34.18 67,744 33.95 34.14 -1.0 0.990 0.984, 0.997 0.0067
groupb
Paracetamolc 40,240 14.40 14.44 32,346 16.21 16.29 ?2.5 1.025 1.009, 1.041 0.0022
COX-2 18,281 6.54 6.54 11,051 5.54 5.61 -4.8 0.952 0.944, 0.961 \0.0001
inhibitord
NS-NSAIDe 10,836 3.88 3.87 6,675 3.35 3.32 -4.9 0.951 0.940, 0.962 \0.0001
Selected opioidsf 25,526 9.14 9.13 17,377 8.71 8.76 -2.1 0.979 0.973, 0.985 \0.0001
Whole study Analgesicsg 97,429 34.27 34.27 72,273 34.31 34.44 -0.8 0.992 0.985, 0.999 0.0276
population
CI confidence interval, COX-2 cyclo-oxygenase-2, CR crude rate, GI gastrointestinal, MI myocardial infarction NS-NSAID non-selective non-
steroidal anti-inflammatory drug, SR age-standardized rate, SRR age-standardized rate ratio
a
p values for the ratio of the rate in 1 year to the rate in the previous year were calculated using a Poisson regression model accounting for over-/
under-dispersion in the data when necessary
b
Any analgesic group includes the use of paracetamol, NS-NSAIDs, COX-2 inhibitors, tramadol, oxycodone, and paracetamolcodeine
combination products, including concurrent use
c
Paracetamol includes paracetamol-only use without any other analgesics
d
COX-2 inhibitor includes COX-2 use with or without paracetamol, but excluding concomitant use with NS-NSAIDs or selected opioids
e
NS-NSAID includes NS-NSAID use with or without paracetamol, but excluding concomitant use with COX-2s or selected opioids
f
Selected opioids include tramadol, codeine with paracetamol, and oxycodone (including concurrent use with paracetamol, COX-2 inhibitors
and NS-NSAIDs)
g
Analgesics includes all analgesics in the individual drug classes, including concurrent use

history of stroke. However, COX-2 inhibitor use was still
higher than NS-NSAID use in the stroke and MI cohorts.
We found a lower use of paracetamol and NSAIDs when
compared with data from an Australian survey of medicine
use among those aged 50 years and over, which reported
past-month paracetamol use of 42.9 % and NSAID use of
16.5 % [33]. The difference may be attributed to the fact
that the survey included both prescription and over-the-
counter medicines. We could locate no comparable inter-
national studies that reported trends in analgesic use in the
same time period. Based on sales data, a European study
reported paracetamol as the most commonly used analgesic
28 S. Gadzhanova et al.

Paracetamol
Selected opioids
COX-2 inhibitors
NS-NSAIDs
GI cohort MI cohort
24 24

Percentage of participants
Percentage of participants

20 20

16 16

12 12

8 8

4 4

0 0

M -08

N -08

M -09

N -09
Ja 08

9
M -08

Se 8

M 09

Se 9
Ju 8

Ju 9
M -07

N -07
M -08

N -08

M -09

N -09

Ja 07
Ja 08

M -07

Se 7
M -08

Se 8

M 09

Se 9

Ju 7
Ju 8

Ju 9
M -07

N -07
Ja 07
M -07

Se 7
Ju 7

-0
l-0

l-0
-0

-0
-0

l-0
l-0

l-0

-0
-0

-0
l-0
-0

-
n-
-
-
n-
-

ar

ar

p
ov

ov
n

ay

ay
ar

p
ar

ar

ov
ov

ov

n
n

ay
ay

ay
ar

p
ov
n

ay

Ja
Ja

Stroke cohort Low-risk cohort

24 24
Percentage of participants

Percentage of participants
20 20

16 16

12 12

8 8

4 4

0 0

M -08

N -08

M -09

N -09
Ja 08

9
M -08

Se 8

M 09

Se 9
Ju 8

Ju 9
M -07

N -07
Ja 07
M -07

Se 7
Ju 7
M -08

N -08

M -09

N -09
Ja 08

9
M -08

Se 8

M 09

Se 9
Ju 8

Ju 9
M -07

N -07
Ja 07
M -07

Se 7
Ju 7

-0
l-0

l-0
-0

-0
l-0
-0
-0
l-0

l-0
-0

-0
l-0
-0

-
n-
-
-
n-
-

ar

ar

p
ov

ov
n

ay

ay
ar

p
ov
n

ay
ar

ar

p
ov

ov
n

ay

ay
ar

p
ov
n

ay

Ja
Ja

Fig. 2 Age-standardized analgesics use in the risk (GI, MI, and
stroke) and low-risk cohorts during 20072009. The group of
participants dispensed paracetamol does not include those also
dispensed NS-NSAIDs, COX-2 inhibitors or selected opioids. The
group of participants dispensed COX-2 inhibitors does not include
those also dispensed NS-NSAIDs or selected opioids. The group of
participants dispensed NS-NSAIDs does not include those also
dispensed COX-2 inhibitors or selected opioids. Selected opioids are
defined as tramadol, codeine with paracetamol, and oxycodone. COX-
2 cyclo-oxygenase-2, GI gastrointestinal, MI myocardial infarction,
NS-NSAID non-selective non-steroidal anti-inflammatory drug

in Norway and Denmark in 2009, while in Finland, Esto-
nia, Czech Republic and Slovakia, ibuprofen was the most
commonly used analgesic [34].
A Danish study reported a rapid decrease in the number
of COX-2 inhibitors used among the whole Danish popu-
lation between 2002 and 2005, to almost 0 % in 2005 [35].
Similarly, a previous study on NSAID use in our study
population showed a decrease until July 2007 [24]. The
current results demonstrate that, between 2007 and 2009,
the decline continued in all our cohorts, albeit at more
modest rates. COX-2 inhibitors continued to decrease in all
cohorts except the MI cohort, in which COX-2 inhibitor
use was already low at the beginning of our study. These
findings are encouraging and it seems that the majority of
pharmacological pain management is consistent with
guideline recommendations.
Our data also show seasonal effects in analgesic use.
The seasonal variation is due to the Australian Safety Net
where, once annual threshold costs are exceeded, medi-
cines become less expensive or free for the rest of the
calendar year. This causes the safety net effect where
more medicines are collected from pharmacies late in the
calendar year while at the beginning of the year the dis-
pensing rate declines [36].
Our study was based on a large nationwide dataset of
almost 300,000 Australian veterans and their dependants.
We had a long follow-up period with monthly data on
medicine dispensing. DVA subsidises all prescribed med-
icines to their clients, which means that we captured all
prescribed analgesic use for the study population.
The veteran population have slightly more visits to
general practitioners (GPs) [relative risk (RR) 1.17;
NSAID Use in Those at Risk of Adverse Events
p \ 0.05] and hospitalizations (RR 1.21; p \ 0.05) annu-
ally than the general Australian population aged C40 years
[25]. Due to the higher rate of GP visits, veterans receive
slightly more prescriptions per year than the general pop-
ulation (RR 1.13; p \ 0.05), suggesting that our study
results may slightly overestimate the prescribing rates over
those for the general population. However, because of our
inability to include over-the-counter medicine use, our
study may slightly underestimate overall rates of use.
Compared with the Australian population aged 55 years
and over, men are over-represented in the veteran popu-
lation (48 % vs. 60 %) [37].
One limitation that affects all prescription claims data is
that neither the actual use of dispensed medicine nor
whether the medicine was prescribed for regular or as-
needed use is known. We defined the risk cohorts on the
basis of hospitalization data for GI events, MI and stroke.
We have no information on events that may have occurred
without hospital admission, for example, milder GI events.
Thus, our results reflect use in those with more severe
disease and at high risk of NSAID-related adverse events.
5 Conclusions
In this study, COX-2 inhibitors were the preferred NSAID in
those with a prior hospitalization for a GI event, and in the
2.5 % who were still using NS-NSAIDs, a PPI was co-dis-
pensed for the majority. COX-2 inhibitors were used at
slightly higher rates than NS-NSAIDs in those with a history
of MI or stroke, which is not consistent with guidelines rec-
ommending NS-NSAIDs. While the majority of analgesic use
was consistent with guidelines, greater awareness of preferred
use of NS-NSAIDs in those with CVD is still required.
Acknowledgments Svetla Gadzhanova and Elizabeth E. Roughead
were responsible for study design. Svetla Gadzhanova, Jenni Ilomaki
and Elizabeth E. Roughead were responsible for the analyses, inter-
pretation of the results and preparation of the manuscript. This study
was funded by NPS Better choices Better health. The authors wish to
thank Mark Bartlett from NPS for his comments on the manuscript.
We acknowledge the support of the Australian Government Depart-
ment of Veterans Affairs, which provided data for the conduct of
these analyses. The authors have no conflicts of interest.
References
1. Schnitzer TJ. Update on guidelines for treatment of chronic
musculoskeletal pain. Clin Rheumatol. 2006;25(Suppl. 1):S229.
2. Therapeutic Goods Administration, Department of Health and
Ageing. Expanded information on Cox-2 inhibitors for doctors and
pharmacists. Department of Health and Ageing. 2005. http://
www.tga.gov.au/archive/media-2005-cox2-050210.htm. Accessed
24 Jan 2012.
29
3. European Medicines Agency. Scientific conclusions for the
amendment of the marketing authorisationCOX-2 inhibitors.
European Medicines Agency. 2005. http://www.ema.europa.eu/
docs/en_GB/document_library/Referrals_document/Celecoxib_31/
WC500012417.pdf. Accessed 24 Jan 2012.
4. Food and Drug Administration. COX-2 selective (includes Bex-
tra, Celebrex, and Vioxx) and Non-Selective Non-Steroidal Anti-
Inflammatory Drugs (NSAIDs). Food and Drug Administration.
2005. http://www.fda.gov/drugs/drugsafety/postmarketdrugsafety
informationforpatientsandproviders/ucm103420.htm. Accessed
24 Jan 2011.
5. Andersohn F, Suissa S, Garbe E. Use of first- and second-gen-
eration cyclooxygenase-2-selective nonsteroidal antiinflamma-
tory drugs and risk of acute myocardial infarction. Circulation.
2006;113(16):19507.
6. Bombardier C, Laine L, Reicin A, et al. Comparison of upper
gastrointestinal toxicity of rofecoxib and naproxen in patients
with rheumatoid arthritis. VIGOR Study Group. N Engl J Med.
2000;343(32):15208.
7. Bresalier RS, Sandler RS, Quan H, et al. Cardiovascular events
associated with rofecoxib in a colorectal adenoma chemopre-
vention trial. N Engl J Med. 2005;352(11):1092102.
8. Solomon SD, McMurray JJ, Pfeffer MA, et al. Cardiovascular
risk associated with celecoxib in a clinical trial for colorectal
adenoma prevention. N Engl J Med. 2005;352(11):107180.
9. McGettigan P, Henry D. Cardiovascular risk and inhibition of
cyclooxygenase: a systematic review of the observational studies
of selective and nonselective inhibitors of cyclooxygenase 2.
JAMA. 2006;296(13):163344.
10. Hernandez-Daz S, Varas-Lorenzo C, Garca Rodrguez LA, et al.
Non-steroidal antiinflammatory drugs and the risk of acute
myocardial infarction. Basic Clin Pharmacol Toxicol. 2006;
98(3):26674.
11. WHO Collaborating Centre for Drug Statistics Methodology. The
Anatomical Therapeutic Chemical Classification System. Oslo:
WHO Collaborating Centre for Drug Statistics Methodology.
Norwegian Institute of Public Health. 2012. http://www.whocc.
no/atcddd/. Accessed 21 May 2012.
12. Warner TD, Mitchell JA. Cyclooxygenases: new forms, new
inhibitors, and lessons from the clinic. FASEB J. 2004;18(7):
790804.
13. Ray WA, Varas-Lorenzo C, Chung CP, et al. Cardiovascular risks
of nonsteroidal antiinflammatory drugs in patients after hospi-
talization for serious coronary heart disease. Circ Cardiovasc
Qual Outcomes. 2009;2(3):15563.
14. Motsko SP, Rascati KL, Busti AJ, et al. Temporal relationship
between use of NSAIDs, including selective COX-2 inhibitors,
and cardiovascular risk. Drug Saf. 2006;29(7):62132.
15. Hernandez-Daz S, Rodrguez LA. Association between nonste-
roidal anti-inflammatory drugs and upper gastrointestinal tract
bleeding/perforation: an overview of epidemiologic studies pub-
lished in the 1990s. Arch Intern Med. 2000;160(14):20939.
16. Moore RA, Derry S, Makinson GT, et al. Tolerability and adverse
events in clinical trials of celecoxib in osteoarthritis and rheu-
matoid arthritis: systematic review and meta-analysis of infor-
mation from company clinical trial reports. Arthritis Res Ther.
2005;7(3):R64465.
17. Silverstein FE, Faich G, Goldstein JL, et al. Gastrointestinal
toxicity with celecoxib vs nonsteroidal anti-inflammatory drugs
for osteoarthritis and rheumatoid arthritis: the CLASS study: a
randomized controlled trial. Celecoxib Long-term Arthritis
Safety Study. JAMA. 2000;284(10):124755.
18. The Royal Australian College of General Practitioners. Guideline
for the non-surgical management of hip and knee osteoarthritis.
Melbourne: The Royal Australian College of General Practitio-
ners; 2009.
30
19. Rheumatology Expert Group. Therapeutic guidelines: rheuma-
tology. Version 1. Melbourne: Therapeutic Guidelines Limited;
2006.
20. Analgesic Expert Group. Therapeutic guidelines: analgesics.
Version 5. Melbourne: Therapeutic Guidelines Limited; 2007.
21. Gastroenterological Society of Australia, Digestive Health
Foundation. NSAIDs and the Gastronitestinal Tract. Mulgrave:
The Gastroenterological Society of Australia; 2009.
22. Australian Bureau of Statistics. National Health Survey: sum-
mary of results, 20072008. Canberra: Australian Bureau of
Statistics; 2010.
23. Barkin RL, Beckerman M, Blum SL, et al. Should nonsteroidal
anti-inflammatory drugs (NSAIDs) be prescribed to the older
adult? Drugs Aging. 2010;27(10):77589.
24. Roughead EE, Ramsay E, Pratt N, et al. NSAID use in individuals
at risk of renal adverse events: an observational study to inves-
tigate trends in Australian veterans. Drug Saf. 2008;31(11):
9971003.
25. Lloyd J, Anderson P, Australian Institute of Health and Welfare.
Veterans use of health services. Canberra: AIHW; 2008.
26. Australian Government Department of Veterans Affairs. Treat-
ment population statistics December 2011. Canberra: Australian
Government Department of Veterans Affairs; 2011.
27. Australian Government Department of Human Services. Pharma-
ceutical Benefits Scheme (PBS). Canberra: Australian Government
Department of Human Services, 2011. http://www.medicareaustralia.
gov.au/provider/pbs/index.jsp. Accessed 20 Apr 2012.
28. National Centre for Classification in Health. The international
statistical classification of diseases and related health problems,
10th revision, Australian modification (ICD-10-AM). 6th ed.
S. Gadzhanova et al.
Sydney: National Centre for Classification in Health, Faculty of
Health Sciences, University of Sydney; 2005.
29. National Prescribing Service Limited. Analgesic choices in per-
sistent pain. Report No.: 35. Sydney: National Prescribing Ser-
vice Limited; 2006.
30. Therapeutic Goods Administration, Department of Health and
Ageing. Regulator takes tough action on arthritis drugs (*amen-
ded). Woden: Department of Health and Ageing, 2005.
http://www.tga.gov.au/archive/media-2005-arthritis-050210.htm.
Accessed 24 Jan 2012.
31. Palliative Care Expert Group. Therapeutic guidelines: palliative
care. Version 3. Melbourne: Therapeutic Guidelines Ltd; 2010.
32. Reville B, Axelrod D, Maury R. Palliative care for the cancer
patient. Prim Care. 2009;36(4):781810.
33. Morgan TK, Williamson M, Pirotta M, et al. A national census of
medicines use: a 24-hour snapshot of Australians aged 50 years
and older. Med J Aust. 2012;196(1):503.
34. Hudec R, Bozekova L, Tisonova J. Consumption of three most
widely used analgesics in six European countries. J Clin Pharm
Ther. 2012;37(1):7880.
35. Fosbl EL, Gislason GH, Jacobsen S, et al. The pattern of use of
non-steroidal anti-inflammatory drugs (NSAIDs) from 1997 to
2005: a nationwide study on 4.6 million people. Pharmacoepi-
demiol Drug Saf. 2008;17(8):82233.
36. Donnelly N, McManus P, Dudley J, et al. Impact of increasing the
re-supply interval on the seasonality of subsidised prescription
use in Australia. Aust N Z J Public Health. 2000;6(24):6036.
37. Australian Bureau of Statistics. Population by age and sex,
Australian states and territories, June 2010. Contract no.: ABS
cat. no. 3201.0. Canberra: Australian Bureau of Statistics; 2010.
Copyright of Drugs & Aging is the property of ADIS International Limited and its content may not be copied or
emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.
However, users may print, download, or email articles for individual use.