Sei sulla pagina 1di 6

Alimentary Pharmacology & Therapeutics

Nonsteroidal anti-inflammatory drug use as a risk factor for


gastro-oesophageal reflux disease: an observational study
P. RUSZNIEWSKI*, C. SOUFFLET  & P. BARTHÉLÉMY 

*Department of Gastroenterology, SUMMARY


Beaujon Hospital, Clichy, France;
 Laboratoire AstraZénéca, Rueil- Background
Malmaison, France
Although the associations between nonsteroidal anti-inflammatory
Correspondence to: drugs (NSAIDs) and peptic ulcer disease or dyspepsia are well estab-
Prof. P. Ruszniewski, Department of lished, fewer data exist concerning the relationship between NSAIDs
Gastroenterology, Beaujon Hospital, and gastro-oesophageal reflux disease (GERD).
100 Bd Général Leclerc, 92110 Clichy,
France. Aim
E-mail: philippe.ruszniewski@
bjn.aphp.fr
To examine the relationship between NSAIDs and GERD.

Methods
Publication data A self-administered questionnaire covering NSAID use and GERD symp-
Submitted 9 May 2008
toms (heartburn and acid regurgitation) was sent to a representative
First decision 31 May 2008
Resubmitted 22 July 2008 national sample of 10 000 French adults (‡18 years) between 14
Accepted 28 July 2008 October and 21 November 2005. Risk factors associated with GERD were
Epub Accepted Article 30 July 2008 identified by logistic regression analysis in respondents who were not
taking aspirin or proton pump inhibitors.

Findings
A total of 7259 completed questionnaires were returned of which 6823
were evaluable. Overall, 2262 respondents (33%) reported using NSAIDs
during the previous 3 months. The lifetime and 3-month prevalence
rates of GERD symptoms were 37% and 21% respectively. GERD symp-
toms were significantly more common among NSAID users than among
non-users (27% vs. 19%, P £ 0.001) and a similar trend was seen for
aspirin use. Proton pump inhibitors were received by 31% of respon-
dents who reported experiencing GERD symptoms within the previous
3 months compared with 6% of those without symptoms (P < 0.01);
however, only 20% of NSAID-treated respondents were receiving proton
pump inhibitors. NSAID use, age and female gender were independent
predictors of GERD symptoms.

Conclusion
NSAID or aspirin use is a significant risk factor for GERD symptoms.

Aliment Pharmacol Ther 28, 1134–1139

1134 ª 2008 The Authors


Journal compilation ª 2008 Blackwell Publishing Ltd
doi:10.1111/j.1365-2036.2008.03821.x
N S A I D U S E A S A R I S K F A C T O R F O R G E R D 1135

residents of the Paris region. A computational weight-


INTRODUCTION
ing was therefore applied to the sample to fit a refer-
Upper gastrointestinal symptoms such as dyspepsia ence population.12 The efficiency of the imputation
occur in approximately 15–25% of patients treated was excellent (96.2%), with maximum and minimum
with nonsteroidal anti-inflammatory drugs (NSAIDs) weightings of 0.53 and 1.72 respectively.
including the selective cyclooxygenase (COX)-2 inhibi- Quantitative variables were described by the mean
tors.1, 2 Such symptoms have been shown to be a risk and standard deviation. Categorical variables were
factor for NSAID-related ulcer complications,3–5 but described by the numbers and relative proportions in
there is no correlation between symptom severity and the relevant categories. Where appropriate, differences
the risk of such complications.6–8 between groups were compared by the Student’s t-test
Nonsteroidal anti-inflammatory drugs have (if the size of the groups to be compared was >30) for
also been implicated in the pathogenesis of gastro- quantitative data or by the Z-test for categorical data.
oesophageal reflux disease (GERD).9 However, while To avoid potential bias resulting from the use of anti-
the associations between NSAID use and peptic ulcer secretory drugs, a further analysis was performed,
disease or dyspepsia are well documented, there are which excluded patients who were currently receiving
fewer data on the relationship between NSAIDs and proton pump inhibitors.
GERD. In a prospective case–control study, the preva- Logistic regression analysis was performed using
lence of erosive oesophagitis in NSAID-treated patients the LOGISTIC procedure in SAS version 9.1.3 (SAS
with osteoarthritis was 21%,10 while a prospective Institute Inc., Cary, NC, USA) to identify factors asso-
endoscopy study found that the use of nonselective ciated with GERD symptoms in patients who were
NSAIDs was a significant predictor of GERD.11 We not receiving daily treatment with aspirin or proton
performed an observational study to investigate the pump inhibitors. Odds ratios and the corresponding
relationship between NSAID use and GERD symptoms. 95% confidence intervals (CIs) were calculated for the
presence or absence of GERD symptoms in partici-
pants who had experienced such symptoms within
METHODS
the previous 3 months compared with those without
The study was a cross-sectional questionnaire survey GERD symptoms in the previous 3 months.
conducted in France by TNS Healthcare, Montrouge,
France. Between 14 October and 21 November 2005, a
Role of the funding source
four-page self-administered questionnaire was sent by
post to a representative sample of 10 000 adults (age The authors designed the study and contributed to the
18 years and above). The sample was selected to be analysis and interpretation of data. The TNS Health-
representative of the general French population in care Sofres institute carried out the statistical analysis.
terms of gender, age, occupation and regional distribu- The study was funded by AstraZeneca France.
tion (including urban and rural areas). Participants
were asked about their use of NSAIDs and aspirin dur-
RESULTS
ing the previous 3 months and about the presence of
GERD symptoms (heartburn, acid regurgitation or A total of 7259 (72.6%) recipients returned the ques-
both) during this time. tionnaire and 6823 of the completed questionnaires
Three months prior to the study, the questionnaire were considered evaluable for NSAID use and GERD
was tested on 1000 subjects during a pilot phase with symptoms. Demographic characteristics of the partici-
a response rate of 73.3%. The questionnaire was then pants with evaluable questionnaires are summarized in
modified to account for difficulties in the wording ⁄ Table 1. There were no significant differences between
understanding of the questions. the characteristics of these participants and those of
the overall study sample.
Overall, 2262 respondents (33%) reported using at
Statistical analysis
least one NSAID during the previous 3 months. Of
Analysis of the demographic characteristics of non- these, 84% had received noncontinuous treatment and
responders (see below) showed certain biases including 6% had received continuous treatment; data were
under-representation of males, young people and missing in 10% of respondents. In total, 50% of

ª 2008 The Authors, Aliment Pharmacol Ther 28, 1134–1139


Journal compilation ª 2008 Blackwell Publishing Ltd
1136 P . R U Z N I E W S K I et al.

this period and further 166 (4%) reported daily use of


Table 1. Demographic characteristics of participants with
evaluable questionnaires (n = 6823) aspirin. Demographic characteristics of these groups
are summarized in Table 2. A total of 781 respondents
Males ⁄ females 3251 ⁄ 3572 (11%) reported using proton pump inhibitors during
(48% ⁄ 52%)
the previous 3 months.
Age (years) (mean  s.d.) 48  18
18–34 1956 (29%)
35–44 1442 (21%) Prevalence of GERD symptoms
45–64 1911 (28%)
‡65 1514 (22%) Overall, 2501 respondents (37%) reported experiencing
Employed ⁄ unemployed 4046 ⁄ 2777 GERD symptoms at some time in their life and 1463
(59% ⁄ 41%)
(21%) had experienced such symptoms during the pre-
Location of residence
Rural 1532 (23%) vious 3 months. Among the latter respondents, 22%
Urban (2000–20 000 inhabitants) 1042 (15%) rated their symptoms as severe or very severe. GERD
Urban (20 000–100 000 inhabitants) 757 (11%) symptoms were significantly more common among
Urban (>100 000 inhabitants) 2208 (32%) NSAID users than among non-users (27% vs. 19%
Paris conurbation 1285 (19%)
respectively; P £ 0.001). The odds ratio for GERD
symptoms in NSAID users compared with non-users
was 1.61 (95% CI: 1.42, 1.80). A similar pattern was
NSAID users had received NSAIDs for 6 days or less seen in respondents who were not receiving proton
during the previous 3 months and 21% had been trea- pump inhibitors: the prevalence of GERD symptoms
ted for 1–2 weeks. The most widely used agents were was 23% in NSAID users and 15% in non-users
ketoprofen (24%), diclofenac (19%), solubilized ibupro- (P £ 0.01) giving an odds ratio of 1.65 (95% CI: 1.43,
fen (19%) and ibuprofen (17%); 4% of respondents 1.91). The duration of NSAID use was longer in
were using COX-2 inhibitors. Among the respondents respondents with GERD symptoms within the previous
who did not report any use of NSAIDs during the pre- 3 months than in those without such symptoms: 48%
vious 3 months and who returned completed question- of respondents with GERD symptoms had used NSAIDs
naires for ASA use and GERD symptoms (n = 4512), for 6 days or more compared with 37% of those
1504 (33%) reported using aspirin at some time during without symptoms (P < 0.01).

Table 2. Demographic characteristics of the aspirin groups

Daily use of aspirin Using aspirin at some No use of aspirin


during the last time during the last during the last
3 months (n = 166) 3 months (n = 1504) 3 months (n = 2842)

Male ⁄ female 96 ⁄ 69 (58% ⁄ 42%)  774 ⁄ 730 (51% ⁄ 49%) 1399 ⁄ 1443 (49% ⁄ 51%)
Age (years) (mean  s.d.) 65  15**,   44  16 50  18
18–34 9 (5%)**,   524 (35%) 742 (26%)
35–44 10 (6%)**,   373 (25%) 539 (19%)
45–64 52 (31%) 402 (27%) 834 (29%)
‡ 65 95 (58%)**,   206 (13%) 728 (26%)
Employed ⁄ unemployed 42 ⁄ 124 (25% ⁄ 75%)**,   1043 ⁄ 461 (69% ⁄ 31%) 1590 ⁄ 1252 (56% ⁄ 44%)
Location of residence
Rural 27 (16%)*,  348 (23%) 660 (23%)
Urban (2000–20 000 inhabitants) 26 (16%) 200 (13%) 445 (16%)
Urban (20 000–100 000 inhabitants) 22 (13%) 156 (10%) 332 (12%)
Urban (>100 000 inhabitants) 51 (31%) 473 (32%) 918 (32%)
Paris conurbation 39 (24%)  326 (22%) 487 (17%)

* P < 0.05 and ** P < 0.01 vs. group; using aspirin at some time during the last 3 months.
  P < 0.05 and    P < 0.01 vs. group; no use of aspirin during the last 3 months.

ª 2008 The Authors, Aliment Pharmacol Ther 28, 1134–1139


Journal compilation ª 2008 Blackwell Publishing Ltd
N S A I D U S E A S A R I S K F A C T O R F O R G E R D 1137

Proton pump inhibitors were received by 31% of


DISCUSSION
respondents who reported experiencing GERD symp-
toms within the previous 3 months compared with 6% The associations between NSAID use and peptic ulcer
of those without symptoms (P < 0.01); however, only disease13–15 and dyspepsia2, 16, 17 are well-established.
20% of NSAID-treated respondents were receiving pro- In contrast, data concerning the relationship between
ton pump inhibitors. Respondents aged 50 years and NSAID use and GERD are more limited. In a study of
above with GERD symptoms were more likely to be 163 000 Medicaid patients in the US, the relative risk
receiving proton pump inhibitors than younger of GERD in NSAID users compared with non-users
respondents (43% vs. 19% respectively). was 2.11,18 while a large community-based study in
Among respondents receiving daily aspirin treat- the UK showed that GERD symptoms were more com-
ment, the prevalence of GERD symptoms was 25% mon in NSAID users than in non-users.19 Similarly,
compared with 18% in respondents who were not the risk of oesophageal disorders in patients with
receiving daily aspirin (P £ 0.05); the corresponding rheumatoid arthritis has been reported to be twice as
odds ratio was 1.5 (95% CI: 1.05, 2.15). high in NSAID users as in non-users.20 Evidence for
Acid regurgitation was the most common GERD an association between NSAID use and GERD comes
symptom being reported by 27% of respondents with from a prospective endoscopy study that found that
any lifetime experience of GERD symptoms and 14% NSAID use was an independent risk factor for GERD
of those experiencing symptoms within the previous (odds ratio 2.0; CI: 1.3, 3.0; P < 0.001).11 Similarly, in
3 months. Heartburn was reported by 19% and 11% of a case–control study using data from the UK General
respondents respectively. Among the respondents who Practice Research Database, the odds ratio for GERD
reported experiencing symptoms within the previous associated with current NSAID use was 1.5 (95% CI:
3 months, 22–48% (depending on the symptom or 1.3, 1.7).21
symptoms reported) experienced symptoms on fewer Information on the mechanism through which
than 1–3 days ⁄ month: 9–13% reported experiencing NSAIDs may cause GERD is even more limited,
symptoms once per week, and 6–12% reported daily although data from one study suggest that they may act
symptoms. to increase the duration of acid reflux.22 This was a pro-
Gastrointestinal oesophageal reflux disease symp- spective double-blind, randomized, placebo-controlled
toms within the previous 3 months were significantly cross-over study in 17 patients with symptomatic
(P < 0.01) more common in respondents aged 65 years GERD, which used 24-h oesophageal pH-monitoring at
or above (27%) than in those aged 45–64 years (22%) baseline and during treatment with ibuprofen or pla-
or 18–44 years (19%). The prevalence was also signifi- cebo. The results showed that ibuprofen significantly
cantly higher in women than in men (23% vs. 20% increased the percentage of time during the 24-h period
respectively; P < 0.01). Logistic regression analysis in that the oesophageal pH was below 4 compared with
respondents who were not receiving aspirin or proton both baseline and placebo treatment (P < 0.05 in both
pump inhibitors showed that NSAID use, age and cases). Among these patients, who were already suffer-
female gender were independent predictors of GERD ing from GERD symptoms, this increased duration of
symptoms (Table 3). acid reflux was not associated with a significant
increase in heartburn frequency or intensity.22 Never-
theless, there is an established relationship between the
percentage of time the oesophageal pH is below 4 and
Table 3. Odds ratios and 95% confidence intervals for
the frequency of heartburn and acid regurgitation.23
independent predictors of GERD symptoms, identified by
linear regression analysis Therefore, in a patient who previously did not suffer
from GERD, any increased oesophageal acid exposure
Odds ratio (95% associated with NSAID therapy could result in GERD
confidence interval)
symptoms.
This study in a large French population confirms
NSAID use within 1.64 (1.36, 1.98)
previous 3 months
that NSAID use is a significant predictor of GERD
Age ‡65 years 1.34 (1.04, 1.73) symptoms that increases the risk of such symptoms by
Female gender 1.22 (1.02, 1.46) approximately 60%: indeed, in this study, it was the
strongest risk factor to be identified. Similar findings

ª 2008 The Authors, Aliment Pharmacol Ther 28, 1134–1139


Journal compilation ª 2008 Blackwell Publishing Ltd
1138 P . R U Z N I E W S K I et al.

were obtained when patients receiving treatment with for GERD symptoms. NSAID-related upper gastrointes-
proton pump inhibitors were excluded from the analy- tinal symptoms can markedly impair patients’ quality
sis. Daily use of aspirin was also associated with a sig- of life26, 27 and may necessitate dose reduction or dis-
nificantly increased risk of GERD symptoms. A continuation of treatment.28, 29
majority of NSAID users in this study were receiving
nonselective agents: only 4% were receiving selective
ACKNOWLEDGEMENTS
COX-2 inhibitors. Although it has traditionally been
assumed that selective COX-2 inhibitors should offer a Declaration of personal interests: Professor Ruszniewski
more favourable gastrointestinal toxicity profile than acts as a consultant to AstraZeneca, France.
nonselective agents, recent data indicate that there is Declaration of funding interests: The authors would
little clinically significant difference between the like to thank AstraZeneca France, which funded the
incidence of dyspepsia with the two classes of drugs.24 study, and Ann McIlhinney for her assistance in
It therefore seems unlikely that the high prevalence writing this manuscript, which was also funded by
of GERD symptoms among NSAID users in this study AstraZeneca France.
was biased by the predominant use of nonselective
agents.
SUPPORTING INFORMATION
The overall lifetime prevalence of GERD symptoms
in this study (37%) was significantly (P £ 0.01) higher Additional supporting information may be found in
than that reported in a similar survey in France in the online version of this article.
2003,25 in which 31% of participants reported experi- Questionnaire. You and your health over the last
encing GERD symptoms at least once. This increase 3 months.
was because of a significantly higher proportion of Please note: Wiley-Blackwell are not responsible for
patients who reported experiencing symptoms less the content or functionality of any supporting materi-
than once a month (24% vs. 18%, P £ 0.01). als supplied by the authors. Any queries (other than
In conclusion, this observational study has shown missing material) should be directed to the corre-
that NSAID or aspirin use is a significant risk factor sponding author for the article.

REFERENCES upper gastrointestinal clinical events: 10 Avidan B, Sonnenberg A, Schnell TG,


results of a double-blind outcomes study Budiman-Mak E, Sontag SJ. Risk factors
1 Buttgereit F, Burmester GR, Simon LS. in patients with rheumatoid arthritis. of oesophagitis in arthritic patients.
Gastrointestinal toxic side effects of non- Gastroenterology 2002; 123: 1006–12. Eur J Gastroenterol Hepatol 2001; 13:
steroidal anti-inflammatory drugs and 6 Larkai EN, Smith JL, Lidsky MD, Sessoms 1095–9.
cyclooxygenase-2-specific inhibitors. Am SL, Graham DY. Dyspepsia in NSAID 11 Voutilainen M, Sipponen P, Mecklin JP,
J Med 2001; 110(Suppl. 3A): 13S–9S. users: the size of the problem. J Clin Juhola M, Färkkilä M. Gastroesophageal
2 Ofman JJ, Maclean CH, Straus WL, et al. Gastroenterol 1989; 11: 158–62. reflux disease: prevalence, clinical, endo-
Meta-analysis of dyspepsia and nonsteroi- 7 Graham DY, White RH, Moreland LW, scopic and histopathological findings in
dal antiinflammatory drugs. Arthritis et al. Duodenal and gastric ulcer preven- 1,128 consecutive patients referred for
Rheum 2003; 49: 508–18. tion with misoprostol in arthritis patients endoscopy due to dyspeptic and reflux
3 Aalykke C, Lauritsen JM, Hallas J, Rein- taking NSAIDs. Misoprostol Study Group. symptoms. Digestion 2000; 61: 6–13.
holdt S, Krogfelt K, Lauritsen K. Helico- Ann Intern Med 1993; 119: 257–62. 12 Deville J-C, Sarndal C-E, Sautory O. Gen-
bacter pylori and risk of ulcer bleeding 8 Silverstein FE, Graham DY, Senior JR, eralized raking procedures in survey sam-
among users of nonsteroidal anti-inflam- et al. Misoprostol reduces serious pling. J Am Statist Assoc 1993; 88:
matory drugs: a case-control study. gastrointestinal complications in patients 1013–20.
Gastroenterology 1999; 116: 1305–9. with rheumatoid arthritis receiving 13 Hawkey CJ. Non-steroidal anti-inflamma-
4 Hansen JM, Hallas J, Lauritsen JM, Bytzer nonsteroidal anti-inflammatory drugs. A tory drugs and peptic ulcers. BMJ 1990;
P. Non-steroidal anti-inflammatory drugs randomized, double-blind, placebo-con- 300: 278–84.
and ulcer complications: a risk factor trolled trial. Ann Intern Med 1995; 123: 14 Singh G, Ramey DR, Morfeld D, Shi H,
analysis for clinical decision-making. 241–9. Hatoum HT, Fries JF. Gastrointestinal
Scand J Gastroenterol 1996; 31: 126–30. 9 Mason JC. NSAIDs and the oesophagus. tract complications of nonsteroidal anti-
5 Laine L, Bombardier C, Hawkey CJ, et al. Eur J Gastroenterol Hepatol 1999; 11: inflammatory drug treatment in rheuma-
Stratifying the risk of NSAID-related 369–73. toid arthritis. A prospective observational

ª 2008 The Authors, Aliment Pharmacol Ther 28, 1134–1139


Journal compilation ª 2008 Blackwell Publishing Ltd
N S A I D U S E A S A R I S K F A C T O R F O R G E R D 1139

cohort study. Arch Intern Med 1996; 156: 19 Harvey RF, McCune A, Lane A, et al. upper gastrointestinal diseases. Drugs
1530–6. Effect of NSAIDs and aspirin on heart- 2006; 66(Suppl. 1): 15–21.
15 Singh G. Recent considerations in non- burn and gastro-oesophageal reflux in the 25 Bretagne JF, Richard-Molard B, Honnorat
steroidal anti-inflammatory drug gastro- community: the Bristol Helicobacter Pro- C, Caekaert A, Barthélémy P. Gastro-
pathy. Am J Med 1998; 105(Suppl. 1B): ject. Gut 2003; 52(Suppl. 6): A15. esophageal reflux in the French general
S31–8. 20 Singh G, Ramey DR, Terry R, Khraishi M, population: national survey of 8000
16 Strauss WL, Ofman JJ, MacLean C, et al. Triadafilopoulos G. NSAID-related effects adults. Presse Med 2006; 35: 23–31.
Do NSAIDs cause dyspepsia? A meta- on the GI tract: an ever widening spectrum. 26 Wiklund I. Quality of life in arthritis
analysis evaluating alternative dyspepsia Arthritis Rheum 1997; 40(Suppl. 9): S93. patients using nonsteroidal anti-inflam-
definitions. Am J Gastroenterol 2002; 97: 21 Ruigómez A, Garcı́a-Rodrı́guez LA, matory drugs. Can J Gastroenterol 1999;
1951–8. Wallender M-A, Johansson S, Graffner H, 13: 129–33.
17 Wildner-Christensen M, Møller Hansen J, Dent J. Natural history of gastro-oesopha- 27 Wolfe F, Kong SX, Watson DJ. Gastroin-
Schaffalitzky de Muckadell OB. Risk geal reflux disease diagnosed in general testinal symptoms and health related
factors for dyspepsia in a general practice. Aliment Pharmacol Ther 2004; quality of life in patients with arthritis.
population: non-steroidal anti-inflamma- 20: 751–60. J Rheumatol 2000; 27: 1373–8.
tory drugs, cigarette smoking and 22 Cryer B, Spechler SJ. Effects of 28 Hirschowitz BI. Nonsteroidal anti-inflam-
unemployment are more important than non-steroidal anti-inflammatory drugs matory drugs and the gastrointestinal
Helicobacter pylori infection. Scand (NSAIDs) on acid reflux in patients with tract. Gastroenterologist 1994; 2: 207–23.
J Gastroenterol 2006; 41: 149–54. gastroesophageal reflux disease (GERD). 29 Singh G, Rosen Ramey D. NSAID-induced
18 Kotzan J, Wade W, Yu HH. Assessing Gastroenterology 2000; 4(Suppl. 2): A862. gastrointestinal complications: the AR-
NSAID prescription use as a predisposing 23 Joelsson B, Johnsson F. Heartburn – the AMIS perspective – 1997. Arthritis, Rheu-
factor for gastroesophageal reflux disease acid test. Gut 1989; 30: 1523–5. matism, and Aging Medical Information
in a Medicaid population. Pharm Res 24 Scheiman JM. Unmet needs in non-steroi- System. J Rheumatol 1998; 25(Suppl. 51):
2001; 18: 1367–72. dal anti-inflammatory drug-induced 8–16.

ª 2008 The Authors, Aliment Pharmacol Ther 28, 1134–1139


Journal compilation ª 2008 Blackwell Publishing Ltd

Potrebbero piacerti anche